Classes
class	TessBaseAPI
class	TessResultRenderer
class	TessTextRenderer
class	TessHOcrRenderer
class	TessPDFRenderer
class	TessUnlvRenderer
class	TessBoxTextRenderer
class	TessOsdRenderer
class	CubeRecoContext
class	CubeClassifier
class	CubeTessClassifier
struct	DocQualCallbacks
class	EquationDetect
class	LTRResultIterator
class	ChoiceIterator
class	MutableIterator
class	PageIterator
struct	BlobData
class	UnicodeSpanSkipper
struct	Cluster
class	SimpleClusterer
struct	GeometricClassifierState
struct	Interval
class	RowInfo
struct	LineHypothesis
class	RowScratchRegisters
class	ParagraphTheory
class	ParagraphModelSmearer
class	ResultIterator
class	TesseractCubeCombiner
struct	TesseractStats
struct	WordData
class	Tesseract
class	ImageThresholder
class	BoxWord
class	CCStruct
class	DetLineFit
class	DPPoint
struct	ScoredFont
struct	FontSpacingInfo
struct	FontInfo
struct	FontSet
class	FontInfoTable
class	WordFeature
struct	FloatWordFeature
class	ImageData
class	DocumentData
class	DocumentCache
struct	ParamsTrainingHypothesis
class	ParamsTrainingBundle
class	UnicharIdArrayUtils
class	AmbigSpec
class	UnicharAmbigs
class	BitVector
class	CCUtilMutex
class	CCUtil
class	DoublePtr
class	GenericHeap
class	PointerVector
class	TRand
class	IndexMap
class	IndexMapBiDi
struct	KDPair
struct	KDPairInc
struct	KDPairDec
class	KDPtrPair
struct	KDPtrPairInc
struct	KDPtrPairDec
class	KDVector
class	ObjectCache
struct	ParamsVectors
class	ParamUtils
class	Param
class	IntParam
class	BoolParam
class	StringParam
class	DoubleParam
class	TFile
class	TessdataManager
class	Classify
class	ErrorCounter
class	IntFeatureDist
class	IntFeatureMap
class	IntFeatureSpace
class	ClassPruner
struct	ShapeDist
class	MasterTrainer
class	SampleIterator
class	ShapeClassifier
struct	UnicharRating
struct	ShapeRating
struct	ShapeQueueEntry
struct	UnicharAndFonts
class	Shape
class	ShapeTable
class	TessClassifier
class	TrainingSample
class	TrainingSampleSet
class	AltList
class	BeamSearch
class	Bmp8
class	CachedFile
class	CharAltList
struct	Bigram
struct	CharBigram
struct	CharBigramTable
class	CharBigrams
class	CharSamp
class	CharSampEnum
class	CharSampSet
class	CharSet
class	CharClassifier
class	CharClassifierFactory
class	ConCompPt
class	ConComp
class	ConvNetCharClassifier
class	CubeLineObject
class	CubeLineSegmenter
class	CubeObject
class	CubeSearchObject
class	CubeTuningParams
class	CubeUtils
class	FeatureBase
class	FeatureBmp
class	FeatureChebyshev
class	FeatureHybrid
class	HybridNeuralNetCharClassifier
class	LangModEdge
class	LangModel
class	SearchColumn
class	SearchNode
class	SearchNodeHashTable
class	SearchObject
class	TessLangModEdge
class	TessLangModel
class	TuningParams
class	WordAltList
class	WordListLangModel
struct	PairSizeInfo
struct	FontPairSizeInfo
class	WordSizeModel
class	WordUnigrams
class	CUtil
struct	NodeChild
class	Dawg
struct	DawgPosition
class	DawgPositionVector
class	SquishedDawg
struct	DawgLoader
class	DawgCache
struct	DawgArgs
class	Dict
class	Trie
class	InputFileBuffer
class	NeuralNet
class	Neuron
struct	AlignedBlobParams
class	AlignedBlob
class	BaselineRow
class	BaselineBlock
class	BaselineDetect
class	GridBase
class	IntGrid
class	BBGrid
struct	PtrHash
class	GridSearch
class	TabEventHandler
class	BlobGrid
class	CCNonTextDetect
class	ColumnFinder
class	ColPartition
class	ColPartitionGrid
class	ColPartitionSet
class	PixelHistogram
class	ShiroRekhaSplitter
class	EquationDetectBase
class	ImageFind
class	LineFinder
class	StrokeWidth
class	TabFind
class	ColSegment
class	TableFinder
class	StructuredTable
class	TableRecognizer
class	TabConstraint
class	TabVector
class	TextlineProjection
class	WordWithBox
class	Textord
struct	BlockGroup
class	WorkingPartSet
class	BoxChar
struct	BoxCharPtrSort
class	File
class	InputBuffer
class	OutputBuffer
class	IcuErrorCode
class	LigatureTable
class	PangoFontInfo
class	FontUtils
class	StringRenderer
struct	SpacingProperties
struct	AssociateStats
class	AssociateUtils
class	LanguageModel
struct	LMConsistencyInfo
class	LMPainPoints
struct	LanguageModelDawgInfo
struct	LanguageModelNgramInfo
struct	ViterbiStateEntry
struct	LanguageModelState
	Struct to store information maintained by various language model components. More...
struct	BestChoiceBundle
	Bundle together all the things pertaining to the best choice/state. More...
class	ParamsModel
class	SegSearchPending
class	FRAGMENT
class	Wordrec
Typedefs
typedef int(Dict::*	DictFunc )(void *void_dawg_args, UNICHAR_ID unichar_id, bool word_end) const
typedef double(Dict::*	ProbabilityInContextFunc )(const char lang, const char context, int context_bytes, const char *character, int character_bytes)
typedef float(Dict::*	ParamsModelClassifyFunc )(const char lang, void path)
typedef void(Wordrec::*	FillLatticeFunc )(const MATRIX &ratings, const WERD_CHOICE_LIST &best_choices, const UNICHARSET &unicharset, BlamerBundle *blamer_bundle)
typedef TessCallback4< const UNICHARSET &, int, PageIterator , Pix >	TruthCallback
typedef GenericVectorEqEq < const ParagraphModel * >	SetOfModels
typedef void(Tesseract::*	WordRecognizer )(const WordData &word_data, WERD_RES *in_word, PointerVector< WERD_RES > out_words)
typedef GenericVector < ParamsTrainingHypothesis >	ParamsTrainingHypothesisList
typedef GenericVector< UNICHAR_ID >	UnicharIdVector
typedef GenericVector < AmbigSpec_LIST * >	UnicharAmbigsVector
typedef bool(*	FileReader )(const STRING &filename, GenericVector< char > *data)
typedef bool(*	FileWriter )(const GenericVector< char > &data, const STRING &filename)
typedef KDPairInc< int, int >	IntKDPair
typedef GenericHeap < ShapeQueueEntry >	ShapeQueue
typedef signed int	char_32
typedef basic_string< char_32 >	string_32
typedef GenericVector< NodeChild >	NodeChildVector
typedef GenericVector< int >	SuccessorList
typedef GenericVector < SuccessorList * >	SuccessorListsVector
typedef GenericVector< Dawg * >	DawgVector
typedef GridSearch< BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT >	BlobGridSearch
typedef GridSearch < ColPartition, ColPartition_CLIST, ColPartition_C_IT >	ColPartitionGridSearch
typedef GenericVector < ColPartitionSet * >	PartSetVector
typedef TessResultCallback1 < bool, int >	WidthCallback
typedef BBGrid< ColSegment, ColSegment_CLIST, ColSegment_C_IT >	ColSegmentGrid
typedef GridSearch< ColSegment, ColSegment_CLIST, ColSegment_C_IT >	ColSegmentGridSearch
typedef BBGrid< WordWithBox, WordWithBox_CLIST, WordWithBox_C_IT >	WordGrid
typedef GridSearch < WordWithBox, WordWithBox_CLIST, WordWithBox_C_IT >	WordSearch
typedef hash_map< string, string, StringHash >	LigHash
typedef GenericHeap < MatrixCoordPair >	PainPointHeap
typedef unsigned char	LanguageModelFlagsType
	Used for expressing various language model flags.
Enumerations
enum	LineType { LT_START = 'S', LT_BODY = 'C', LT_UNKNOWN = 'U', LT_MULTIPLE = 'M' }
enum	CMD_EVENTS { ACTION_1_CMD_EVENT, RECOG_WERDS, RECOG_PSEUDO, ACTION_2_CMD_EVENT }
enum	NormalizationMode { NM_BASELINE = -3, NM_CHAR_ISOTROPIC = -2, NM_CHAR_ANISOTROPIC = -1 }
enum	kParamsTrainingFeatureType { PTRAIN_DIGITS_SHORT, PTRAIN_DIGITS_MED, PTRAIN_DIGITS_LONG, PTRAIN_NUM_SHORT, PTRAIN_NUM_MED, PTRAIN_NUM_LONG, PTRAIN_DOC_SHORT, PTRAIN_DOC_MED, PTRAIN_DOC_LONG, PTRAIN_DICT_SHORT, PTRAIN_DICT_MED, PTRAIN_DICT_LONG, PTRAIN_FREQ_SHORT, PTRAIN_FREQ_MED, PTRAIN_FREQ_LONG, PTRAIN_SHAPE_COST_PER_CHAR, PTRAIN_NGRAM_COST_PER_CHAR, PTRAIN_NUM_BAD_PUNC, PTRAIN_NUM_BAD_CASE, PTRAIN_XHEIGHT_CONSISTENCY, PTRAIN_NUM_BAD_CHAR_TYPE, PTRAIN_NUM_BAD_SPACING, PTRAIN_NUM_BAD_FONT, PTRAIN_RATING_PER_CHAR, PTRAIN_NUM_FEATURE_TYPES }
enum	Orientation { ORIENTATION_PAGE_UP = 0, ORIENTATION_PAGE_RIGHT = 1, ORIENTATION_PAGE_DOWN = 2, ORIENTATION_PAGE_LEFT = 3 }
enum	WritingDirection { WRITING_DIRECTION_LEFT_TO_RIGHT = 0, WRITING_DIRECTION_RIGHT_TO_LEFT = 1, WRITING_DIRECTION_TOP_TO_BOTTOM = 2 }
enum	TextlineOrder { TEXTLINE_ORDER_LEFT_TO_RIGHT = 0, TEXTLINE_ORDER_RIGHT_TO_LEFT = 1, TEXTLINE_ORDER_TOP_TO_BOTTOM = 2 }
enum	PageSegMode { PSM_OSD_ONLY, PSM_AUTO_OSD, PSM_AUTO_ONLY, PSM_AUTO, PSM_SINGLE_COLUMN, PSM_SINGLE_BLOCK_VERT_TEXT, PSM_SINGLE_BLOCK, PSM_SINGLE_LINE, PSM_SINGLE_WORD, PSM_CIRCLE_WORD, PSM_SINGLE_CHAR, PSM_SPARSE_TEXT, PSM_SPARSE_TEXT_OSD, PSM_RAW_LINE, PSM_COUNT }
enum	PageIteratorLevel { RIL_BLOCK, RIL_PARA, RIL_TEXTLINE, RIL_WORD, RIL_SYMBOL }
enum	ParagraphJustification { JUSTIFICATION_UNKNOWN, JUSTIFICATION_LEFT, JUSTIFICATION_CENTER, JUSTIFICATION_RIGHT }
enum	OcrEngineMode { OEM_TESSERACT_ONLY, OEM_CUBE_ONLY, OEM_TESSERACT_CUBE_COMBINED, OEM_DEFAULT }
enum	ScriptPos { SP_NORMAL, SP_SUBSCRIPT, SP_SUPERSCRIPT, SP_DROPCAP }
enum	AmbigType { NOT_AMBIG, REPLACE_AMBIG, DEFINITE_AMBIG, SIMILAR_AMBIG, CASE_AMBIG, AMBIG_TYPE_COUNT }
enum	SetParamConstraint { SET_PARAM_CONSTRAINT_NONE, SET_PARAM_CONSTRAINT_DEBUG_ONLY, SET_PARAM_CONSTRAINT_NON_DEBUG_ONLY, SET_PARAM_CONSTRAINT_NON_INIT_ONLY }
enum	TessdataType { TESSDATA_LANG_CONFIG, TESSDATA_UNICHARSET, TESSDATA_AMBIGS, TESSDATA_INTTEMP, TESSDATA_PFFMTABLE, TESSDATA_NORMPROTO, TESSDATA_PUNC_DAWG, TESSDATA_SYSTEM_DAWG, TESSDATA_NUMBER_DAWG, TESSDATA_FREQ_DAWG, TESSDATA_FIXED_LENGTH_DAWGS, TESSDATA_CUBE_UNICHARSET, TESSDATA_CUBE_SYSTEM_DAWG, TESSDATA_SHAPE_TABLE, TESSDATA_BIGRAM_DAWG, TESSDATA_UNAMBIG_DAWG, TESSDATA_PARAMS_MODEL, TESSDATA_NUM_ENTRIES }
enum	CharSegmentationType { CST_FRAGMENT, CST_WHOLE, CST_IMPROPER, CST_NGRAM }
enum	CountTypes { CT_UNICHAR_TOP_OK, CT_UNICHAR_TOP1_ERR, CT_UNICHAR_TOP2_ERR, CT_UNICHAR_TOPN_ERR, CT_UNICHAR_TOPTOP_ERR, CT_OK_MULTI_UNICHAR, CT_OK_JOINED, CT_OK_BROKEN, CT_REJECT, CT_FONT_ATTR_ERR, CT_OK_MULTI_FONT, CT_NUM_RESULTS, CT_RANK, CT_REJECTED_JUNK, CT_ACCEPTED_JUNK, CT_SIZE }
enum	DawgType { DAWG_TYPE_PUNCTUATION, DAWG_TYPE_WORD, DAWG_TYPE_NUMBER, DAWG_TYPE_PATTERN, DAWG_TYPE_COUNT }
enum	XHeightConsistencyEnum { XH_GOOD, XH_SUBNORMAL, XH_INCONSISTENT }
enum	ColumnSpanningType { CST_NOISE, CST_FLOWING, CST_HEADING, CST_PULLOUT, CST_COUNT }
enum	NeighbourPartitionType { NPT_HTEXT, NPT_VTEXT, NPT_WEAK_HTEXT, NPT_WEAK_VTEXT, NPT_IMAGE, NPT_COUNT }
enum	LeftOrRight { LR_LEFT, LR_RIGHT }
enum	PartitionFindResult { PFR_OK, PFR_SKEW, PFR_NOISE }
enum	ColSegType { COL_UNKNOWN, COL_TEXT, COL_TABLE, COL_MIXED, COL_COUNT }
enum	TabAlignment { TA_LEFT_ALIGNED, TA_LEFT_RAGGED, TA_CENTER_JUSTIFIED, TA_RIGHT_ALIGNED, TA_RIGHT_RAGGED, TA_SEPARATOR, TA_COUNT }
enum	LMPainPointsType { LM_PPTYPE_BLAMER, LM_PPTYPE_AMBIG, LM_PPTYPE_PATH, LM_PPTYPE_SHAPE, LM_PPTYPE_NUM }
Functions
STRING	HOcrEscape (const char *text)
double	prec (double x)
long	dist2 (int x1, int y1, int x2, int y2)
void	GetWordBaseline (int writing_direction, int ppi, int height, int word_x1, int word_y1, int word_x2, int word_y2, int line_x1, int line_y1, int line_x2, int line_y2, double x0, double y0, double *length)
void	AffineMatrix (int writing_direction, int line_x1, int line_y1, int line_x2, int line_y2, double a, double b, double c, double d)
void	ClipBaseline (int ppi, int x1, int y1, int x2, int y2, int line_x1, int line_y1, int line_x2, int line_y2)
bool	IsTextOrEquationType (PolyBlockType type)
bool	IsLeftIndented (const EquationDetect::IndentType type)
bool	IsRightIndented (const EquationDetect::IndentType type)
STRING	RtlEmbed (const STRING &word, bool rtlify)
bool	IsLatinLetter (int ch)
bool	IsDigitLike (int ch)
bool	IsOpeningPunct (int ch)
bool	IsTerminalPunct (int ch)
const char *	SkipChars (const char str, const char toskip)
const char *	SkipChars (const char str, bool(skip)(int))
const char *	SkipOne (const char str, const char toskip)
bool	LikelyListNumeral (const STRING &word)
bool	LikelyListMark (const STRING &word)
bool	AsciiLikelyListItem (const STRING &word)
int	UnicodeFor (const UNICHARSET u, const WERD_CHOICE werd, int pos)
bool	LikelyListMarkUnicode (int ch)
bool	UniLikelyListItem (const UNICHARSET u, const WERD_CHOICE werd)
void	LeftWordAttributes (const UNICHARSET unicharset, const WERD_CHOICE werd, const STRING &utf8, bool is_list, bool starts_idea, bool *ends_idea)
void	RightWordAttributes (const UNICHARSET unicharset, const WERD_CHOICE werd, const STRING &utf8, bool is_list, bool starts_idea, bool *ends_idea)
int	ClosestCluster (const GenericVector< Cluster > &clusters, int value)
void	CalculateTabStops (GenericVector< RowScratchRegisters > rows, int row_start, int row_end, int tolerance, GenericVector< Cluster > left_tabs, GenericVector< Cluster > *right_tabs)
void	MarkRowsWithModel (GenericVector< RowScratchRegisters > rows, int row_start, int row_end, const ParagraphModel model, bool ltr, int eop_threshold)
void	GeometricClassifyThreeTabStopTextBlock (int debug_level, GeometricClassifierState &s, ParagraphTheory *theory)
void	GeometricClassify (int debug_level, GenericVector< RowScratchRegisters > rows, int row_start, int row_end, ParagraphTheory theory)
bool	ValidFirstLine (const GenericVector< RowScratchRegisters > rows, int row, const ParagraphModel model)
bool	ValidBodyLine (const GenericVector< RowScratchRegisters > rows, int row, const ParagraphModel model)
bool	CrownCompatible (const GenericVector< RowScratchRegisters > rows, int a, int b, const ParagraphModel model)
void	DiscardUnusedModels (const GenericVector< RowScratchRegisters > &rows, ParagraphTheory *theory)
void	DowngradeWeakestToCrowns (int debug_level, ParagraphTheory theory, GenericVector< RowScratchRegisters > rows)
void	RecomputeMarginsAndClearHypotheses (GenericVector< RowScratchRegisters > *rows, int start, int end, int percentile)
int	InterwordSpace (const GenericVector< RowScratchRegisters > &rows, int row_start, int row_end)
bool	FirstWordWouldHaveFit (const RowScratchRegisters &before, const RowScratchRegisters &after, tesseract::ParagraphJustification justification)
bool	FirstWordWouldHaveFit (const RowScratchRegisters &before, const RowScratchRegisters &after)
bool	TextSupportsBreak (const RowScratchRegisters &before, const RowScratchRegisters &after)
bool	LikelyParagraphStart (const RowScratchRegisters &before, const RowScratchRegisters &after)
bool	LikelyParagraphStart (const RowScratchRegisters &before, const RowScratchRegisters &after, tesseract::ParagraphJustification j)
ParagraphModel	InternalParagraphModelByOutline (const GenericVector< RowScratchRegisters > rows, int start, int end, int tolerance, bool consistent)
ParagraphModel	ParagraphModelByOutline (int debug_level, const GenericVector< RowScratchRegisters > *rows, int start, int end, int tolerance)
bool	RowsFitModel (const GenericVector< RowScratchRegisters > rows, int start, int end, const ParagraphModel model)
void	MarkStrongEvidence (GenericVector< RowScratchRegisters > *rows, int row_start, int row_end)
void	ModelStrongEvidence (int debug_level, GenericVector< RowScratchRegisters > rows, int row_start, int row_end, bool allow_flush_models, ParagraphTheory theory)
void	StrongEvidenceClassify (int debug_level, GenericVector< RowScratchRegisters > rows, int row_start, int row_end, ParagraphTheory theory)
void	SeparateSimpleLeaderLines (GenericVector< RowScratchRegisters > rows, int row_start, int row_end, ParagraphTheory theory)
void	ConvertHypothesizedModelRunsToParagraphs (int debug_level, const GenericVector< RowScratchRegisters > &rows, GenericVector< PARA * > row_owners, ParagraphTheory theory)
bool	RowIsStranded (const GenericVector< RowScratchRegisters > &rows, int row)
void	LeftoverSegments (const GenericVector< RowScratchRegisters > &rows, GenericVector< Interval > *to_fix, int row_start, int row_end)
void	CanonicalizeDetectionResults (GenericVector< PARA * > row_owners, PARA_LIST paragraphs)
void	DetectParagraphs (int debug_level, GenericVector< RowInfo > row_infos, GenericVector< PARA > row_owners, PARA_LIST paragraphs, GenericVector< ParagraphModel * > *models)
void	InitializeTextAndBoxesPreRecognition (const MutableIterator &it, RowInfo *info)
void	InitializeRowInfo (bool after_recognition, const MutableIterator &it, RowInfo *info)
void	DetectParagraphs (int debug_level, bool after_text_recognition, const MutableIterator block_start, GenericVector< ParagraphModel > *models)
bool	StrongModel (const ParagraphModel *model)
bool	read_t (PAGE_RES_IT page_res_it, TBOX tbox)
void	YOutlierPieces (WERD_RES word, int rebuilt_blob_index, int super_y_bottom, int sub_y_top, ScriptPos leading_pos, int num_leading_outliers, ScriptPos trailing_pos, int *num_trailing_outliers)
bool	CompareFontInfo (const FontInfo &fi1, const FontInfo &fi2)
bool	CompareFontSet (const FontSet &fs1, const FontSet &fs2)
void	FontInfoDeleteCallback (FontInfo f)
void	FontSetDeleteCallback (FontSet fs)
bool	read_info (FILE f, FontInfo fi, bool swap)
bool	write_info (FILE *f, const FontInfo &fi)
bool	read_spacing_info (FILE f, FontInfo fi, bool swap)
bool	write_spacing_info (FILE *f, const FontInfo &fi)
bool	read_set (FILE f, FontSet fs, bool swap)
bool	write_set (FILE *f, const FontSet &fs)
usr src packages BUILD tesseract ccstruct otsuthr cpp int	OtsuThreshold (Pix src_pix, int left, int top, int width, int height, int thresholds, int *hi_values)
void	HistogramRect (Pix src_pix, int channel, int left, int top, int width, int height, int histogram)
int	OtsuStats (const int histogram, int H_out, int *omega0_out)
int	ParamsTrainingFeatureByName (const char *name)
bool	PSM_OSD_ENABLED (int pageseg_mode)
bool	PSM_ORIENTATION_ENABLED (int pageseg_mode)
bool	PSM_COL_FIND_ENABLED (int pageseg_mode)
bool	PSM_SPARSE (int pageseg_mode)
bool	PSM_BLOCK_FIND_ENABLED (int pageseg_mode)
bool	PSM_LINE_FIND_ENABLED (int pageseg_mode)
bool	PSM_WORD_FIND_ENABLED (int pageseg_mode)
const char *	ScriptPosToString (enum ScriptPos script_pos)
	ELISTIZE (AmbigSpec)
	ELISTIZEH (AmbigSpec)
bool	LoadDataFromFile (const STRING &filename, GenericVector< char > *data)
bool	SaveDataToFile (const GenericVector< char > &data, const STRING &filename)
template<typename T >
bool	cmp_eq (T const &t1, T const &t2)
template<typename T >
int	sort_cmp (const void t1, const void t2)
template<typename T >
int	sort_ptr_cmp (const void t1, const void t2)
void	ExtractFontName (const STRING &filename, STRING *fontname)
TrainingSample *	BlobToTrainingSample (const TBLOB &blob, bool nonlinear_norm, INT_FX_RESULT_STRUCT fx_info, GenericVector< INT_FEATURE_STRUCT > bl_features)
uinT8	NormalizeDirection (uinT8 dir, const FCOORD &unnormed_pos, const DENORM &denorm, const DENORM *root_denorm)
void	ClearFeatureSpaceWindow (NORM_METHOD norm_method, ScrollView *window)
void	CallWithUTF8 (TessCallback1< const char * > cb, const WERD_CHOICE wc)
Pix *	GridReducedPix (const TBOX &box, int gridsize, ICOORD bleft, int left, int bottom)
Pix *	TraceOutlineOnReducedPix (C_OUTLINE outline, int gridsize, ICOORD bleft, int left, int *bottom)
Pix *	TraceBlockOnReducedPix (BLOCK block, int gridsize, ICOORD bleft, int left, int *bottom)
template<class BBC >
int	SortByBoxLeft (const void void1, const void void2)
template<class BBC >
int	SortRightToLeft (const void void1, const void void2)
template<class BBC >
int	SortByBoxBottom (const void void1, const void void2)
template<typename T >
void	DeleteObject (T *object)
void	SetBlobStrokeWidth (Pix pix, BLOBNBOX blob)
void	assign_blobs_to_blocks2 (Pix pix, BLOCK_LIST blocks, TO_BLOCK_LIST *port_blocks)
void	ParseCommandLineFlags (const char usage, int argc, char ***argv, const bool remove_flags)
ShapeTable *	LoadShapeTable (const STRING &file_prefix)
void	WriteShapeTable (const STRING &file_prefix, const ShapeTable &shape_table)
MasterTrainer *	LoadTrainingData (int argc, const char const argv, bool replication, ShapeTable *shape_table, STRING file_prefix)
Pix *	DegradeImage (Pix input, int exposure, TRand randomizer, float *rotation)
usr src packages BUILD tesseract training normstrngs cpp void	UTF8ToUTF32 (const char utf8_str, GenericVector< char32 > str32)
void	UTF32ToUTF8 (const GenericVector< char32 > &str32, STRING *utf8_str)
bool	is_hyphen_punc (const char32 ch)
bool	is_single_quote (const char32 ch)
bool	is_double_quote (const char32 ch)
STRING	NormalizeUTF8String (const char *str8)
void	NormalizeChar32 (char32 ch, GenericVector< char32 > *str)
char32	OCRNormalize (char32 ch)
bool	IsOCREquivalent (char32 ch1, char32 ch2)
bool	IsValidCodepoint (const char32 ch)
bool	IsWhitespace (const char32 ch)
bool	IsUTF8Whitespace (const char *text)
int	SpanUTF8Whitespace (const char *text)
int	SpanUTF8NotWhitespace (const char *text)
bool	IsInterchangeValid (const char32 ch)
bool	IsInterchangeValid7BitAscii (const char32 ch)
char32	FullwidthToHalfwidth (const char32 ch)
Pix *	CairoARGB32ToPixFormat (cairo_surface_t *surface)
void	ExtractFontProperties (const string &utf8_text, StringRenderer *render, const string &output_base)
bool	MakeIndividualGlyphs (Pix pix, const vector< BoxChar > &vbox, const int input_tiff_page)
void	SetupBasicProperties (bool report_errors, UNICHARSET *unicharset)
void	SetPropertiesForInputFile (const string &script_dir, const string &input_unicharset_file, const string &output_unicharset_file, const string &output_xheights_file)
	ELISTIZE (ViterbiStateEntry)
	ELISTIZEH (ViterbiStateEntry)
template<class BLOB_CHOICE >
int	SortByUnicharID (const void void1, const void void2)
template<class BLOB_CHOICE >
int	SortByRating (const void void1, const void void2)
convert_prob_to_tess_certainty
Normalize a probability in the range [0.0, 1.0] to a tesseract certainty in the range [-20.0, 0.0]
char_box_to_tbox
Create a TBOX from a character bounding box. If nonzero, the x_offset accounts for any additional padding of the word box that should be taken into account.
TBOX	char_box_to_tbox (Box *char_box, TBOX word_box, int x_offset)
Variables
const int	kBasicBufSize = 2048
const int	kCharWidth = 2
const float	kMathDigitDensityTh1 = 0.25
const float	kMathDigitDensityTh2 = 0.1
const float	kMathItalicDensityTh = 0.5
const float	kUnclearDensityTh = 0.25
const int	kSeedBlobsCountTh = 10
const int	kLeftIndentAlignmentCountTh = 1
const int	kMaxCharTopRange = 48
const int	kMinCredibleResolution = 70
	Minimum believable resolution.
const int	kDefaultResolution = 300
	Default resolution used if input in not believable.
const int	kMaxCircleErosions = 8
usr src packages BUILD tesseract ccmain paragraphs cpp usr src packages BUILD tesseract ccmain paragraphs cpp usr src packages BUILD tesseract ccmain paragraphs cpp const ParagraphModel *	kCrownLeft = reinterpret_cast<ParagraphModel *>(0xDEAD111F)
const ParagraphModel *	kCrownRight = reinterpret_cast<ParagraphModel *>(0xDEAD888F)
const inT16	kMaxBoxEdgeDiff = 2
const int	kBoxClipTolerance = 2
usr src packages BUILD tesseract ccstruct detlinefit cpp const int	kNumEndPoints = 3
const int	kMinPointsForErrorCount = 16
const int	kMaxRealDistance = 2.0
const int	kFeaturePadding = 2
const int	kImagePadding = 4
const int	kNumPagesPerMiniBatch = 100
const int	kHistogramSize = 256
usr src packages BUILD tesseract ccutil ambigs cpp usr src packages BUILD tesseract ccutil ambigs cpp const int	kMaxAmbigStringSize = UNICHAR_LEN * (MAX_AMBIG_SIZE + 1)
CCUtilMutex	tprintfMutex
usr src packages BUILD tesseract ccutil unicodes cpp const char *	kUTF8LineSeparator = "\u2028"
const char *	kUTF8ParagraphSeparator = "\u2029"
const char *	kLRM = "\u200E"
const char *	kRLM = "\u200F"
const char *	kRLE = "\u202A"
const char *	kPDF = "\u202C"
const char *	kHyphenLikeUTF8 []
const char *	kApostropheLikeUTF8 []
const char	kUniversalAmbigsFile []
const int	ksizeofUniversalAmbigsFile = sizeof(kUniversalAmbigsFile)
usr src packages BUILD tesseract classify errorcounter cpp usr src packages BUILD tesseract classify errorcounter cpp usr src packages BUILD tesseract classify errorcounter cpp const double	kRatingEpsilon = 1.0 / 32
const int	kMaxOffsetDist = 32
const double	kMinPCLengthIncrease = 1.0 / 1024
usr src packages BUILD tesseract classify mastertrainer cpp const int	kMinClusteredShapes = 1
const int	kMaxUnicharsPerCluster = 2000
const float	kFontMergeDistance = 0.025
const float	kInfiniteDist = 999.0f
const int	kRandomizingCenter = 128
const int	kTestChar = -1
const int	kSquareLimit = 25
const int	kPrime1 = 17
const int	kPrime2 = 13
const int	kMinOutlierSamples = 5
const int	kStateCnt = 4
const int	kNumLiteralCnt = 5
const int	case_state_table [6][4]
const char	kDoNotReverse [] = "RRP_DO_NO_REVERSE"
const char	kReverseIfHasRTL [] = "RRP_REVERSE_IF_HAS_RTL"
const char	kForceReverse [] = "RRP_FORCE_REVERSE"
const char *const	RTLReversePolicyNames []
const double	kAlignedFraction = 0.03125
const double	kRaggedFraction = 2.5
const double	kAlignedGapFraction = 0.75
const double	kRaggedGapFraction = 1.0
const int	kVLineAlignment = 3
const int	kVLineGutter = 1
const int	kVLineSearchSize = 150
const int	kMinRaggedTabs = 5
const int	kMinAlignedTabs = 4
const int	kVLineMinLength = 500
const double	kMinTabGradient = 4.0
const int	kMaxSkewFactor = 15
const char *	kTextordDebugPix = "psdebug_pix"
usr src packages BUILD tesseract textord ccnontextdetect cpp usr src packages BUILD tesseract textord ccnontextdetect cpp usr src packages BUILD tesseract textord ccnontextdetect cpp const double	kMaxSmallNeighboursPerPix = 1.0 / 32
const int	kMaxLargeOverlapsWithSmall = 3
const int	kMaxMediumOverlapsWithSmall = 12
const int	kMaxLargeOverlapsWithMedium = 12
const int	kOriginalNoiseMultiple = 8
const int	kNoisePadding = 4
const double	kPhotoOffsetFraction = 0.375
const double	kMinGoodTextPARatio = 1.5
usr src packages BUILD tesseract textord colfind cpp usr src packages BUILD tesseract textord colfind cpp usr src packages BUILD tesseract textord colfind cpp usr src packages BUILD tesseract textord colfind cpp const int	kMinColumnWidth = 100
const int	kMaxIncompatibleColumnCount = 2
const double	kMarginOverlapFraction = 0.25
const double	kHorizontalGapMergeFraction = 0.5
const double	kMinNonNoiseFraction = 0.5
const double	kMinGutterWidthGrid = 0.5
const double	kMaxDistToPartSizeRatio = 1.5
bool	textord_tabfind_show_initial_partitions = false
bool	textord_tabfind_show_reject_blobs = false
int	textord_tabfind_show_partitions = 0
bool	textord_tabfind_show_columns = false
bool	textord_tabfind_show_blocks = false
bool	textord_tabfind_find_tables = true
const double	kMaxSpacingDrift = 1.0 / 72
const double	kMaxTopSpacingFraction = 0.25
const double	kMaxSameBlockLineSpacing = 3
const double	kMaxSizeRatio = 1.5
const double	kMaxLeaderGapFractionOfMax = 0.25
const double	kMaxLeaderGapFractionOfMin = 0.5
const int	kMinLeaderCount = 5
const int	kMinStrongTextValue = 6
const int	kMinChainTextValue = 3
const int	kHorzStrongTextlineCount = 8
const int	kHorzStrongTextlineHeight = 10
const int	kHorzStrongTextlineAspect = 5
const double	kMaxBaselineError = 0.4375
const double	kMinBaselineCoverage = 0.5
const int	kMaxRMSColorNoise = 128
const int	kMaxColorDistance = 900
const int	kRGBRMSColors = 4
bool	textord_tabfind_show_color_fit = false
const int	kMaxPadFactor = 6
const int	kMaxNeighbourDistFactor = 4
const int	kMaxCaptionLines = 7
const double	kMinCaptionGapRatio = 2.0
const double	kMinCaptionGapHeightRatio = 0.5
const double	kBigPartSizeRatio = 1.75
const double	kTinyEnoughTextlineOverlapFraction = 0.25
const double	kMaxPartitionSpacing = 1.75
const int	kSmoothDecisionMargin = 4
const double	kMinRectangularFraction = 0.125
const double	kMaxRectangularFraction = 0.75
const double	kMaxRectangularGradient = 0.1
const int	kMinImageFindSize = 100
const double	kRMSFitScaling = 8.0
const int	kMinColorDifference = 16
const int	kThinLineFraction = 20
	Denominator of resolution makes max pixel width to allow thin lines.
const int	kMinLineLengthFraction = 4
	Denominator of resolution makes min pixels to demand line lengths to be.
const int	kCrackSpacing = 100
	Spacing of cracks across the page to break up tall vertical lines.
const int	kLineFindGridSize = 50
	Grid size used by line finder. Not very critical.
const int	kMinThickLineWidth = 12
const int	kMaxLineResidue = 6
const double	kThickLengthMultiple = 0.75
const double	kMaxNonLineDensity = 0.25
const double	kMaxStaveHeight = 1.0
const double	kMinMusicPixelFraction = 0.75
int	textord_tabfind_show_strokewidths = 0
bool	textord_tabfind_only_strokewidths = false
const double	kStrokeWidthFractionTolerance = 0.125
const double	kStrokeWidthTolerance = 1.5
const double	kStrokeWidthFractionCJK = 0.25
const double	kStrokeWidthCJK = 2.0
const int	kCJKRadius = 2
const double	kCJKBrokenDistanceFraction = 0.25
const int	kCJKMaxComponents = 8
const double	kCJKAspectRatio = 1.25
const double	kCJKAspectRatioIncrease = 1.0625
const int	kMaxCJKSizeRatio = 5
const double	kBrokenCJKIterationFraction = 0.125
const double	kDiacriticXPadRatio = 7.0
const double	kDiacriticYPadRatio = 1.75
const double	kMinDiacriticSizeRatio = 1.0625
const double	kMaxDiacriticDistanceRatio = 1.25
const double	kMaxDiacriticGapToBaseCharHeight = 1.0
const int	kSearchRadius = 2
const int	kLineTrapLongest = 4
const int	kLineTrapShortest = 2
const int	kMostlyOneDirRatio = 3
const double	kLineResidueAspectRatio = 8.0
const int	kLineResiduePadRatio = 3
const double	kLineResidueSizeRatio = 1.75
const float	kSizeRatioToReject = 2.0
const int	kMaxLargeOverlaps = 3
const double	kNeighbourSearchFactor = 2.5
const double	kNoiseOverlapGrowthFactor = 4.0
const double	kNoiseOverlapAreaFactor = 1.0 / 512
const double	kShapePerimeterRatio = 3.0
const int	kTabRadiusFactor = 5
const int	kMinVerticalSearch = 3
const int	kMaxVerticalSearch = 12
const int	kMaxRaggedSearch = 25
const int	kMinLinesInColumn = 10
const double	kMinFractionalLinesInColumn = 0.125
const double	kMinGutterWidthAbsolute = 0.02
const double	kMaxGutterWidthAbsolute = 2.00
const int	kRaggedGutterMultiple = 5
const double	kLineFragmentAspectRatio = 10.0
const double	kSmoothFactor = 0.25
const double	kCharVerticalOverlapFraction = 0.375
const double	kMaxHorizontalGap = 3.0
const int	kMinEvaluatedTabs = 3
const int	kMaxTextLineBlobRatio = 5
const int	kMinTextLineBlobRatio = 3
const double	kMinImageArea = 0.5
const double	kCosMaxSkewAngle = 0.866025
bool	textord_tabfind_show_initialtabs = false
bool	textord_tabfind_show_finaltabs = false
const int	kColumnWidthFactor = 20
usr src packages BUILD tesseract textord tablefind cpp const int	kMaxVerticalSpacing = 500
const int	kMaxBlobWidth = 500
const double	kSplitPartitionSize = 2.0
const double	kAllowTextHeight = 0.5
const double	kAllowTextWidth = 0.6
const double	kAllowTextArea = 0.8
const double	kAllowBlobHeight = 0.3
const double	kAllowBlobWidth = 0.4
const double	kAllowBlobArea = 0.05
const int	kMinBoxesInTextPartition = 10
const int	kMaxBoxesInDataPartition = 20
const double	kMaxGapInTextPartition = 4.0
const double	kMinMaxGapInTextPartition = 0.5
const double	kMaxBlobOverlapFactor = 4.0
const double	kMaxTableCellXheight = 2.0
const int	kMaxColumnHeaderDistance = 4
const double	kTableColumnThreshold = 3.0
const int	kRulingVerticalMargin = 3
const double	kMinOverlapWithTable = 0.6
const int	kSideSpaceMargin = 10
const double	kSmallTableProjectionThreshold = 0.35
const double	kLargeTableProjectionThreshold = 0.45
const int	kLargeTableRowCount = 6
const int	kMinRowsInTable = 3
const double	kRequiredFullJustifiedSpacing = 4.0
const int	kAdjacentLeaderSearchPadding = 2
const double	kParagraphEndingPreviousLineRatio = 1.3
const double	kMaxParagraphEndingLeftSpaceMultiple = 3.0
const double	kMinParagraphEndingTextToWhitespaceRatio = 3.0
const double	kMaxXProjectionGapFactor = 2.0
const double	kStrokeWidthFractionalTolerance = 0.25
const double	kStrokeWidthConstantTolerance = 2.0
bool	textord_dump_table_images = false
bool	textord_show_tables = false
bool	textord_tablefind_show_mark = false
bool	textord_tablefind_show_stats = false
bool	textord_tablefind_recognize_tables = false
const double	kHorizontalSpacing = 0.30
const double	kVerticalSpacing = -0.2
const int	kCellSplitRowThreshold = 0
const int	kCellSplitColumnThreshold = 0
const int	kLinedTableMinVerticalLines = 3
const int	kLinedTableMinHorizontalLines = 3
const double	kRequiredColumns = 0.7
const double	kMarginFactor = 1.1
const double	kMaxRowSize = 2.5
const double	kGoodRowNumberOfColumnsSmall [] = { 2, 2, 2, 2, 2, 3, 3 }
const int	kGoodRowNumberOfColumnsSmallSize
const double	kGoodRowNumberOfColumnsLarge = 0.7
const double	kMinFilledArea = 0.35
const int	kGutterMultiple = 4
const int	kGutterToNeighbourRatio = 3
const int	kSimilarVectorDist = 10
const int	kSimilarRaggedDist = 50
const int	kMaxFillinMultiple = 11
const double	kMinGutterFraction = 0.5
const double	kLineCountReciprocal = 4.0
const double	kMinAlignedGutter = 0.25
const double	kMinRaggedGutter = 1.5
double	textord_tabvector_vertical_gap_fraction = 0.5
double	textord_tabvector_vertical_box_ratio = 0.5
const char *	kAlignmentNames []
const int	kMaxLineLength = 1024
usr src packages BUILD tesseract training degradeimage cpp const float	kRotationRange = 0.02f
const int	kExposureFactor = 16
const int	kSaltnPepper = 5
const int	kMinRampSize = 1000
const int	kMinLigature = 0xfb00
const int	kMaxLigature = 0xfb17

Detailed Description

The box file is assumed to contain box definitions, one per line, of the following format for blob-level boxes:

 *   <UTF8 str> <left> <bottom> <right> <top> <page id>
 *

and for word/line-level boxes:

 *   WordStr <left> <bottom> <right> <top> <page id> #<space-delimited word str>
 *

NOTES: The boxes use tesseract coordinates, i.e. 0,0 is at BOTTOM-LEFT.

<page id>=""> is 0-based, and the page number is used for multipage input (tiff).

In the blob-level form, each line represents a recognizable unit, which may be several UTF-8 bytes, but there is a bounding box around each recognizable unit, and no classifier is needed to train in this mode (bootstrapping.)

In the word/line-level form, the line begins with the literal "WordStr", and the bounding box bounds either a whole line or a whole word. The recognizable units in the word/line are listed after the # at the end of the line and are space delimited, ignoring any original spaces on the line. Eg.

 * word -> #w o r d
 * multi word line -> #m u l t i w o r d l i n e
 *

The recognizable units must be space-delimited in order to allow multiple unicodes to be used for a single recognizable unit, eg Hindi.

In this mode, the classifier must have been pre-trained with the desired character set, or it will not be able to find the character segmentations.

Make a word from the selected blobs and run Tess on them.

Parameters:

page_res	recognise blobs
selection_box	within this box

fp_eval_word_spacing() Evaluation function for fixed pitch word lists.

Basically, count the number of "nice" characters - those which are in tess acceptable words or in dict words and are not rejected. Penalise any potential noise chars

build_menu()

Construct the menu tree used by the command window

process_cmd_win_event()

Process a command returned from the command window (Just call the appropriate command handler)

word_blank_and_set_display() Word processor

Blank display of word then redisplay word according to current display mode settings

---------------------------------------------------------------------------- Public Function Prototypes ----------------------------------------------------------------------------

---------------------------------------------------------------------------- Include Files and Type Defines ----------------------------------------------------------------------------

Typedef Documentation

typedef GridSearch<BLOBNBOX, BLOBNBOX_CLIST, BLOBNBOX_C_IT> tesseract::BlobGridSearch

Definition at line 31 of file blobgrid.h.

typedef signed int tesseract::char_32

Definition at line 40 of file string_32.h.

typedef GridSearch<ColPartition, ColPartition_CLIST, ColPartition_C_IT> tesseract::ColPartitionGridSearch

Definition at line 913 of file colpartition.h.

typedef BBGrid<ColSegment, ColSegment_CLIST, ColSegment_C_IT> tesseract::ColSegmentGrid

Definition at line 118 of file tablefind.h.

typedef GridSearch<ColSegment, ColSegment_CLIST, ColSegment_C_IT> tesseract::ColSegmentGridSearch

Definition at line 121 of file tablefind.h.

typedef GenericVector<Dawg *> tesseract::DawgVector

Definition at line 50 of file dict.h.

typedef int(Dict::* tesseract::DictFunc)(void *void_dawg_args, UNICHAR_ID unichar_id, bool word_end) const

Definition at line 83 of file baseapi.h.

typedef bool(* tesseract::FileReader)(const STRING &filename, GenericVector< char > *data)

Definition at line 349 of file genericvector.h.

typedef bool(* tesseract::FileWriter)(const GenericVector< char > &data, const STRING &filename)

Definition at line 352 of file genericvector.h.

typedef void(Wordrec::* tesseract::FillLatticeFunc)(const MATRIX &ratings, const WERD_CHOICE_LIST &best_choices, const UNICHARSET &unicharset, BlamerBundle *blamer_bundle)

Definition at line 92 of file baseapi.h.

typedef KDPairInc<int, int> tesseract::IntKDPair

Definition at line 179 of file kdpair.h.

typedef unsigned char tesseract::LanguageModelFlagsType

Used for expressing various language model flags.

Definition at line 37 of file lm_state.h.

typedef hash_map<string, string, StringHash> tesseract::LigHash

Definition at line 32 of file ligature_table.h.

typedef GenericVector<NodeChild> tesseract::NodeChildVector

Definition at line 67 of file dawg.h.

typedef GenericHeap<MatrixCoordPair> tesseract::PainPointHeap

Definition at line 34 of file lm_pain_points.h.

typedef float(Dict::* tesseract::ParamsModelClassifyFunc)(const char *lang, void *path)

Definition at line 90 of file baseapi.h.

typedef GenericVector<ParamsTrainingHypothesis> tesseract::ParamsTrainingHypothesisList

Definition at line 122 of file params_training_featdef.h.

typedef GenericVector<ColPartitionSet*> tesseract::PartSetVector

Definition at line 33 of file colpartitionset.h.

typedef double(Dict::* tesseract::ProbabilityInContextFunc)(const char *lang, const char *context, int context_bytes, const char *character, int character_bytes)

Definition at line 85 of file baseapi.h.

typedef GenericVectorEqEq<const ParagraphModel *> tesseract::SetOfModels

Definition at line 94 of file paragraphs_internal.h.

typedef GenericHeap<ShapeQueueEntry> tesseract::ShapeQueue

Definition at line 156 of file shapetable.h.

typedef basic_string<char_32> tesseract::string_32

Definition at line 41 of file string_32.h.

typedef GenericVector<int> tesseract::SuccessorList

Definition at line 68 of file dawg.h.

typedef GenericVector<SuccessorList *> tesseract::SuccessorListsVector

Definition at line 69 of file dawg.h.

typedef TessCallback4<const UNICHARSET &, int, PageIterator *, Pix *> tesseract::TruthCallback

Definition at line 97 of file baseapi.h.

typedef GenericVector<AmbigSpec_LIST *> tesseract::UnicharAmbigsVector

Definition at line 142 of file ambigs.h.

typedef GenericVector<UNICHAR_ID> tesseract::UnicharIdVector

Definition at line 34 of file ambigs.h.

typedef TessResultCallback1<bool, int> tesseract::WidthCallback

Definition at line 36 of file tabfind.h.

typedef BBGrid<WordWithBox, WordWithBox_CLIST, WordWithBox_C_IT> tesseract::WordGrid

Definition at line 65 of file textord.h.

typedef void(Tesseract::* tesseract::WordRecognizer)(const WordData &word_data, WERD_RES **in_word, PointerVector< WERD_RES > *out_words)

Definition at line 166 of file tesseractclass.h.

typedef GridSearch<WordWithBox, WordWithBox_CLIST, WordWithBox_C_IT> tesseract::WordSearch

Definition at line 66 of file textord.h.

Enumeration Type Documentation

enum tesseract::AmbigType

Enumerator:

NOT_AMBIG
REPLACE_AMBIG
DEFINITE_AMBIG
SIMILAR_AMBIG
CASE_AMBIG
AMBIG_TYPE_COUNT

Definition at line 44 of file ambigs.h.

               {
  NOT_AMBIG,        // the ngram pair is not ambiguous
  REPLACE_AMBIG,    // ocred ngram should always be substituted with correct
  DEFINITE_AMBIG,   // add correct ngram to the classifier results (1-1)
  SIMILAR_AMBIG,    // use pairwise classifier for ocred/correct pair (1-1)
  CASE_AMBIG,       // this is a case ambiguity (1-1)

  AMBIG_TYPE_COUNT  // number of enum entries
};

enum tesseract::CharSegmentationType

Enumerator:

CST_FRAGMENT
CST_WHOLE
CST_IMPROPER
CST_NGRAM

Definition at line 54 of file classify.h.

                          {
  CST_FRAGMENT,  // A partial character.
  CST_WHOLE,     // A correctly segmented character.
  CST_IMPROPER,  // More than one but less than 2 characters.
  CST_NGRAM      // Multiple characters.
};

enum tesseract::CMD_EVENTS

Enumerator:

ACTION_1_CMD_EVENT
RECOG_WERDS
RECOG_PSEUDO
ACTION_2_CMD_EVENT

Definition at line 482 of file tessedit.cpp.

{
  ACTION_1_CMD_EVENT,
  RECOG_WERDS,
  RECOG_PSEUDO,
  ACTION_2_CMD_EVENT
};

enum tesseract::ColSegType

Enumerator:

COL_UNKNOWN
COL_TEXT
COL_TABLE
COL_MIXED
COL_COUNT

Definition at line 30 of file tablefind.h.

                {
  COL_UNKNOWN,
  COL_TEXT,
  COL_TABLE,
  COL_MIXED,
  COL_COUNT
};

enum tesseract::ColumnSpanningType

Enumerator:

CST_NOISE
CST_FLOWING
CST_HEADING
CST_PULLOUT
CST_COUNT

Definition at line 47 of file colpartition.h.

                        {
  CST_NOISE,        // Strictly between columns.
  CST_FLOWING,      // Strictly within a single column.
  CST_HEADING,      // Spans multiple columns.
  CST_PULLOUT,      // Touches multiple columns, but doesn't span them.
  CST_COUNT         // Number of entries.
};

enum tesseract::CountTypes

Enumerator:

CT_UNICHAR_TOP_OK
CT_UNICHAR_TOP1_ERR
CT_UNICHAR_TOP2_ERR
CT_UNICHAR_TOPN_ERR
CT_UNICHAR_TOPTOP_ERR
CT_OK_MULTI_UNICHAR
CT_OK_JOINED
CT_OK_BROKEN
CT_REJECT
CT_FONT_ATTR_ERR
CT_OK_MULTI_FONT
CT_NUM_RESULTS
CT_RANK
CT_REJECTED_JUNK
CT_ACCEPTED_JUNK
CT_SIZE

Definition at line 69 of file errorcounter.h.

                {
  CT_UNICHAR_TOP_OK,     // Top shape contains correct unichar id.
  // The rank of the results in TOP1, TOP2, TOPN is determined by a gap of
  // kRatingEpsilon from the first result in each group. The real top choice
  // is measured using TOPTOP.
  CT_UNICHAR_TOP1_ERR,   // Top shape does not contain correct unichar id.
  CT_UNICHAR_TOP2_ERR,   // Top 2 shapes don't contain correct unichar id.
  CT_UNICHAR_TOPN_ERR,   // No output shape contains correct unichar id.
  CT_UNICHAR_TOPTOP_ERR,   // Very top choice not correct.
  CT_OK_MULTI_UNICHAR,   // Top shape id has correct unichar id, and others.
  CT_OK_JOINED,          // Top shape id is correct but marked joined.
  CT_OK_BROKEN,          // Top shape id is correct but marked broken.
  CT_REJECT,             // Classifier hates this.
  CT_FONT_ATTR_ERR,      // Top unichar OK, but font attributes incorrect.
  CT_OK_MULTI_FONT,      // CT_FONT_ATTR_OK but there are multiple font attrs.
  CT_NUM_RESULTS,        // Number of answers produced.
  CT_RANK,               // Rank of correct answer.
  CT_REJECTED_JUNK,      // Junk that was correctly rejected.
  CT_ACCEPTED_JUNK,      // Junk that was incorrectly classified otherwise.

  CT_SIZE                // Number of types for array sizing.
};

enum tesseract::DawgType

Enumerator:

DAWG_TYPE_PUNCTUATION
DAWG_TYPE_WORD
DAWG_TYPE_NUMBER
DAWG_TYPE_PATTERN
DAWG_TYPE_COUNT

Definition at line 71 of file dawg.h.

              {
  DAWG_TYPE_PUNCTUATION,
  DAWG_TYPE_WORD,
  DAWG_TYPE_NUMBER,
  DAWG_TYPE_PATTERN,

  DAWG_TYPE_COUNT  // number of enum entries
};

enum tesseract::kParamsTrainingFeatureType

Enumerator:

PTRAIN_DIGITS_SHORT
PTRAIN_DIGITS_MED
PTRAIN_DIGITS_LONG
PTRAIN_NUM_SHORT
PTRAIN_NUM_MED
PTRAIN_NUM_LONG
PTRAIN_DOC_SHORT
PTRAIN_DOC_MED
PTRAIN_DOC_LONG
PTRAIN_DICT_SHORT
PTRAIN_DICT_MED
PTRAIN_DICT_LONG
PTRAIN_FREQ_SHORT
PTRAIN_FREQ_MED
PTRAIN_FREQ_LONG
PTRAIN_SHAPE_COST_PER_CHAR
PTRAIN_NGRAM_COST_PER_CHAR
PTRAIN_NUM_BAD_PUNC
PTRAIN_NUM_BAD_CASE
PTRAIN_XHEIGHT_CONSISTENCY
PTRAIN_NUM_BAD_CHAR_TYPE
PTRAIN_NUM_BAD_SPACING
PTRAIN_NUM_BAD_FONT
PTRAIN_RATING_PER_CHAR
PTRAIN_NUM_FEATURE_TYPES

Definition at line 39 of file params_training_featdef.h.

                                {
  // Digits
  PTRAIN_DIGITS_SHORT,             // 0
  PTRAIN_DIGITS_MED,               // 1
  PTRAIN_DIGITS_LONG,              // 2
  // Number or pattern (NUMBER_PERM, USER_PATTERN_PERM)
  PTRAIN_NUM_SHORT,                // 3
  PTRAIN_NUM_MED,                  // 4
  PTRAIN_NUM_LONG,                 // 5
  // Document word (DOC_DAWG_PERM)
  PTRAIN_DOC_SHORT,                // 6
  PTRAIN_DOC_MED,                  // 7
  PTRAIN_DOC_LONG,                 // 8
  // Word (SYSTEM_DAWG_PERM, USER_DAWG_PERM, COMPOUND_PERM)
  PTRAIN_DICT_SHORT,               // 9
  PTRAIN_DICT_MED,                 // 10
  PTRAIN_DICT_LONG,                // 11
  // Frequent word (FREQ_DAWG_PERM)
  PTRAIN_FREQ_SHORT,               // 12
  PTRAIN_FREQ_MED,                 // 13
  PTRAIN_FREQ_LONG,                // 14
  PTRAIN_SHAPE_COST_PER_CHAR,      // 15
  PTRAIN_NGRAM_COST_PER_CHAR,      // 16
  PTRAIN_NUM_BAD_PUNC,             // 17
  PTRAIN_NUM_BAD_CASE,             // 18
  PTRAIN_XHEIGHT_CONSISTENCY,      // 19
  PTRAIN_NUM_BAD_CHAR_TYPE,        // 20
  PTRAIN_NUM_BAD_SPACING,          // 21
  PTRAIN_NUM_BAD_FONT,             // 22
  PTRAIN_RATING_PER_CHAR,          // 23

  PTRAIN_NUM_FEATURE_TYPES
};

enum tesseract::LeftOrRight

Enumerator:

LR_LEFT
LR_RIGHT

Definition at line 39 of file strokewidth.h.

                 {
  LR_LEFT,
  LR_RIGHT
};

enum tesseract::LineType

Enumerator:

LT_START
LT_BODY
LT_UNKNOWN
LT_MULTIPLE

Definition at line 54 of file paragraphs_internal.h.

              {
  LT_START = 'S',     // First line of a paragraph.
  LT_BODY = 'C',      // Continuation line of a paragraph.
  LT_UNKNOWN = 'U',   // No clues.
  LT_MULTIPLE = 'M',  // Matches for both LT_START and LT_BODY.
};

enum tesseract::LMPainPointsType

Enumerator:

LM_PPTYPE_BLAMER
LM_PPTYPE_AMBIG
LM_PPTYPE_PATH
LM_PPTYPE_SHAPE
LM_PPTYPE_NUM

Definition at line 37 of file lm_pain_points.h.

                      {
  LM_PPTYPE_BLAMER,
  LM_PPTYPE_AMBIG,
  LM_PPTYPE_PATH,
  LM_PPTYPE_SHAPE,

  LM_PPTYPE_NUM
};

enum tesseract::NeighbourPartitionType

Enumerator:

NPT_HTEXT
NPT_VTEXT
NPT_WEAK_HTEXT
NPT_WEAK_VTEXT
NPT_IMAGE
NPT_COUNT

Definition at line 1550 of file colpartitiongrid.cpp.

                            {
  NPT_HTEXT,       // Definite horizontal text.
  NPT_VTEXT,       // Definite vertical text.
  NPT_WEAK_HTEXT,  // Weakly horizontal text. Counts as HTEXT for HTEXT, but
                   // image for image and VTEXT.
  NPT_WEAK_VTEXT,  // Weakly vertical text. Counts as VTEXT for VTEXT, but
                   // image for image and HTEXT.
  NPT_IMAGE,       // Defininte non-text.
  NPT_COUNT        // Number of array elements.
};

enum tesseract::NormalizationMode

Enumerator:

NM_BASELINE
NM_CHAR_ISOTROPIC
NM_CHAR_ANISOTROPIC

Definition at line 44 of file normalis.h.

                       {
  NM_BASELINE = -3,         // The original BL normalization mode.
  NM_CHAR_ISOTROPIC = -2,   // Character normalization but isotropic.
  NM_CHAR_ANISOTROPIC = -1  // The original CN normalization mode.
};

enum tesseract::OcrEngineMode

When Tesseract/Cube is initialized we can choose to instantiate/load/run only the Tesseract part, only the Cube part or both along with the combiner. The preference of which engine to use is stored in tessedit_ocr_engine_mode.

ATTENTION: When modifying this enum, please make sure to make the appropriate changes to all the enums mirroring it (e.g. OCREngine in cityblock/workflow/detection/detection_storage.proto). Such enums will mention the connection to OcrEngineMode in the comments.

Enumerator:

OEM_TESSERACT_ONLY
OEM_CUBE_ONLY
OEM_TESSERACT_CUBE_COMBINED
OEM_DEFAULT

Definition at line 256 of file publictypes.h.

                   {
  OEM_TESSERACT_ONLY,           // Run Tesseract only - fastest
  OEM_CUBE_ONLY,                // Run Cube only - better accuracy, but slower
  OEM_TESSERACT_CUBE_COMBINED,  // Run both and combine results - best accuracy
  OEM_DEFAULT                   // Specify this mode when calling init_*(),
                                // to indicate that any of the above modes
                                // should be automatically inferred from the
                                // variables in the language-specific config,
                                // command-line configs, or if not specified
                                // in any of the above should be set to the
                                // default OEM_TESSERACT_ONLY.
};

enum tesseract::Orientation

+------------------+ Orientation Example: | 1 Aaaa Aaaa Aaaa | ==================== | Aaa aa aaa aa | To left is a diagram of some (1) English and | aaaaaa A aa aaa. | (2) Chinese text and a (3) photo credit. | 2 | | ####### c c C | Upright Latin characters are represented as A and a. | ####### c c c | '<' represents a latin character rotated | < ####### c c c | anti-clockwise 90 degrees. | < ####### c c | | < ####### . c | Upright Chinese characters are represented C and c. | 3 ####### c | +------------------+ NOTA BENE: enum values here should match goodoc.proto

If you orient your head so that "up" aligns with Orientation, then the characters will appear "right side up" and readable.

In the example above, both the English and Chinese paragraphs are oriented so their "up" is the top of the page (page up). The photo credit is read with one's head turned leftward ("up" is to page left).

The values of this enum match the convention of Tesseract's osdetect.h

Enumerator:

ORIENTATION_PAGE_UP
ORIENTATION_PAGE_RIGHT
ORIENTATION_PAGE_DOWN
ORIENTATION_PAGE_LEFT

Definition at line 108 of file publictypes.h.

                 {
  ORIENTATION_PAGE_UP = 0,
  ORIENTATION_PAGE_RIGHT = 1,
  ORIENTATION_PAGE_DOWN = 2,
  ORIENTATION_PAGE_LEFT = 3,
};

enum tesseract::PageIteratorLevel

enum of the elements of the page hierarchy, used in ResultIterator to provide functions that operate on each level without having to have 5x as many functions.

Enumerator:

RIL_BLOCK
RIL_PARA
RIL_TEXTLINE
RIL_WORD
RIL_SYMBOL

Definition at line 207 of file publictypes.h.

                       {
  RIL_BLOCK,     // Block of text/image/separator line.
  RIL_PARA,      // Paragraph within a block.
  RIL_TEXTLINE,  // Line within a paragraph.
  RIL_WORD,      // Word within a textline.
  RIL_SYMBOL     // Symbol/character within a word.
};

enum tesseract::PageSegMode

Possible modes for page layout analysis. These *must* be kept in order of decreasing amount of layout analysis to be done, except for OSD_ONLY, so that the inequality test macros below work.

Enumerator:

PSM_OSD_ONLY	Orientation and script detection only.
PSM_AUTO_OSD	Automatic page segmentation with orientation and script detection. (OSD)
PSM_AUTO_ONLY	Automatic page segmentation, but no OSD, or OCR.
PSM_AUTO	Fully automatic page segmentation, but no OSD.
PSM_SINGLE_COLUMN	Assume a single column of text of variable sizes.
PSM_SINGLE_BLOCK_VERT_TEXT	Assume a single uniform block of vertically aligned text.
PSM_SINGLE_BLOCK	Assume a single uniform block of text. (Default.)
PSM_SINGLE_LINE	Treat the image as a single text line.
PSM_SINGLE_WORD	Treat the image as a single word.
PSM_CIRCLE_WORD	Treat the image as a single word in a circle.
PSM_SINGLE_CHAR	Treat the image as a single character.
PSM_SPARSE_TEXT	Find as much text as possible in no particular order.
PSM_SPARSE_TEXT_OSD	Sparse text with orientation and script det.
PSM_RAW_LINE	Treat the image as a single text line, bypassing hacks that are Tesseract-specific.
PSM_COUNT	Number of enum entries.

Definition at line 151 of file publictypes.h.

                 {
  PSM_OSD_ONLY,       
  PSM_AUTO_OSD,       

  PSM_AUTO_ONLY,      
  PSM_AUTO,           
  PSM_SINGLE_COLUMN,  
  PSM_SINGLE_BLOCK_VERT_TEXT,  

  PSM_SINGLE_BLOCK,   
  PSM_SINGLE_LINE,    
  PSM_SINGLE_WORD,    
  PSM_CIRCLE_WORD,    
  PSM_SINGLE_CHAR,    
  PSM_SPARSE_TEXT,    
  PSM_SPARSE_TEXT_OSD,  
  PSM_RAW_LINE,       


  PSM_COUNT           
};

enum tesseract::ParagraphJustification

JUSTIFICATION_UNKNONW The alignment is not clearly one of the other options. This could happen for example if there are only one or two lines of text or the text looks like source code or poetry.

NOTA BENE: Fully justified paragraphs (text aligned to both left and right margins) are marked by Tesseract with JUSTIFICATION_LEFT if their text is written with a left-to-right script and with JUSTIFICATION_RIGHT if their text is written in a right-to-left script.

Interpretation for text read in vertical lines: "Left" is wherever the starting reading position is.

JUSTIFICATION_LEFT Each line, except possibly the first, is flush to the same left tab stop.

JUSTIFICATION_CENTER The text lines of the paragraph are centered about a line going down through their middle of the text lines.

JUSTIFICATION_RIGHT Each line, except possibly the first, is flush to the same right tab stop.

Enumerator:

JUSTIFICATION_UNKNOWN
JUSTIFICATION_LEFT
JUSTIFICATION_CENTER
JUSTIFICATION_RIGHT

Definition at line 239 of file publictypes.h.

                            {
  JUSTIFICATION_UNKNOWN,
  JUSTIFICATION_LEFT,
  JUSTIFICATION_CENTER,
  JUSTIFICATION_RIGHT,
};

enum tesseract::PartitionFindResult

Enumerator:

PFR_OK
PFR_SKEW
PFR_NOISE

Definition at line 46 of file strokewidth.h.

                         {
  PFR_OK,    // Everything is OK.
  PFR_SKEW,  // Skew was detected and rotated.
  PFR_NOISE  // Noise was detected and removed.
};

enum tesseract::ScriptPos

Enumerator:

SP_NORMAL
SP_SUBSCRIPT
SP_SUPERSCRIPT
SP_DROPCAP

Definition at line 260 of file ratngs.h.

               {
  SP_NORMAL,
  SP_SUBSCRIPT,
  SP_SUPERSCRIPT,
  SP_DROPCAP
};

enum tesseract::SetParamConstraint

Enumerator:

SET_PARAM_CONSTRAINT_NONE
SET_PARAM_CONSTRAINT_DEBUG_ONLY
SET_PARAM_CONSTRAINT_NON_DEBUG_ONLY
SET_PARAM_CONSTRAINT_NON_INIT_ONLY

Definition at line 36 of file params.h.

                        {
  SET_PARAM_CONSTRAINT_NONE,
  SET_PARAM_CONSTRAINT_DEBUG_ONLY,
  SET_PARAM_CONSTRAINT_NON_DEBUG_ONLY,
  SET_PARAM_CONSTRAINT_NON_INIT_ONLY,
};

enum tesseract::TabAlignment

Enumerator:

TA_LEFT_ALIGNED
TA_LEFT_RAGGED
TA_CENTER_JUSTIFIED
TA_RIGHT_ALIGNED
TA_RIGHT_RAGGED
TA_SEPARATOR
TA_COUNT

Definition at line 43 of file tabvector.h.

                  {
  TA_LEFT_ALIGNED,
  TA_LEFT_RAGGED,
  TA_CENTER_JUSTIFIED,
  TA_RIGHT_ALIGNED,
  TA_RIGHT_RAGGED,
  TA_SEPARATOR,
  TA_COUNT
};

enum tesseract::TessdataType

Enumerator:

TESSDATA_LANG_CONFIG
TESSDATA_UNICHARSET
TESSDATA_AMBIGS
TESSDATA_INTTEMP
TESSDATA_PFFMTABLE
TESSDATA_NORMPROTO
TESSDATA_PUNC_DAWG
TESSDATA_SYSTEM_DAWG
TESSDATA_NUMBER_DAWG
TESSDATA_FREQ_DAWG
TESSDATA_FIXED_LENGTH_DAWGS
TESSDATA_CUBE_UNICHARSET
TESSDATA_CUBE_SYSTEM_DAWG
TESSDATA_SHAPE_TABLE
TESSDATA_BIGRAM_DAWG
TESSDATA_UNAMBIG_DAWG
TESSDATA_PARAMS_MODEL
TESSDATA_NUM_ENTRIES

Definition at line 53 of file tessdatamanager.h.

                  {
  TESSDATA_LANG_CONFIG,         // 0
  TESSDATA_UNICHARSET,          // 1
  TESSDATA_AMBIGS,              // 2
  TESSDATA_INTTEMP,             // 3
  TESSDATA_PFFMTABLE,           // 4
  TESSDATA_NORMPROTO,           // 5
  TESSDATA_PUNC_DAWG,           // 6
  TESSDATA_SYSTEM_DAWG,         // 7
  TESSDATA_NUMBER_DAWG,         // 8
  TESSDATA_FREQ_DAWG,           // 9
  TESSDATA_FIXED_LENGTH_DAWGS,  // 10  // deprecated
  TESSDATA_CUBE_UNICHARSET,     // 11
  TESSDATA_CUBE_SYSTEM_DAWG,    // 12
  TESSDATA_SHAPE_TABLE,         // 13
  TESSDATA_BIGRAM_DAWG,         // 14
  TESSDATA_UNAMBIG_DAWG,        // 15
  TESSDATA_PARAMS_MODEL,        // 16

  TESSDATA_NUM_ENTRIES
};

enum tesseract::TextlineOrder

The text lines are read in the given sequence.

In English, the order is top-to-bottom. In Chinese, vertical text lines are read right-to-left. Mongolian is written in vertical columns top to bottom like Chinese, but the lines order left-to right.

Note that only some combinations make sense. For example, WRITING_DIRECTION_LEFT_TO_RIGHT implies TEXTLINE_ORDER_TOP_TO_BOTTOM

Enumerator:

TEXTLINE_ORDER_LEFT_TO_RIGHT
TEXTLINE_ORDER_RIGHT_TO_LEFT
TEXTLINE_ORDER_TOP_TO_BOTTOM

Definition at line 140 of file publictypes.h.

                   {
  TEXTLINE_ORDER_LEFT_TO_RIGHT = 0,
  TEXTLINE_ORDER_RIGHT_TO_LEFT = 1,
  TEXTLINE_ORDER_TOP_TO_BOTTOM = 2,
};

enum tesseract::WritingDirection

The grapheme clusters within a line of text are laid out logically in this direction, judged when looking at the text line rotated so that its Orientation is "page up".

For English text, the writing direction is left-to-right. For the Chinese text in the above example, the writing direction is top-to-bottom.

Enumerator:

WRITING_DIRECTION_LEFT_TO_RIGHT
WRITING_DIRECTION_RIGHT_TO_LEFT
WRITING_DIRECTION_TOP_TO_BOTTOM

Definition at line 123 of file publictypes.h.

                      {
  WRITING_DIRECTION_LEFT_TO_RIGHT = 0,
  WRITING_DIRECTION_RIGHT_TO_LEFT = 1,
  WRITING_DIRECTION_TOP_TO_BOTTOM = 2,
};

enum tesseract::XHeightConsistencyEnum

Enumerator:

XH_GOOD
XH_SUBNORMAL
XH_INCONSISTENT

Definition at line 75 of file dict.h.

{XH_GOOD, XH_SUBNORMAL, XH_INCONSISTENT};

Function Documentation

void tesseract::AffineMatrix	(	int	writing_direction,
		int	line_x1,
		int	line_y1,
		int	line_x2,
		int	line_y2,
		double *	a,
		double *	b,
		double *	c,
		double *	d
	)

Definition at line 246 of file pdfrenderer.cpp.

                                                              {
  double theta = atan2(static_cast<double>(line_y1 - line_y2),
                       static_cast<double>(line_x2 - line_x1));
  *a = cos(theta);
  *b = sin(theta);
  *c = -sin(theta);
  *d = cos(theta);
  switch(writing_direction) {
    case WRITING_DIRECTION_RIGHT_TO_LEFT:
      *a = -*a;
      *b = -*b;
      break;
    case WRITING_DIRECTION_TOP_TO_BOTTOM:
      // TODO(jbreiden) Consider using the vertical PDF writing mode.
      break;
    default:
      break;
  }
}

bool tesseract::AsciiLikelyListItem ( const STRING & word )

Definition at line 270 of file paragraphs.cpp.

void tesseract::assign_blobs_to_blocks2	(	Pix *	pix,
		BLOCK_LIST *	blocks,
		TO_BLOCK_LIST *	port_blocks
	)

Definition at line 157 of file tordmain.cpp.

                                                         {  // output list
  BLOCK *block;                  // current block
  BLOBNBOX *newblob;             // created blob
  C_BLOB *blob;                  // current blob
  BLOCK_IT block_it = blocks;
  C_BLOB_IT blob_it;             // iterator
  BLOBNBOX_IT port_box_it;       // iterator
                                 // destination iterator
  TO_BLOCK_IT port_block_it = port_blocks;
  TO_BLOCK *port_block;          // created block

  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
    block = block_it.data();
    port_block = new TO_BLOCK(block);

    // Convert the good outlines to block->blob_list
    port_box_it.set_to_list(&port_block->blobs);
    blob_it.set_to_list(block->blob_list());
    for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
      blob = blob_it.extract();
      newblob = new BLOBNBOX(blob);  // Convert blob to BLOBNBOX.
      SetBlobStrokeWidth(pix, newblob);
      port_box_it.add_after_then_move(newblob);
    }

    // Put the rejected outlines in block->noise_blobs, which allows them to
    // be reconsidered and sorted back into rows and recover outlines mistakenly
    // rejected.
    port_box_it.set_to_list(&port_block->noise_blobs);
    blob_it.set_to_list(block->reject_blobs());
    for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
      blob = blob_it.extract();
      newblob = new BLOBNBOX(blob);  // Convert blob to BLOBNBOX.
      SetBlobStrokeWidth(pix, newblob);
      port_box_it.add_after_then_move(newblob);
    }

    port_block_it.add_after_then_move(port_block);
  }
}

TrainingSample * tesseract::BlobToTrainingSample	(	const TBLOB &	blob,
		bool	nonlinear_norm,
		INT_FX_RESULT_STRUCT *	fx_info,
		GenericVector< INT_FEATURE_STRUCT > *	bl_features
	)

Definition at line 81 of file intfx.cpp.

                                                    {
  GenericVector<INT_FEATURE_STRUCT> cn_features;
  Classify::ExtractFeatures(blob, nonlinear_norm, bl_features,
                            &cn_features, fx_info, NULL);
  // TODO(rays) Use blob->PreciseBoundingBox() instead.
  TBOX box = blob.bounding_box();
  TrainingSample* sample = NULL;
  int num_features = fx_info->NumCN;
  if (num_features > 0) {
    sample = TrainingSample::CopyFromFeatures(*fx_info, box, &cn_features[0],
                                              num_features);
  }
  if (sample != NULL) {
    // Set the bounding box (in original image coordinates) in the sample.
    TPOINT topleft, botright;
    topleft.x = box.left();
    topleft.y = box.top();
    botright.x = box.right();
    botright.y = box.bottom();
    TPOINT original_topleft, original_botright;
    blob.denorm().DenormTransform(NULL, topleft, &original_topleft);
    blob.denorm().DenormTransform(NULL, botright, &original_botright);
    sample->set_bounding_box(TBOX(original_topleft.x, original_botright.y,
                                  original_botright.x, original_topleft.y));
  }
  return sample;
}

Pix* tesseract::CairoARGB32ToPixFormat ( cairo_surface_t * surface )

Definition at line 79 of file stringrenderer.cpp.

                                                                      {
    printf("Unexpected surface format %d\n",
           cairo_image_surface_get_format(surface));
    return NULL;
  }
  const int width = cairo_image_surface_get_width(surface);
  const int height = cairo_image_surface_get_height(surface);
  Pix* pix = pixCreate(width, height, 32);
  int byte_stride = cairo_image_surface_get_stride(surface);

  for (int i = 0; i < height; ++i) {
    memcpy(reinterpret_cast<unsigned char*>(pix->data + i * pix->wpl) + 1,
           cairo_image_surface_get_data(surface) + i * byte_stride,
           byte_stride - ((i == height - 1) ? 1 : 0));
  }
  return pix;
}

void tesseract::CalculateTabStops	(	GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end,
		int	tolerance,
		GenericVector< Cluster > *	left_tabs,
		GenericVector< Cluster > *	right_tabs
	)

Definition at line 694 of file paragraphs.cpp.

                                                           {
  if (!AcceptableRowArgs(0, 1, __func__, rows, row_start, row_end))
    return;
  // First pass: toss all left and right indents into clusterers.
  SimpleClusterer initial_lefts(tolerance);
  SimpleClusterer initial_rights(tolerance);
  GenericVector<Cluster> initial_left_tabs;
  GenericVector<Cluster> initial_right_tabs;
  for (int i = row_start; i < row_end; i++) {
    initial_lefts.Add((*rows)[i].lindent_);
    initial_rights.Add((*rows)[i].rindent_);
  }
  initial_lefts.GetClusters(&initial_left_tabs);
  initial_rights.GetClusters(&initial_right_tabs);

  // Second pass: cluster only lines that are not "stray"
  //   An example of a stray line is a page number -- a line whose start
  //   and end tab-stops are far outside the typical start and end tab-stops
  //   for the block.
  //   Put another way, we only cluster data from lines whose start or end
  //   tab stop is frequent.
  SimpleClusterer lefts(tolerance);
  SimpleClusterer rights(tolerance);

  // Outlier elimination.  We might want to switch this to test outlier-ness
  // based on how strange a position an outlier is in instead of or in addition
  // to how rare it is.  These outliers get re-added if we end up having too
  // few tab stops, to work with, however.
  int infrequent_enough_to_ignore = 0;
  if (row_end - row_start >= 8) infrequent_enough_to_ignore = 1;
  if (row_end - row_start >= 20) infrequent_enough_to_ignore = 2;

  for (int i = row_start; i < row_end; i++) {
    int lidx = ClosestCluster(initial_left_tabs, (*rows)[i].lindent_);
    int ridx = ClosestCluster(initial_right_tabs, (*rows)[i].rindent_);
    if (initial_left_tabs[lidx].count > infrequent_enough_to_ignore ||
        initial_right_tabs[ridx].count > infrequent_enough_to_ignore) {
      lefts.Add((*rows)[i].lindent_);
      rights.Add((*rows)[i].rindent_);
    }
  }
  lefts.GetClusters(left_tabs);
  rights.GetClusters(right_tabs);

  if ((left_tabs->size() == 1 && right_tabs->size() >= 4) ||
      (right_tabs->size() == 1 && left_tabs->size() >= 4)) {
    // One side is really ragged, and the other only has one tab stop,
    // so those "insignificant outliers" are probably important, actually.
    // This often happens on a page of an index.  Add back in the ones
    // we omitted in the first pass.
    for (int i = row_start; i < row_end; i++) {
      int lidx = ClosestCluster(initial_left_tabs, (*rows)[i].lindent_);
      int ridx = ClosestCluster(initial_right_tabs, (*rows)[i].rindent_);
      if (!(initial_left_tabs[lidx].count > infrequent_enough_to_ignore ||
            initial_right_tabs[ridx].count > infrequent_enough_to_ignore)) {
        lefts.Add((*rows)[i].lindent_);
        rights.Add((*rows)[i].rindent_);
      }
    }
  }
  lefts.GetClusters(left_tabs);
  rights.GetClusters(right_tabs);

  // If one side is almost a two-indent aligned side, and the other clearly
  // isn't, try to prune out the least frequent tab stop from that side.
  if (left_tabs->size() == 3 && right_tabs->size() >= 4) {
    int to_prune = -1;
    for (int i = left_tabs->size() - 1; i >= 0; i--) {
      if (to_prune < 0 ||
          (*left_tabs)[i].count < (*left_tabs)[to_prune].count) {
        to_prune = i;
      }
    }
    if (to_prune >= 0 &&
        (*left_tabs)[to_prune].count <= infrequent_enough_to_ignore) {
      left_tabs->remove(to_prune);
    }
  }
  if (right_tabs->size() == 3 && left_tabs->size() >= 4) {
    int to_prune = -1;
    for (int i = right_tabs->size() - 1; i >= 0; i--) {
      if (to_prune < 0 ||
          (*right_tabs)[i].count < (*right_tabs)[to_prune].count) {
        to_prune = i;
      }
    }
    if (to_prune >= 0 &&
        (*right_tabs)[to_prune].count <= infrequent_enough_to_ignore) {
      right_tabs->remove(to_prune);
    }
  }
}

// Given a paragraph model mark rows[row_start, row_end) as said model
// start or body lines.

void tesseract::CallWithUTF8	(	TessCallback1< const char * > *	cb,
		const WERD_CHOICE *	wc
	)

Definition at line 112 of file dawg.cpp.

                                                                          {
  STRING s;
  wc->string_and_lengths(&s, NULL);
  cb->Run(s.string());
}

void tesseract::CanonicalizeDetectionResults	(	GenericVector< PARA * > *	row_owners,
		PARA_LIST *	paragraphs
	)

Definition at line 2235 of file paragraphs.cpp.

                                        {
    if (rows[i] == NULL) {
      if (i == 0 || rows[i - 1] != formerly_null) {
        rows[i] = formerly_null = new PARA();
      } else {
        rows[i] = formerly_null;
        continue;
      }
    } else if (i > 0 && rows[i - 1] == rows[i]) {
      continue;
    }
    out.add_after_then_move(rows[i]);
  }
}

// Main entry point for Paragraph Detection Algorithm.
//

TBOX tesseract::char_box_to_tbox	(	Box *	char_box,
		TBOX	word_box,
		int	x_offset
	)

Definition at line 44 of file cube_control.cpp.

void tesseract::ClearFeatureSpaceWindow	(	NORM_METHOD	norm_method,
		ScrollView *	window
	)

Clears the given window and draws the featurespace guides for the appropriate normalization method.

Definition at line 1104 of file intproto.cpp.

                                                                          {
  window->Clear();

  window->Pen(ScrollView::GREY);
  // Draw the feature space limit rectangle.
  window->Rectangle(0, 0, INT_MAX_X, INT_MAX_Y);
  if (norm_method == baseline) {
    window->SetCursor(0, INT_DESCENDER);
    window->DrawTo(INT_MAX_X, INT_DESCENDER);
    window->SetCursor(0, INT_BASELINE);
    window->DrawTo(INT_MAX_X, INT_BASELINE);
    window->SetCursor(0, INT_XHEIGHT);
    window->DrawTo(INT_MAX_X, INT_XHEIGHT);
    window->SetCursor(0, INT_CAPHEIGHT);
    window->DrawTo(INT_MAX_X, INT_CAPHEIGHT);
  } else {
    window->Rectangle(INT_XCENTER - INT_XRADIUS, INT_YCENTER - INT_YRADIUS,
                      INT_XCENTER + INT_XRADIUS, INT_YCENTER + INT_YRADIUS);
  }
}

void tesseract::ClipBaseline	(	int	ppi,
		int	x1,
		int	y1,
		int	x2,
		int	y2,
		int *	line_x1,
		int *	line_y1,
		int *	line_x2,
		int *	line_y2
	)

Definition at line 275 of file pdfrenderer.cpp.

                                              {
  *line_x1 = x1;
  *line_y1 = y1;
  *line_x2 = x2;
  *line_y2 = y2;
  double rise = abs(y2 - y1) * 72 / ppi;
  double run = abs(x2 - x1) * 72 / ppi;
  if (rise < 2.0 && 2.0 < run)
    *line_y1 = *line_y2 = (y1 + y2) / 2;
}

int tesseract::ClosestCluster	(	const GenericVector< Cluster > &	clusters,
		int	value
	)

Definition at line 668 of file paragraphs.cpp.

                                                                  {
  clusters->clear();

template<typename T >

bool tesseract::cmp_eq	(	T const &	t1,
		T const &	t2
	)

Definition at line 382 of file genericvector.h.

                                        {
  return t1 == t2;
}

bool tesseract::CompareFontInfo	(	const FontInfo &	fi1,
		const FontInfo &	fi2
	)

Definition at line 120 of file fontinfo.cpp.

                                                               {
  // The font properties are required to be the same for two font with the same
  // name, so there is no need to test them.
  // Consequently, querying the table with only its font name as information is
  // enough to retrieve its properties.
  return strcmp(fi1.name, fi2.name) == 0;
}

bool tesseract::CompareFontSet	(	const FontSet &	fs1,
		const FontSet &	fs2
	)

Definition at line 128 of file fontinfo.cpp.

                                                            {
  if (fs1.size != fs2.size)
    return false;
  for (int i = 0; i < fs1.size; ++i) {
    if (fs1.configs[i] != fs2.configs[i])
      return false;
  }
  return true;
}

void tesseract::ConvertHypothesizedModelRunsToParagraphs	(	int	debug_level,
		const GenericVector< RowScratchRegisters > &	rows,
		GenericVector< PARA * > *	row_owners,
		ParagraphTheory *	theory
	)

Definition at line 2044 of file paragraphs.cpp.

                             {
  int end = rows.size();
  int start;
  for (; end > 0; end = start) {
    start = end - 1;
    const ParagraphModel *model = NULL;
    // TODO(eger): Be smarter about dealing with multiple hypotheses.
    bool single_line_paragraph = false;
    SetOfModels models;
    rows[start].NonNullHypotheses(&models);
    if (models.size() > 0) {
      model = models[0];
      if (rows[start].GetLineType(model) != LT_BODY)
        single_line_paragraph = true;
    }
    if (model && !single_line_paragraph) {
      // walk back looking for more body lines and then a start line.
      while (--start > 0 && rows[start].GetLineType(model) == LT_BODY) {
        // do nothing
      }
      if (start < 0 || rows[start].GetLineType(model) != LT_START) {
        model = NULL;
      }
    }
    if (model == NULL) {
      continue;
    }
    // rows[start, end) should be a paragraph.
    PARA *p = new PARA();
    if (model == kCrownLeft || model == kCrownRight) {
      p->is_very_first_or_continuation = true;
      // Crown paragraph.
      //   If we can find an existing ParagraphModel that fits, use it,
      //   else create a new one.
      for (int row = end; row < rows.size(); row++) {
        if ((*row_owners)[row] &&
            (ValidBodyLine(&rows, start, (*row_owners)[row]->model) &&
            (start == 0 ||
             ValidFirstLine(&rows, start, (*row_owners)[row]->model)))) {
          model = (*row_owners)[row]->model;
          break;
        }
      }
      if (model == kCrownLeft) {
        // No subsequent model fits, so cons one up.
        model = theory->AddModel(ParagraphModel(
            JUSTIFICATION_LEFT, rows[start].lmargin_ + rows[start].lindent_,
            0, 0, Epsilon(rows[start].ri_->average_interword_space)));
      } else if (model == kCrownRight) {
        // No subsequent model fits, so cons one up.
        model = theory->AddModel(ParagraphModel(
            JUSTIFICATION_RIGHT, rows[start].rmargin_ + rows[start].rmargin_,
            0, 0, Epsilon(rows[start].ri_->average_interword_space)));
      }
    }
    rows[start].SetUnknown();
    rows[start].AddStartLine(model);
    for (int i = start + 1; i < end; i++) {
      rows[i].SetUnknown();
      rows[i].AddBodyLine(model);
    }
    p->model = model;
    p->has_drop_cap = rows[start].ri_->has_drop_cap;
    p->is_list_item =
        model->justification() == JUSTIFICATION_RIGHT
            ? rows[start].ri_->rword_indicates_list_item
            : rows[start].ri_->lword_indicates_list_item;
    for (int row = start; row < end; row++) {
      if ((*row_owners)[row] != NULL) {
        tprintf("Memory leak! ConvertHypothesizeModelRunsToParagraphs() called "
                "more than once!\n");
      }
      (*row_owners)[row] = p;
    }
  }
}

struct Interval {
  Interval() : begin(0), end(0) {}

bool tesseract::CrownCompatible	(	const GenericVector< RowScratchRegisters > *	rows,
		int	a,
		int	b,
		const ParagraphModel *	model
	)

Definition at line 1291 of file paragraphs.cpp.

                            {
    return NearlyEqual(row_a.rindent_ + row_a.rmargin_,
                       row_b.rindent_ + row_b.rmargin_,
                       Epsilon(row_a.ri_->average_interword_space));
  }
  return NearlyEqual(row_a.lindent_ + row_a.lmargin_,
                     row_b.lindent_ + row_b.lmargin_,
                     Epsilon(row_a.ri_->average_interword_space));
}


// =============== Implementation of ParagraphModelSmearer ====================

struct Pix * tesseract::DegradeImage	(	Pix *	input,
		int	exposure,
		TRand *	randomizer,
		float *	rotation
	)		`[read]`

Definition at line 66 of file degradeimage.cpp.

                                   {
  Pix* pix = pixConvertTo8(input, false);
  pixDestroy(&input);
  input = pix;
  int width = pixGetWidth(input);
  int height = pixGetHeight(input);
  if (exposure >= 2) {
    // An erosion simulates the spreading darkening of a dark copy.
    // This is backwards to binary morphology,
    // see http://www.leptonica.com/grayscale-morphology.html
    pix = input;
    input = pixErodeGray(pix, 3, 3);
    pixDestroy(&pix);
  }
  // A convolution is essential to any mode as no scanner produces an
  // image as sharp as the electronic image.
  pix = pixBlockconv(input, 1, 1);
  pixDestroy(&input);
  // A small random rotation helps to make the edges jaggy in a realistic way.
  if (rotation != NULL) {
    float radians_clockwise = 0.0f;
    if (*rotation) {
      radians_clockwise = *rotation;
    } else if (randomizer != NULL) {
      radians_clockwise = randomizer->SignedRand(kRotationRange);
    }

    input = pixRotate(pix, radians_clockwise,
                      L_ROTATE_AREA_MAP, L_BRING_IN_WHITE,
                      0, 0);
    // Rotate the boxes to match.
    *rotation = radians_clockwise;
    pixDestroy(&pix);
  } else {
    input = pix;
  }

  if (exposure >= 3 || exposure == 1) {
    // Erosion after the convolution is not as heavy as before, so it is
    // good for level 1 and in addition as a level 3.
    // This is backwards to binary morphology,
    // see http://www.leptonica.com/grayscale-morphology.html
    pix = input;
    input = pixErodeGray(pix, 3, 3);
    pixDestroy(&pix);
  }
  // The convolution really needed to be 2x2 to be realistic enough, but
  // we only have 3x3, so we have to bias the image darker or lose thin
  // strokes.
  int erosion_offset = 0;
  // For light and 0 exposure, there is no dilation, so compensate for the
  // convolution with a big darkening bias which is undone for lighter
  // exposures.
  if (exposure <= 0)
    erosion_offset = -3 * kExposureFactor;
  // Add in a general offset of the greyscales for the exposure level so
  // a threshold of 128 gives a reasonable binary result.
  erosion_offset -= exposure * kExposureFactor;
  // Add a gradual fade over the page and a small amount of salt and pepper
  // noise to simulate noise in the sensor/paper fibres and varying
  // illumination.
  l_uint32* data = pixGetData(input);
  for (int y = 0; y < height; ++y) {
    for (int x = 0; x < width; ++x) {
      int pixel = GET_DATA_BYTE(data, x);
      if (randomizer != NULL)
        pixel += randomizer->IntRand() % (kSaltnPepper*2 + 1) - kSaltnPepper;
      if (height + width > kMinRampSize)
        pixel -= (2*x + y) * 32 / (height + width);
      pixel += erosion_offset;
      if (pixel < 0)
        pixel = 0;
      if (pixel > 255)
        pixel = 255;
      SET_DATA_BYTE(data, x, pixel);
    }
    data += input->wpl;
  }
  return input;
}

template<typename T >

void tesseract::DeleteObject ( T * object )

Definition at line 166 of file tablefind.cpp.

void tesseract::DetectParagraphs	(	int	debug_level,
		GenericVector< RowInfo > *	row_infos,
		GenericVector< PARA * > *	row_owners,
		PARA_LIST *	paragraphs,
		GenericVector< ParagraphModel * > *	models
	)

Definition at line 2267 of file paragraphs.cpp.

                                                               {
  GenericVector<RowScratchRegisters> rows;
  ParagraphTheory theory(models);

  // Initialize row_owners to be a bunch of NULL pointers.
  row_owners->init_to_size(row_infos->size(), NULL);

  // Set up row scratch registers for the main algorithm.
  rows.init_to_size(row_infos->size(), RowScratchRegisters());
  for (int i = 0; i < row_infos->size(); i++) {
    rows[i].Init((*row_infos)[i]);
  }

  // Pass 1:
  //   Detect sequences of lines that all contain leader dots (.....)
  //   These are likely Tables of Contents.  If there are three text lines in
  //   a row with leader dots, it's pretty safe to say the middle one should
  //   be a paragraph of its own.
  SeparateSimpleLeaderLines(&rows, 0, rows.size(), &theory);

  DebugDump(debug_level > 1, "End of Pass 1", theory, rows);

  GenericVector<Interval> leftovers;
  LeftoverSegments(rows, &leftovers, 0, rows.size());
  for (int i = 0; i < leftovers.size(); i++) {
    // Pass 2a:
    //   Find any strongly evidenced start-of-paragraph lines.  If they're
    //   followed by two lines that look like body lines, make a paragraph
    //   model for that and see if that model applies throughout the text
    //   (that is, "smear" it).
    StrongEvidenceClassify(debug_level, &rows,
                           leftovers[i].begin, leftovers[i].end, &theory);

    // Pass 2b:
    //   If we had any luck in pass 2a, we got part of the page and didn't
    //   know how to classify a few runs of rows. Take the segments that
    //   didn't find a model and reprocess them individually.
    GenericVector<Interval> leftovers2;
    LeftoverSegments(rows, &leftovers2, leftovers[i].begin, leftovers[i].end);
    bool pass2a_was_useful = leftovers2.size() > 1 ||
        (leftovers2.size() == 1 &&
         (leftovers2[0].begin != 0 || leftovers2[0].end != rows.size()));
    if (pass2a_was_useful) {
      for (int j = 0; j < leftovers2.size(); j++) {
        StrongEvidenceClassify(debug_level, &rows,
                               leftovers2[j].begin, leftovers2[j].end,
                               &theory);
      }
    }
  }

  DebugDump(debug_level > 1, "End of Pass 2", theory, rows);

  // Pass 3:
  //   These are the dregs for which we didn't have enough strong textual
  //   and geometric clues to form matching models for.  Let's see if
  //   the geometric clues are simple enough that we could just use those.
  LeftoverSegments(rows, &leftovers, 0, rows.size());
  for (int i = 0; i < leftovers.size(); i++) {
    GeometricClassify(debug_level, &rows,
                      leftovers[i].begin, leftovers[i].end, &theory);
  }

  // Undo any flush models for which there's little evidence.
  DowngradeWeakestToCrowns(debug_level, &theory, &rows);

  DebugDump(debug_level > 1, "End of Pass 3", theory, rows);

  // Pass 4:
  //   Take everything that's still not marked up well and clear all markings.
  LeftoverSegments(rows, &leftovers, 0, rows.size());
  for (int i = 0; i < leftovers.size(); i++) {
    for (int j = leftovers[i].begin; j < leftovers[i].end; j++) {
      rows[j].SetUnknown();
    }
  }

  DebugDump(debug_level > 1, "End of Pass 4", theory, rows);

  // Convert all of the unique hypothesis runs to PARAs.
  ConvertHypothesizedModelRunsToParagraphs(debug_level, rows, row_owners,
                                           &theory);

  DebugDump(debug_level > 0, "Final Paragraph Segmentation", theory, rows);

  // Finally, clean up any dangling NULL row paragraph parents.
  CanonicalizeDetectionResults(row_owners, paragraphs);
}

// ============ Code interfacing with the rest of Tesseract ==================

void tesseract::DetectParagraphs	(	int	debug_level,
		bool	after_text_recognition,
		const MutableIterator *	block_start,
		GenericVector< ParagraphModel * > *	models
	)

Definition at line 2512 of file paragraphs.cpp.

                                                               {
  // Clear out any preconceived notions.
  if (block_start->Empty(RIL_TEXTLINE)) {
    return;
  }
  BLOCK *block = block_start->PageResIt()->block()->block;
  block->para_list()->clear();
  bool is_image_block = block->poly_block() && !block->poly_block()->IsText();

  // Convert the Tesseract structures to RowInfos
  // for the paragraph detection algorithm.
  MutableIterator row(*block_start);
  if (row.Empty(RIL_TEXTLINE))
    return;  // end of input already.

  GenericVector<RowInfo> row_infos;
  do {
    if (!row.PageResIt()->row())
      continue;  // empty row.
    row.PageResIt()->row()->row->set_para(NULL);
    row_infos.push_back(RowInfo());
    RowInfo &ri = row_infos.back();
    InitializeRowInfo(after_text_recognition, row, &ri);
  } while (!row.IsAtFinalElement(RIL_BLOCK, RIL_TEXTLINE) &&
           row.Next(RIL_TEXTLINE));

  // If we're called before text recognition, we might not have
  // tight block bounding boxes, so trim by the minimum on each side.
  if (row_infos.size() > 0) {
    int min_lmargin = row_infos[0].pix_ldistance;
    int min_rmargin = row_infos[0].pix_rdistance;
    for (int i = 1; i < row_infos.size(); i++) {
      if (row_infos[i].pix_ldistance < min_lmargin)
        min_lmargin = row_infos[i].pix_ldistance;
      if (row_infos[i].pix_rdistance < min_rmargin)
        min_rmargin = row_infos[i].pix_rdistance;
    }
    if (min_lmargin > 0 || min_rmargin > 0) {
      for (int i = 0; i < row_infos.size(); i++) {
        row_infos[i].pix_ldistance -= min_lmargin;
        row_infos[i].pix_rdistance -= min_rmargin;
      }
    }
  }

  // Run the paragraph detection algorithm.
  GenericVector<PARA *> row_owners;
  GenericVector<PARA *> the_paragraphs;
  if (!is_image_block) {
    DetectParagraphs(debug_level, &row_infos, &row_owners, block->para_list(),
                     models);
  } else {
    row_owners.init_to_size(row_infos.size(), NULL);
    CanonicalizeDetectionResults(&row_owners, block->para_list());
  }

  // Now stitch in the row_owners into the rows.
  row = *block_start;
  for (int i = 0; i < row_owners.size(); i++) {
    while (!row.PageResIt()->row())
      row.Next(RIL_TEXTLINE);
    row.PageResIt()->row()->row->set_para(row_owners[i]);
    row.Next(RIL_TEXTLINE);
  }
}

}  // namespace

void tesseract::DiscardUnusedModels	(	const GenericVector< RowScratchRegisters > &	rows,
		ParagraphTheory *	theory
	)

Definition at line 1458 of file paragraphs.cpp.

                                        {
    rows[i].StrongHypotheses(&used_models);
  }
  theory->DiscardUnusedModels(used_models);
}

// DowngradeWeakestToCrowns:
//   Forget any flush-{left, right} models unless we see two or more

long tesseract::dist2	(	int	x1,
		int	y1,
		int	x2,
		int	y2
	)

Definition at line 192 of file pdfrenderer.cpp.

                                           {
  return (x2 - x1) * (x2 - x1) + (y2 - y1) * (y2 - y1);
}

void tesseract::DowngradeWeakestToCrowns	(	int	debug_level,
		ParagraphTheory *	theory,
		GenericVector< RowScratchRegisters > *	rows
	)

Definition at line 1491 of file paragraphs.cpp.

                                                     {
    // Search back for a body line of a unique type.
    const ParagraphModel *model = NULL;
    while (end > 0 &&
           (model = (*rows)[end - 1].UniqueBodyHypothesis()) == NULL) {
      end--;
    }
    if (end == 0) break;
    start = end - 1;
    while (start >= 0 && (*rows)[start].UniqueBodyHypothesis() == model) {
      start--;  // walk back to the first line that is not the same body type.
    }
    if (start >= 0 && (*rows)[start].UniqueStartHypothesis() == model &&
        StrongModel(model) &&
        NearlyEqual(model->first_indent(), model->body_indent(),
                    model->tolerance())) {
        start--;
    }
    start++;
    // Now rows[start, end) is a sequence of unique body hypotheses of model.
    if (StrongModel(model) && model->justification() == JUSTIFICATION_CENTER)
      continue;
    if (!StrongModel(model)) {
      while (start > 0 &&
             CrownCompatible(rows, start - 1, start, model))
        start--;
    }
    if (start == 0 ||
        (!StrongModel(model)) ||
        (StrongModel(model) && !ValidFirstLine(rows, start - 1, model))) {
      // crownify rows[start, end)
      const ParagraphModel *crown_model = model;
      if (StrongModel(model)) {
          if (model->justification() == JUSTIFICATION_LEFT)
            crown_model = kCrownLeft;
          else
            crown_model = kCrownRight;
      }
      (*rows)[start].SetUnknown();
      (*rows)[start].AddStartLine(crown_model);
      for (int row = start + 1; row < end; row++) {
        (*rows)[row].SetUnknown();
        (*rows)[row].AddBodyLine(crown_model);
      }
    }
  }
  DiscardUnusedModels(*rows, theory);
}


// Clear all hypotheses about lines [start, end) and reset margins.

tesseract::ELISTIZE ( AmbigSpec )

tesseract::ELISTIZE ( ViterbiStateEntry )

tesseract::ELISTIZEH ( AmbigSpec )

tesseract::ELISTIZEH ( ViterbiStateEntry )

void tesseract::ExtractFontName	(	const STRING &	filename,
		STRING *	fontname
	)

---------------------------------------------------------------------------- Public Code ----------------------------------------------------------------------------

Definition at line 46 of file blobclass.cpp.

                                                               {
  *fontname = classify_font_name;
  if (*fontname == kUnknownFontName) {
    // filename is expected to be of the form [lang].[fontname].exp[num]
    // The [lang], [fontname] and [num] fields should not have '.' characters.
    const char *basename = strrchr(filename.string(), '/');
    const char *firstdot = strchr(basename ? basename : filename.string(), '.');
    const char *lastdot  = strrchr(filename.string(), '.');
    if (firstdot != lastdot && firstdot != NULL && lastdot != NULL) {
      ++firstdot;
      *fontname = firstdot;
      fontname->truncate_at(lastdot - firstdot);
    }
  }
}

void tesseract::ExtractFontProperties	(	const string &	utf8_text,
		StringRenderer *	render,
		const string &	output_base
	)

Definition at line 212 of file text2image.cpp.

                                                      {
  map<string, SpacingProperties> spacing_map;
  map<string, SpacingProperties>::iterator spacing_map_it0;
  map<string, SpacingProperties>::iterator spacing_map_it1;
  int x_bearing, x_advance;
  int len = utf8_text.length();
  int offset = 0;
  const char* text = utf8_text.c_str();
  while (offset < len) {
    offset += render->RenderToImage(text + offset, strlen(text + offset), NULL);
    const vector<BoxChar*> &boxes = render->GetBoxes();

    // If the page break split a bigram, correct the offset so we try the bigram
    // on the next iteration.
    if (boxes.size() > 2 && !IsWhitespaceBox(boxes[boxes.size() - 1]) &&
        IsWhitespaceBox(boxes[boxes.size() - 2])) {
      if (boxes.size() > 3) {
        tprintf("WARNING: Adjusting to bad page break after '%s%s'\n",
                boxes[boxes.size() - 4]->ch().c_str(),
                boxes[boxes.size() - 3]->ch().c_str());
      }
      offset -= boxes[boxes.size() - 1]->ch().size();
    }

    for (int b = 0; b < boxes.size(); b += 2) {
      while (b < boxes.size() && IsWhitespaceBox(boxes[b])) ++b;
      if (b + 1 >= boxes.size()) break;
      const string &ch0 = boxes[b]->ch();
      // We encountered a ligature. This happens in at least two scenarios:
      // One is when the rendered bigram forms a grapheme cluster (eg. the
      // second character in the bigram is a combining vowel), in which case we
      // correctly output only one bounding box.
      // A second far less frequent case is when caused some fonts like 'DejaVu
      // Sans Ultra-Light' force Pango to render a ligatured character even if
      // the input consists of the separated characters.  NOTE(ranjith): As per
      // behdad@ this is not currently controllable at the level of the Pango
      // API.
      // Safeguard against these cases here by just skipping the bigram.
      if (IsWhitespaceBox(boxes[b+1])) {
        continue;
      }
      int xgap = (boxes[b+1]->box()->x -
                  (boxes[b]->box()->x + boxes[b]->box()->w));
      spacing_map_it0 = spacing_map.find(ch0);
      int ok_count = 0;
      if (spacing_map_it0 == spacing_map.end() &&
          render->font().GetSpacingProperties(ch0, &x_bearing, &x_advance)) {
        spacing_map[ch0] = SpacingProperties(
            x_bearing, x_advance - x_bearing - boxes[b]->box()->w);
        spacing_map_it0 = spacing_map.find(ch0);
        ++ok_count;
      }
      const string &ch1 = boxes[b+1]->ch();
      tlog(3, "%s%s\n", ch0.c_str(), ch1.c_str());
      spacing_map_it1 = spacing_map.find(ch1);
      if (spacing_map_it1 == spacing_map.end() &&
          render->font().GetSpacingProperties(ch1, &x_bearing, &x_advance)) {
        spacing_map[ch1] = SpacingProperties(
            x_bearing, x_advance - x_bearing - boxes[b+1]->box()->w);
        spacing_map_it1 = spacing_map.find(ch1);
        ++ok_count;
      }
      if (ok_count == 2 && xgap != (spacing_map_it0->second.x_gap_after +
                                    spacing_map_it1->second.x_gap_before)) {
        spacing_map_it0->second.kerned_x_gaps[ch1] = xgap;
      }
    }
    render->ClearBoxes();
  }
  string output_string;
  const int kBufSize = 1024;
  char buf[kBufSize];
  snprintf(buf, kBufSize, "%d\n", static_cast<int>(spacing_map.size()));
  output_string.append(buf);
  map<string, SpacingProperties>::const_iterator spacing_map_it;
  for (spacing_map_it = spacing_map.begin();
       spacing_map_it != spacing_map.end(); ++spacing_map_it) {
    snprintf(buf, kBufSize,
             "%s %d %d %d", spacing_map_it->first.c_str(),
             spacing_map_it->second.x_gap_before,
             spacing_map_it->second.x_gap_after,
             static_cast<int>(spacing_map_it->second.kerned_x_gaps.size()));
    output_string.append(buf);
    map<string, int>::const_iterator kern_it;
    for (kern_it = spacing_map_it->second.kerned_x_gaps.begin();
         kern_it != spacing_map_it->second.kerned_x_gaps.end(); ++kern_it) {
      snprintf(buf, kBufSize,
               " %s %d", kern_it->first.c_str(), kern_it->second);
      output_string.append(buf);
    }
    output_string.append("\n");
  }
  File::WriteStringToFileOrDie(output_string, output_base + ".fontinfo");
}

bool tesseract::FirstWordWouldHaveFit	(	const RowScratchRegisters &	before,
		const RowScratchRegisters &	after,
		tesseract::ParagraphJustification	justification
	)

Definition at line 1624 of file paragraphs.cpp.

                                              {
    tprintf("Don't call FirstWordWouldHaveFit(r, s, JUSTIFICATION_UNKNOWN).\n");
  }
  int available_space;
  if (justification == JUSTIFICATION_CENTER) {
    available_space = before.lindent_ + before.rindent_;
  } else {
    available_space = before.OffsideIndent(justification);
  }
  available_space -= before.ri_->average_interword_space;

  if (before.ri_->ltr)
    return after.ri_->lword_box.width() < available_space;
  return after.ri_->rword_box.width() < available_space;
}

// Return whether the first word on the after line can fit in the space at
// the end of the before line (not knowing which way the text goes) in a left

bool tesseract::FirstWordWouldHaveFit	(	const RowScratchRegisters &	before,
		const RowScratchRegisters &	after
	)

Definition at line 1649 of file paragraphs.cpp.

void tesseract::FontInfoDeleteCallback ( FontInfo f )

Definition at line 139 of file fontinfo.cpp.

                                        {
  if (f.spacing_vec != NULL) {
    f.spacing_vec->delete_data_pointers();
    delete f.spacing_vec;
  }
  delete[] f.name;
}

void tesseract::FontSetDeleteCallback ( FontSet fs )

Definition at line 146 of file fontinfo.cpp.

                                       {
  delete[] fs.configs;
}

char32 tesseract::FullwidthToHalfwidth ( const char32 ch )

Definition at line 240 of file normstrngs.cpp.

                                                           {
    if (ch != 0x3000) return ch;
  }
  // Special case for fullwidth left and right "white parentheses".
  if (ch == 0xFF5F) return 0x2985;
  if (ch == 0xFF60) return 0x2986;
  // Construct a full-to-half width transliterator.
  IcuErrorCode error_code;
  icu::UnicodeString uch_str(static_cast<UChar32>(ch));
  const icu::Transliterator* fulltohalf = icu::Transliterator::createInstance(
      "Fullwidth-Halfwidth", UTRANS_FORWARD, error_code);
  error_code.assertSuccess();
  error_code.reset();

  fulltohalf->transliterate(uch_str);
  delete fulltohalf;
  ASSERT_HOST(uch_str.length() != 0);
  return uch_str[0];
}

void tesseract::GeometricClassify	(	int	debug_level,
		GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end,
		ParagraphTheory *	theory
	)

Definition at line 1080 of file paragraphs.cpp.

                                                {
  if (!AcceptableRowArgs(debug_level, 4, __func__, rows, row_start, row_end))
    return;
  if (debug_level > 1) {
    tprintf("###############################################\n");
    tprintf("##### GeometricClassify( rows[%d:%d) )   ####\n",
            row_start, row_end);
    tprintf("###############################################\n");
  }
  RecomputeMarginsAndClearHypotheses(rows, row_start, row_end, 10);

  GeometricClassifierState s(debug_level, rows, row_start, row_end);
  if (s.left_tabs.size() > 2 && s.right_tabs.size() > 2) {
    s.Fail(2, "Too much variety for simple outline classification.");
    return;
  }
  if (s.left_tabs.size() <= 1 && s.right_tabs.size() <= 1) {
    s.Fail(1, "Not enough variety for simple outline classification.");
    return;
  }
  if (s.left_tabs.size() + s.right_tabs.size() == 3) {
    GeometricClassifyThreeTabStopTextBlock(debug_level, s, theory);
    return;
  }

  // At this point, we know that one side has at least two tab stops, and the
  // other side has one or two tab stops.
  // Left to determine:
  //   (1) Which is the body indent and which is the first line indent?
  //   (2) Is the text fully justified?

  // If one side happens to have three or more tab stops, assume that side
  // is opposite of the aligned side.
  if (s.right_tabs.size() > 2) {
    s.AssumeLeftJustification();
  } else if (s.left_tabs.size() > 2) {
    s.AssumeRightJustification();
  } else if (s.ltr) {  // guess based on script direction
    s.AssumeLeftJustification();
  } else {
    s.AssumeRightJustification();
  }

  if (s.AlignTabs().size() == 2) {
    // For each tab stop on the aligned side, how many of them appear
    // to be paragraph start lines?  [first lines]
    int firsts[2] = {0, 0};
    // Count the first line as a likely paragraph start line.
    firsts[s.AlignsideTabIndex(s.row_start)]++;
    // For each line, if the first word would have fit on the previous
    // line count it as a likely paragraph start line.
    bool jam_packed = true;
    for (int i = s.row_start + 1; i < s.row_end; i++) {
      if (s.FirstWordWouldHaveFit(i - 1, i)) {
        firsts[s.AlignsideTabIndex(i)]++;
        jam_packed = false;
      }
    }
    // Make an extra accounting for the last line of the paragraph just
    // in case it's the only short line in the block.  That is, take its
    // first word as typical and see if this looks like the *last* line
    // of a paragraph.  If so, mark the *other* indent as probably a first.
    if (jam_packed && s.FirstWordWouldHaveFit(s.row_end - 1, s.row_end - 1)) {
      firsts[1 - s.AlignsideTabIndex(s.row_end - 1)]++;
    }

    int percent0firsts, percent1firsts;
    percent0firsts = (100 * firsts[0]) / s.AlignTabs()[0].count;
    percent1firsts = (100 * firsts[1]) / s.AlignTabs()[1].count;

    // TODO(eger): Tune these constants if necessary.
    if ((percent0firsts < 20 && 30 < percent1firsts) ||
        percent0firsts + 30 < percent1firsts) {
      s.first_indent = s.AlignTabs()[1].center;
      s.body_indent = s.AlignTabs()[0].center;
    } else if ((percent1firsts < 20 && 30 < percent0firsts) ||
               percent1firsts + 30 < percent0firsts) {
      s.first_indent = s.AlignTabs()[0].center;
      s.body_indent = s.AlignTabs()[1].center;
    } else {
      // Ambiguous! Probably lineated (poetry)
      if (debug_level > 1) {
        tprintf("# Cannot determine %s indent likely to start paragraphs.\n",
                s.just == tesseract::JUSTIFICATION_LEFT ? "left" : "right");
        tprintf("# Indent of %d looks like a first line %d%% of the time.\n",
                s.AlignTabs()[0].center, percent0firsts);
        tprintf("# Indent of %d looks like a first line %d%% of the time.\n",
                s.AlignTabs()[1].center, percent1firsts);
        s.PrintRows();
      }
      return;
    }
  } else {
    // There's only one tab stop for the "aligned to" side.
    s.first_indent = s.body_indent = s.AlignTabs()[0].center;
  }

  // At this point, we have our model.
  const ParagraphModel *model = theory->AddModel(s.Model());

  // Now all we have to do is figure out if the text is fully justified or not.
  // eop_threshold: default to fully justified unless we see evidence below.
  //    See description on MarkRowsWithModel()
  s.eop_threshold =
      (s.OffsideTabs()[0].center + s.OffsideTabs()[1].center) / 2;
  // If the text is not fully justified, re-set the eop_threshold to 0.
  if (s.AlignTabs().size() == 2) {
    // Paragraphs with a paragraph-start indent.
    for (int i = s.row_start; i < s.row_end - 1; i++) {
      if (ValidFirstLine(s.rows, i + 1, model) &&
          !NearlyEqual(s.OffsideTabs()[0].center,
                       (*s.rows)[i].OffsideIndent(s.just), s.tolerance)) {
        // We found a non-end-of-paragraph short line: not fully justified.
        s.eop_threshold = 0;
        break;
      }
    }
  } else {
    // Paragraphs with no paragraph-start indent.
    for (int i = s.row_start; i < s.row_end - 1; i++) {
      if (!s.FirstWordWouldHaveFit(i, i + 1) &&
          !NearlyEqual(s.OffsideTabs()[0].center,
                       (*s.rows)[i].OffsideIndent(s.just), s.tolerance)) {
        // We found a non-end-of-paragraph short line: not fully justified.
        s.eop_threshold = 0;
        break;
      }
    }
  }
  MarkRowsWithModel(rows, row_start, row_end, model, s.ltr, s.eop_threshold);
}

// =============== Implementation of ParagraphTheory =====================

void tesseract::GeometricClassifyThreeTabStopTextBlock	(	int	debug_level,
		GeometricClassifierState &	s,
		ParagraphTheory *	theory
	)

Definition at line 988 of file paragraphs.cpp.

                             {
  int num_rows = s.row_end - s.row_start;
  int num_full_rows = 0;
  int last_row_full = 0;
  for (int i = s.row_start; i < s.row_end; i++) {
    if (s.IsFullRow(i)) {
      num_full_rows++;
      if (i == s.row_end - 1) last_row_full++;
    }
  }

  if (num_full_rows < 0.7 * num_rows) {
    s.Fail(1, "Not enough full lines to know which lines start paras.");
    return;
  }

  // eop_threshold gets set if we're fully justified; see MarkRowsWithModel()
  s.eop_threshold = 0;

  if (s.ltr) {
    s.AssumeLeftJustification();
  } else {
    s.AssumeRightJustification();
  }

  if (debug_level > 0) {
    tprintf("# Not enough variety for clear outline classification. "
            "Guessing these are %s aligned based on script.\n",
            s.ltr ? "left" : "right");
    s.PrintRows();
  }

  if (s.AlignTabs().size() == 2) {  // case A1 or A2
    s.first_indent = s.AlignTabs()[1].center;
    s.body_indent = s.AlignTabs()[0].center;
  } else {                      // case B1 or B2
    if (num_rows - 1 == num_full_rows - last_row_full) {
      // case B2
      const ParagraphModel *model = s.ltr ? kCrownLeft : kCrownRight;
      (*s.rows)[s.row_start].AddStartLine(model);
      for (int i = s.row_start + 1; i < s.row_end; i++) {
        (*s.rows)[i].AddBodyLine(model);
      }
      return;
    } else {
      // case B1
      s.first_indent = s.body_indent = s.AlignTabs()[0].center;
      s.eop_threshold = (s.OffsideTabs()[0].center +
                         s.OffsideTabs()[1].center) / 2;
    }
  }
  const ParagraphModel *model = theory->AddModel(s.Model());
  MarkRowsWithModel(s.rows, s.row_start, s.row_end, model,
                    s.ltr, s.eop_threshold);
  return;
}

// This function is called if strong textual clues were not available, but
// the caller hopes that the paragraph breaks will be super obvious just

void tesseract::GetWordBaseline	(	int	writing_direction,
		int	ppi,
		int	height,
		int	word_x1,
		int	word_y1,
		int	word_x2,
		int	word_y2,
		int	line_x1,
		int	line_y1,
		int	line_x2,
		int	line_y2,
		double *	x0,
		double *	y0,
		double *	length
	)

Definition at line 204 of file pdfrenderer.cpp.

                                                             {
  if (writing_direction == WRITING_DIRECTION_RIGHT_TO_LEFT) {
    Swap(&word_x1, &word_x2);
    Swap(&word_y1, &word_y2);
  }
  double word_length;
  double x, y;
  {
    int px = word_x1;
    int py = word_y1;
    double l2 = dist2(line_x1, line_y1, line_x2, line_y2);
    if (l2 == 0) {
      x = line_x1;
      y = line_y1;
    } else {
      double t = ((px - line_x2) * (line_x2 - line_x1) +
                  (py - line_y2) * (line_y2 - line_y1)) / l2;
      x = line_x2 + t * (line_x2 - line_x1);
      y = line_y2 + t * (line_y2 - line_y1);
    }
    word_length = sqrt(static_cast<double>(dist2(word_x1, word_y1,
                                                 word_x2, word_y2)));
    word_length = word_length * 72.0 / ppi;
    x = x * 72 / ppi;
    y = height - (y * 72.0 / ppi);
  }
  *x0 = x;
  *y0 = y;
  *length = word_length;
}

Pix* tesseract::GridReducedPix	(	const TBOX &	box,
		int	gridsize,
		ICOORD	bleft,
		int *	left,
		int *	bottom
	)

Definition at line 212 of file bbgrid.cpp.

                                                          {
  // Compute grid bounds of the outline and pad all round by 1.
  int grid_left = (box.left() - bleft.x()) / gridsize - 1;
  int grid_bottom = (box.bottom() - bleft.y()) / gridsize - 1;
  int grid_right = (box.right() - bleft.x()) / gridsize + 1;
  int grid_top = (box.top() - bleft.y()) / gridsize + 1;
  *left = grid_left;
  *bottom = grid_bottom;
  return pixCreate(grid_right - grid_left + 1,
                   grid_top - grid_bottom + 1,
                   1);
}

void tesseract::HistogramRect	(	Pix *	src_pix,
		int	channel,
		int	left,
		int	top,
		int	width,
		int	height,
		int *	histogram
	)

Definition at line 158 of file otsuthr.cpp.

                                   {
  PERF_COUNT_START("HistogramRect")
  int num_channels = pixGetDepth(src_pix) / 8;
  channel = ClipToRange(channel, 0, num_channels - 1);
  int bottom = top + height;
  memset(histogram, 0, sizeof(*histogram) * kHistogramSize);
  int src_wpl = pixGetWpl(src_pix);
  l_uint32* srcdata = pixGetData(src_pix);
  for (int y = top; y < bottom; ++y) {
    const l_uint32* linedata = srcdata + y * src_wpl;
    for (int x = 0; x < width; ++x) {
      int pixel = GET_DATA_BYTE(const_cast<void*>(
          reinterpret_cast<const void *>(linedata)),
          (x + left) * num_channels + channel);
      ++histogram[pixel];
    }
  }
  PERF_COUNT_END
}

STRING tesseract::HOcrEscape ( const char * text )

Escape a char string - remove &<>"' with HTML codes.

void tesseract::InitializeRowInfo	(	bool	after_recognition,
		const MutableIterator &	it,
		RowInfo *	info
	)

Definition at line 2414 of file paragraphs.cpp.

                                     {
    ROW *row = it.PageResIt()->row()->row;
    info->pix_ldistance = row->lmargin();
    info->pix_rdistance = row->rmargin();
    info->average_interword_space =
        row->space() > 0 ? row->space() : MAX(row->x_height(), 1);
    info->pix_xheight = row->x_height();
    info->has_leaders = false;
    info->has_drop_cap = row->has_drop_cap();
    info->ltr = true;  // set below depending on word scripts
  } else {
    info->pix_ldistance = info->pix_rdistance = 0;
    info->average_interword_space = 1;
    info->pix_xheight = 1.0;
    info->has_leaders = false;
    info->has_drop_cap = false;
    info->ltr = true;
  }

  info->num_words = 0;
  info->lword_indicates_list_item = false;
  info->lword_likely_starts_idea = false;
  info->lword_likely_ends_idea = false;
  info->rword_indicates_list_item = false;
  info->rword_likely_starts_idea = false;
  info->rword_likely_ends_idea = false;
  info->has_leaders = false;
  info->ltr = 1;

  if (!after_recognition) {
    InitializeTextAndBoxesPreRecognition(it, info);
    return;
  }
  info->text = "";
  char *text = it.GetUTF8Text(RIL_TEXTLINE);
  int trailing_ws_idx = strlen(text);  // strip trailing space
  while (trailing_ws_idx > 0 &&
         // isspace() only takes ASCII
         ((text[trailing_ws_idx - 1] & 0x80) == 0) &&
         isspace(text[trailing_ws_idx - 1]))
    trailing_ws_idx--;
  if (trailing_ws_idx > 0) {
    int lspaces = info->pix_ldistance / info->average_interword_space;
    for (int i = 0; i < lspaces; i++)
      info->text += ' ';
    for (int i = 0; i < trailing_ws_idx; i++)
      info->text += text[i];
  }
  delete []text;

  if (info->text.size() == 0) {
    return;
  }

  PAGE_RES_IT page_res_it = *it.PageResIt();
  GenericVector<WERD_RES *> werds;
  WERD_RES *word_res = page_res_it.restart_row();
  ROW_RES *this_row = page_res_it.row();
  int num_leaders = 0;
  int ltr = 0;
  int rtl = 0;
  do {
    if (word_res && word_res->best_choice->unichar_string().length() > 0) {
      werds.push_back(word_res);
      ltr += word_res->AnyLtrCharsInWord() ? 1 : 0;
      rtl += word_res->AnyRtlCharsInWord() ? 1 : 0;
      if (word_res->word->flag(W_REP_CHAR)) num_leaders++;
    }
    word_res = page_res_it.forward();
  } while (page_res_it.row() == this_row);
  info->ltr = ltr >= rtl;
  info->has_leaders = num_leaders > 3;
  info->num_words = werds.size();
  if (werds.size() > 0) {
    WERD_RES *lword = werds[0], *rword = werds[werds.size() - 1];
    info->lword_text = lword->best_choice->unichar_string().string();
    info->rword_text = rword->best_choice->unichar_string().string();
    info->lword_box = lword->word->bounding_box();
    info->rword_box = rword->word->bounding_box();
    LeftWordAttributes(lword->uch_set, lword->best_choice,
                       info->lword_text,
                       &info->lword_indicates_list_item,
                       &info->lword_likely_starts_idea,
                       &info->lword_likely_ends_idea);
    RightWordAttributes(rword->uch_set, rword->best_choice,
                        info->rword_text,
                        &info->rword_indicates_list_item,
                        &info->rword_likely_starts_idea,
                        &info->rword_likely_ends_idea);
  }
}

// This is called after rows have been identified and words are recognized.
// Much of this could be implemented before word recognition, but text helps

void tesseract::InitializeTextAndBoxesPreRecognition	(	const MutableIterator &	it,
		RowInfo *	info
	)

Definition at line 2362 of file paragraphs.cpp.

                            {
    do {
      fake_text += "x";
      if (first_word) info->lword_text += "x";
      info->rword_text += "x";
      if (pit.IsAtFinalElement(RIL_WORD, RIL_SYMBOL) &&
          !pit.IsAtFinalElement(RIL_TEXTLINE, RIL_SYMBOL)) {
        fake_text += " ";
        info->rword_text = "";
        first_word = false;
      }
    } while (!pit.IsAtFinalElement(RIL_TEXTLINE, RIL_SYMBOL) &&
             pit.Next(RIL_SYMBOL));
  }
  if (fake_text.size() == 0) return;

  int lspaces = info->pix_ldistance / info->average_interword_space;
  for (int i = 0; i < lspaces; i++) {
    info->text += ' ';
  }
  info->text += fake_text;

  // Set up lword_box, rword_box, and num_words.
  PAGE_RES_IT page_res_it = *it.PageResIt();
  WERD_RES *word_res = page_res_it.restart_row();
  ROW_RES *this_row = page_res_it.row();

  WERD_RES *lword = NULL;
  WERD_RES *rword = NULL;
  info->num_words = 0;
  do {
    if (word_res) {
      if (!lword) lword = word_res;
      if (rword != word_res) info->num_words++;
      rword = word_res;
    }
    word_res = page_res_it.forward();
  } while (page_res_it.row() == this_row);

  if (lword) info->lword_box = lword->word->bounding_box();
  if (rword) info->rword_box = rword->word->bounding_box();
}


// Given a Tesseract Iterator pointing to a text line, fill in the paragraph

ParagraphModel tesseract::InternalParagraphModelByOutline	(	const GenericVector< RowScratchRegisters > *	rows,
		int	start,
		int	end,
		int	tolerance,
		bool *	consistent
	)

Definition at line 1695 of file paragraphs.cpp.

                                    {
    ltr_line_count += static_cast<int>((*rows)[i].ri_->ltr);
  }
  bool ltr = (ltr_line_count >= (end - start) / 2);

  *consistent = true;
  if (!AcceptableRowArgs(0, 2, __func__, rows, start, end))
    return ParagraphModel();

  // Ensure the caller only passed us a region with a common rmargin and
  // lmargin.
  int lmargin = (*rows)[start].lmargin_;
  int rmargin = (*rows)[start].rmargin_;
  int lmin, lmax, rmin, rmax, cmin, cmax;
  lmin = lmax = (*rows)[start + 1].lindent_;
  rmin = rmax = (*rows)[start + 1].rindent_;
  cmin = cmax = 0;
  for (int i = start + 1; i < end; i++) {
    if ((*rows)[i].lmargin_ != lmargin || (*rows)[i].rmargin_ != rmargin) {
      tprintf("Margins don't match! Software error.\n");
      *consistent = false;
      return ParagraphModel();
    }
    UpdateRange((*rows)[i].lindent_, &lmin, &lmax);
    UpdateRange((*rows)[i].rindent_, &rmin, &rmax);
    UpdateRange((*rows)[i].rindent_ - (*rows)[i].lindent_, &cmin, &cmax);
  }
  int ldiff = lmax - lmin;
  int rdiff = rmax - rmin;
  int cdiff = cmax - cmin;
  if (rdiff > tolerance && ldiff > tolerance) {
    if (cdiff < tolerance * 2) {
      if (end - start < 3)
        return ParagraphModel();
      return ParagraphModel(JUSTIFICATION_CENTER, 0, 0, 0, tolerance);
    }
    *consistent = false;
    return ParagraphModel();
  }
  if (end - start < 3)  // Don't return a model for two line paras.
    return ParagraphModel();

  // These booleans keep us from saying something is aligned left when the body
  // left variance is too large.
  bool body_admits_left_alignment = ldiff < tolerance;
  bool body_admits_right_alignment = rdiff < tolerance;

  ParagraphModel left_model =
      ParagraphModel(JUSTIFICATION_LEFT, lmargin, (*rows)[start].lindent_,
                     (lmin + lmax) / 2, tolerance);
  ParagraphModel right_model =
      ParagraphModel(JUSTIFICATION_RIGHT, rmargin, (*rows)[start].rindent_,
                     (rmin + rmax) / 2, tolerance);

  // These booleans keep us from having an indent on the "wrong side" for the
  // first line.
  bool text_admits_left_alignment = ltr || left_model.is_flush();
  bool text_admits_right_alignment = !ltr || right_model.is_flush();

  // At least one of the edges is less than tolerance in variance.
  // If the other is obviously ragged, it can't be the one aligned to.
  // [Note the last line is included in this raggedness.]
  if (tolerance < rdiff) {
    if (body_admits_left_alignment && text_admits_left_alignment)
      return left_model;
    *consistent = false;
    return ParagraphModel();
  }
  if (tolerance < ldiff) {
    if (body_admits_right_alignment && text_admits_right_alignment)
      return right_model;
    *consistent = false;
    return ParagraphModel();
  }

  // At this point, we know the body text doesn't vary much on either side.

  // If the first line juts out oddly in one direction or the other,
  // that likely indicates the side aligned to.
  int first_left = (*rows)[start].lindent_;
  int first_right = (*rows)[start].rindent_;

  if (ltr && body_admits_left_alignment &&
      (first_left < lmin || first_left > lmax))
    return left_model;
  if (!ltr && body_admits_right_alignment &&
      (first_right < rmin || first_right > rmax))
    return right_model;

  *consistent = false;
  return ParagraphModel();
}

// Examine rows[start, end) and try to determine what sort of ParagraphModel
// would fit them as a single paragraph.   If nothing fits,

int tesseract::InterwordSpace	(	const GenericVector< RowScratchRegisters > &	rows,
		int	row_start,
		int	row_end
	)

Definition at line 1601 of file paragraphs.cpp.

                                            {
    if (rows[i].ri_->num_words > 1) {
      spacing_widths.add(rows[i].ri_->average_interword_space, 1);
    }
  }
  int minimum_reasonable_space = word_height / 3;
  if (minimum_reasonable_space < 2)
    minimum_reasonable_space = 2;
  int median = spacing_widths.median();
  return (median > minimum_reasonable_space)
      ? median : minimum_reasonable_space;
}

// Return whether the first word on the after line can fit in the space at
// the end of the before line (knowing which way the text is aligned and read).

bool tesseract::is_double_quote ( const char32 ch )

Definition at line 98 of file normstrngs.cpp.

                                                                      {
    '"',
    0x201C,  // left double quotation mark (English, others)
    0x201D,  // right double quotation mark (Danish, Finnish, Swedish, Norw.)
    0x201F,  // double high-reversed-9 quotation mark (PropList.txt)
    0x2033,  // double prime
    0x301D,  // reversed double prime quotation mark (East Asian langs, horiz.)
    0x301E,  // close double prime (East Asian languages written horizontally)
    0xFF02,  // fullwidth quotation mark
  };
  for (int i = 0; i < kNumDoubleQuoteUnicodes; ++i) {
    if (kDoubleQuoteUnicodes[i] == ch)
      return true;
  }
  return false;
}

bool tesseract::is_hyphen_punc ( const char32 ch )

Definition at line 59 of file normstrngs.cpp.

                                                                    {
    '-',
    0x2010, 0x2011, 0x2012, 0x2013, 0x2014, 0x2015,  // hyphen..horizontal bar
    0x207b,  // superscript minus
    0x208b,  // subscript minus
    0x2212,  // minus sign
    0xfe58,  // small em dash
    0xfe63,  // small hyphen-minus
    0xff0d,  // fullwidth hyphen-minus
  };
  for (int i = 0; i < kNumHyphenPuncUnicodes; ++i) {
    if (kHyphenPuncUnicodes[i] == ch)
      return true;
  }
  return false;
}

bool tesseract::is_single_quote ( const char32 ch )

Definition at line 78 of file normstrngs.cpp.

                                                                      {
    '\'',
    '`',
    0x2018,  // left single quotation mark (English, others)
    0x2019,  // right single quotation mark (Danish, Finnish, Swedish, Norw.)
             // We may have to introduce a comma set with 0x201a
    0x201B,  // single high-reveresed-9 quotation mark (PropList.txt)
    0x2032,  // prime
    0x300C,  // left corner bracket (East Asian languages)
    0xFF07,  // fullwidth apostrophe
  };
  for (int i = 0; i < kNumSingleQuoteUnicodes; ++i) {
    if (kSingleQuoteUnicodes[i] == ch)
      return true;
  }
  return false;
}

bool tesseract::IsDigitLike ( int ch )

Definition at line 200 of file paragraphs.cpp.

                            {
  return strchr("'\"({[", ch) != NULL;

bool tesseract::IsInterchangeValid ( const char32 ch )

Definition at line 209 of file normstrngs.cpp.

bool tesseract::IsInterchangeValid7BitAscii ( const char32 ch )

Definition at line 233 of file normstrngs.cpp.

bool tesseract::IsLatinLetter ( int ch )

Definition at line 196 of file paragraphs.cpp.

                         {
  return ch == 'o' || ch == 'O' || ch == 'l' || ch == 'I';

bool tesseract::IsLeftIndented ( const EquationDetect::IndentType type ) [inline]

Definition at line 95 of file equationdetect.cpp.

                                                                {
  return type == EquationDetect::LEFT_INDENT ||
      type == EquationDetect::BOTH_INDENT;
}

bool tesseract::IsOCREquivalent	(	char32	ch1,
		char32	ch2
	)

Definition at line 167 of file normstrngs.cpp.

bool tesseract::IsOpeningPunct ( int ch )

Definition at line 204 of file paragraphs.cpp.

                             {
  return strchr(":'\".?!]})", ch) != NULL;

bool tesseract::IsRightIndented ( const EquationDetect::IndentType type ) [inline]

Definition at line 100 of file equationdetect.cpp.

                                                                 {
  return type == EquationDetect::RIGHT_INDENT ||
      type == EquationDetect::BOTH_INDENT;
}

bool tesseract::IsTerminalPunct ( int ch )

Definition at line 208 of file paragraphs.cpp.

bool tesseract::IsTextOrEquationType ( PolyBlockType type ) [inline]

Definition at line 91 of file equationdetect.cpp.

                                                     {
  return PTIsTextType(type) || type == PT_EQUATION;
}

bool tesseract::IsUTF8Whitespace ( const char * text )

Definition at line 183 of file normstrngs.cpp.

bool tesseract::IsValidCodepoint ( const char32 ch )

Definition at line 171 of file normstrngs.cpp.

bool tesseract::IsWhitespace ( const char32 ch )

Definition at line 177 of file normstrngs.cpp.

                                            : 0x%x\n", ch);
  return u_isUWhiteSpace(static_cast<UChar32>(ch));
}

void tesseract::LeftoverSegments	(	const GenericVector< RowScratchRegisters > &	rows,
		GenericVector< Interval > *	to_fix,
		int	row_start,
		int	row_end
	)

Definition at line 2184 of file paragraphs.cpp.

                                            {
    bool needs_fixing = false;

    SetOfModels models;
    SetOfModels models_w_crowns;
    rows[i].StrongHypotheses(&models);
    rows[i].NonNullHypotheses(&models_w_crowns);
    if (models.empty() && models_w_crowns.size() > 0) {
      // Crown paragraph.  Is it followed by a modeled line?
      for (int end = i + 1; end < rows.size(); end++) {
        SetOfModels end_models;
        SetOfModels strong_end_models;
        rows[end].NonNullHypotheses(&end_models);
        rows[end].StrongHypotheses(&strong_end_models);
        if (end_models.size() == 0) {
          needs_fixing = true;
          break;
        } else if (strong_end_models.size() > 0) {
          needs_fixing = false;
          break;
        }
      }
    } else if (models.empty() && rows[i].ri_->num_words > 0) {
      // No models at all.
      needs_fixing = true;
    }

    if (!needs_fixing && !models.empty()) {
      needs_fixing = RowIsStranded(rows, i);
    }

    if (needs_fixing) {
      if (!to_fix->empty() && to_fix->back().end == i - 1)
        to_fix->back().end = i;
      else
        to_fix->push_back(Interval(i, i));
    }
  }
  // Convert inclusive intervals to half-open intervals.
  for (int i = 0; i < to_fix->size(); i++) {
    (*to_fix)[i].end = (*to_fix)[i].end + 1;
  }
}

// Given a set of row_owners pointing to PARAs or NULL (no paragraph known),
// normalize each row_owner to point to an actual PARA, and output the

void tesseract::LeftWordAttributes	(	const UNICHARSET *	unicharset,
		const WERD_CHOICE *	werd,
		const STRING &	utf8,
		bool *	is_list,
		bool *	starts_idea,
		bool *	ends_idea
	)

Definition at line 397 of file paragraphs.cpp.

                                                                 {  // Empty
    *ends_idea = true;
    return;
  }

  if (unicharset && werd) {  // We have a proper werd and unicharset so use it.
    if (UniLikelyListItem(unicharset, werd)) {
      *is_list = true;
      *starts_idea = true;
      *ends_idea = true;
    }
    if (unicharset->get_isupper(werd->unichar_id(0))) {
      *starts_idea = true;
    }
    if (unicharset->get_ispunctuation(werd->unichar_id(0))) {
      *starts_idea = true;
      *ends_idea = true;
    }
  } else {  // Assume utf8 is mostly ASCII
    if (AsciiLikelyListItem(utf8)) {
      *is_list = true;
      *starts_idea = true;
    }
    int start_letter = utf8[0];
    if (IsOpeningPunct(start_letter)) {
      *starts_idea = true;
    }
    if (IsTerminalPunct(start_letter)) {
      *ends_idea = true;
    }
    if (start_letter >= 'A' && start_letter <= 'Z') {
      *starts_idea = true;
    }
  }
}

// Given the rightmost word of a line either as a Tesseract unicharset + werd
// or a utf8 string, set the following attributes for it:

bool tesseract::LikelyListMark ( const STRING & word )

Definition at line 265 of file paragraphs.cpp.

                                             {
  return LikelyListMark(word) || LikelyListNumeral(word);

bool tesseract::LikelyListMarkUnicode ( int ch )

Definition at line 331 of file paragraphs.cpp.

              {
    // TODO(eger) expand this list of unicodes as needed.
    case 0x00B0:  // degree sign
    case 0x2022:  // bullet
    case 0x25E6:  // white bullet
    case 0x00B7:  // middle dot
    case 0x25A1:  // white square
    case 0x25A0:  // black square
    case 0x25AA:  // black small square
    case 0x2B1D:  // black very small square
    case 0x25BA:  // black right-pointing pointer
    case 0x25CF:  // black circle
    case 0x25CB:  // white circle
      return true;
    default:
      break;  // fall through
  }
  return false;
}

// Return whether it is very likely that this is a numeral marker that could
// start a list item.  Some examples include:

bool tesseract::LikelyListNumeral ( const STRING & word )

Definition at line 231 of file paragraphs.cpp.

                        {(";
  const char *kSep = ":;-.,";
  const char *kClose = "]})";

  int num_segments = 0;
  const char *pos = word.string();
  while (*pos != '\0' && num_segments < 3) {
    // skip up to two open parens.
    const char *numeral_start = SkipOne(SkipOne(pos, kOpen), kOpen);
    const char *numeral_end = SkipChars(numeral_start, kRomans);
    if (numeral_end != numeral_start) {
      // Got Roman Numeral. Great.
    } else {
      numeral_end = SkipChars(numeral_start, kDigits);
      if (numeral_end == numeral_start) {
        // If there's a single latin letter, we can use that.
        numeral_end = SkipChars(numeral_start, IsLatinLetter);
        if (numeral_end - numeral_start != 1)
          break;
      }
    }
    // We got some sort of numeral.
    num_segments++;
    // Skip any trailing parens or punctuation.
    pos = SkipChars(SkipChars(numeral_end, kClose), kSep);
    if (pos == numeral_end)
      break;
  }
  return *pos == '\0';
}

bool LikelyListMark(const STRING &word) {
  const char *kListMarks = "0Oo*.,+.";

bool tesseract::LikelyParagraphStart	(	const RowScratchRegisters &	before,
		const RowScratchRegisters &	after
	)

Definition at line 1675 of file paragraphs.cpp.

bool tesseract::LikelyParagraphStart	(	const RowScratchRegisters &	before,
		const RowScratchRegisters &	after,
		tesseract::ParagraphJustification	j
	)

Definition at line 1682 of file paragraphs.cpp.

bool tesseract::LoadDataFromFile	(	const STRING &	filename,
		GenericVector< char > *	data
	)		`[inline]`

Definition at line 356 of file genericvector.h.

                                                        {
  FILE* fp = fopen(filename.string(), "rb");
  if (fp == NULL) return false;
  fseek(fp, 0, SEEK_END);
  size_t size = ftell(fp);
  fseek(fp, 0, SEEK_SET);
  // Pad with a 0, just in case we treat the result as a string.
  data->init_to_size((int)size + 1, 0);
  bool result = fread(&(*data)[0], 1, size, fp) == size;
  fclose(fp);
  return result;
}

ShapeTable * tesseract::LoadShapeTable ( const STRING & file_prefix )

Definition at line 118 of file commontraining.cpp.

                                                      {
  ShapeTable* shape_table = NULL;
  STRING shape_table_file = file_prefix;
  shape_table_file += kShapeTableFileSuffix;
  FILE* shape_fp = fopen(shape_table_file.string(), "rb");
  if (shape_fp != NULL) {
    shape_table = new ShapeTable;
    if (!shape_table->DeSerialize(false, shape_fp)) {
      delete shape_table;
      shape_table = NULL;
      tprintf("Error: Failed to read shape table %s\n",
              shape_table_file.string());
    } else {
      int num_shapes = shape_table->NumShapes();
      tprintf("Read shape table %s of %d shapes\n",
              shape_table_file.string(), num_shapes);
    }
    fclose(shape_fp);
  } else {
    tprintf("Warning: No shape table file present: %s\n",
            shape_table_file.string());
  }
  return shape_table;
}

MasterTrainer * tesseract::LoadTrainingData	(	int	argc,
		const char const	argv,
		bool	replication,
		ShapeTable **	shape_table,
		STRING *	file_prefix
	)

Creates a MasterTraininer and loads the training data into it: Initializes feature_defs and IntegerFX. Loads the shape_table if shape_table != NULL. Loads initial unicharset from -U command-line option. If FLAGS_T is set, loads the majority of data from there, else:

Loads font info from -F option.
Loads xheights from -X option.
Loads samples from .tr files in remaining command-line args.
Deletes outliers and computes canonical samples.
If FLAGS_output_trainer is set, saves the trainer for future use. Computes canonical and cloud features. If shape_table is not NULL, but failed to load, make a fake flat one, as shape clustering was not run.

Definition at line 175 of file commontraining.cpp.

                                                     {
  InitFeatureDefs(&feature_defs);
  InitIntegerFX();
  *file_prefix = "";
  if (!FLAGS_D.empty()) {
    *file_prefix += FLAGS_D.c_str();
    *file_prefix += "/";
  }
  // If we are shape clustering (NULL shape_table) or we successfully load
  // a shape_table written by a previous shape clustering, then
  // shape_analysis will be true, meaning that the MasterTrainer will replace
  // some members of the unicharset with their fragments.
  bool shape_analysis = false;
  if (shape_table != NULL) {
    *shape_table = LoadShapeTable(*file_prefix);
    if (*shape_table != NULL)
      shape_analysis = true;
  } else {
    shape_analysis = true;
  }
  MasterTrainer* trainer = new MasterTrainer(NM_CHAR_ANISOTROPIC,
                                             shape_analysis,
                                             replication,
                                             FLAGS_debug_level);
  IntFeatureSpace fs;
  fs.Init(kBoostXYBuckets, kBoostXYBuckets, kBoostDirBuckets);
  if (FLAGS_T.empty()) {
    trainer->LoadUnicharset(FLAGS_U.c_str());
    // Get basic font information from font_properties.
    if (!FLAGS_F.empty()) {
      if (!trainer->LoadFontInfo(FLAGS_F.c_str())) {
        delete trainer;
        return NULL;
      }
    }
    if (!FLAGS_X.empty()) {
      if (!trainer->LoadXHeights(FLAGS_X.c_str())) {
        delete trainer;
        return NULL;
      }
    }
    trainer->SetFeatureSpace(fs);
    const char* page_name;
    // Load training data from .tr files on the command line.
    while ((page_name = GetNextFilename(argc, argv)) != NULL) {
      tprintf("Reading %s ...\n", page_name);
      trainer->ReadTrainingSamples(page_name, feature_defs, false);

      // If there is a file with [lang].[fontname].exp[num].fontinfo present,
      // read font spacing information in to fontinfo_table.
      int pagename_len = strlen(page_name);
      char *fontinfo_file_name = new char[pagename_len + 7];
      strncpy(fontinfo_file_name, page_name, pagename_len - 2);  // remove "tr"
      strcpy(fontinfo_file_name + pagename_len - 2, "fontinfo");  // +"fontinfo"
      trainer->AddSpacingInfo(fontinfo_file_name);
      delete[] fontinfo_file_name;

      // Load the images into memory if required by the classifier.
      if (FLAGS_load_images) {
        STRING image_name = page_name;
        // Chop off the tr and replace with tif. Extension must be tif!
        image_name.truncate_at(image_name.length() - 2);
        image_name += "tif";
        trainer->LoadPageImages(image_name.string());
      }
    }
    trainer->PostLoadCleanup();
    // Write the master trainer if required.
    if (!FLAGS_output_trainer.empty()) {
      FILE* fp = fopen(FLAGS_output_trainer.c_str(), "wb");
      if (fp == NULL) {
        tprintf("Can't create saved trainer data!\n");
      } else {
        trainer->Serialize(fp);
        fclose(fp);
      }
    }
  } else {
    bool success = false;
    tprintf("Loading master trainer from file:%s\n",
            FLAGS_T.c_str());
    FILE* fp = fopen(FLAGS_T.c_str(), "rb");
    if (fp == NULL) {
      tprintf("Can't read file %s to initialize master trainer\n",
              FLAGS_T.c_str());
    } else {
      success = trainer->DeSerialize(false, fp);
      fclose(fp);
    }
    if (!success) {
      tprintf("Deserialize of master trainer failed!\n");
      delete trainer;
      return NULL;
    }
    trainer->SetFeatureSpace(fs);
  }
  trainer->PreTrainingSetup();
  if (!FLAGS_O.empty() &&
      !trainer->unicharset().save_to_file(FLAGS_O.c_str())) {
    fprintf(stderr, "Failed to save unicharset to file %s\n", FLAGS_O.c_str());
    delete trainer;
    return NULL;
  }
  if (shape_table != NULL) {
    // If we previously failed to load a shapetable, then shape clustering
    // wasn't run so make a flat one now.
    if (*shape_table == NULL) {
      *shape_table = new ShapeTable;
      trainer->SetupFlatShapeTable(*shape_table);
      tprintf("Flat shape table summary: %s\n",
              (*shape_table)->SummaryStr().string());
    }
    (*shape_table)->set_unicharset(trainer->unicharset());
  }
  return trainer;
}

bool tesseract::MakeIndividualGlyphs	(	Pix *	pix,
		const vector< BoxChar * > &	vbox,
		const int	input_tiff_page
	)

Definition at line 309 of file text2image.cpp.

                                                     {
  // If checks fail, return false without exiting text2image
  if (!pix) {
    tprintf("ERROR: MakeIndividualGlyphs(): Input Pix* is NULL\n");
    return false;
  } else if (FLAGS_glyph_resized_size <= 0) {
    tprintf("ERROR: --glyph_resized_size must be positive\n");
    return false;
  } else if (FLAGS_glyph_num_border_pixels_to_pad < 0) {
    tprintf("ERROR: --glyph_num_border_pixels_to_pad must be 0 or positive\n");
    return false;
  }

  const int n_boxes = vbox.size();
  int n_boxes_saved = 0;
  int current_tiff_page = 0;
  int y_previous = 0;
  static int glyph_count = 0;
  for (int i = 0; i < n_boxes; i++) {
    // Get one bounding box
    Box* b = vbox[i]->mutable_box();
    if (!b) continue;
    const int x = b->x;
    const int y = b->y;
    const int w = b->w;
    const int h = b->h;
    // Check present tiff page (for multipage tiff)
    if (y < y_previous-pixGetHeight(pix)/10) {
      tprintf("ERROR: Wrap-around encountered, at i=%d\n", i);
      current_tiff_page++;
    }
    if (current_tiff_page < input_tiff_page) continue;
    else if (current_tiff_page > input_tiff_page) break;
    // Check box validity
    if (x < 0 || y < 0 ||
        (x+w-1) >= pixGetWidth(pix) ||
        (y+h-1) >= pixGetHeight(pix)) {
      tprintf("ERROR: MakeIndividualGlyphs(): Index out of range, at i=%d"
              " (x=%d, y=%d, w=%d, h=%d\n)", i, x, y, w, h);
      continue;
    } else if (w < FLAGS_glyph_num_border_pixels_to_pad &&
               h < FLAGS_glyph_num_border_pixels_to_pad) {
      tprintf("ERROR: Input image too small to be a character, at i=%d\n", i);
      continue;
    }
    // Crop the boxed character
    Pix* pix_glyph = pixClipRectangle(pix, b, NULL);
    if (!pix_glyph) {
      tprintf("ERROR: MakeIndividualGlyphs(): Failed to clip, at i=%d\n", i);
      continue;
    }
    // Resize to square
    Pix* pix_glyph_sq = pixScaleToSize(pix_glyph,
                                       FLAGS_glyph_resized_size,
                                       FLAGS_glyph_resized_size);
    if (!pix_glyph_sq) {
      tprintf("ERROR: MakeIndividualGlyphs(): Failed to resize, at i=%d\n", i);
      continue;
    }
    // Zero-pad
    Pix* pix_glyph_sq_pad = pixAddBorder(pix_glyph_sq,
                                         FLAGS_glyph_num_border_pixels_to_pad,
                                         0);
    if (!pix_glyph_sq_pad) {
      tprintf("ERROR: MakeIndividualGlyphs(): Failed to zero-pad, at i=%d\n",
              i);
      continue;
    }
    // Write out
    Pix* pix_glyph_sq_pad_8 = pixConvertTo8(pix_glyph_sq_pad, false);
    char filename[1024];
    snprintf(filename, 1024, "%s_%d.jpg", FLAGS_outputbase.c_str(),
             glyph_count++);
    if (pixWriteJpeg(filename, pix_glyph_sq_pad_8, 100, 0)) {
      tprintf("ERROR: MakeIndividualGlyphs(): Failed to write JPEG to %s,"
              " at i=%d\n", filename, i);
      continue;
    }

    pixDestroy(&pix_glyph);
    pixDestroy(&pix_glyph_sq);
    pixDestroy(&pix_glyph_sq_pad);
    pixDestroy(&pix_glyph_sq_pad_8);
    n_boxes_saved++;
    y_previous = y;
  }
  if (n_boxes_saved == 0) {
    return false;
  } else {
    tprintf("Total number of characters saved = %d\n", n_boxes_saved);
    return true;
  }
}

void tesseract::MarkRowsWithModel	(	GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end,
		const ParagraphModel *	model,
		bool	ltr,
		int	eop_threshold
	)

Definition at line 810 of file paragraphs.cpp.

                                          {
  if (!AcceptableRowArgs(0, 0, __func__, rows, row_start, row_end))
    return;
  for (int row = row_start; row < row_end; row++) {
    bool valid_first = ValidFirstLine(rows, row, model);
    bool valid_body = ValidBodyLine(rows, row, model);
    if (valid_first && !valid_body) {
      (*rows)[row].AddStartLine(model);
    } else if (valid_body && !valid_first) {
      (*rows)[row].AddBodyLine(model);
    } else if (valid_body && valid_first) {
      bool after_eop = (row == row_start);
      if (row > row_start) {
        if (eop_threshold > 0) {
          if (model->justification() == JUSTIFICATION_LEFT) {
            after_eop = (*rows)[row - 1].rindent_ > eop_threshold;
          } else {
            after_eop = (*rows)[row - 1].lindent_ > eop_threshold;
          }
        } else {
         after_eop = FirstWordWouldHaveFit((*rows)[row - 1], (*rows)[row],
                                           model->justification());
        }
      }
      if (after_eop) {
        (*rows)[row].AddStartLine(model);
      } else {
        (*rows)[row].AddBodyLine(model);
      }
    } else {
      // Do nothing. Stray row.
    }
  }
}

// GeometricClassifierState holds all of the information we'll use while
// trying to determine a paragraph model for the text lines in a block of

void tesseract::MarkStrongEvidence	(	GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end
	)

Definition at line 1833 of file paragraphs.cpp.

                                                {
    const RowScratchRegisters &prev = (*rows)[i - 1];
    RowScratchRegisters &curr = (*rows)[i];
    tesseract::ParagraphJustification typical_justification =
        prev.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT;
    if (!curr.ri_->rword_likely_starts_idea &&
        !curr.ri_->lword_likely_starts_idea &&
        !FirstWordWouldHaveFit(prev, curr, typical_justification)) {
      curr.SetBodyLine();
    }
  }

  // Record patently obvious start paragraph lines.
  //
  // It's an extremely good signal of the start of a paragraph that
  // the first word would have fit on the end of the previous line.
  // However, applying just that signal would have us mark random
  // start lines of lineated text (poetry and source code) and some
  // centered headings as paragraph start lines.  Therefore, we use
  // a second qualification for a paragraph start: Not only should
  // the first word of this line have fit on the previous line,
  // but also, this line should go full to the right of the block,
  // disallowing a subsequent word from having fit on this line.

  // First row:
  {
    RowScratchRegisters &curr = (*rows)[row_start];
    RowScratchRegisters &next = (*rows)[row_start + 1];
    tesseract::ParagraphJustification j =
        curr.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT;
    if (curr.GetLineType() == LT_UNKNOWN &&
        !FirstWordWouldHaveFit(curr, next, j) &&
        (curr.ri_->lword_likely_starts_idea ||
         curr.ri_->rword_likely_starts_idea)) {
      curr.SetStartLine();
    }
  }
  // Middle rows
  for (int i = row_start + 1; i < row_end - 1; i++) {
    RowScratchRegisters &prev = (*rows)[i - 1];
    RowScratchRegisters &curr = (*rows)[i];
    RowScratchRegisters &next = (*rows)[i + 1];
    tesseract::ParagraphJustification j =
        curr.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT;
    if (curr.GetLineType() == LT_UNKNOWN &&
        !FirstWordWouldHaveFit(curr, next, j) &&
        LikelyParagraphStart(prev, curr, j)) {
      curr.SetStartLine();
    }
  }
  // Last row
  {  // the short circuit at the top means we have at least two lines.
    RowScratchRegisters &prev = (*rows)[row_end - 2];
    RowScratchRegisters &curr = (*rows)[row_end - 1];
    tesseract::ParagraphJustification j =
        curr.ri_->ltr ? JUSTIFICATION_LEFT : JUSTIFICATION_RIGHT;
    if (curr.GetLineType() == LT_UNKNOWN &&
        !FirstWordWouldHaveFit(curr, curr, j) &&
        LikelyParagraphStart(prev, curr, j)) {
      curr.SetStartLine();
    }
  }
}

// Look for sequences of a start line followed by some body lines in
// rows[row_start, row_end) and create ParagraphModels for them if

void tesseract::ModelStrongEvidence	(	int	debug_level,
		GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end,
		bool	allow_flush_models,
		ParagraphTheory *	theory
	)

Definition at line 1903 of file paragraphs.cpp.

                                                  {
  if (!AcceptableRowArgs(debug_level, 2, __func__, rows, row_start, row_end))
    return;

  int start = row_start;
  while (start < row_end) {
    while (start < row_end && (*rows)[start].GetLineType() != LT_START)
      start++;
    if (start >= row_end - 1)
      break;

    int tolerance = Epsilon((*rows)[start + 1].ri_->average_interword_space);
    int end = start;
    ParagraphModel last_model;
    bool next_consistent;
    do {
      ++end;
      // rows[row, end) was consistent.
      // If rows[row, end + 1) is not consistent,
      //   just model rows[row, end)
      if (end < row_end - 1) {
        RowScratchRegisters &next = (*rows)[end];
        LineType lt = next.GetLineType();
        next_consistent = lt == LT_BODY ||
            (lt == LT_UNKNOWN &&
             !FirstWordWouldHaveFit((*rows)[end - 1], (*rows)[end]));
      } else {
        next_consistent = false;
      }
      if (next_consistent) {
        ParagraphModel next_model = InternalParagraphModelByOutline(
            rows, start, end + 1, tolerance, &next_consistent);
        if (((*rows)[start].ri_->ltr &&
             last_model.justification() == JUSTIFICATION_LEFT &&
             next_model.justification() != JUSTIFICATION_LEFT) ||
            (!(*rows)[start].ri_->ltr &&
             last_model.justification() == JUSTIFICATION_RIGHT &&
             next_model.justification() != JUSTIFICATION_RIGHT)) {
          next_consistent = false;
        }
        last_model = next_model;
      } else {
        next_consistent = false;
      }
    } while (next_consistent && end < row_end);
    // At this point, rows[start, end) looked like it could have been a
    // single paragraph.  If we can make a good ParagraphModel for it,
    // do so and mark this sequence with that model.
    if (end > start + 1) {
      // emit a new paragraph if we have more than one line.
      const ParagraphModel *model = NULL;
      ParagraphModel new_model = ParagraphModelByOutline(
          debug_level, rows, start, end,
          Epsilon(InterwordSpace(*rows, start, end)));
      if (new_model.justification() == JUSTIFICATION_UNKNOWN) {
        // couldn't create a good model, oh well.
      } else if (new_model.is_flush()) {
        if (end == start + 2) {
          // It's very likely we just got two paragraph starts in a row.
          end = start + 1;
        } else if (start == row_start) {
          // Mark this as a Crown.
          if (new_model.justification() == JUSTIFICATION_LEFT) {
            model = kCrownLeft;
          } else {
            model = kCrownRight;
          }
        } else if (allow_flush_models) {
          model = theory->AddModel(new_model);
        }
      } else {
        model = theory->AddModel(new_model);
      }
      if (model) {
        (*rows)[start].AddStartLine(model);
        for (int i = start + 1; i < end; i++) {
          (*rows)[i].AddBodyLine(model);
        }
      }
    }
    start = end;
  }
}

// We examine rows[row_start, row_end) and do the following:
//   (1) Clear all existing hypotheses for the rows being considered.

void tesseract::NormalizeChar32	(	char32	ch,
		GenericVector< char32 > *	str
	)

Definition at line 132 of file normstrngs.cpp.

                                              {
    // If any spaces were added by NFKC, pretend normalization is a nop.
    if (norm_str[i] == ' ') {
      str->clear();
      str->push_back(ch);
      break;
    } else {
      str->push_back(OCRNormalize(static_cast<char32>(norm_str[i])));
    }
  }
}

uinT8 tesseract::NormalizeDirection	(	uinT8	dir,
		const FCOORD &	unnormed_pos,
		const DENORM &	denorm,
		const DENORM *	root_denorm
	)

Definition at line 171 of file intfx.cpp.

                                                                          {
  // Convert direction to a vector.
  FCOORD unnormed_end;
  unnormed_end.from_direction(dir);
  unnormed_end += unnormed_pos;
  FCOORD normed_pos, normed_end;
  denorm.NormTransform(root_denorm, unnormed_pos, &normed_pos);
  denorm.NormTransform(root_denorm, unnormed_end, &normed_end);
  normed_end -= normed_pos;
  return normed_end.to_direction();
}

STRING tesseract::NormalizeUTF8String ( const char * str8 )

Definition at line 117 of file normstrngs.cpp.

                                           {
    norm_str.clear();
    NormalizeChar32(str32[i], &norm_str);
    for (int j = 0; j < norm_str.length(); ++j) {
      out_str32.push_back(norm_str[j]);
    }
  }
  STRING out_str8;
  UTF32ToUTF8(out_str32, &out_str8);
  return out_str8;
}

char32 tesseract::OCRNormalize ( char32 ch )

Definition at line 157 of file normstrngs.cpp.

int tesseract::OtsuStats	(	const int *	histogram,
		int *	H_out,
		int *	omega0_out
	)

Definition at line 183 of file otsuthr.cpp.

                                           {
    H += histogram[i];
    mu_T += static_cast<double>(i) * histogram[i];
  }

  // Now maximize sig_sq_B over t.
  // http://www.ctie.monash.edu.au/hargreave/Cornall_Terry_328.pdf
  int best_t = -1;
  int omega_0, omega_1;
  int best_omega_0 = 0;
  double best_sig_sq_B = 0.0;
  double mu_0, mu_1, mu_t;
  omega_0 = 0;
  mu_t = 0.0;
  for (int t = 0; t < kHistogramSize - 1; ++t) {
    omega_0 += histogram[t];
    mu_t += t * static_cast<double>(histogram[t]);
    if (omega_0 == 0)
      continue;
    omega_1 = H - omega_0;
    if (omega_1 == 0)
      break;
    mu_0 = mu_t / omega_0;
    mu_1 = (mu_T - mu_t) / omega_1;
    double sig_sq_B = mu_1 - mu_0;
    sig_sq_B *= sig_sq_B * omega_0 * omega_1;
    if (best_t < 0 || sig_sq_B > best_sig_sq_B) {
      best_sig_sq_B = sig_sq_B;
      best_t = t;
      best_omega_0 = omega_0;
    }
  }
  if (H_out != NULL) *H_out = H;
  if (omega0_out != NULL) *omega0_out = best_omega_0;
  return best_t;
}

int tesseract::OtsuThreshold	(	Pix *	src_pix,
		int	left,
		int	top,
		int	width,
		int	height,
		int **	thresholds,
		int **	hi_values
	)

Definition at line 40 of file otsuthr.cpp.

                                                     {
  int num_channels = pixGetDepth(src_pix) / 8;
  // Of all channels with no good hi_value, keep the best so we can always
  // produce at least one answer.
  PERF_COUNT_START("OtsuThreshold")
  int best_hi_value = 1;
  int best_hi_index = 0;
  bool any_good_hivalue = false;
  double best_hi_dist = 0.0;
  *thresholds = new int[num_channels];
  *hi_values = new int[num_channels];
  // all of channel 0 then all of channel 1...
  int *histogramAllChannels = new int[kHistogramSize * num_channels];

  // only use opencl if compiled w/ OpenCL and selected device is opencl
#ifdef USE_OPENCL
    // Calculate Histogram on GPU
    OpenclDevice od;
    if (od.selectedDeviceIsOpenCL() &&
        (num_channels == 1 || num_channels == 4) && top == 0 && left == 0 ) {
      od.HistogramRectOCL(
          (const unsigned char*)pixGetData(src_pix),
          num_channels,
          pixGetWpl(src_pix) * 4,
          left,
          top,
          width,
          height,
          kHistogramSize,
          histogramAllChannels);

    // Calculate Threshold from Histogram on cpu
    for (int ch = 0; ch < num_channels; ++ch) {
      (*thresholds)[ch] = -1;
      (*hi_values)[ch] = -1;
      int *histogram = &histogramAllChannels[kHistogramSize * ch];
      int H;
      int best_omega_0;
      int best_t = OtsuStats(histogram, &H, &best_omega_0);
      if (best_omega_0 == 0 || best_omega_0 == H) {
         // This channel is empty.
         continue;
       }
      // To be a convincing foreground we must have a small fraction of H
      // or to be a convincing background we must have a large fraction of H.
      // In between we assume this channel contains no thresholding information.
      int hi_value = best_omega_0 < H * 0.5;
      (*thresholds)[ch] = best_t;
      if (best_omega_0 > H * 0.75) {
        any_good_hivalue = true;
        (*hi_values)[ch] = 0;
      } else if (best_omega_0 < H * 0.25) {
        any_good_hivalue = true;
        (*hi_values)[ch] = 1;
      } else {
        // In case all channels are like this, keep the best of the bad lot.
        double hi_dist = hi_value ? (H - best_omega_0) : best_omega_0;
        if (hi_dist > best_hi_dist) {
          best_hi_dist = hi_dist;
          best_hi_value = hi_value;
          best_hi_index = ch;
        }
      }
    }
  } else {
#endif
    for (int ch = 0; ch < num_channels; ++ch) {
      (*thresholds)[ch] = -1;
      (*hi_values)[ch] = -1;
      // Compute the histogram of the image rectangle.
      int histogram[kHistogramSize];
      HistogramRect(src_pix, ch, left, top, width, height, histogram);
      int H;
      int best_omega_0;
      int best_t = OtsuStats(histogram, &H, &best_omega_0);
      if (best_omega_0 == 0 || best_omega_0 == H) {
         // This channel is empty.
         continue;
       }
      // To be a convincing foreground we must have a small fraction of H
      // or to be a convincing background we must have a large fraction of H.
      // In between we assume this channel contains no thresholding information.
      int hi_value = best_omega_0 < H * 0.5;
      (*thresholds)[ch] = best_t;
      if (best_omega_0 > H * 0.75) {
        any_good_hivalue = true;
        (*hi_values)[ch] = 0;
      } else if (best_omega_0 < H * 0.25) {
        any_good_hivalue = true;
        (*hi_values)[ch] = 1;
      } else {
        // In case all channels are like this, keep the best of the bad lot.
        double hi_dist = hi_value ? (H - best_omega_0) : best_omega_0;
        if (hi_dist > best_hi_dist) {
          best_hi_dist = hi_dist;
          best_hi_value = hi_value;
          best_hi_index = ch;
        }
      }
    }
#ifdef USE_OPENCL
  }
#endif  // USE_OPENCL
  delete[] histogramAllChannels;

  if (!any_good_hivalue) {
    // Use the best of the ones that were not good enough.
    (*hi_values)[best_hi_index] = best_hi_value;
  }
  PERF_COUNT_END
  return num_channels;
}

ParagraphModel tesseract::ParagraphModelByOutline	(	int	debug_level,
		const GenericVector< RowScratchRegisters > *	rows,
		int	start,
		int	end,
		int	tolerance
	)

Definition at line 1796 of file paragraphs.cpp.

                                       {
  bool unused_consistent;
  ParagraphModel retval = InternalParagraphModelByOutline(
      rows, start, end, tolerance, &unused_consistent);
  if (debug_level >= 2 && retval.justification() == JUSTIFICATION_UNKNOWN) {
    tprintf("Could not determine a model for this paragraph:\n");
    PrintRowRange(*rows, start, end);
  }
  return retval;
}

// Do rows[start, end) form a single instance of the given paragraph model?
bool RowsFitModel(const GenericVector<RowScratchRegisters> *rows,

int tesseract::ParamsTrainingFeatureByName ( const char * name )

Definition at line 26 of file params_training_featdef.cpp.

                                                  {
  if (name == NULL)
    return -1;
  int array_size = sizeof(kParamsTrainingFeatureTypeName) /
    sizeof(kParamsTrainingFeatureTypeName[0]);
  for (int i = 0; i < array_size; i++) {
    if (kParamsTrainingFeatureTypeName[i] == NULL)
      continue;
    if (strcmp(name, kParamsTrainingFeatureTypeName[i]) == 0)
      return i;
  }
  return -1;
}

void tesseract::ParseCommandLineFlags	(	const char *	usage,
		int *	argc,
		char ***	argv,
		const bool	remove_flags
	)

Definition at line 312 of file commandlineflags.cpp.

                                                    {
  InitGoogle(usage, argc, argv, remove_flags);
}

double tesseract::prec ( double x )

Definition at line 184 of file pdfrenderer.cpp.

                      {
  double kPrecision = 1000.0;
  double a = round(x * kPrecision) / kPrecision;
  if (a == -0)
    return 0;
  return a;
}

bool tesseract::PSM_BLOCK_FIND_ENABLED ( int pageseg_mode ) [inline]

Definition at line 191 of file publictypes.h.

                                                     {
  return pageseg_mode >= PSM_AUTO_OSD && pageseg_mode <= PSM_SINGLE_COLUMN;
}

bool tesseract::PSM_COL_FIND_ENABLED ( int pageseg_mode ) [inline]

Definition at line 185 of file publictypes.h.

                                                   {
  return pageseg_mode >= PSM_AUTO_OSD && pageseg_mode <= PSM_AUTO;
}

bool tesseract::PSM_LINE_FIND_ENABLED ( int pageseg_mode ) [inline]

Definition at line 194 of file publictypes.h.

                                                    {
  return pageseg_mode >= PSM_AUTO_OSD && pageseg_mode <= PSM_SINGLE_BLOCK;
}

bool tesseract::PSM_ORIENTATION_ENABLED ( int pageseg_mode ) [inline]

Definition at line 182 of file publictypes.h.

                                                      {
  return pageseg_mode <= PSM_AUTO || pageseg_mode == PSM_SPARSE_TEXT_OSD;
}

bool tesseract::PSM_OSD_ENABLED ( int pageseg_mode ) [inline]

Inline functions that act on a PageSegMode to determine whether components of layout analysis are enabled. *Depend critically on the order of elements of PageSegMode.* NOTE that arg is an int for compatibility with INT_PARAM.

Definition at line 179 of file publictypes.h.

                                              {
  return pageseg_mode <= PSM_AUTO_OSD || pageseg_mode == PSM_SPARSE_TEXT_OSD;
}

bool tesseract::PSM_SPARSE ( int pageseg_mode ) [inline]

Definition at line 188 of file publictypes.h.

                                         {
  return pageseg_mode == PSM_SPARSE_TEXT || pageseg_mode == PSM_SPARSE_TEXT_OSD;
}

bool tesseract::PSM_WORD_FIND_ENABLED ( int pageseg_mode ) [inline]

Definition at line 197 of file publictypes.h.

                                                    {
  return (pageseg_mode >= PSM_AUTO_OSD && pageseg_mode <= PSM_SINGLE_LINE) ||
      pageseg_mode == PSM_SPARSE_TEXT || pageseg_mode == PSM_SPARSE_TEXT_OSD;
}

bool tesseract::read_info	(	FILE *	f,
		FontInfo *	fi,
		bool	swap
	)

Definition at line 152 of file fontinfo.cpp.

                                                 {
  inT32 size;
  if (fread(&size, sizeof(size), 1, f) != 1) return false;
  if (swap)
    Reverse32(&size);
  char* font_name = new char[size + 1];
  fi->name = font_name;
  if (static_cast<int>(fread(font_name, sizeof(*font_name), size, f)) != size)
    return false;
  font_name[size] = '\0';
  if (fread(&fi->properties, sizeof(fi->properties), 1, f) != 1) return false;
  if (swap)
    Reverse32(&fi->properties);
  return true;
}

bool tesseract::read_set	(	FILE *	f,
		FontSet *	fs,
		bool	swap
	)

Definition at line 240 of file fontinfo.cpp.

                                               {
  if (fread(&fs->size, sizeof(fs->size), 1, f) != 1) return false;
  if (swap)
    Reverse32(&fs->size);
  fs->configs = new int[fs->size];
  for (int i = 0; i < fs->size; ++i) {
    if (fread(&fs->configs[i], sizeof(fs->configs[i]), 1, f) != 1) return false;
    if (swap)
      Reverse32(&fs->configs[i]);
  }
  return true;
}

bool tesseract::read_spacing_info	(	FILE *	f,
		FontInfo *	fi,
		bool	swap
	)

Definition at line 177 of file fontinfo.cpp.

                                                         {
  inT32 vec_size, kern_size;
  if (fread(&vec_size, sizeof(vec_size), 1, f) != 1) return false;
  if (swap) Reverse32(&vec_size);
  ASSERT_HOST(vec_size >= 0);
  if (vec_size == 0) return true;
  fi->init_spacing(vec_size);
  for (int i = 0; i < vec_size; ++i) {
    FontSpacingInfo *fs = new FontSpacingInfo();
    if (fread(&fs->x_gap_before, sizeof(fs->x_gap_before), 1, f) != 1 ||
        fread(&fs->x_gap_after, sizeof(fs->x_gap_after), 1, f) != 1 ||
        fread(&kern_size, sizeof(kern_size), 1, f) != 1) {
      delete fs;
      return false;
    }
    if (swap) {
      ReverseN(&(fs->x_gap_before), sizeof(fs->x_gap_before));
      ReverseN(&(fs->x_gap_after), sizeof(fs->x_gap_after));
      Reverse32(&kern_size);
    }
    if (kern_size < 0) {  // indication of a NULL entry in fi->spacing_vec
      delete fs;
      continue;
    }
    if (kern_size > 0 && (!fs->kerned_unichar_ids.DeSerialize(swap, f) ||
                          !fs->kerned_x_gaps.DeSerialize(swap, f))) {
      delete fs;
      return false;
    }
    fi->add_spacing(i, fs);
  }
  return true;
}

bool tesseract::read_t	(	PAGE_RES_IT *	page_res_it,
		TBOX *	tbox
	)

Definition at line 53 of file recogtraining.cpp.

                                                  {
  while (page_res_it->block() != NULL && page_res_it->word() == NULL)
    page_res_it->forward();

  if (page_res_it->word() != NULL) {
    *tbox = page_res_it->word()->word->bounding_box();

    // If tbox->left() is negative, the training image has vertical text and
    // all the coordinates of bounding boxes of page_res are rotated by 90
    // degrees in a counterclockwise direction. We need to rotate the TBOX back
    // in order to compare with the TBOXes of box files.
    if (tbox->left() < 0) {
      tbox->rotate(FCOORD(0.0, -1.0));
    }

    return true;
  } else {
    return false;
  }
}

void tesseract::RecomputeMarginsAndClearHypotheses	(	GenericVector< RowScratchRegisters > *	rows,
		int	start,
		int	end,
		int	percentile
	)

Definition at line 1561 of file paragraphs.cpp.

                                    {
    RowScratchRegisters &sr = (*rows)[i];
    sr.SetUnknown();
    if (sr.ri_->num_words == 0)
      continue;
    UpdateRange(sr.lmargin_ + sr.lindent_, &lmin, &lmax);
    UpdateRange(sr.rmargin_ + sr.rindent_, &rmin, &rmax);
  }
  STATS lefts(lmin, lmax + 1);
  STATS rights(rmin, rmax + 1);
  for (int i = start; i < end; i++) {
    RowScratchRegisters &sr = (*rows)[i];
    if (sr.ri_->num_words == 0)
      continue;
    lefts.add(sr.lmargin_ + sr.lindent_, 1);
    rights.add(sr.rmargin_ + sr.rindent_, 1);
  }
  int ignorable_left = lefts.ile(ClipToRange(percentile, 0, 100) / 100.0);
  int ignorable_right = rights.ile(ClipToRange(percentile, 0, 100) / 100.0);
  for (int i = start; i < end; i++) {
    RowScratchRegisters &sr = (*rows)[i];
    int ldelta = ignorable_left - sr.lmargin_;
    sr.lmargin_ += ldelta;
    sr.lindent_ -= ldelta;
    int rdelta = ignorable_right - sr.rmargin_;
    sr.rmargin_ += rdelta;
    sr.rindent_ -= rdelta;
  }
}

// Return the median inter-word space in rows[row_start, row_end).
int InterwordSpace(const GenericVector<RowScratchRegisters> &rows,

void tesseract::RightWordAttributes	(	const UNICHARSET *	unicharset,
		const WERD_CHOICE *	werd,
		const STRING &	utf8,
		bool *	is_list,
		bool *	starts_idea,
		bool *	ends_idea
	)

Definition at line 444 of file paragraphs.cpp.

                                                                 {  // Empty
    *ends_idea = true;
    return;
  }

  if (unicharset && werd) {  // We have a proper werd and unicharset so use it.
    if (UniLikelyListItem(unicharset, werd)) {
      *is_list = true;
      *starts_idea = true;
    }
    UNICHAR_ID last_letter = werd->unichar_id(werd->length() - 1);
    if (unicharset->get_ispunctuation(last_letter)) {
      *ends_idea = true;
    }
  } else {  // Assume utf8 is mostly ASCII
    if (AsciiLikelyListItem(utf8)) {
      *is_list = true;
      *starts_idea = true;
    }
    int last_letter = utf8[utf8.size() - 1];
    if (IsOpeningPunct(last_letter) || IsTerminalPunct(last_letter)) {
      *ends_idea = true;
    }
  }
}

// =============== Implementation of RowScratchRegisters =====================
/* static */

bool tesseract::RowIsStranded	(	const GenericVector< RowScratchRegisters > &	rows,
		int	row
	)

Definition at line 2142 of file paragraphs.cpp.

                                              {
    bool all_starts = rows[row].GetLineType();
    int run_length = 1;
    bool continues = true;
    for (int i = row - 1; i >= 0 && continues; i--) {
      SetOfModels models;
      rows[i].NonNullHypotheses(&models);
      switch (rows[i].GetLineType(row_models[m])) {
        case LT_START: run_length++; break;
        case LT_MULTIPLE:  // explicit fall-through
        case LT_BODY: run_length++; all_starts = false; break;
        case LT_UNKNOWN:  // explicit fall-through
        default: continues = false;
      }
    }
    continues = true;
    for (int i = row + 1; i < rows.size() && continues; i++) {
      SetOfModels models;
      rows[i].NonNullHypotheses(&models);
      switch (rows[i].GetLineType(row_models[m])) {
        case LT_START: run_length++; break;
        case LT_MULTIPLE:  // explicit fall-through
        case LT_BODY: run_length++; all_starts = false; break;
        case LT_UNKNOWN:  // explicit fall-through
        default: continues = false;
      }
    }
    if (run_length > 2 || (!all_starts && run_length > 1)) return false;
  }
  return true;
}

// Go through rows[row_start, row_end) and gather up sequences that need better
// classification.

bool tesseract::RowsFitModel	(	const GenericVector< RowScratchRegisters > *	rows,
		int	start,
		int	end,
		const ParagraphModel *	model
	)

Definition at line 1811 of file paragraphs.cpp.

                                         {
    if (!ValidBodyLine(rows, i, model)) return false;
  }
  return true;
}

// Examine rows[row_start, row_end) as an independent section of text,
// and mark rows that are exceptionally clear as start-of-paragraph

STRING tesseract::RtlEmbed	(	const STRING &	word,
		bool	rtlify
	)

Definition at line 124 of file paragraphs.cpp.

bool tesseract::SaveDataToFile	(	const GenericVector< char > &	data,
		const STRING &	filename
	)		`[inline]`

Definition at line 371 of file genericvector.h.

                                                   {
  FILE* fp = fopen(filename.string(), "wb");
  if (fp == NULL) return false;
  bool result =
      static_cast<int>(fwrite(&data[0], 1, data.size(), fp)) == data.size();
  fclose(fp);
  return result;
}

const char * tesseract::ScriptPosToString ( enum ScriptPos script_pos )

Definition at line 180 of file ratngs.cpp.

                                                         {
  switch (script_pos) {
    case SP_NORMAL: return "NORM";
    case SP_SUBSCRIPT: return "SUB";
    case SP_SUPERSCRIPT: return "SUPER";
    case SP_DROPCAP: return "DROPC";
  }
  return "SP_UNKNOWN";
}

void tesseract::SeparateSimpleLeaderLines	(	GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end,
		ParagraphTheory *	theory
	)

Definition at line 2028 of file paragraphs.cpp.

                                                    {
    if ((*rows)[i - 1].ri_->has_leaders &&
        (*rows)[i].ri_->has_leaders &&
        (*rows)[i + 1].ri_->has_leaders) {
      const ParagraphModel *model = theory->AddModel(
          ParagraphModel(JUSTIFICATION_UNKNOWN, 0, 0, 0, 0));
      (*rows)[i].AddStartLine(model);
    }
  }
}

// Collect sequences of unique hypotheses in row registers and create proper
// paragraphs for them, referencing the paragraphs in row_owners.

void tesseract::SetBlobStrokeWidth	(	Pix *	pix,
		BLOBNBOX *	blob
	)

Definition at line 58 of file tordmain.cpp.

                                                  {
  // Cut the blob rectangle into a Pix.
  int pix_height = pixGetHeight(pix);
  const TBOX& box = blob->bounding_box();
  int width = box.width();
  int height = box.height();
  Box* blob_pix_box = boxCreate(box.left(), pix_height - box.top(),
                                width, height);
  Pix* pix_blob = pixClipRectangle(pix, blob_pix_box, NULL);
  boxDestroy(&blob_pix_box);
  Pix* dist_pix = pixDistanceFunction(pix_blob, 4, 8, L_BOUNDARY_BG);
  pixDestroy(&pix_blob);
  // Compute the stroke widths.
  uinT32* data = pixGetData(dist_pix);
  int wpl = pixGetWpl(dist_pix);
  // Horizontal width of stroke.
  STATS h_stats(0, width + 1);
  for (int y = 0; y < height; ++y) {
    uinT32* pixels = data + y*wpl;
    int prev_pixel = 0;
    int pixel = GET_DATA_BYTE(pixels, 0);
    for (int x = 1; x < width; ++x) {
      int next_pixel = GET_DATA_BYTE(pixels, x);
      // We are looking for a pixel that is equal to its vertical neighbours,
      // yet greater than its left neighbour.
      if (prev_pixel < pixel &&
          (y == 0 || pixel == GET_DATA_BYTE(pixels - wpl, x - 1)) &&
          (y == height - 1 || pixel == GET_DATA_BYTE(pixels + wpl, x - 1))) {
        if (pixel > next_pixel) {
          // Single local max, so an odd width.
          h_stats.add(pixel * 2 - 1, 1);
        } else if (pixel == next_pixel && x + 1 < width &&
                 pixel > GET_DATA_BYTE(pixels, x + 1)) {
          // Double local max, so an even width.
          h_stats.add(pixel * 2, 1);
        }
      }
      prev_pixel = pixel;
      pixel = next_pixel;
    }
  }
  // Vertical width of stroke.
  STATS v_stats(0, height + 1);
  for (int x = 0; x < width; ++x) {
    int prev_pixel = 0;
    int pixel = GET_DATA_BYTE(data, x);
    for (int y = 1; y < height; ++y) {
      uinT32* pixels = data + y*wpl;
      int next_pixel = GET_DATA_BYTE(pixels, x);
      // We are looking for a pixel that is equal to its horizontal neighbours,
      // yet greater than its upper neighbour.
      if (prev_pixel < pixel &&
          (x == 0 || pixel == GET_DATA_BYTE(pixels - wpl, x - 1)) &&
          (x == width - 1 || pixel == GET_DATA_BYTE(pixels - wpl, x + 1))) {
        if (pixel > next_pixel) {
          // Single local max, so an odd width.
          v_stats.add(pixel * 2 - 1, 1);
        } else if (pixel == next_pixel && y + 1 < height &&
                 pixel > GET_DATA_BYTE(pixels + wpl, x)) {
          // Double local max, so an even width.
          v_stats.add(pixel * 2, 1);
        }
      }
      prev_pixel = pixel;
      pixel = next_pixel;
    }
  }
  pixDestroy(&dist_pix);
  // Store the horizontal and vertical width in the blob, keeping both
  // widths if there is enough information, otherwse only the one with
  // the most samples.
  // If there are insufficent samples, store zero, rather than using
  // 2*area/perimeter, as the numbers that gives do not match the numbers
  // from the distance method.
  if (h_stats.get_total() >= (width + height) / 4) {
    blob->set_horz_stroke_width(h_stats.ile(0.5f));
    if (v_stats.get_total() >= (width + height) / 4)
      blob->set_vert_stroke_width(v_stats.ile(0.5f));
    else
      blob->set_vert_stroke_width(0.0f);
  } else {
    if (v_stats.get_total() >= (width + height) / 4 ||
        v_stats.get_total() > h_stats.get_total()) {
      blob->set_horz_stroke_width(0.0f);
      blob->set_vert_stroke_width(v_stats.ile(0.5f));
    } else {
      blob->set_horz_stroke_width(h_stats.get_total() > 2 ? h_stats.ile(0.5f)
                                                          : 0.0f);
      blob->set_vert_stroke_width(0.0f);
    }
  }
}

void tesseract::SetPropertiesForInputFile	(	const string &	script_dir,
		const string &	input_unicharset_file,
		const string &	output_unicharset_file,
		const string &	output_xheights_file
	)

Definition at line 148 of file unicharset_training_utils.cpp.

                                                                   {
  UNICHARSET unicharset;

  // Load the input unicharset
  unicharset.load_from_file(input_unicharset_file.c_str());
  tprintf("Loaded unicharset of size %d from file %s\n", unicharset.size(),
          input_unicharset_file.c_str());

  // Set unichar properties
  tprintf("Setting unichar properties\n");
  SetupBasicProperties(true, &unicharset);
  string xheights_str;
  for (int s = 0; s < unicharset.get_script_table_size(); ++s) {
    // Load the unicharset for the script if available.
    string filename = script_dir + "/" +
        unicharset.get_script_from_script_id(s) + ".unicharset";
    UNICHARSET script_set;
    if (script_set.load_from_file(filename.c_str())) {
      unicharset.SetPropertiesFromOther(script_set);
    }
    // Load the xheights for the script if available.
    filename = script_dir + "/" + unicharset.get_script_from_script_id(s) +
        ".xheights";
    string script_heights;
    if (File::ReadFileToString(filename, &script_heights))
      xheights_str += script_heights;
  }
  if (!output_xheights_file.empty())
    File::WriteStringToFileOrDie(xheights_str, output_xheights_file);
  for (int c = SPECIAL_UNICHAR_CODES_COUNT; c < unicharset.size(); ++c) {
    if (unicharset.PropertiesIncomplete(c)) {
      tprintf("Warning: properties incomplete for index %d = %s\n",
              c, unicharset.id_to_unichar(c));
    }
  }

  // Write the output unicharset
  tprintf("Writing unicharset to file %s\n", output_unicharset_file.c_str());
  unicharset.save_to_file(output_unicharset_file.c_str());
}

void tesseract::SetupBasicProperties	(	bool	report_errors,
		UNICHARSET *	unicharset
	)

Definition at line 40 of file unicharset_training_utils.cpp.

                                                                      {
  for (int unichar_id = 0; unichar_id < unicharset->size(); ++unichar_id) {
    // Convert any custom ligatures.
    const char* unichar_str = unicharset->id_to_unichar(unichar_id);
    for (int i = 0; UNICHARSET::kCustomLigatures[i][0] != NULL; ++i) {
      if (!strcmp(UNICHARSET::kCustomLigatures[i][1], unichar_str)) {
        unichar_str = UNICHARSET::kCustomLigatures[i][0];
        break;
      }
    }

    // Convert the unichar to UTF32 representation
    GenericVector<char32> uni_vector;
    tesseract::UTF8ToUTF32(unichar_str, &uni_vector);

    // Assume that if the property is true for any character in the string,
    // then it holds for the whole "character".
    bool unichar_isalpha = false;
    bool unichar_islower = false;
    bool unichar_isupper = false;
    bool unichar_isdigit = false;
    bool unichar_ispunct = false;

    for (int i = 0; i < uni_vector.size(); ++i) {
      if (u_isalpha(uni_vector[i]))
        unichar_isalpha = true;
      if (u_islower(uni_vector[i]))
        unichar_islower = true;
      if (u_isupper(uni_vector[i]))
        unichar_isupper = true;
      if (u_isdigit(uni_vector[i]))
        unichar_isdigit = true;
      if (u_ispunct(uni_vector[i]))
        unichar_ispunct = true;
    }

    unicharset->set_isalpha(unichar_id, unichar_isalpha);
    unicharset->set_islower(unichar_id, unichar_islower);
    unicharset->set_isupper(unichar_id, unichar_isupper);
    unicharset->set_isdigit(unichar_id, unichar_isdigit);
    unicharset->set_ispunctuation(unichar_id, unichar_ispunct);

    tesseract::IcuErrorCode err;
    unicharset->set_script(unichar_id, uscript_getName(
        uscript_getScript(uni_vector[0], err)));

    const int num_code_points = uni_vector.size();
    // Obtain the lower/upper case if needed and record it in the properties.
    unicharset->set_other_case(unichar_id, unichar_id);
    if (unichar_islower || unichar_isupper) {
      GenericVector<char32> other_case(num_code_points, 0);
      for (int i = 0; i < num_code_points; ++i) {
        // TODO(daria): Ideally u_strToLower()/ustrToUpper() should be used.
        // However since they deal with UChars (so need a conversion function
        // from char32 or UTF8string) and require a meaningful locale string,
        // for now u_tolower()/u_toupper() are used.
        other_case[i] = unichar_islower ? u_toupper(uni_vector[i]) :
          u_tolower(uni_vector[i]);
      }
      STRING other_case_uch;
      tesseract::UTF32ToUTF8(other_case, &other_case_uch);
      UNICHAR_ID other_case_id =
          unicharset->unichar_to_id(other_case_uch.c_str());
      if (other_case_id != INVALID_UNICHAR_ID) {
        unicharset->set_other_case(unichar_id, other_case_id);
      } else if (unichar_id >= SPECIAL_UNICHAR_CODES_COUNT && report_errors) {
        tprintf("Other case %s of %s is not in unicharset\n",
                other_case_uch.c_str(), unichar_str);
      }
    }

    // Set RTL property and obtain mirror unichar ID from ICU.
    GenericVector<char32> mirrors(num_code_points, 0);
    for (int i = 0; i < num_code_points; ++i) {
      mirrors[i] = u_charMirror(uni_vector[i]);
      if (i == 0) {  // set directionality to that of the 1st code point
        unicharset->set_direction(unichar_id,
                                  static_cast<UNICHARSET::Direction>(
                                      u_charDirection(uni_vector[i])));
      }
    }
    STRING mirror_uch;
    tesseract::UTF32ToUTF8(mirrors, &mirror_uch);
    UNICHAR_ID mirror_uch_id = unicharset->unichar_to_id(mirror_uch.c_str());
    if (mirror_uch_id != INVALID_UNICHAR_ID) {
      unicharset->set_mirror(unichar_id, mirror_uch_id);
    } else if (report_errors) {
      tprintf("Mirror %s of %s is not in unicharset\n",
              mirror_uch.c_str(), unichar_str);
    }

    // Record normalized version of this unichar.
    STRING normed_str = tesseract::NormalizeUTF8String(unichar_str);
    if (unichar_id != 0 && normed_str.length() > 0) {
      unicharset->set_normed(unichar_id, normed_str.c_str());
    } else {
      unicharset->set_normed(unichar_id, unichar_str);
    }
    ASSERT_HOST(unicharset->get_other_case(unichar_id) < unicharset->size());
  }
  unicharset->post_load_setup();
}

const char* tesseract::SkipChars	(	const char *	str,
		bool(*)(int)	skip
	)

Definition at line 218 of file paragraphs.cpp.

                                                         {
  if (*str != '\0' && strchr(toskip, *str)) return str + 1;

const char* tesseract::SkipChars	(	const char *	str,
		const char *	toskip
	)

Definition at line 213 of file paragraphs.cpp.

                                                          {
  while (*str != '\0' && skip(*str)) { str++; }

const char* tesseract::SkipOne	(	const char *	str,
		const char *	toskip
	)

Definition at line 223 of file paragraphs.cpp.

template<typename T >

int tesseract::sort_cmp	(	const void *	t1,
		const void *	t2
	)

Definition at line 391 of file genericvector.h.

                                             {
  const T* a = static_cast<const T *> (t1);
  const T* b = static_cast<const T *> (t2);
  if (*a < *b) {
    return -1;
  } else if (*b < *a) {
    return 1;
  } else {
    return 0;
  }
}

template<typename T >

int tesseract::sort_ptr_cmp	(	const void *	t1,
		const void *	t2
	)

Definition at line 408 of file genericvector.h.

                                                 {
  const T* a = *reinterpret_cast<T * const *>(t1);
  const T* b = *reinterpret_cast<T * const *>(t2);
  if (*a < *b) {
    return -1;
  } else if (*b < *a) {
    return 1;
  } else {
    return 0;
  }
}

template<class BBC >

int tesseract::SortByBoxBottom	(	const void *	void1,
		const void *	void2
	)

Definition at line 408 of file bbgrid.h.

                                                          {
  // The void*s are actually doubly indirected, so get rid of one level.
  const BBC* p1 = *reinterpret_cast<const BBC* const *>(void1);
  const BBC* p2 = *reinterpret_cast<const BBC* const *>(void2);
  int result = p1->bounding_box().bottom() - p2->bounding_box().bottom();
  if (result != 0)
    return result;
  result =  p1->bounding_box().top() - p2->bounding_box().top();
  if (result != 0)
    return result;
  result = p1->bounding_box().left() - p2->bounding_box().left();
  if (result != 0)
    return result;
  return p1->bounding_box().right() - p2->bounding_box().right();
}

template<class BBC >

int tesseract::SortByBoxLeft	(	const void *	void1,
		const void *	void2
	)

Definition at line 372 of file bbgrid.h.

                                                        {
  // The void*s are actually doubly indirected, so get rid of one level.
  const BBC* p1 = *reinterpret_cast<const BBC* const *>(void1);
  const BBC* p2 = *reinterpret_cast<const BBC* const *>(void2);
  int result = p1->bounding_box().left() - p2->bounding_box().left();
  if (result != 0)
    return result;
  result = p1->bounding_box().right() - p2->bounding_box().right();
  if (result != 0)
    return result;
  result = p1->bounding_box().bottom() - p2->bounding_box().bottom();
  if (result != 0)
    return result;
  return p1->bounding_box().top() - p2->bounding_box().top();
}

template<class BLOB_CHOICE >

int tesseract::SortByRating	(	const void *	void1,
		const void *	void2
	)

Definition at line 86 of file pieces.cpp.

                                                       {
  const BLOB_CHOICE *p1 = *reinterpret_cast<const BLOB_CHOICE * const *>(void1);
  const BLOB_CHOICE *p2 = *reinterpret_cast<const BLOB_CHOICE * const *>(void2);

  if (p1->rating() < p2->rating())
    return 1;
  return -1;
}

template<class BLOB_CHOICE >

int tesseract::SortByUnicharID	(	const void *	void1,
		const void *	void2
	)

Definition at line 78 of file pieces.cpp.

                                                          {
  const BLOB_CHOICE *p1 = *reinterpret_cast<const BLOB_CHOICE * const *>(void1);
  const BLOB_CHOICE *p2 = *reinterpret_cast<const BLOB_CHOICE * const *>(void2);

  return p1->unichar_id() - p2->unichar_id();
}

template<class BBC >

int tesseract::SortRightToLeft	(	const void *	void1,
		const void *	void2
	)

Definition at line 390 of file bbgrid.h.

                                                          {
  // The void*s are actually doubly indirected, so get rid of one level.
  const BBC* p1 = *reinterpret_cast<const BBC* const *>(void1);
  const BBC* p2 = *reinterpret_cast<const BBC* const *>(void2);
  int result = p2->bounding_box().right() - p1->bounding_box().right();
  if (result != 0)
    return result;
  result = p2->bounding_box().left() - p1->bounding_box().left();
  if (result != 0)
    return result;
  result = p1->bounding_box().bottom() - p2->bounding_box().bottom();
  if (result != 0)
    return result;
  return p1->bounding_box().top() - p2->bounding_box().top();
}

int tesseract::SpanUTF8NotWhitespace ( const char * text )

Definition at line 198 of file normstrngs.cpp.

             {
    if (IsWhitespace(*it)) break;
    n_notwhite += it.utf8_len();
  }
  return n_notwhite;
}

int tesseract::SpanUTF8Whitespace ( const char * text )

Definition at line 187 of file normstrngs.cpp.

             {
    if (!IsWhitespace(*it)) break;
    n_white += it.utf8_len();
  }
  return n_white;
}

void tesseract::StrongEvidenceClassify	(	int	debug_level,
		GenericVector< RowScratchRegisters > *	rows,
		int	row_start,
		int	row_end,
		ParagraphTheory *	theory
	)

Definition at line 1998 of file paragraphs.cpp.

                                                     {
  if (!AcceptableRowArgs(debug_level, 2, __func__, rows, row_start, row_end))
    return;

  if (debug_level > 1) {
    tprintf("#############################################\n");
    tprintf("# StrongEvidenceClassify( rows[%d:%d) )\n", row_start, row_end);
    tprintf("#############################################\n");
  }

  RecomputeMarginsAndClearHypotheses(rows, row_start, row_end, 10);
  MarkStrongEvidence(rows, row_start, row_end);

  DebugDump(debug_level > 2, "Initial strong signals.", *theory, *rows);

  // Create paragraph models.
  ModelStrongEvidence(debug_level, rows, row_start, row_end, false, theory);

  DebugDump(debug_level > 2, "Unsmeared hypotheses.s.", *theory, *rows);

  // At this point, some rows are marked up as paragraphs with model numbers,
  // and some rows are marked up as either LT_START or LT_BODY.  Now let's
  // smear any good paragraph hypotheses forward and backward.
  ParagraphModelSmearer smearer(rows, row_start, row_end, theory);
  smearer.Smear();
}

void SeparateSimpleLeaderLines(GenericVector<RowScratchRegisters> *rows,
                               int row_start, int row_end,

bool tesseract::StrongModel ( const ParagraphModel * model ) [inline]

Definition at line 75 of file paragraphs_internal.h.

                                                     {
  return model != NULL && model != kCrownLeft && model != kCrownRight;
}

bool tesseract::TextSupportsBreak	(	const RowScratchRegisters &	before,
		const RowScratchRegisters &	after
	)

Definition at line 1664 of file paragraphs.cpp.

         {
    return before.ri_->lword_likely_ends_idea &&
           after.ri_->rword_likely_starts_idea;
  }
}

bool LikelyParagraphStart(const RowScratchRegisters &before,
                          const RowScratchRegisters &after) {

Pix * tesseract::TraceBlockOnReducedPix	(	BLOCK *	block,
		int	gridsize,
		ICOORD	bleft,
		int *	left,
		int *	bottom
	)

Definition at line 258 of file bbgrid.cpp.

                                                                  {
  TBOX box = block->bounding_box();
  Pix* pix = GridReducedPix(box, gridsize, bleft, left, bottom);
  int wpl = pixGetWpl(pix);
  l_uint32* data = pixGetData(pix);
  ICOORDELT_IT it(block->poly_block()->points());
  for (it.mark_cycle_pt(); !it.cycled_list();) {
    ICOORD pos = *it.data();
    it.forward();
    ICOORD next_pos = *it.data();
    ICOORD line_vector = next_pos - pos;
    int major, minor;
    ICOORD major_step, minor_step;
    line_vector.setup_render(&major_step, &minor_step, &major, &minor);
    int accumulator = major / 2;
    while (pos != next_pos) {
      int grid_x = (pos.x() - bleft.x()) / gridsize - *left;
      int grid_y = (pos.y() - bleft.y()) / gridsize - *bottom;
      SET_DATA_BIT(data + grid_y * wpl, grid_x);
      pos += major_step;
      accumulator += minor;
      if (accumulator >= major) {
        accumulator -= major;
        pos += minor_step;
      }
    }
  }
  return pix;
}

Pix * tesseract::TraceOutlineOnReducedPix	(	C_OUTLINE *	outline,
		int	gridsize,
		ICOORD	bleft,
		int *	left,
		int *	bottom
	)

Definition at line 232 of file bbgrid.cpp.

                                                                    {
  TBOX box = outline->bounding_box();
  Pix* pix = GridReducedPix(box, gridsize, bleft, left, bottom);
  int wpl = pixGetWpl(pix);
  l_uint32* data = pixGetData(pix);
  int length = outline->pathlength();
  ICOORD pos = outline->start_pos();
  for (int i = 0; i < length; ++i) {
    int grid_x = (pos.x() - bleft.x()) / gridsize - *left;
    int grid_y = (pos.y() - bleft.y()) / gridsize - *bottom;
    SET_DATA_BIT(data + grid_y * wpl, grid_x);
    pos += outline->step(i);
  }
  return pix;
}

int tesseract::UnicodeFor	(	const UNICHARSET *	u,
		const WERD_CHOICE *	werd,
		int	pos
	)

Definition at line 277 of file paragraphs.cpp.

bool tesseract::UniLikelyListItem	(	const UNICHARSET *	u,
		const WERD_CHOICE *	werd
	)

Definition at line 360 of file paragraphs.cpp.

                                                   {
    int numeral_start = m.SkipPunc(pos);
    if (numeral_start > pos + 1) break;
    int numeral_end = m.SkipRomans(numeral_start);
    if (numeral_end == numeral_start) {
      numeral_end = m.SkipDigits(numeral_start);
      if (numeral_end == numeral_start) {
        // If there's a single latin letter, we can use that.
        numeral_end = m.SkipAlpha(numeral_start);
        if (numeral_end - numeral_start != 1)
          break;
      }
    }
    // We got some sort of numeral.
    num_segments++;
    // Skip any trailing punctuation.
    pos = m.SkipPunc(numeral_end);
    if (pos == numeral_end)
      break;
  }
  return pos == werd->length();
}

// ========= Brain Dead Language Model (combined entry points) ================

void tesseract::UTF32ToUTF8	(	const GenericVector< char32 > &	str32,
		STRING *	utf8_str
	)

Definition at line 46 of file normstrngs.cpp.

                                           {
    UNICHAR uni_ch(str32[i]);
    char *utf8 = uni_ch.utf8_str();
    if (utf8 != NULL) {
      (*utf8_str) += utf8;
      delete[] utf8;
    }
  }
}

void tesseract::UTF8ToUTF32	(	const char *	utf8_str,
		GenericVector< char32 > *	str32
	)

Definition at line 32 of file normstrngs.cpp.

                                         {
    step = UNICHAR::utf8_step(utf8_str + ch);
    if (step > 0) {
      UNICHAR uni_ch(utf8_str + ch, step);
      (*str32) += uni_ch.first_uni();
    }
  }
}

bool tesseract::ValidBodyLine	(	const GenericVector< RowScratchRegisters > *	rows,
		int	row,
		const ParagraphModel *	model
	)

Definition at line 1280 of file paragraphs.cpp.

bool tesseract::ValidFirstLine	(	const GenericVector< RowScratchRegisters > *	rows,
		int	row,
		const ParagraphModel *	model
	)

Definition at line 1269 of file paragraphs.cpp.

bool tesseract::write_info	(	FILE *	f,
		const FontInfo &	fi
	)

Definition at line 168 of file fontinfo.cpp.

                                             {
  inT32 size = strlen(fi.name);
  if (fwrite(&size, sizeof(size), 1, f) != 1) return false;
  if (static_cast<int>(fwrite(fi.name, sizeof(*fi.name), size, f)) != size)
    return false;
  if (fwrite(&fi.properties, sizeof(fi.properties), 1, f) != 1) return false;
  return true;
}

bool tesseract::write_set	(	FILE *	f,
		const FontSet &	fs
	)

Definition at line 253 of file fontinfo.cpp.

                                           {
  if (fwrite(&fs.size, sizeof(fs.size), 1, f) != 1) return false;
  for (int i = 0; i < fs.size; ++i) {
    if (fwrite(&fs.configs[i], sizeof(fs.configs[i]), 1, f) != 1) return false;
  }
  return true;
}

bool tesseract::write_spacing_info	(	FILE *	f,
		const FontInfo &	fi
	)

Definition at line 211 of file fontinfo.cpp.

                                                     {
  inT32 vec_size = (fi.spacing_vec == NULL) ? 0 : fi.spacing_vec->size();
  if (fwrite(&vec_size,  sizeof(vec_size), 1, f) != 1) return false;
  inT16 x_gap_invalid = -1;
  for (int i = 0; i < vec_size; ++i) {
    FontSpacingInfo *fs = fi.spacing_vec->get(i);
    inT32 kern_size = (fs == NULL) ? -1 : fs->kerned_x_gaps.size();
    if (fs == NULL) {
      // Valid to have the identical fwrites. Writing invalid x-gaps.
      if (fwrite(&(x_gap_invalid), sizeof(x_gap_invalid), 1, f) != 1 ||
          fwrite(&(x_gap_invalid), sizeof(x_gap_invalid), 1, f) != 1 ||
          fwrite(&kern_size, sizeof(kern_size), 1, f) != 1) {
        return false;
      }
    } else {
      if (fwrite(&(fs->x_gap_before), sizeof(fs->x_gap_before), 1, f) != 1 ||
          fwrite(&(fs->x_gap_after), sizeof(fs->x_gap_after), 1, f) != 1 ||
          fwrite(&kern_size, sizeof(kern_size), 1, f) != 1) {
        return false;
      }
    }
    if (kern_size > 0 && (!fs->kerned_unichar_ids.Serialize(f) ||
                          !fs->kerned_x_gaps.Serialize(f))) {
      return false;
    }
  }
  return true;
}

void tesseract::WriteShapeTable	(	const STRING &	file_prefix,
		const ShapeTable &	shape_table
	)

Definition at line 144 of file commontraining.cpp.

                                                                               {
  STRING shape_table_file = file_prefix;
  shape_table_file += kShapeTableFileSuffix;
  FILE* fp = fopen(shape_table_file.string(), "wb");
  if (fp != NULL) {
    if (!shape_table.Serialize(fp)) {
      fprintf(stderr, "Error writing shape table: %s\n",
              shape_table_file.string());
    }
    fclose(fp);
  } else {
    fprintf(stderr, "Error creating shape table: %s\n",
            shape_table_file.string());
  }
}

void tesseract::YOutlierPieces	(	WERD_RES *	word,
		int	rebuilt_blob_index,
		int	super_y_bottom,
		int	sub_y_top,
		ScriptPos *	leading_pos,
		int *	num_leading_outliers,
		ScriptPos *	trailing_pos,
		int *	num_trailing_outliers
	)

Given a recognized blob, see if a contiguous collection of sub-pieces (chopped blobs) starting at its left might qualify as being a subscript or superscript letter based only on y position. Also do this for the right side.

Definition at line 46 of file superscript.cpp.

                                                                         {
  ScriptPos sp_unused1, sp_unused2;
  int unused1, unused2;
  if (!leading_pos) leading_pos = &sp_unused1;
  if (!num_leading_outliers) num_leading_outliers = &unused1;
  if (!trailing_pos) trailing_pos = &sp_unused2;
  if (!num_trailing_outliers) num_trailing_outliers = &unused2;

  *num_leading_outliers = *num_trailing_outliers = 0;
  *leading_pos = *trailing_pos = SP_NORMAL;

  int chopped_start = LeadingUnicharsToChopped(word, rebuilt_blob_index);
  int num_chopped_pieces = word->best_state[rebuilt_blob_index];
  ScriptPos last_pos = SP_NORMAL;
  int trailing_outliers = 0;
  for (int i = 0; i < num_chopped_pieces; i++) {
    TBOX box = word->chopped_word->blobs[chopped_start + i]->bounding_box();
    ScriptPos pos = SP_NORMAL;
    if (box.bottom() >= super_y_bottom) {
      pos = SP_SUPERSCRIPT;
    } else if (box.top() <= sub_y_top) {
      pos = SP_SUBSCRIPT;
    }
    if (pos == SP_NORMAL) {
      if (trailing_outliers == i) {
        *num_leading_outliers = trailing_outliers;
        *leading_pos = last_pos;
      }
      trailing_outliers = 0;
    } else {
      if (pos == last_pos) {
        trailing_outliers++;
      } else {
        trailing_outliers = 1;
      }
    }
    last_pos = pos;
  }
  *num_trailing_outliers = trailing_outliers;
  *trailing_pos = last_pos;
}

Variable Documentation

const int tesseract::case_state_table[6][4]

Initial value:

 { {
                                  
    
                                  
      0, 1, 5, 4
    },
    {                            
      0, 3, 2, 4
    },
    {                            
      0, -1, 2, -1
    },
    {                            
      0, 3, -1, 4
    },
    {                            
      0, -1, -1, 4
    },
    {                            
      5, -1, 2, -1
    },
  }

Definition at line 36 of file context.cpp.

const int tesseract::kAdjacentLeaderSearchPadding = 2

Definition at line 126 of file tablefind.cpp.

const double tesseract::kAlignedFraction = 0.03125

Definition at line 39 of file alignedblob.cpp.

const double tesseract::kAlignedGapFraction = 0.75

Definition at line 43 of file alignedblob.cpp.

const char* tesseract::kAlignmentNames[]

Initial value:

 {
  "Left Aligned",
  "Left Ragged",
  "Center",
  "Right Aligned",
  "Right Ragged",
  "Separator"
}

Definition at line 515 of file tabvector.cpp.

const double tesseract::kAllowBlobArea = 0.05

Definition at line 62 of file tablefind.cpp.

const double tesseract::kAllowBlobHeight = 0.3

Definition at line 60 of file tablefind.cpp.

const double tesseract::kAllowBlobWidth = 0.4

Definition at line 61 of file tablefind.cpp.

const double tesseract::kAllowTextArea = 0.8

Definition at line 55 of file tablefind.cpp.

const double tesseract::kAllowTextHeight = 0.5

Definition at line 53 of file tablefind.cpp.

const double tesseract::kAllowTextWidth = 0.6

Definition at line 54 of file tablefind.cpp.

const char * tesseract::kApostropheLikeUTF8

Initial value:

 {
  "'",       
  "`",       
  "\u2018",  
  "\u2019",  
  "\u2032",  
  NULL,      
}

Definition at line 49 of file unicodes.cpp.

const int tesseract::kBasicBufSize = 2048

Definition at line 155 of file pdfrenderer.cpp.

const double tesseract::kBigPartSizeRatio = 1.75

Definition at line 47 of file colpartitiongrid.cpp.

const int tesseract::kBoxClipTolerance = 2

Definition at line 31 of file boxword.cpp.

const double tesseract::kBrokenCJKIterationFraction = 0.125

Definition at line 71 of file strokewidth.cpp.

const int tesseract::kCellSplitColumnThreshold = 0

Definition at line 40 of file tablerecog.cpp.

const int tesseract::kCellSplitRowThreshold = 0

Definition at line 39 of file tablerecog.cpp.

const double tesseract::kCharVerticalOverlapFraction = 0.375

Definition at line 62 of file tabfind.cpp.

const int tesseract::kCharWidth = 2

Definition at line 158 of file pdfrenderer.cpp.

const double tesseract::kCJKAspectRatio = 1.25

Definition at line 65 of file strokewidth.cpp.

const double tesseract::kCJKAspectRatioIncrease = 1.0625

Definition at line 67 of file strokewidth.cpp.

const double tesseract::kCJKBrokenDistanceFraction = 0.25

Definition at line 61 of file strokewidth.cpp.

const int tesseract::kCJKMaxComponents = 8

Definition at line 63 of file strokewidth.cpp.

const int tesseract::kCJKRadius = 2

Definition at line 59 of file strokewidth.cpp.

const int tesseract::kColumnWidthFactor = 20

Pixel resolution of column width estimates.

Definition at line 42 of file tabfind.h.

const double tesseract::kCosMaxSkewAngle = 0.866025

Definition at line 81 of file tabfind.cpp.

const int tesseract::kCrackSpacing = 100

Spacing of cracks across the page to break up tall vertical lines.

Definition at line 45 of file linefind.cpp.

const ParagraphModel * tesseract::kCrownLeft = reinterpret_cast<ParagraphModel *>(0xDEAD111F)

Definition at line 48 of file paragraphs.cpp.

const ParagraphModel * tesseract::kCrownRight = reinterpret_cast<ParagraphModel *>(0xDEAD888F)

Definition at line 50 of file paragraphs.cpp.

const int tesseract::kDefaultResolution = 300

Default resolution used if input in not believable.

Definition at line 60 of file pagesegmain.cpp.

const double tesseract::kDiacriticXPadRatio = 7.0

Definition at line 74 of file strokewidth.cpp.

const double tesseract::kDiacriticYPadRatio = 1.75

Definition at line 77 of file strokewidth.cpp.

const char tesseract::kDoNotReverse[] = "RRP_DO_NO_REVERSE"

Definition at line 44 of file trie.cpp.

const int tesseract::kExposureFactor = 16

Definition at line 33 of file degradeimage.cpp.

const int tesseract::kFeaturePadding = 2

Definition at line 34 of file imagedata.h.

const float tesseract::kFontMergeDistance = 0.025

Definition at line 53 of file mastertrainer.cpp.

const char tesseract::kForceReverse[] = "RRP_FORCE_REVERSE"

Definition at line 46 of file trie.cpp.

const double tesseract::kGoodRowNumberOfColumnsLarge = 0.7

Definition at line 58 of file tablerecog.cpp.

const double tesseract::kGoodRowNumberOfColumnsSmall[] = { 2, 2, 2, 2, 2, 3, 3 }

Definition at line 54 of file tablerecog.cpp.

const int tesseract::kGoodRowNumberOfColumnsSmallSize

Initial value:

 
    sizeof(kGoodRowNumberOfColumnsSmall) / sizeof(double) - 1

Definition at line 55 of file tablerecog.cpp.

const int tesseract::kGutterMultiple = 4

Definition at line 38 of file tabvector.cpp.

const int tesseract::kGutterToNeighbourRatio = 3

Definition at line 40 of file tabvector.cpp.

const int tesseract::kHistogramSize = 256

Definition at line 27 of file otsuthr.h.

const double tesseract::kHorizontalGapMergeFraction = 0.5

Definition at line 61 of file colfind.cpp.

const double tesseract::kHorizontalSpacing = 0.30

Definition at line 33 of file tablerecog.cpp.

const int tesseract::kHorzStrongTextlineAspect = 5

Definition at line 69 of file colpartition.cpp.

const int tesseract::kHorzStrongTextlineCount = 8

Definition at line 65 of file colpartition.cpp.

const int tesseract::kHorzStrongTextlineHeight = 10

Definition at line 67 of file colpartition.cpp.

const char * tesseract::kHyphenLikeUTF8

Initial value:

 {
  "-",       
  "\u05BE",  
  "\u2010",  
  "\u2011",  
  "\u2012",  
  "\u2013",  
  "\u2014",  
  "\u2015",  
  "\u2212",  
  "\uFE58",  
  "\uFE63",  
  "\uFF0D",  
  NULL,      
}

The following are confusable internal word punctuation symbols which we normalize to the first variant when matching in dawgs.

Definition at line 33 of file unicodes.cpp.

const int tesseract::kImagePadding = 4

Definition at line 36 of file imagedata.h.

const float tesseract::kInfiniteDist = 999.0f

Definition at line 912 of file mastertrainer.cpp.

const double tesseract::kLargeTableProjectionThreshold = 0.45

Definition at line 111 of file tablefind.cpp.

const int tesseract::kLargeTableRowCount = 6

Definition at line 113 of file tablefind.cpp.

const int tesseract::kLeftIndentAlignmentCountTh = 1

Definition at line 88 of file equationdetect.cpp.

const double tesseract::kLineCountReciprocal = 4.0

Definition at line 51 of file tabvector.cpp.

const int tesseract::kLinedTableMinHorizontalLines = 3

Definition at line 43 of file tablerecog.cpp.

const int tesseract::kLinedTableMinVerticalLines = 3

Definition at line 42 of file tablerecog.cpp.

const int tesseract::kLineFindGridSize = 50

Grid size used by line finder. Not very critical.

Definition at line 47 of file linefind.cpp.

const double tesseract::kLineFragmentAspectRatio = 10.0

Definition at line 56 of file tabfind.cpp.

const double tesseract::kLineResidueAspectRatio = 8.0

Definition at line 100 of file strokewidth.cpp.

const int tesseract::kLineResiduePadRatio = 3

Definition at line 102 of file strokewidth.cpp.

const double tesseract::kLineResidueSizeRatio = 1.75

Definition at line 104 of file strokewidth.cpp.

const int tesseract::kLineTrapLongest = 4

Definition at line 93 of file strokewidth.cpp.

const int tesseract::kLineTrapShortest = 2

Definition at line 95 of file strokewidth.cpp.

const char * tesseract::kLRM = "\u200E"

Definition at line 28 of file unicodes.cpp.

const double tesseract::kMarginFactor = 1.1

Definition at line 48 of file tablerecog.cpp.

const double tesseract::kMarginOverlapFraction = 0.25

Definition at line 58 of file colfind.cpp.

const float tesseract::kMathDigitDensityTh1 = 0.25

Definition at line 83 of file equationdetect.cpp.

const float tesseract::kMathDigitDensityTh2 = 0.1

Definition at line 84 of file equationdetect.cpp.

const float tesseract::kMathItalicDensityTh = 0.5

Definition at line 85 of file equationdetect.cpp.

usr src packages BUILD tesseract ccutil ambigs cpp usr src packages BUILD tesseract ccutil ambigs cpp const int tesseract::kMaxAmbigStringSize = UNICHAR_LEN * (MAX_AMBIG_SIZE + 1)

Definition at line 42 of file ambigs.cpp.

const double tesseract::kMaxBaselineError = 0.4375

Definition at line 72 of file colpartition.cpp.

const double tesseract::kMaxBlobOverlapFactor = 4.0

Definition at line 81 of file tablefind.cpp.

const int tesseract::kMaxBlobWidth = 500

Definition at line 44 of file tablefind.cpp.

const inT16 tesseract::kMaxBoxEdgeDiff = 2

Definition at line 32 of file recogtraining.cpp.

const int tesseract::kMaxBoxesInDataPartition = 20

Definition at line 70 of file tablefind.cpp.

const int tesseract::kMaxCaptionLines = 7

Definition at line 39 of file colpartitiongrid.cpp.

const int tesseract::kMaxCharTopRange = 48

Definition at line 66 of file fixxht.cpp.

const int tesseract::kMaxCircleErosions = 8

Definition at line 62 of file pagesegmain.cpp.

const int tesseract::kMaxCJKSizeRatio = 5

Definition at line 69 of file strokewidth.cpp.

const int tesseract::kMaxColorDistance = 900

Definition at line 79 of file colpartition.cpp.

const int tesseract::kMaxColumnHeaderDistance = 4

Definition at line 89 of file tablefind.cpp.

const double tesseract::kMaxDiacriticDistanceRatio = 1.25

Definition at line 83 of file strokewidth.cpp.

const double tesseract::kMaxDiacriticGapToBaseCharHeight = 1.0

Definition at line 86 of file strokewidth.cpp.

const double tesseract::kMaxDistToPartSizeRatio = 1.5

Definition at line 68 of file colfind.cpp.

const int tesseract::kMaxFillinMultiple = 11

Definition at line 47 of file tabvector.cpp.

const double tesseract::kMaxGapInTextPartition = 4.0

Definition at line 73 of file tablefind.cpp.

const double tesseract::kMaxGutterWidthAbsolute = 2.00

Definition at line 51 of file tabfind.cpp.

const double tesseract::kMaxHorizontalGap = 3.0

Definition at line 64 of file tabfind.cpp.

const int tesseract::kMaxIncompatibleColumnCount = 2

Definition at line 56 of file colfind.cpp.

const int tesseract::kMaxLargeOverlaps = 3

Definition at line 109 of file strokewidth.cpp.

const int tesseract::kMaxLargeOverlapsWithMedium = 12

Definition at line 47 of file ccnontextdetect.cpp.

const int tesseract::kMaxLargeOverlapsWithSmall = 3

Definition at line 38 of file ccnontextdetect.cpp.

const double tesseract::kMaxLeaderGapFractionOfMax = 0.25

Definition at line 55 of file colpartition.cpp.

const double tesseract::kMaxLeaderGapFractionOfMin = 0.5

Definition at line 57 of file colpartition.cpp.

const int tesseract::kMaxLigature = 0xfb17

Definition at line 49 of file ligature_table.cpp.

const int tesseract::kMaxLineLength = 1024

Definition at line 290 of file boxchar.cpp.

const int tesseract::kMaxLineResidue = 6

Definition at line 53 of file linefind.cpp.

const int tesseract::kMaxMediumOverlapsWithSmall = 12

Definition at line 43 of file ccnontextdetect.cpp.

const int tesseract::kMaxNeighbourDistFactor = 4

Definition at line 37 of file colpartitiongrid.cpp.

const double tesseract::kMaxNonLineDensity = 0.25

Definition at line 58 of file linefind.cpp.

const int tesseract::kMaxOffsetDist = 32

Definition at line 32 of file intfeaturemap.cpp.

const int tesseract::kMaxPadFactor = 6

Definition at line 34 of file colpartitiongrid.cpp.

const double tesseract::kMaxParagraphEndingLeftSpaceMultiple = 3.0

Definition at line 135 of file tablefind.cpp.

const double tesseract::kMaxPartitionSpacing = 1.75

Definition at line 62 of file colpartitiongrid.cpp.

const int tesseract::kMaxRaggedSearch = 25

Definition at line 39 of file tabfind.cpp.

const int tesseract::kMaxRealDistance = 2.0

Definition at line 38 of file detlinefit.cpp.

const double tesseract::kMaxRectangularFraction = 0.75

Definition at line 46 of file imagefind.cpp.

const double tesseract::kMaxRectangularGradient = 0.1

Definition at line 49 of file imagefind.cpp.

const int tesseract::kMaxRMSColorNoise = 128

Definition at line 76 of file colpartition.cpp.

const double tesseract::kMaxRowSize = 2.5

Definition at line 51 of file tablerecog.cpp.

const double tesseract::kMaxSameBlockLineSpacing = 3

Definition at line 51 of file colpartition.cpp.

const double tesseract::kMaxSizeRatio = 1.5

Definition at line 53 of file colpartition.cpp.

const int tesseract::kMaxSkewFactor = 15

Definition at line 65 of file alignedblob.cpp.

usr src packages BUILD tesseract textord ccnontextdetect cpp usr src packages BUILD tesseract textord ccnontextdetect cpp usr src packages BUILD tesseract textord ccnontextdetect cpp const double tesseract::kMaxSmallNeighboursPerPix = 1.0 / 32

Definition at line 35 of file ccnontextdetect.cpp.

const double tesseract::kMaxSpacingDrift = 1.0 / 72

Definition at line 45 of file colpartition.cpp.

const double tesseract::kMaxStaveHeight = 1.0

Definition at line 60 of file linefind.cpp.

const double tesseract::kMaxTableCellXheight = 2.0

Definition at line 85 of file tablefind.cpp.

const int tesseract::kMaxTextLineBlobRatio = 5

Definition at line 72 of file tabfind.cpp.

const double tesseract::kMaxTopSpacingFraction = 0.25

Definition at line 48 of file colpartition.cpp.

const int tesseract::kMaxUnicharsPerCluster = 2000

Definition at line 51 of file mastertrainer.cpp.

const int tesseract::kMaxVerticalSearch = 12

Definition at line 38 of file tabfind.cpp.

usr src packages BUILD tesseract textord tablefind cpp const int tesseract::kMaxVerticalSpacing = 500

Definition at line 42 of file tablefind.cpp.

const double tesseract::kMaxXProjectionGapFactor = 2.0

Definition at line 145 of file tablefind.cpp.

const double tesseract::kMinAlignedGutter = 0.25

Definition at line 53 of file tabvector.cpp.

const int tesseract::kMinAlignedTabs = 4

Definition at line 55 of file alignedblob.cpp.

const double tesseract::kMinBaselineCoverage = 0.5

Definition at line 74 of file colpartition.cpp.

const int tesseract::kMinBoxesInTextPartition = 10

Definition at line 67 of file tablefind.cpp.

const double tesseract::kMinCaptionGapHeightRatio = 0.5

Definition at line 43 of file colpartitiongrid.cpp.

const double tesseract::kMinCaptionGapRatio = 2.0

Definition at line 41 of file colpartitiongrid.cpp.

const int tesseract::kMinChainTextValue = 3

Definition at line 63 of file colpartition.cpp.

usr src packages BUILD tesseract classify mastertrainer cpp const int tesseract::kMinClusteredShapes = 1

Definition at line 49 of file mastertrainer.cpp.

const int tesseract::kMinColorDifference = 16

Definition at line 55 of file imagefind.cpp.

const int tesseract::kMinColumnWidth = 100

Definition at line 53 of file colfind.cpp.

const int tesseract::kMinCredibleResolution = 70

Minimum believable resolution.

Definition at line 58 of file pagesegmain.cpp.

const double tesseract::kMinDiacriticSizeRatio = 1.0625

Definition at line 80 of file strokewidth.cpp.

const int tesseract::kMinEvaluatedTabs = 3

Definition at line 69 of file tabfind.cpp.

const double tesseract::kMinFilledArea = 0.35

Definition at line 61 of file tablerecog.cpp.

const double tesseract::kMinFractionalLinesInColumn = 0.125

Definition at line 45 of file tabfind.cpp.

const double tesseract::kMinGoodTextPARatio = 1.5

Definition at line 63 of file ccnontextdetect.cpp.

const double tesseract::kMinGutterFraction = 0.5

Definition at line 49 of file tabvector.cpp.

const double tesseract::kMinGutterWidthAbsolute = 0.02

Definition at line 49 of file tabfind.cpp.

const double tesseract::kMinGutterWidthGrid = 0.5

Definition at line 65 of file colfind.cpp.

const double tesseract::kMinImageArea = 0.5

Definition at line 77 of file tabfind.cpp.

const int tesseract::kMinImageFindSize = 100

Definition at line 51 of file imagefind.cpp.

const int tesseract::kMinLeaderCount = 5

Definition at line 59 of file colpartition.cpp.

const int tesseract::kMinLigature = 0xfb00

Definition at line 48 of file ligature_table.cpp.

const int tesseract::kMinLineLengthFraction = 4

Denominator of resolution makes min pixels to demand line lengths to be.

Definition at line 43 of file linefind.cpp.

const int tesseract::kMinLinesInColumn = 10

Definition at line 41 of file tabfind.cpp.

const double tesseract::kMinMaxGapInTextPartition = 0.5

Definition at line 77 of file tablefind.cpp.

const double tesseract::kMinMusicPixelFraction = 0.75

Definition at line 62 of file linefind.cpp.

const double tesseract::kMinNonNoiseFraction = 0.5

Definition at line 63 of file colfind.cpp.

const int tesseract::kMinOutlierSamples = 5

Definition at line 37 of file trainingsampleset.cpp.

const double tesseract::kMinOverlapWithTable = 0.6

Definition at line 101 of file tablefind.cpp.

const double tesseract::kMinParagraphEndingTextToWhitespaceRatio = 3.0

Definition at line 141 of file tablefind.cpp.

const double tesseract::kMinPCLengthIncrease = 1.0 / 1024

Definition at line 33 of file intfeaturemap.cpp.

const int tesseract::kMinPointsForErrorCount = 16

Definition at line 35 of file detlinefit.cpp.

const double tesseract::kMinRaggedGutter = 1.5

Definition at line 55 of file tabvector.cpp.

const int tesseract::kMinRaggedTabs = 5

Definition at line 53 of file alignedblob.cpp.

const int tesseract::kMinRampSize = 1000

Definition at line 37 of file degradeimage.cpp.

const double tesseract::kMinRectangularFraction = 0.125

Definition at line 44 of file imagefind.cpp.

const int tesseract::kMinRowsInTable = 3

Definition at line 116 of file tablefind.cpp.

const int tesseract::kMinStrongTextValue = 6

Definition at line 61 of file colpartition.cpp.

const double tesseract::kMinTabGradient = 4.0

Definition at line 61 of file alignedblob.cpp.

const int tesseract::kMinTextLineBlobRatio = 3

Definition at line 75 of file tabfind.cpp.

const int tesseract::kMinThickLineWidth = 12

Definition at line 49 of file linefind.cpp.

const int tesseract::kMinVerticalSearch = 3

Definition at line 37 of file tabfind.cpp.

const int tesseract::kMostlyOneDirRatio = 3

Definition at line 98 of file strokewidth.cpp.

const double tesseract::kNeighbourSearchFactor = 2.5

Definition at line 111 of file strokewidth.cpp.

const double tesseract::kNoiseOverlapAreaFactor = 1.0 / 512

Definition at line 116 of file strokewidth.cpp.

const double tesseract::kNoiseOverlapGrowthFactor = 4.0

Definition at line 113 of file strokewidth.cpp.

const int tesseract::kNoisePadding = 4

Definition at line 54 of file ccnontextdetect.cpp.

usr src packages BUILD tesseract ccstruct detlinefit cpp const int tesseract::kNumEndPoints = 3

Definition at line 29 of file detlinefit.cpp.

const int tesseract::kNumLiteralCnt = 5

Definition at line 36 of file tess_lang_model.h.

const int tesseract::kNumPagesPerMiniBatch = 100

Definition at line 38 of file imagedata.h.

const int tesseract::kOriginalNoiseMultiple = 8

Definition at line 50 of file ccnontextdetect.cpp.

const double tesseract::kParagraphEndingPreviousLineRatio = 1.3

Definition at line 131 of file tablefind.cpp.

const char * tesseract::kPDF = "\u202C"

Definition at line 31 of file unicodes.cpp.

const double tesseract::kPhotoOffsetFraction = 0.375

Definition at line 57 of file ccnontextdetect.cpp.

const int tesseract::kPrime1 = 17

Definition at line 34 of file trainingsampleset.cpp.

const int tesseract::kPrime2 = 13

Definition at line 35 of file trainingsampleset.cpp.

const double tesseract::kRaggedFraction = 2.5

Definition at line 41 of file alignedblob.cpp.

const double tesseract::kRaggedGapFraction = 1.0

Definition at line 45 of file alignedblob.cpp.

const int tesseract::kRaggedGutterMultiple = 5

Definition at line 53 of file tabfind.cpp.

const int tesseract::kRandomizingCenter = 128

Definition at line 35 of file trainingsample.cpp.

usr src packages BUILD tesseract classify errorcounter cpp usr src packages BUILD tesseract classify errorcounter cpp usr src packages BUILD tesseract classify errorcounter cpp const double tesseract::kRatingEpsilon = 1.0 / 32

Definition at line 34 of file errorcounter.cpp.

const double tesseract::kRequiredColumns = 0.7

Definition at line 46 of file tablerecog.cpp.

const double tesseract::kRequiredFullJustifiedSpacing = 4.0

Definition at line 121 of file tablefind.cpp.

const char tesseract::kReverseIfHasRTL[] = "RRP_REVERSE_IF_HAS_RTL"

Definition at line 45 of file trie.cpp.

const int tesseract::kRGBRMSColors = 4

Definition at line 36 of file colpartition.h.

const char * tesseract::kRLE = "\u202A"

Definition at line 30 of file unicodes.cpp.

const char * tesseract::kRLM = "\u200F"

Definition at line 29 of file unicodes.cpp.

const double tesseract::kRMSFitScaling = 8.0

Definition at line 53 of file imagefind.cpp.

usr src packages BUILD tesseract training degradeimage cpp const float tesseract::kRotationRange = 0.02f

Definition at line 31 of file degradeimage.cpp.

const int tesseract::kRulingVerticalMargin = 3

Definition at line 97 of file tablefind.cpp.

const int tesseract::kSaltnPepper = 5

Definition at line 35 of file degradeimage.cpp.

const int tesseract::kSearchRadius = 2

Definition at line 88 of file strokewidth.cpp.

const int tesseract::kSeedBlobsCountTh = 10

Definition at line 87 of file equationdetect.cpp.

const double tesseract::kShapePerimeterRatio = 3.0

Definition at line 118 of file strokewidth.cpp.

const int tesseract::kSideSpaceMargin = 10

Definition at line 106 of file tablefind.cpp.

const int tesseract::kSimilarRaggedDist = 50

Definition at line 45 of file tabvector.cpp.

const int tesseract::kSimilarVectorDist = 10

Definition at line 42 of file tabvector.cpp.

const int tesseract::ksizeofUniversalAmbigsFile = sizeof(kUniversalAmbigsFile)

Definition at line 24 of file universalambigs.h.

const float tesseract::kSizeRatioToReject = 2.0

Definition at line 106 of file strokewidth.cpp.

const double tesseract::kSmallTableProjectionThreshold = 0.35

Definition at line 110 of file tablefind.cpp.

const int tesseract::kSmoothDecisionMargin = 4

Definition at line 65 of file colpartitiongrid.cpp.

const double tesseract::kSmoothFactor = 0.25

Definition at line 58 of file tabfind.cpp.

const double tesseract::kSplitPartitionSize = 2.0

Definition at line 48 of file tablefind.cpp.

const int tesseract::kSquareLimit = 25

Definition at line 32 of file trainingsampleset.cpp.

const int tesseract::kStateCnt = 4

Definition at line 35 of file tess_lang_model.h.

const double tesseract::kStrokeWidthCJK = 2.0

Definition at line 56 of file strokewidth.cpp.

const double tesseract::kStrokeWidthConstantTolerance = 2.0

Definition at line 150 of file tablefind.cpp.

const double tesseract::kStrokeWidthFractionalTolerance = 0.25

Definition at line 149 of file tablefind.cpp.

const double tesseract::kStrokeWidthFractionCJK = 0.25

Definition at line 55 of file strokewidth.cpp.

const double tesseract::kStrokeWidthFractionTolerance = 0.125

Allowed proportional change in stroke width to be the same font.

Definition at line 48 of file strokewidth.cpp.

const double tesseract::kStrokeWidthTolerance = 1.5

Allowed constant change in stroke width to be the same font. Really 1.5 pixels.

Definition at line 53 of file strokewidth.cpp.

const double tesseract::kTableColumnThreshold = 3.0

Definition at line 93 of file tablefind.cpp.

const int tesseract::kTabRadiusFactor = 5

Definition at line 35 of file tabfind.cpp.

const int tesseract::kTestChar = -1

Definition at line 30 of file trainingsampleset.cpp.

const char* tesseract::kTextordDebugPix = "psdebug_pix"

Definition at line 68 of file alignedblob.cpp.

const double tesseract::kThickLengthMultiple = 0.75

Definition at line 56 of file linefind.cpp.

const int tesseract::kThinLineFraction = 20

Denominator of resolution makes max pixel width to allow thin lines.

Definition at line 41 of file linefind.cpp.

const double tesseract::kTinyEnoughTextlineOverlapFraction = 0.25

Definition at line 49 of file colpartitiongrid.cpp.

const float tesseract::kUnclearDensityTh = 0.25

Definition at line 86 of file equationdetect.cpp.

const char tesseract::kUniversalAmbigsFile

Definition at line 23 of file universalambigs.h.

const char * tesseract::kUTF8LineSeparator = "\u2028"

Definition at line 26 of file unicodes.cpp.

const char * tesseract::kUTF8ParagraphSeparator = "\u2029"

Definition at line 27 of file unicodes.cpp.

const double tesseract::kVerticalSpacing = -0.2

Definition at line 36 of file tablerecog.cpp.

const int tesseract::kVLineAlignment = 3

Definition at line 47 of file alignedblob.cpp.

const int tesseract::kVLineGutter = 1

Definition at line 49 of file alignedblob.cpp.

const int tesseract::kVLineMinLength = 500

Definition at line 57 of file alignedblob.cpp.

const int tesseract::kVLineSearchSize = 150

Definition at line 51 of file alignedblob.cpp.

const char* const tesseract::RTLReversePolicyNames[]

Initial value:

 {
  kDoNotReverse,
  kReverseIfHasRTL,
  kForceReverse
}

Definition at line 48 of file trie.cpp.

bool tesseract::textord_dump_table_images = false

"Paint table detection output"

Definition at line 152 of file tablefind.cpp.

bool tesseract::textord_show_tables = false

"Show table regions"

Definition at line 153 of file tablefind.cpp.

bool tesseract::textord_tabfind_find_tables = true

"run table detection"

Definition at line 78 of file colfind.cpp.

bool tesseract::textord_tabfind_only_strokewidths = false

"Only run stroke widths"

Definition at line 45 of file strokewidth.cpp.

bool tesseract::textord_tabfind_show_blocks = false

"Show final block bounds"

Definition at line 77 of file colfind.cpp.

bool tesseract::textord_tabfind_show_color_fit = false

"Show stroke widths"

Definition at line 30 of file colpartitiongrid.cpp.

bool tesseract::textord_tabfind_show_columns = false

"Show column bounds"

Definition at line 76 of file colfind.cpp.

bool tesseract::textord_tabfind_show_finaltabs = false

"Show tab vectors"

Definition at line 84 of file tabfind.cpp.

bool tesseract::textord_tabfind_show_initial_partitions = false

"Show partition bounds"

Definition at line 71 of file colfind.cpp.

bool tesseract::textord_tabfind_show_initialtabs = false

"Show tab candidates"

Definition at line 83 of file tabfind.cpp.

int tesseract::textord_tabfind_show_partitions = 0

"Show partition bounds, waiting if >1"

Definition at line 75 of file colfind.cpp.

bool tesseract::textord_tabfind_show_reject_blobs = false

"Show blobs rejected as noise"

Definition at line 73 of file colfind.cpp.

int tesseract::textord_tabfind_show_strokewidths = 0

"Show stroke widths"

Definition at line 44 of file strokewidth.cpp.

bool tesseract::textord_tablefind_recognize_tables = false

"Enables the table recognizer for table layout and filtering."

Definition at line 159 of file tablefind.cpp.

bool tesseract::textord_tablefind_show_mark = false

"Debug table marking steps in detail"

Definition at line 155 of file tablefind.cpp.

bool tesseract::textord_tablefind_show_stats = false

"Show page stats used in table finding"

Definition at line 157 of file tablefind.cpp.

double tesseract::textord_tabvector_vertical_box_ratio = 0.5

"Fraction of box matches required to declare a line vertical"

Definition at line 61 of file tabvector.cpp.

double tesseract::textord_tabvector_vertical_gap_fraction = 0.5

"max fraction of mean blob width allowed for vertical gaps in vertical text"

"Max fraction of mean blob width allowed for vertical gaps in vertical text"

Definition at line 58 of file tabvector.cpp.

CCUtilMutex tesseract::tprintfMutex

Definition at line 52 of file ccutil.cpp.

tesseract Namespace Reference

Classes

Typedefs

Enumerations

Functions

Variables

Detailed Description

Typedef Documentation

Enumeration Type Documentation

Function Documentation

Variable Documentation