|
tesseract 3.04.01
|
Public Member Functions | |
| UnicodeSpanSkipper (const UNICHARSET *unicharset, const WERD_CHOICE *word) | |
| int | SkipPunc (int pos) |
| int | SkipDigits (int pos) |
| int | SkipRomans (int pos) |
| int | SkipAlpha (int pos) |
Definition at line 285 of file paragraphs.cpp.
| tesseract::UnicodeSpanSkipper::UnicodeSpanSkipper | ( | const UNICHARSET * | unicharset, |
| const WERD_CHOICE * | word | ||
| ) | [inline] |
Definition at line 287 of file paragraphs.cpp.
: const UNICHARSET *u_;
| int tesseract::UnicodeSpanSkipper::SkipAlpha | ( | int | pos | ) |
Definition at line 326 of file paragraphs.cpp.
{
if (ch < 0x80) {
| int tesseract::UnicodeSpanSkipper::SkipDigits | ( | int | pos | ) |
Definition at line 310 of file paragraphs.cpp.
{
const char *kRomans = "ivxlmdIVXLMD";
| int tesseract::UnicodeSpanSkipper::SkipPunc | ( | int | pos | ) |
Definition at line 305 of file paragraphs.cpp.
{
while (pos < wordlen_ && (u_->get_isdigit(word_->unichar_id(pos)) ||
| int tesseract::UnicodeSpanSkipper::SkipRomans | ( | int | pos | ) |
Definition at line 316 of file paragraphs.cpp.
{
while (pos < wordlen_ && u_->get_isalpha(word_->unichar_id(pos))) pos++;