tesseract::UnicodeSpanSkipper Class Reference

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)

Detailed Description

Definition at line 283 of file paragraphs.cpp.

Constructor & Destructor Documentation

UnicodeSpanSkipper()

tesseract::UnicodeSpanSkipper::UnicodeSpanSkipper ( const UNICHARSET *  unicharset, const WERD_CHOICE *  word  )

inline

Definition at line 285 of file paragraphs.cpp.

      : u_(unicharset), word_(word) { wordlen_ = word->length(); }

Member Function Documentation

SkipAlpha()

int tesseract::UnicodeSpanSkipper::SkipAlpha

(

int

pos

)

Definition at line 324 of file paragraphs.cpp.

                                         {
  while (pos < wordlen_ && u_->get_isalpha(word_->unichar_id(pos))) pos++;
  return pos;
}

SkipDigits()

int tesseract::UnicodeSpanSkipper::SkipDigits

(

int

pos

)

Definition at line 308 of file paragraphs.cpp.

                                          {
  while (pos < wordlen_ && (u_->get_isdigit(word_->unichar_id(pos)) ||
                            IsDigitLike(UnicodeFor(u_, word_, pos)))) pos++;
  return pos;
}

SkipPunc()

int tesseract::UnicodeSpanSkipper::SkipPunc

(

int

pos

)

Definition at line 303 of file paragraphs.cpp.

                                        {
  while (pos < wordlen_ && u_->get_ispunctuation(word_->unichar_id(pos))) pos++;
  return pos;
}

SkipRomans()

int tesseract::UnicodeSpanSkipper::SkipRomans

(

int

pos

)

Definition at line 314 of file paragraphs.cpp.

                                          {
  const char *kRomans = "ivxlmdIVXLMD";
  while (pos < wordlen_) {
    int ch = UnicodeFor(u_, word_, pos);
    if (ch >= 0xF0 || strchr(kRomans, ch) == 0) break;
    pos++;
  }
  return pos;
}

---

The documentation for this class was generated from the following file:

• /home/stefan/src/github/tesseract-ocr/tesseract/ccmain/paragraphs.cpp