tesseract v5.3.3.20231005
validate_khmer.h
Go to the documentation of this file.
1#ifndef TESSERACT_TRAINING_VALIDATE_KHMER_H_
2#define TESSERACT_TRAINING_VALIDATE_KHMER_H_
3
4#include "validator.h"
5
6namespace tesseract {
7
8// Subclass of Validator that validates and segments Khmer.
9class ValidateKhmer : public Validator {
10public:
11 ValidateKhmer(ViramaScript script, bool report_errors) : Validator(script, report_errors) {}
12 ~ValidateKhmer() override = default;
13
14protected:
15 // Returns whether codes matches the pattern for an Khmer Grapheme.
16 // Consumes the next Grapheme in codes_[codes_used_++...] and copies it to
17 // parts_ and output_. Returns true if a valid Grapheme was consumed,
18 // otherwise does not increment codes_used_.
19 bool ConsumeGraphemeIfValid() override;
20 // Returns the CharClass corresponding to the given Unicode ch.
21 CharClass UnicodeToCharClass(char32 ch) const override;
22};
23
24} // namespace tesseract
25
26#endif // TESSERACT_TRAINING_VALIDATE_KHMER_H_
signed int char32
Definition: unichar.h:49
bool ConsumeGraphemeIfValid() override
~ValidateKhmer() override=default
CharClass UnicodeToCharClass(char32 ch) const override
ValidateKhmer(ViramaScript script, bool report_errors)