tesseract  v4.0.0-17-g361f3264
Open Source OCR Engine
validate_khmer.h
1 #ifndef TESSERACT_TRAINING_VALIDATE_KHMER_H_
2 #define TESSERACT_TRAINING_VALIDATE_KHMER_H_
3 
4 #include "validator.h"
5 
6 namespace tesseract {
7 
8 // Subclass of Validator that validates and segments Khmer.
9 class ValidateKhmer : public Validator {
10  public:
11  ValidateKhmer(ViramaScript script, bool report_errors)
12  : Validator(script, report_errors) {}
14 
15  protected:
16  // Returns whether codes matches the pattern for an Khmer Grapheme.
17  // Consumes the next Grapheme in codes_[codes_used_++...] and copies it to
18  // parts_ and output_. Returns true if a valid Grapheme was consumed,
19  // otherwise does not increment codes_used_.
20  bool ConsumeGraphemeIfValid() override;
21  // Returns the CharClass corresponding to the given Unicode ch.
22  CharClass UnicodeToCharClass(char32 ch) const override;
23 };
24 
25 } // namespace tesseract
26 
27 #endif // TESSERACT_TRAINING_VALIDATE_KHMER_H_
signed int char32
Definition: unichar.h:52
CharClass UnicodeToCharClass(char32 ch) const override
Definition: validate_khmer.cpp:88
Definition: validate_khmer.h:9
Definition: baseapi.cpp:94
ViramaScript
Definition: validator.h:53
CharClass
Definition: validator.h:112
~ValidateKhmer()
Definition: validate_khmer.h:13
ValidateKhmer(ViramaScript script, bool report_errors)
Definition: validate_khmer.h:11
bool ConsumeGraphemeIfValid() override
Definition: validate_khmer.cpp:20
Definition: validator.h:72