#include <validate_myanmar.h>

Inheritance diagram for tesseract::ValidateMyanmar:

Public Member Functions
	ValidateMyanmar (ViramaScript script, bool report_errors)

	~ValidateMyanmar ()

Public Member Functions inherited from tesseract::Validator
virtual	~Validator ()

Protected Member Functions
bool	ConsumeGraphemeIfValid () override

Validator::CharClass	UnicodeToCharClass (char32 ch) const override

Protected Member Functions inherited from tesseract::Validator
	Validator (ViramaScript script, bool report_errors)

bool	ValidateCleanAndSegmentInternal (GraphemeNormMode g_mode, const std::vector< char32 > &src, std::vector< std::vector< char32 >> *dest)

void	MoveResultsToDest (GraphemeNormMode g_mode, std::vector< std::vector< char32 >> *dest)

bool	IsSubscriptScript () const

bool	CodeOnlyToOutput ()

void	MultiCodePart (int length)

bool	UseMultiCode (int length)

void	ComputeClassCodes (const std::vector< char32 > &text)

void	Clear ()

Additional Inherited Members
Static Public Member Functions inherited from tesseract::Validator
static bool	ValidateCleanAndSegment (GraphemeNormMode g_mode, bool report_errors, const std::vector< char32 > &src, std::vector< std::vector< char32 >> *dest)

static bool	IsZeroWidthMark (char32 ch)

Static Public Attributes inherited from tesseract::Validator
static const char32	kZeroWidthSpace = 0x200B

static const char32	kZeroWidthNonJoiner = 0x200C

static const char32	kZeroWidthJoiner = 0x200D

static const char32	kLeftToRightMark = 0x200E

static const char32	kRightToLeftMark = 0x200F

static const char32	kInvalid = 0xfffd

Protected Types inherited from tesseract::Validator
enum	CharClass { CharClass::kConsonant = 'C', CharClass::kVowel = 'V', CharClass::kVirama = 'H', CharClass::kMatra = 'M', CharClass::kMatraPiece = 'P', CharClass::kVowelModifier = 'D', CharClass::kZeroWidthNonJoiner = 'z', CharClass::kZeroWidthJoiner = 'Z', CharClass::kVedicMark = 'v', CharClass::kNukta = 'N', CharClass::kRobat = 'R', CharClass::kOther = 'O', CharClass::kWhitespace = ' ', CharClass::kCombiner = 'c' }

using	IndicPair = std::pair< CharClass, char32 >

Static Protected Member Functions inherited from tesseract::Validator
static std::unique_ptr< Validator >	ScriptValidator (ViramaScript script, bool report_errors)

static ViramaScript	MostFrequentViramaScript (const std::vector< char32 > &utf32)

static bool	IsVirama (char32 unicode)

static bool	IsVedicAccent (char32 unicode)

Protected Attributes inherited from tesseract::Validator
ViramaScript	script_

std::vector< IndicPair >	codes_

std::vector< std::vector< char32 > >	parts_

std::vector< char32 >	output_

int	codes_used_

int	output_used_

bool	report_errors_

Static Protected Attributes inherited from tesseract::Validator
static const int	kIndicCodePageSize = 128

static const char32	kMinIndicUnicode = 0x900

static const char32	kMaxSinhalaUnicode = 0xdff

static const char32	kMaxViramaScriptUnicode = 0x17ff

static const char32	kSinhalaVirama = 0xdca

static const char32	kMyanmarVirama = 0x1039

static const char32	kKhmerVirama = 0x17d2

static const char32	kJavaneseVirama = 0xa9c0

static const char32	kMaxJavaneseUnicode = 0xa9df

Detailed Description

Definition at line 9 of file validate_myanmar.h.

Constructor & Destructor Documentation

◆ ValidateMyanmar()

tesseract::ValidateMyanmar::ValidateMyanmar	(	ViramaScript	script,
		bool	report_errors
	)

inline

Definition at line 11 of file validate_myanmar.h.

12 : Validator(script, report_errors) {}

tesseract::Validator::Validator

Validator(ViramaScript script, bool report_errors)

Definition: validator.h:136

◆ ~ValidateMyanmar()

tesseract::ValidateMyanmar::~ValidateMyanmar ( )

inline

Definition at line 13 of file validate_myanmar.h.

13 {}

Member Function Documentation

◆ ConsumeGraphemeIfValid()

bool tesseract::ValidateMyanmar::ConsumeGraphemeIfValid ( )

overrideprotectedvirtual

Implements tesseract::Validator.

Definition at line 13 of file validate_myanmar.cpp.

                                              {
   int num_codes = codes_.size();
   if (codes_used_ == num_codes) return true;
   // Other.
   if (IsMyanmarOther(codes_[codes_used_].second)) {
     UseMultiCode(1);
     return true;
   }
   // Kinzi.
   if (codes_used_ + 2 < num_codes && codes_[codes_used_].second == 0x1004 &&
       codes_[codes_used_ + 1].second == kMyanmarAsat &&
       codes_[codes_used_ + 2].second == kMyanmarVirama) {
     ASSERT_HOST(!CodeOnlyToOutput());
     ASSERT_HOST(!CodeOnlyToOutput());
     if (UseMultiCode(3)) return true;
   }
   // Base consonant/vowel. NOTE that since everything in Myanmar appears to be
   // optional, except the base, this is the only place where invalid input can
   // be detected and false returned.
   if (IsMyanmarLetter(codes_[codes_used_].second)) {
     if (UseMultiCode(1)) return true;
   } else {
     if (report_errors_) {
       tprintf("Invalid start of Myanmar syllable:0x%x\n",
               codes_[codes_used_].second);
     }
     return false;  // One of these is required.
   }
   if (ConsumeSubscriptIfPresent()) return true;
   ConsumeOptionalSignsIfPresent();
   // What we have consumed so far is a valid syllable.
   return true;
 }

◆ UnicodeToCharClass()

Validator::CharClass tesseract::ValidateMyanmar::UnicodeToCharClass ( char32 ch ) const

overrideprotectedvirtual

Implements tesseract::Validator.

Definition at line 55 of file validate_myanmar.cpp.

                                                                       {
   if (IsMyanmarLetter(ch)) return CharClass::kConsonant;
   return CharClass::kOther;
 }

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

/usr/src/tesseract-ocr.master/src/training/validate_myanmar.h
/usr/src/tesseract-ocr.master/src/training/validate_myanmar.cpp

Public Member Functions

Protected Member Functions