tesseract 5.2.0
|
#include <validator.h>
Public Member Functions | |
virtual | ~Validator () |
Static Public Member Functions | |
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 | |
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 | |
enum class | CharClass { kConsonant = 'C' , kVowel = 'V' , kVirama = 'H' , kMatra = 'M' , kMatraPiece = 'P' , kVowelModifier = 'D' , kZeroWidthNonJoiner = 'z' , kZeroWidthJoiner = 'Z' , kVedicMark = 'v' , kNukta = 'N' , kRobat = 'R' , kOther = 'O' , kWhitespace = ' ' , kCombiner = 'c' } |
using | IndicPair = std::pair< CharClass, char32 > |
Protected Member Functions | |
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 (unsigned length) |
bool | UseMultiCode (unsigned length) |
virtual bool | ConsumeGraphemeIfValid ()=0 |
void | ComputeClassCodes (const std::vector< char32 > &text) |
virtual CharClass | UnicodeToCharClass (char32 ch) const =0 |
void | Clear () |
Static Protected Member Functions | |
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 | |
ViramaScript | script_ |
std::vector< IndicPair > | codes_ |
std::vector< std::vector< char32 > > | parts_ |
std::vector< char32 > | output_ |
unsigned | codes_used_ |
unsigned | output_used_ |
bool | report_errors_ |
Static Protected Attributes | |
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 |
Definition at line 74 of file validator.h.
|
protected |
Definition at line 135 of file validator.h.
|
strongprotected |
Enumerator | |
---|---|
kConsonant | |
kVowel | |
kVirama | |
kMatra | |
kMatraPiece | |
kVowelModifier | |
kZeroWidthNonJoiner | |
kZeroWidthJoiner | |
kVedicMark | |
kNukta | |
kRobat | |
kOther | |
kWhitespace | |
kCombiner |
Definition at line 113 of file validator.h.
|
virtualdefault |
|
inlineprotected |
Definition at line 137 of file validator.h.
|
protected |
Definition at line 198 of file validator.cpp.
|
inlineprotected |
Definition at line 166 of file validator.h.
|
protected |
Definition at line 190 of file validator.cpp.
|
protectedpure virtual |
|
protected |
Definition at line 184 of file validator.cpp.
|
staticprotected |
Definition at line 178 of file validator.cpp.
|
staticprotected |
Definition at line 169 of file validator.cpp.
|
inlinestatic |
Definition at line 89 of file validator.h.
|
staticprotected |
Definition at line 135 of file validator.cpp.
|
protected |
Definition at line 106 of file validator.cpp.
|
inlineprotected |
Definition at line 176 of file validator.h.
|
staticprotected |
Definition at line 71 of file validator.cpp.
|
inlineprotected |
Definition at line 189 of file validator.h.
|
static |
Definition at line 40 of file validator.cpp.
|
protected |
Definition at line 89 of file validator.cpp.
|
protected |
Definition at line 225 of file validator.h.
|
protected |
Definition at line 231 of file validator.h.
|
staticprotected |
Definition at line 207 of file validator.h.
|
static |
Definition at line 101 of file validator.h.
|
staticprotected |
Definition at line 219 of file validator.h.
|
staticprotected |
Definition at line 217 of file validator.h.
|
static |
Definition at line 99 of file validator.h.
|
staticprotected |
Definition at line 220 of file validator.h.
|
staticprotected |
Definition at line 211 of file validator.h.
|
staticprotected |
Definition at line 213 of file validator.h.
|
staticprotected |
Definition at line 209 of file validator.h.
|
staticprotected |
Definition at line 216 of file validator.h.
|
static |
Definition at line 100 of file validator.h.
|
staticprotected |
Definition at line 215 of file validator.h.
|
static |
Definition at line 98 of file validator.h.
|
static |
Definition at line 97 of file validator.h.
|
static |
Definition at line 96 of file validator.h.
|
protected |
Definition at line 229 of file validator.h.
|
protected |
Definition at line 233 of file validator.h.
|
protected |
Definition at line 227 of file validator.h.
|
protected |
Definition at line 235 of file validator.h.
|
protected |
Definition at line 223 of file validator.h.