28 #ifndef DISABLED_LEGACY_ENGINE
92 #ifndef DISABLED_LEGACY_ENGINE
98 int score1 = 0, score2 = 0;
101 for (
int f = 0; f < fonts_.
size(); ++f) {
102 if (fonts_[f].score > score1) {
104 fontinfo_id2_ = fontinfo_id_;
105 score1 = fonts_[f].score;
106 fontinfo_id_ = fonts_[f].fontinfo_id;
107 }
else if (fonts_[f].score > score2) {
108 score2 = fonts_[f].score;
109 fontinfo_id2_ = fonts_[f].fontinfo_id;
142 unichar_id_ = newunichar_id;
151 script_id_ = newscript_id;
154 matrix_cell_.
col = col;
155 matrix_cell_.
row = row;
171 tprintf(
"r%.2f c%.2f x[%g,%g]: %d %s",
173 min_xheight_, max_xheight_, unichar_id_,
174 (unicharset ==
nullptr) ?
"" :
179 tprintf(
" script=%d, font1=%d, font2=%d, yshift=%g, classifier=%d\n",
180 script_id_, fontinfo_id_, fontinfo_id2_, yshift_, classifier_);
186 return (bc1->rating_ < bc2->rating_) ? -1 : 1;
194 #ifndef DISABLED_LEGACY_ENGINE
198 int16_t fontinfo_id_;
199 int16_t fontinfo_id2_;
273 const char *src_lengths,
276 uint8_t src_permuter,
279 this->
init(src_string, src_lengths, src_rating,
280 src_certainty, src_permuter);
284 :
ELIST_LINK(word), unicharset_(word.unicharset_) {
297 return adjust_factor_;
300 adjust_factor_ = factor;
306 assert(index < length_);
307 return unichar_ids_[index];
310 return state_[index];
313 if (index < 0 || index >= length_)
315 return script_pos_[index];
324 return certainties_[index];
327 return min_x_height_;
330 return max_x_height_;
333 min_x_height_ = min_height;
334 max_x_height_ = max_height;
350 assert(index < length_);
354 return dangerous_ambig_found_;
357 dangerous_ambig_found_ = value;
363 certainty_ = new_val;
380 reserved_, unichar_ids_);
382 reserved_, script_pos_);
386 reserved_, certainties_);
392 certainties_ =
new float[1];
399 inline void init(
int reserved) {
400 reserved_ = reserved;
404 state_ =
new int[reserved];
405 certainties_ =
new float[reserved];
407 unichar_ids_ =
nullptr;
408 script_pos_ =
nullptr;
410 certainties_ =
nullptr;
413 adjust_factor_ = 1.0f;
415 certainty_ = FLT_MAX;
416 min_x_height_ = 0.0f;
417 max_x_height_ = FLT_MAX;
419 unichars_in_script_order_ =
false;
420 dangerous_ambig_found_ =
false;
428 void init(
const char *src_string,
const char *src_lengths,
429 float src_rating,
float src_certainty,
430 uint8_t src_permuter);
436 certainty_ = -FLT_MAX;
445 assert(reserved_ > length_);
456 assert(index < length_);
458 state_[index] = blob_count;
497 for (
int i = 0; i < length_; ++i) {
498 word_str += unicharset_->
debug_str(unichar_ids_[i]);
505 for (
int i = 0; i < length_; ++i) {
512 for (
int i = 0; i < length_; ++i) {
522 return unichars_in_script_order_ = in_script_order;
526 return unichars_in_script_order_;
533 return unichar_string_;
540 return unichar_lengths_;
556 const TBOX& blob_box,
571 void print(
const char *msg)
const;
608 float adjust_factor_;
622 bool unichars_in_script_order_;
624 bool dangerous_ambig_found_;
628 mutable STRING unichar_string_;
629 mutable STRING unichar_lengths_;
644 BLOB_CHOICE_LIST *ratings,
void print_ratings_list(const char *msg, BLOB_CHOICE_LIST *ratings, const UNICHARSET ¤t_unicharset)
bool EqualIgnoringCaseAndTerminalPunct(const WERD_CHOICE &word1, const WERD_CHOICE &word2)
BLOB_CHOICE * FindMatchingChoice(UNICHAR_ID char_id, BLOB_CHOICE_LIST *bc_list)
#define ELISTIZEH(CLASSNAME)
DLLSYM void tprintf(const char *format,...)
const char * ScriptPosToString(enum ScriptPos script_pos)
static T * double_the_size_memcpy(int current_size, T *data)
float max_xheight() const
void set_script(int newscript_id)
void set_rating(float newrat)
void set_fonts(const GenericVector< tesseract::ScoredFont > &fonts)
void set_classifier(BlobChoiceClassifier classifier)
int16_t fontinfo_id2() const
bool PosAndSizeAgree(const BLOB_CHOICE &other, float x_height, bool debug) const
int16_t fontinfo_id() const
void print(const UNICHARSET *unicharset) const
const GenericVector< tesseract::ScoredFont > & fonts() const
BlobChoiceClassifier classifier() const
void set_matrix_cell(int col, int row)
const MATRIX_COORD & matrix_cell()
UNICHAR_ID unichar_id() const
float min_xheight() const
bool IsClassified() const
void set_certainty(float newrat)
static BLOB_CHOICE * deep_copy(const BLOB_CHOICE *src)
static int SortByRating(const void *p1, const void *p2)
void set_unichar_id(UNICHAR_ID newunichar_id)
MATRIX_COORD MatrixCoord(int index) const
void set_x_heights(float min_height, float max_height)
void set_certainty(float new_val)
WERD_CHOICE shallow_copy(int start, int end) const
WERD_CHOICE(const WERD_CHOICE &word)
float certainty(int index) const
const STRING debug_string() const
bool set_unichars_in_script_order(bool in_script_order)
void remove_unichar_id(int index)
int state(int index) const
float adjust_factor() const
void SetScriptPositions(bool small_caps, TWERD *word, int debug=0)
void UpdateStateForSplit(int blob_position)
int TotalOfStates() const
const UNICHARSET * unicharset() const
void set_adjust_factor(float factor)
void set_unichar_id(UNICHAR_ID unichar_id, int blob_count, float rating, float certainty, int index)
void remove_last_unichar_id()
void double_the_size()
Make more space in unichar_id_ and fragment_lengths_ arrays.
bool unichars_in_script_order() const
static tesseract::ScriptPos ScriptPositionOf(bool print_debug, const UNICHARSET &unicharset, const TBOX &blob_box, UNICHAR_ID unichar_id)
void string_and_lengths(STRING *word_str, STRING *word_lengths_str) const
int GetTopScriptID() const
WERD_CHOICE & operator=(const WERD_CHOICE &source)
void punct_stripped(int *start_core, int *end_core) const
bool dangerous_ambig_found() const
void print_state(const char *msg) const
UNICHAR_ID unichar_id(int index) const
void set_unichar_id(UNICHAR_ID unichar_id, int index)
bool contains_unichar_id(UNICHAR_ID unichar_id) const
void set_rating(float new_val)
const char * permuter_name() const
WERD_CHOICE & operator+=(const WERD_CHOICE &second)
void DisplaySegmentation(TWERD *word)
void SetAllScriptPositions(tesseract::ScriptPos position)
void set_permuter(uint8_t perm)
BLOB_CHOICE_LIST * blob_choices(int index, MATRIX *ratings) const
const UNICHAR_ID * unichar_ids() const
static const float kBadRating
void make_bad()
Set the fields in this choice to be default (bad) values.
bool has_rtl_unichar_id() const
bool ContainsAnyNonSpaceDelimited() const
void remove_unichar_ids(int index, int num)
void set_blob_choice(int index, int blob_count, const BLOB_CHOICE *blob_choice)
void reverse_and_mirror_unichar_ids()
float min_x_height() const
WERD_CHOICE(const UNICHARSET *unicharset)
tesseract::ScriptPos BlobPosition(int index) const
float max_x_height() const
const STRING & unichar_lengths() const
const STRING & unichar_string() const
void set_dangerous_ambig_found_(bool value)
void GetNonSuperscriptSpan(int *start, int *end) const
void append_unichar_id(UNICHAR_ID unichar_id, int blob_count, float rating, float certainty)
void append_unichar_id_space_allocated(UNICHAR_ID unichar_id, int blob_count, float rating, float certainty)
WERD_CHOICE(const UNICHARSET *unicharset, int reserved)
WERD_CHOICE(const char *src_string, const char *src_lengths, float src_rating, float src_certainty, uint8_t src_permuter, const UNICHARSET &unicharset)
const char * string() const
bool IsSpaceDelimited(UNICHAR_ID unichar_id) const
STRING debug_str(UNICHAR_ID id) const