tesseract v5.3.3.20231005
Public Member Functions | Static Public Member Functions | Protected Member Functions | Protected Attributes | Static Protected Attributes | List of all members
tesseract::LigatureTable Class Reference

#include <ligature_table.h>

Public Member Functions

std::string AddLigatures (const std::string &str, const PangoFontInfo *font) const
 
std::string RemoveLigatures (const std::string &str) const
 
std::string RemoveCustomLigatures (const std::string &str) const
 
const LigHash & norm_to_lig_table () const
 
const LigHash & lig_to_norm_table () const
 

Static Public Member Functions

static LigatureTable * Get ()
 

Protected Member Functions

 LigatureTable ()
 
void Init ()
 

Protected Attributes

LigHash norm_to_lig_table_
 
LigHash lig_to_norm_table_
 
int min_lig_length_
 
int max_lig_length_
 
int min_norm_length_
 
int max_norm_length_
 

Static Protected Attributes

static std::unique_ptr< LigatureTable > instance_
 

Detailed Description

Definition at line 38 of file ligature_table.h.

Constructor & Destructor Documentation

◆ LigatureTable()

tesseract::LigatureTable::LigatureTable ( )
protected

Definition at line 59 of file ligature_table.cpp.

60 : min_lig_length_(0), max_lig_length_(0), min_norm_length_(0), max_norm_length_(0) {}

Member Function Documentation

◆ AddLigatures()

std::string tesseract::LigatureTable::AddLigatures ( const std::string &  str,
const PangoFontInfo *  font 
) const

Definition at line 153 of file ligature_table.cpp.

153 {
154 std::string result;
155 int len = str.size();
156 int step = 0;
157 int i = 0;
158 for (i = 0; i < len - min_norm_length_ + 1; i += step) {
159 step = 0;
160 for (int liglen = max_norm_length_; liglen >= min_norm_length_; --liglen) {
161 if (i + liglen <= len) {
162 std::string lig_cand = str.substr(i, liglen);
163 auto it = norm_to_lig_table_.find(lig_cand);
164 if (it != norm_to_lig_table_.end()) {
165 tlog(3, "Considering %s -> %s\n", lig_cand.c_str(), it->second.c_str());
166 if (font) {
167 // Test for renderability.
168 if (!font->CanRenderString(it->second.data(), it->second.length())) {
169 continue; // Not renderable
170 }
171 }
172 // Found a match so convert it.
173 step = liglen;
174 result += it->second;
175 tlog(2, "Substituted %s -> %s\n", lig_cand.c_str(), it->second.c_str());
176 break;
177 }
178 }
179 }
180 if (step == 0) {
181 result += str[i];
182 step = 1;
183 }
184 }
185 result += str.substr(i, len - i);
186 return result;
187}
tlog
#define tlog(level,...)
Definition: tlog.h:36

◆ Get()

LigatureTable * tesseract::LigatureTable::Get ( )
static

Definition at line 51 of file ligature_table.cpp.

51 {
52 if (instance_ == nullptr) {
53 instance_.reset(new LigatureTable());
54 instance_->Init();
55 }
56 return instance_.get();
57}
tesseract::LigatureTable::instance_
static std::unique_ptr< LigatureTable > instance_
Definition: ligature_table.h:65

◆ Init()

void tesseract::LigatureTable::Init ( )
protected

Definition at line 62 of file ligature_table.cpp.

62 {
63 if (norm_to_lig_table_.empty()) {
64 for (char32 lig = kMinLigature; lig <= kMaxLigature; ++lig) {
65 // For each char in the range, convert to utf8, nfc normalize, and if
66 // the strings are different put the both mappings in the hash_maps.
67 std::string lig8 = EncodeAsUTF8(lig);
68 icu::UnicodeString unicode_lig8(static_cast<UChar32>(lig));
69 icu::UnicodeString normed8_result;
70 icu::ErrorCode status;
71 icu::Normalizer::normalize(unicode_lig8, UNORM_NFC, 0, normed8_result, status);
72 std::string normed8;
73 normed8_result.toUTF8String(normed8);
74 int lig_length = lig8.length();
75 int norm_length = normed8.size();
76 if (normed8 != lig8 && lig_length > 1 && norm_length > 1) {
77 norm_to_lig_table_[normed8] = lig8;
78 lig_to_norm_table_[lig8] = normed8;
79 if (min_lig_length_ == 0 || lig_length < min_lig_length_) {
80 min_lig_length_ = lig_length;
81 }
82 if (lig_length > max_lig_length_) {
83 max_lig_length_ = lig_length;
84 }
85 if (min_norm_length_ == 0 || norm_length < min_norm_length_) {
86 min_norm_length_ = norm_length;
87 }
88 if (norm_length > max_norm_length_) {
89 max_norm_length_ = norm_length;
90 }
91 }
92 }
93 // Add custom extra ligatures.
94 for (int i = 0; UNICHARSET::kCustomLigatures[i][0] != nullptr; ++i) {
95 norm_to_lig_table_[UNICHARSET::kCustomLigatures[i][0]] = UNICHARSET::kCustomLigatures[i][1];
96 int norm_length = strlen(UNICHARSET::kCustomLigatures[i][0]);
97 if (min_norm_length_ == 0 || norm_length < min_norm_length_) {
98 min_norm_length_ = norm_length;
99 }
100 if (norm_length > max_norm_length_) {
101 max_norm_length_ = norm_length;
102 }
103
104 lig_to_norm_table_[UNICHARSET::kCustomLigatures[i][1]] = UNICHARSET::kCustomLigatures[i][0];
105 }
106 }
107}
char32
signed int char32
Definition: pango_font_info.h:36
tesseract::kMinLigature
const int kMinLigature
Definition: ligature_table.cpp:44
tesseract::kMaxLigature
const int kMaxLigature
Definition: ligature_table.cpp:45
tesseract::UNICHARSET::kCustomLigatures
static const char * kCustomLigatures[][2]
Definition: unicharset.h:169

◆ lig_to_norm_table()

const LigHash & tesseract::LigatureTable::lig_to_norm_table ( ) const
inline

Definition at line 55 of file ligature_table.h.

55 {
56 return lig_to_norm_table_;
57 }

◆ norm_to_lig_table()

const LigHash & tesseract::LigatureTable::norm_to_lig_table ( ) const
inline

Definition at line 52 of file ligature_table.h.

52 {
53 return norm_to_lig_table_;
54 }

◆ RemoveCustomLigatures()

std::string tesseract::LigatureTable::RemoveCustomLigatures ( const std::string &  str) const

Definition at line 128 of file ligature_table.cpp.

128 {
129 std::string result;
130 UNICHAR::const_iterator it_begin = UNICHAR::begin(str.c_str(), str.length());
131 UNICHAR::const_iterator it_end = UNICHAR::end(str.c_str(), str.length());
132 char tmp[5];
133 int len;
134 int norm_ind;
135 for (UNICHAR::const_iterator it = it_begin; it != it_end; ++it) {
136 len = it.get_utf8(tmp);
137 tmp[len] = '\0';
138 norm_ind = -1;
139 for (int i = 0; UNICHARSET::kCustomLigatures[i][0] != nullptr && norm_ind < 0; ++i) {
140 if (!strcmp(tmp, UNICHARSET::kCustomLigatures[i][1])) {
141 norm_ind = i;
142 }
143 }
144 if (norm_ind >= 0) {
145 result += UNICHARSET::kCustomLigatures[norm_ind][0];
146 } else {
147 result += tmp;
148 }
149 }
150 return result;
151}
tesseract::UNICHAR::begin
static const_iterator begin(const char *utf8_str, int byte_length)
Definition: unichar.cpp:209
tesseract::UNICHAR::end
static const_iterator end(const char *utf8_str, int byte_length)
Definition: unichar.cpp:213

◆ RemoveLigatures()

std::string tesseract::LigatureTable::RemoveLigatures ( const std::string &  str) const

Definition at line 109 of file ligature_table.cpp.

109 {
110 std::string result;
111 UNICHAR::const_iterator it_begin = UNICHAR::begin(str.c_str(), str.length());
112 UNICHAR::const_iterator it_end = UNICHAR::end(str.c_str(), str.length());
113 char tmp[5];
114 int len;
115 for (UNICHAR::const_iterator it = it_begin; it != it_end; ++it) {
116 len = it.get_utf8(tmp);
117 tmp[len] = '\0';
118 auto lig_it = lig_to_norm_table_.find(tmp);
119 if (lig_it != lig_to_norm_table_.end()) {
120 result += lig_it->second;
121 } else {
122 result += tmp;
123 }
124 }
125 return result;
126}

Member Data Documentation

◆ instance_

std::unique_ptr< LigatureTable > tesseract::LigatureTable::instance_
staticprotected

Definition at line 65 of file ligature_table.h.

◆ lig_to_norm_table_

LigHash tesseract::LigatureTable::lig_to_norm_table_
protected

Definition at line 67 of file ligature_table.h.

◆ max_lig_length_

int tesseract::LigatureTable::max_lig_length_
protected

Definition at line 69 of file ligature_table.h.

◆ max_norm_length_

int tesseract::LigatureTable::max_norm_length_
protected

Definition at line 71 of file ligature_table.h.

◆ min_lig_length_

int tesseract::LigatureTable::min_lig_length_
protected

Definition at line 68 of file ligature_table.h.

◆ min_norm_length_

int tesseract::LigatureTable::min_norm_length_
protected

Definition at line 70 of file ligature_table.h.

◆ norm_to_lig_table_

LigHash tesseract::LigatureTable::norm_to_lig_table_
protected

Definition at line 66 of file ligature_table.h.

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