Use single tesseract library

Link with it during compilation.
Bump to 5.2.0.

screen-translator.pro

@@ -8,7 +8,7 @@ DEPS_DIR=$$(ST_DEPS_DIR)
 isEmpty(DEPS_DIR):DEPS_DIR=$$PWD/../deps
 INCLUDEPATH += $$DEPS_DIR/include
 LIBS += -L$$DEPS_DIR/lib
-LIBS += -lhunspell -lleptonica
+LIBS += -lhunspell -lleptonica -ltesseract
 
 win32{
   LIBS += -lUser32

share/ci/appimage.py

@@ -47,8 +47,7 @@ os.environ['VERSION'] = app_version
 flags = '' if os.getenv("DEBUG") is None else '-unsupported-allow-new-glibc'
 
 additional_files = glob(ssl_dir + '/lib/lib*.so.*') + \
-    glob('/usr/lib/x86_64-linux-gnu/nss/*') + \
+    glob('/usr/lib/x86_64-linux-gnu/nss/*')
-    glob(dependencies_dir + '/lib/libtesseract-*.so')
 out_lib_dir = install_dir + '/usr/lib'
 os.makedirs(out_lib_dir, exist_ok=True)
 for f in additional_files:

share/ci/get_tesseract.py

@@ -6,27 +6,12 @@ import platform
 c.print('>> Installing tesseract')
 
 install_dir = dependencies_dir
-url = 'https://github.com/tesseract-ocr/tesseract/archive/5.1.0.tar.gz'
+required_version = '5.2.0'
-required_version = '5.1.0'
+url = 'https://github.com/tesseract-ocr/tesseract/archive/{}.tar.gz'.format(required_version)
 
 build_type_flag = 'Debug' if build_type == 'debug' else 'Release'
 
-# compatibility flags
+cache_file = install_dir + '/tesseract.cache'
-compat_flags = ''
-compat_flags += ' -D DISABLE_LEGACY_ENGINE=ON '
-compat_flags += ' -D DISABLE_ARCHIVE=ON '
-compat_flags += ' -D DISABLE_CURL=ON '
-
-version_tag = os.environ.get('TAG', '')
-if version_tag == 'compatible':
-    compat_flags += ' -D HAVE_AVX2=0 '
-    compat_flags += ' -D HAVE_FMA=0 '
-
-lib_suffix = version_tag
-if len(lib_suffix) > 0:
-    lib_suffix = '-' + lib_suffix
-
-cache_file = install_dir + '/tesseract{}.cache'.format(lib_suffix)
 cache_file_data = required_version + build_type_flag
 
 def check_existing():
@@ -42,22 +27,25 @@ def check_existing():
         return False
 
     if platform.system() == "Windows":
-        lib = install_dir + '/bin/tesseract{}.dll'.format(lib_suffix)
+        file_name_ver = required_version[0] + required_version[2]
-        orig_lib = install_dir + '/bin/tesseract51.dll'
+        dll = install_dir + '/bin/tesseract{}.dll'.format(file_name_ver)
-    elif platform.system() == "Darwin":
+        lib = install_dir + '/lib/tesseract{}.lib'.format(file_name_ver)
-        lib = install_dir + '/lib/libtesseract{}.dylib'.format(lib_suffix)
+        if not os.path.exists(dll) or not os.path.exists(lib):
-        orig_lib = install_dir + '/lib/libtesseract.{}.dylib'.format(required_version)
-    else:
-        lib = install_dir + '/lib/libtesseract{}.so'.format(lib_suffix)
-        orig_lib = install_dir + '/lib/libtesseract.so.{}'.format(required_version)
-
-    if os.path.exists(lib):
-        return True
-    if os.path.exists(orig_lib):
-        os.rename(orig_lib, lib)
-        return True
-
             return False
+        c.symlink(dll, install_dir + '/bin/tesseract.dll')
+        c.symlink(lib, install_dir + '/lib/tesseract.lib')
+    elif platform.system() == "Darwin":
+        lib = install_dir + '/lib/libtesseract.{}.dylib'.format(required_version)
+        if not os.path.exists(lib):
+            return False
+        c.symlink(lib, install_dir + '/lib/libtesseract.dylib')
+    else:
+        lib = install_dir + '/lib/libtesseract.so.{}'.format(required_version)
+        if not os.path.exists(lib):
+            return False
+        c.symlink(lib, install_dir + '/lib/libtesseract.so')
+
+    return True
 
 
 if check_existing() and not 'FORCE' in os.environ:
@@ -71,60 +59,24 @@ src_dir = os.path.abspath('tesseract_src')
 c.extract(archive, '.')
 c.symlink(c.get_archive_top_dir(archive), src_dir)
 
-if platform.system() == "Windows":
-    # workaround for not found 'max'
-    modify_data = ''
-    modify_file = '{}/src/ccmain/thresholder.cpp'.format(src_dir)
-    with open(modify_file, 'r') as f:
-        modify_data = f.read()
-
-    if modify_data.find('<algorithm>') == -1:
-        modify_data = modify_data.replace(
-        '''<tuple>''',
-        '''<tuple>\n#include <algorithm>''')
-
-    with open(modify_file, 'w') as f:
-        f.write(modify_data)
-
-    # ignore libtiff
-    modify_data = ''
-    modify_file = '{}/CMakeLists.txt'.format(src_dir)
-    with open(modify_file, 'r') as f:
-        modify_data = f.read()
-
-    if modify_data.find('#pkg_check_modules(TIFF libtiff-4)') == -1:
-        modify_data = modify_data.replace(
-        '''pkg_check_modules(TIFF libtiff-4)''',
-        '''#pkg_check_modules(TIFF libtiff-4)''')
-
-    with open(modify_file, 'w') as f:
-        f.write(modify_data)
-
-if platform.system() == "Linux":
-    # FIXME fix crash on ubuntu
-    modify_data = ''
-    modify_file = '{}/src/ccmain/tessedit.cpp'.format(src_dir)
-    with open(modify_file, 'r') as f:
-        modify_data = f.read()
-
-    lines = modify_data.split('\n')
-    for line in [250,253,255,256]:
-        if not lines[line].startswith('//'):
-            lines[line] = '// ' + lines[line]
-    modify_data = '\n'.join(lines)
-
-    with open(modify_file, 'w') as f:
-        f.write(modify_data)
-
-
-
 c.ensure_got_path(install_dir)
 
 c.recreate_dir(build_dir)
 os.chdir(build_dir)
 
-cmake_args = '"{0}" -DCMAKE_INSTALL_PREFIX="{1}" -DLeptonica_DIR="{1}/cmake" \
+cmake_args = '"{0}" \
--DBUILD_TRAINING_TOOLS=OFF -DBUILD_TESTS=OFF -DBUILD_SHARED_LIBS=ON -DSW_BUILD=OFF \
+-DCMAKE_INSTALL_PREFIX="{1}" \
+-DLeptonica_DIR="{1}/cmake" \
+-DSW_BUILD=OFF \
+-DBUILD_TRAINING_TOOLS=OFF \
+-DBUILD_TESTS=OFF \
+-DBUILD_SHARED_LIBS=ON \
+-DDISABLE_CURL=ON \
+-DDISABLE_ARCHIVE=ON \
+-DUSE_SYSTEM_ICU=ON \
+-DENABLE_LTO=ON \
+-DGRAPHICS_DISABLED=ON \
+-DDISABLED_LEGACY_ENGINE=ON \
 '.format(src_dir, install_dir)
 
 if platform.system() == "Windows":
@@ -135,10 +87,6 @@ if platform.system() == "Windows":
 c.set_make_threaded()
 c.run('cmake {}'.format(cmake_args))
 
-if len(compat_flags) > 0:
-    c.run('cmake {} .'.format(compat_flags))
-    c.run('cmake {} .'.format(compat_flags))  # for sure :)
-
 c.run('cmake --build . --config {}'.format(build_type_flag))
 c.run('cmake --build . --target install --config {}'.format(build_type_flag))
 

share/ci/release.py

@@ -29,15 +29,8 @@ def r(script):
 r('get_qt.py')
 r('get_qt_ssl.py')
 r('get_leptonica.py')
-
-os.environ['TAG'] = 'optimized'
 r('get_tesseract.py')
 
-os.environ['TAG'] = 'compatible'
-r('get_tesseract.py')
-
-del os.environ['TAG']
-
 r('get_hunspell.py')
 r('test.py')
 r('build.py')

share/translations/screentranslator_ru.ts

@@ -673,12 +673,12 @@ Check for updates to silence this warning</source>
         <translation>неизвестные языки для перевода: %1 или %2</translation>
     </message>
     <message>
-        <location filename="../../src/ocr/tesseract.cpp" line="255"/>
+        <location filename="../../src/ocr/tesseract.cpp" line="234"/>
         <source>init failed</source>
         <translation>ошибка инициалиизации</translation>
     </message>
     <message>
-        <location filename="../../src/ocr/tesseract.cpp" line="306"/>
+        <location filename="../../src/ocr/tesseract.cpp" line="294"/>
         <source>Failed to recognize text or no text selected</source>
         <translation>Ошибка распознавания текста или нет текста в выделенной зоне</translation>
     </message>
@@ -697,12 +697,12 @@ in %1</source>
 в %1</translation>
     </message>
     <message>
-        <location filename="../../src/capture/capturearea.cpp" line="27"/>
+        <location filename="../../src/capture/capturearea.cpp" line="28"/>
         <source>No source language set</source>
         <translation>Не задан исходный язык</translation>
     </message>
     <message>
-        <location filename="../../src/capture/capturearea.cpp" line="34"/>
+        <location filename="../../src/capture/capturearea.cpp" line="35"/>
         <source>No target language set</source>
         <translation>Не задан язык результата</translation>
     </message>
@@ -712,37 +712,37 @@ in %1</source>
         <translation>Не восстанавливать интерфейс пользователя (размер и положения окна и т.д.)</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="197"/>
+        <location filename="../../src/settingseditor.cpp" line="189"/>
         <source>&lt;p&gt;Optical character recognition (OCR) and translation tool&lt;/p&gt;</source>
         <translation>&lt;p&gt;Инструмент оптического распознавания текста (OCR) и перевода&lt;/p&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="199"/>
+        <location filename="../../src/settingseditor.cpp" line="191"/>
         <source>&lt;p&gt;Version: %1&lt;/p&gt;</source>
         <translation>&lt;p&gt;Версия: %1&lt;/p&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="201"/>
+        <location filename="../../src/settingseditor.cpp" line="193"/>
         <source>&lt;p&gt;Setup instructions: &lt;a href=&quot;%1&quot;&gt;%1&lt;/a&gt;&lt;/p&gt;</source>
         <translation>&lt;p&gt;Инструкции по установке: &lt;a href=&quot;%1&quot;&gt;%1&lt;/a&gt;&lt;/p&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="202"/>
+        <location filename="../../src/settingseditor.cpp" line="195"/>
         <source>&lt;p&gt;Changelog: &lt;a href=&quot;%1&quot;&gt;%2&lt;/a&gt;&lt;/p&gt;</source>
         <translation>&lt;p&gt;Список изменений: &lt;a href=&quot;%1&quot;&gt;%2&lt;/a&gt;&lt;/p&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="204"/>
+        <location filename="../../src/settingseditor.cpp" line="197"/>
         <source>&lt;p&gt;License: &lt;a href=&quot;%3&quot;&gt;MIT&lt;/a&gt;&lt;/p&gt;</source>
         <translation>&lt;p&gt;Лицензия: &lt;a href=&quot;%3&quot;&gt;MIT&lt;/a&gt;&lt;/p&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="205"/>
+        <location filename="../../src/settingseditor.cpp" line="198"/>
         <source>&lt;p&gt;Author: Gres (&lt;a href=&quot;mailto:%1&quot;&gt;%1&lt;/a&gt;)&lt;/p&gt;</source>
         <translation>&lt;p&gt;Автор: Gres (&lt;a href=&quot;mailto:%1&quot;&gt;%1&lt;/a&gt;)&lt;/p&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="207"/>
+        <location filename="../../src/settingseditor.cpp" line="200"/>
         <source>&lt;p&gt;Issues: &lt;a href=&quot;%1&quot;&gt;%1&lt;/a&gt;&lt;/p&gt;</source>
         <translation>&lt;p&gt;Поддержка: &lt;a href=&quot;%1&quot;&gt;%1&lt;/a&gt;&lt;/p&gt;</translation>
     </message>
@@ -844,7 +844,7 @@ in %1</source>
         <translation>Повторить захват</translation>
     </message>
     <message>
-        <location filename="../../src/represent/resultwidget.cpp" line="93"/>
+        <location filename="../../src/represent/resultwidget.cpp" line="95"/>
         <source>Without correction:
 </source>
         <translation>Без коррекции:
@@ -919,42 +919,41 @@ in %1</source>
         <translation>сохранять пароль (небезопасно)</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="355"/>
         <source>Library version</source>
-        <translation>Версия</translation>
+        <translation type="vanished">Версия</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="369"/>
+        <location filename="../../src/settingseditor.ui" line="359"/>
         <source>User substitutions</source>
         <translation>Пользовательская коррекция</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="392"/>
+        <location filename="../../src/settingseditor.ui" line="382"/>
         <source>Use auto corrections (hunspell)</source>
         <translation>Использовать автокоррекцию (hunspell)</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="399"/>
+        <location filename="../../src/settingseditor.ui" line="389"/>
         <source>Use user substitutions</source>
         <translation>Использовать пользовательскую коррекцию</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="406"/>
+        <location filename="../../src/settingseditor.ui" line="396"/>
         <source>Hunspell dictionaries path:</source>
         <translation>Путь к словарям Hunspell:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="468"/>
+        <location filename="../../src/settingseditor.ui" line="458"/>
         <source>Language:</source>
         <translation>Язык:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="504"/>
+        <location filename="../../src/settingseditor.ui" line="494"/>
         <source> secs</source>
         <translation> сек</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="444"/>
+        <location filename="../../src/settingseditor.ui" line="434"/>
         <source>Ignore SSL errors</source>
         <translation>Игнорировать ошибки SSL</translation>
     </message>
@@ -984,102 +983,102 @@ in %1</source>
         <translation>Писать логи в файл (отладка)</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="342"/>
+        <location filename="../../src/settingseditor.ui" line="345"/>
         <source>Default language:</source>
         <translation>Язык по умолчанию:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="329"/>
+        <location filename="../../src/settingseditor.ui" line="303"/>
         <source>Tessdata path:</source>
         <translation>Путь к языкам (tessdata):</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="379"/>
+        <location filename="../../src/settingseditor.ui" line="369"/>
         <source>\\ for \ symbol, \n for newline</source>
         <translation>\\ для символа \ , \n для символа новой строки</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="427"/>
+        <location filename="../../src/settingseditor.ui" line="417"/>
         <source>Translators path:</source>
         <translation>Путь к переводчикам:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="434"/>
+        <location filename="../../src/settingseditor.ui" line="424"/>
         <source>Translators</source>
         <translation>Переводчики</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="569"/>
+        <location filename="../../src/settingseditor.ui" line="559"/>
         <source>Result window</source>
         <translation>Окно результата</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="575"/>
+        <location filename="../../src/settingseditor.ui" line="565"/>
         <source>Font:</source>
         <translation>Шрифт:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="585"/>
+        <location filename="../../src/settingseditor.ui" line="575"/>
         <source>Font size:</source>
         <translation>Размер шрифта:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="602"/>
+        <location filename="../../src/settingseditor.ui" line="592"/>
         <source>Font color:</source>
         <translation>Цвет шрифта:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="616"/>
+        <location filename="../../src/settingseditor.ui" line="606"/>
         <source>Background:</source>
         <translation>Фон:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="630"/>
+        <location filename="../../src/settingseditor.ui" line="620"/>
         <source>Show image</source>
         <translation>Показывать изображение</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="637"/>
+        <location filename="../../src/settingseditor.ui" line="627"/>
         <source>Show recognized</source>
         <translation>Показывать распознанное</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="687"/>
+        <location filename="../../src/settingseditor.ui" line="677"/>
         <source>Update check interval (days):</source>
         <translation>Интервал проверки обновления (дней):</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="694"/>
+        <location filename="../../src/settingseditor.ui" line="684"/>
         <source>0 - disabled</source>
         <translation>- отключено</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="454"/>
+        <location filename="../../src/settingseditor.ui" line="444"/>
         <source>Translate text</source>
         <translation>Переводить текст</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="461"/>
+        <location filename="../../src/settingseditor.ui" line="451"/>
         <source>Single translator timeout:</source>
         <translation>Переходить к следующему переводчику после:</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="531"/>
+        <location filename="../../src/settingseditor.ui" line="521"/>
         <source>Result type</source>
         <translation>Тип результата</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="543"/>
+        <location filename="../../src/settingseditor.ui" line="533"/>
         <source>Tray</source>
         <translation>Трей</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="556"/>
+        <location filename="../../src/settingseditor.ui" line="546"/>
         <source>Window</source>
         <translation>Окно</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.ui" line="710"/>
+        <location filename="../../src/settingseditor.ui" line="700"/>
         <source>Check now</source>
         <translation>Проверить сейчас</translation>
     </message>
@@ -1175,32 +1174,29 @@ Hunspell ищет в своем словаре слова, похожие на
         <translation>HTTP</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="126"/>
         <source>Optimized</source>
-        <translation>Оптимизированная</translation>
+        <translation type="vanished">Оптимизированная</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="127"/>
         <source>Compatible</source>
-        <translation>Совместимая</translation>
+        <translation type="vanished">Совместимая</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="131"/>
         <source>Use compatible version if you are experiencing crashes during recognition</source>
-        <translation>Используйте совместимую версию если программа неожиданно завершается во время распознавания</translation>
+        <translation type="vanished">Используйте совместимую версию если программа неожиданно завершается во время распознавания</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="143"/>
+        <location filename="../../src/settingseditor.cpp" line="135"/>
         <source>&lt;b&gt;NOTE! Some translators might require the translation window to be visible. You can make it using the &quot;Show translator&quot; entry in the tray icon&apos;s context menu&lt;/b&gt;</source>
         <translation>&lt;b&gt;ПРИМЕЧАНИЕ! Для работы некоторых переводчиков может потребоваться активное окно перевода. Его можно отобразить при помощи пункта &quot;Показать окно перевода&quot; контекстного меню иконки в трее&lt;/b&gt;</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="151"/>
+        <location filename="../../src/settingseditor.cpp" line="143"/>
         <source>Sample text</source>
         <translation>Текст для проверки</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="215"/>
+        <location filename="../../src/settingseditor.cpp" line="208"/>
         <source>The program workflow consists of the following steps:
 1. Selection on the screen area
 2. Recognition of the selected area
@@ -1223,7 +1219,7 @@ Then set default recognition and translation languages, enable some (or all) tra
 Далее установите языки распознавания и перевода по умолчанию, активируйте некоторые (или все) переводчики и настройку &quot;переводить текст&quot;, если нужно.</translation>
     </message>
     <message>
-        <location filename="../../src/settingseditor.cpp" line="380"/>
+        <location filename="../../src/settingseditor.cpp" line="371"/>
         <source>Portable changed. Apply settings first</source>
         <translation>Portable режим изменен. Сначала примените настройки</translation>
     </message>

src/ocr/recognizer.cpp

@@ -79,9 +79,5 @@ void Recognizer::updateSettings()
   SOFT_ASSERT(!settings_.tessdataPath.isEmpty(), return );
 
   queue_.clear();
-  const auto libName =
+  emit reset(settings_.tessdataPath);
-      (settings_.tesseractVersion == TesseractVersion::Optimized
-           ? "tesseract-optimized"
-           : "tesseract-compatible");
-  emit reset(settings_.tessdataPath, libName);
 }

src/ocr/recognizer.h

@@ -18,7 +18,7 @@ public:
 
 signals:
   void recognizeImpl(const TaskPtr &task);
-  void reset(const QString &tessdataPath, const QString &tesseractLibrary);
+  void reset(const QString &tessdataPath);
 
 private:
   void recognized(const TaskPtr &task);

src/ocr/recognizerworker.cpp

@@ -17,8 +17,8 @@ void RecognizeWorker::handle(const TaskPtr &task)
   if (!engines_.count(task->sourceLanguage)) {
     LTRACE() << "Create OCR engine" << task->sourceLanguage;
 
-    auto engine = std::make_unique<Tesseract>(task->sourceLanguage,
+    auto engine =
-                                              tessdataPath_, tesseractLibrary_);
+        std::make_unique<Tesseract>(task->sourceLanguage, tessdataPath_);
 
     if (!engine->isValid()) {
       result->error = tr("Failed to init OCR engine: %1").arg(engine->error());
@@ -43,14 +43,12 @@ void RecognizeWorker::handle(const TaskPtr &task)
   emit finished(result);
 }
 
-void RecognizeWorker::reset(const QString &tessdataPath,
+void RecognizeWorker::reset(const QString &tessdataPath)
-                            const QString &tesseractLibrary)
 {
-  if (tessdataPath_ == tessdataPath && tesseractLibrary_ == tesseractLibrary)
+  if (tessdataPath_ == tessdataPath)
     return;
 
   tessdataPath_ = tessdataPath;
-  tesseractLibrary_ = tesseractLibrary;
   engines_.clear();
   LTRACE() << "Cleared OCR engines";
 }

src/ocr/recognizerworker.h

@@ -13,7 +13,7 @@ public:
   ~RecognizeWorker();
 
   void handle(const TaskPtr &task);
-  void reset(const QString &tessdataPath, const QString &tesseractLibrary);
+  void reset(const QString &tessdataPath);
 
 signals:
   void finished(const TaskPtr &task);
@@ -24,5 +24,4 @@ private:
   std::map<QString, std::unique_ptr<Tesseract>> engines_;
   std::map<QString, Generation> lastGenerations_;
   QString tessdataPath_;
-  QString tesseractLibrary_;
 };

src/ocr/tesseract.cpp

@@ -4,6 +4,7 @@
 #include "task.h"
 
 #include <leptonica/allheaders.h>
+#include <tesseract/baseapi.h>
 
 #include <QBuffer>
 #include <QDir>
@@ -204,115 +205,7 @@ static Pix *prepareImage(const QImage &image)
   return pix.take();
 }
 
-static void cleanupImage(Pix **image)
+Tesseract::Tesseract(const LanguageId &language, const QString &tessdataPath)
-{
-  pixDestroy(image);
-}
-
-// do not include capi.h from tesseract because it defined BOOL that breaks msvc
-struct TessBaseAPI;
-
-class Tesseract::Wrapper
-{
-  using CreateApi = TessBaseAPI *(*)();
-  using DeleteApi = void (*)(TessBaseAPI *);
-  using InitApi = int (*)(TessBaseAPI *, const char *, const char *, int);
-  using SetImage = void (*)(TessBaseAPI *, struct Pix *);
-  using GetUtf8 = char *(*)(TessBaseAPI *);
-  using ClearApi = void (*)(TessBaseAPI *);
-  using DeleteUtf8 = void (*)(const char *);
-  using SetPageMode = void (*)(TessBaseAPI *, int);
-
-public:
-  explicit Wrapper(const QString &libraryName)
-    : lib(libraryName)
-  {
-    if (!lib.load()) {
-      LERROR() << "Failed to load tesseract library" << libraryName;
-      return;
-    }
-
-    LTRACE() << "Loaded tesseract library" << lib.fileName();
-    auto ok = true;
-    ok &= bool(createApi_ = (CreateApi)lib.resolve("TessBaseAPICreate"));
-    ok &= bool(deleteApi_ = (DeleteApi)lib.resolve("TessBaseAPIDelete"));
-    ok &= bool(initApi_ = (InitApi)lib.resolve("TessBaseAPIInit2"));
-    ok &= bool(setImage_ = (SetImage)lib.resolve("TessBaseAPISetImage2"));
-    ok &= bool(getUtf8_ = (GetUtf8)lib.resolve("TessBaseAPIGetUTF8Text"));
-    ok &= bool(clearApi_ = (ClearApi)lib.resolve("TessBaseAPIClear"));
-    ok &= bool(deleteUtf8_ = (DeleteUtf8)lib.resolve("TessDeleteText"));
-    ok &= bool(setPageMode_ =
-                   (SetPageMode)lib.resolve("TessBaseAPISetPageSegMode"));
-    if (!ok) {
-      LERROR() << "Failed to resolve tesseract functions from" << libraryName;
-      return;
-    }
-    handle_ = createApi_();
-  }
-
-  ~Wrapper()
-  {
-    if (handle_ && deleteApi_) {
-      deleteApi_(handle_);
-    }
-    lib.unload();
-  }
-
-  int Init(const char *datapath, const char *language)
-  {
-    SOFT_ASSERT(handle_, return -1);
-    SOFT_ASSERT(initApi_, return -1);
-
-    const auto mode = 3;  // TessOcrEngineMode::OEM_DEFAULT
-    return initApi_(handle_, datapath, language, mode);
-  }
-
-  QString GetText(Pix *pix)
-  {
-    SOFT_ASSERT(handle_, return {});
-
-    SOFT_ASSERT(setPageMode_, return {});
-    setPageMode_(handle_, 3);  //    PSM_AUTO
-
-    SOFT_ASSERT(setImage_, return {});
-    setImage_(handle_, pix);
-    LTRACE() << "Set Pix to engine";
-
-    char *outText = nullptr;
-
-    SOFT_ASSERT(getUtf8_, return {});
-    outText = getUtf8_(handle_);
-    LTRACE() << "Received recognized text";
-
-    SOFT_ASSERT(clearApi_, return {});
-    clearApi_(handle_);
-    LTRACE() << "Cleared engine";
-
-    const auto result = QString(outText).trimmed();
-
-    SOFT_ASSERT(deleteUtf8_, return {});
-    deleteUtf8_(outText);
-    LTRACE() << "Cleared recognized text buffer";
-
-    return result;
-  }
-
-private:
-  QLibrary lib;
-  CreateApi createApi_{nullptr};
-  DeleteApi deleteApi_{nullptr};
-  InitApi initApi_{nullptr};
-  SetImage setImage_{nullptr};
-  GetUtf8 getUtf8_{nullptr};
-  ClearApi clearApi_{nullptr};
-  DeleteUtf8 deleteUtf8_{nullptr};
-  SetPageMode setPageMode_{nullptr};
-  TessBaseAPI *handle_{nullptr};
-};
-
-Tesseract::Tesseract(const LanguageId &language, const QString &tessdataPath,
-                     const QString &tesseractLibrary)
-  : tesseractLibrary_(tesseractLibrary)
 {
   SOFT_ASSERT(!tessdataPath.isEmpty(), return );
   SOFT_ASSERT(!language.isEmpty(), return );
@@ -324,20 +217,22 @@ Tesseract::~Tesseract() = default;
 
 void Tesseract::init(const LanguageId &language, const QString &tessdataPath)
 {
-  SOFT_ASSERT(!engine_, return );
+  SOFT_ASSERT(!api_, return );
 
-  engine_ = std::make_unique<Wrapper>(tesseractLibrary_);
+  api_ = std::make_unique<tesseract::TessBaseAPI>();
-  LTRACE() << "Created Tesseract api" << engine_.get();
+  LTRACE() << "Created Tesseract api" << api_.get();
 
   const auto tesseractName = LanguageCodes::tesseract(language);
-  auto result =
+  auto result = api_->Init(qPrintable(tessdataPath), qPrintable(tesseractName),
-      engine_->Init(qPrintable(tessdataPath), qPrintable(tesseractName));
+                           tesseract::OcrEngineMode::OEM_DEFAULT);
   LTRACE() << "Inited Tesseract api" << result;
   if (result == 0)
     return;
 
+  api_->SetPageSegMode(tesseract::PageSegMode::PSM_AUTO);
+
   error_ = QObject::tr("init failed");
-  engine_.reset();
+  api_.reset();
   LTRACE() << "Cleared Tesseract api";
 }
 
@@ -372,19 +267,28 @@ QStringList Tesseract::availableLanguageNames(const QString &path)
 
 QString Tesseract::recognize(const QPixmap &source)
 {
-  SOFT_ASSERT(engine_, return {});
+  SOFT_ASSERT(api_, return {});
   SOFT_ASSERT(!source.isNull(), return {});
 
   error_.clear();
 
-  Pix *image = prepareImage(source.toImage());
+  PixGuard image(prepareImage(source.toImage()));
   SOFT_ASSERT(image, return {});
   LTRACE() << "Preprocessed Pix for OCR" << image;
 
-  auto result = engine_->GetText(image);
+  api_->SetImage(image);
+  LTRACE() << "Set Pix to engine";
 
-  cleanupImage(&image);
+  const auto outText = api_->GetUTF8Text();
-  LTRACE() << "Cleared preprocessed Pix";
+  LTRACE() << "Received recognized text";
+
+  api_->Clear();
+  LTRACE() << "Cleared engine";
+
+  const auto result = QString(outText).trimmed();
+
+  delete[] outText;
+  LTRACE() << "Cleared recognized text buffer";
 
   if (result.isEmpty())
     error_ = QObject::tr("Failed to recognize text or no text selected");
@@ -393,5 +297,5 @@ QString Tesseract::recognize(const QPixmap &source)
 
 bool Tesseract::isValid() const
 {
-  return engine_.get();
+  return api_.get();
 }

src/ocr/tesseract.h

@@ -8,12 +8,15 @@
 
 class QPixmap;
 class Task;
+namespace tesseract
+{
+class TessBaseAPI;
+}
 
 class Tesseract
 {
 public:
-  Tesseract(const LanguageId& language, const QString& tessdataPath,
+  Tesseract(const LanguageId& language, const QString& tessdataPath);
-            const QString& tesseractLibrary);
   ~Tesseract();
 
   QString recognize(const QPixmap& source);
@@ -23,10 +26,8 @@ public:
   static QStringList availableLanguageNames(const QString& path);
 
 private:
-  class Wrapper;
   void init(const LanguageId& language, const QString& tessdataPath);
 
-  const QString tesseractLibrary_;
+  std::unique_ptr<tesseract::TessBaseAPI> api_;
-  std::unique_ptr<Wrapper> engine_;
   QString error_;
 };

src/settings.cpp

@@ -35,7 +35,6 @@ const QString qs_showMessageOnStart = "showMessageOnStart";
 
 const QString qs_recogntionGroup = "Recognition";
 const QString qs_ocrLanguage = "language";
-const QString qs_tesseractVersion = "tesseractVersion";
 
 const QString qs_correctionGroup = "Correction";
 const QString qs_userSubstitutions = "userSubstitutions";
@@ -133,36 +132,6 @@ void cleanupOutdated(QSettings& settings)
   settings.endGroup();
 }
 
-#ifdef _MSC_VER
-#include <intrin.h>
-void cpuid(int leaf, int subleaf, std::array<uint, 4>& cpuinfo)
-{
-  __cpuidex(reinterpret_cast<int*>(cpuinfo.data()), leaf, subleaf);
-}
-#else
-#include <cpuid.h>
-void cpuid(int leaf, int subleaf, std::array<uint, 4>& cpuinfo)
-{
-  __get_cpuid_count(leaf, subleaf, &cpuinfo[0], &cpuinfo[1], &cpuinfo[2],
-                    &cpuinfo[3]);
-}
-#endif
-
-bool checkOptimizedTesseractSupport()
-{
-  std::array<uint, 4> cpuinfo{0};
-
-  cpuid(1, 0, cpuinfo);
-  const bool sse4_1 = cpuinfo[2] & (1 << 19);
-  const bool sse4_2 = cpuinfo[2] & (1 << 20);
-  const bool avx = cpuinfo[2] & (1 << 28);
-
-  cpuid(7, 0, cpuinfo);
-  const bool avx2 = cpuinfo[1] & (1 << 5);
-
-  return sse4_1 && sse4_2 && avx && avx2;
-}
-
 }  // namespace
 
 void Settings::save() const
@@ -207,7 +176,6 @@ void Settings::save() const
 
   settings.beginGroup(qs_recogntionGroup);
   settings.setValue(qs_ocrLanguage, sourceLanguage);
-  settings.setValue(qs_tesseractVersion, int(tesseractVersion));
   settings.endGroup();
 
   settings.beginGroup(qs_correctionGroup);
@@ -295,15 +263,6 @@ void Settings::load()
 
   settings.beginGroup(qs_recogntionGroup);
   sourceLanguage = settings.value(qs_ocrLanguage, sourceLanguage).toString();
-  if (!settings.contains(qs_tesseractVersion)) {
-    tesseractVersion = checkOptimizedTesseractSupport()
-                           ? TesseractVersion::Optimized
-                           : TesseractVersion::Compatible;
-  } else {
-    tesseractVersion = TesseractVersion(std::clamp(
-        settings.value(qs_tesseractVersion, int(tesseractVersion)).toInt(),
-        int(TesseractVersion::Optimized), int(TesseractVersion::Compatible)));
-  }
   settings.endGroup();
 
   settings.beginGroup(qs_correctionGroup);

src/settings.h

@@ -19,8 +19,6 @@ using Substitutions = std::multimap<LanguageId, Substitution>;
 
 enum class ProxyType { Disabled, System, Socks5, Http };
 
-enum class TesseractVersion { Optimized, Compatible };
-
 class Settings
 {
 public:
@@ -60,7 +58,6 @@ public:
 
   QString tessdataPath;
   QString sourceLanguage{"eng"};
-  TesseractVersion tesseractVersion{TesseractVersion::Optimized};
 
   bool doTranslation{true};
   bool ignoreSslErrors{false};

src/settingseditor.cpp

@@ -122,14 +122,6 @@ SettingsEditor::SettingsEditor(Manager &manager, update::Updater &updater)
 
   // recognition
   ui->tesseractLangCombo->setModel(models_.sourceLanguageModel());
-  const QMap<TesseractVersion, QString> tesseractVersions{
-      {TesseractVersion::Optimized, tr("Optimized")},
-      {TesseractVersion::Compatible, tr("Compatible")},
-  };
-  ui->tesseractVersion->addItems(tesseractVersions.values());
-  ui->tesseractVersion->setToolTip(
-      tr("Use compatible version if you are experiencing crashes during "
-         "recognition"));
 
   // correction
   ui->userSubstitutionsTable->setEnabled(ui->useUserSubstitutions->isChecked());
@@ -198,7 +190,8 @@ SettingsEditor::SettingsEditor(Manager &manager, update::Updater &updater)
             R"(<p>Optical character recognition (OCR) and translation tool</p>)"),
         QObject::tr(R"(<p>Version: %1</p>)")
             .arg(QApplication::applicationVersion()),
-        QObject::tr(R"(<p>Setup instructions: <a href="%1">%1</a></p>)").arg(help),
+        QObject::tr(R"(<p>Setup instructions: <a href="%1">%1</a></p>)")
+            .arg(help),
         QObject::tr(R"(<p>Changelog: <a href="%1">%2</a></p>)")
             .arg(changelog, QUrl(changelog).fileName()),
         QObject::tr(R"(<p>License: <a href="%3">MIT</a></p>)").arg(license),
@@ -267,8 +260,6 @@ Settings SettingsEditor::settings() const
 
   settings.sourceLanguage =
       LanguageCodes::idForName(ui->tesseractLangCombo->currentText());
-  settings.tesseractVersion =
-      TesseractVersion(ui->tesseractVersion->currentIndex());
 
   settings.useHunspell = ui->useHunspell->isChecked();
   settings.useUserSubstitutions = ui->useUserSubstitutions->isChecked();
@@ -328,7 +319,6 @@ void SettingsEditor::setSettings(const Settings &settings)
 
   ui->tesseractLangCombo->setCurrentText(
       LanguageCodes::name(settings.sourceLanguage));
-  ui->tesseractVersion->setCurrentIndex(int(settings.tesseractVersion));
 
   ui->useHunspell->setChecked(settings.useHunspell);
   ui->hunspellDir->setText(settings.hunspellPath);

src/settingseditor.ui

@@ -221,7 +221,7 @@
            </property>
            <layout class="QGridLayout" name="gridLayout">
             <item row="4" column="1">
-             <widget class="service::KeySequenceEdit" name="clipboardEdit"/>
+             <widget class="service::KeySequenceEdit" name="clipboardEdit" native="true"/>
             </item>
             <item row="2" column="0">
              <widget class="QLabel" name="label_8">
@@ -231,7 +231,7 @@
              </widget>
             </item>
             <item row="2" column="1">
-             <widget class="service::KeySequenceEdit" name="repeatCaptureEdit"/>
+             <widget class="service::KeySequenceEdit" name="repeatCaptureEdit" native="true"/>
             </item>
             <item row="3" column="0">
              <widget class="QLabel" name="label_3">
@@ -241,7 +241,7 @@
              </widget>
             </item>
             <item row="0" column="1">
-             <widget class="service::KeySequenceEdit" name="captureEdit"/>
+             <widget class="service::KeySequenceEdit" name="captureEdit" native="true"/>
             </item>
             <item row="4" column="0">
              <widget class="QLabel" name="label_7">
@@ -251,7 +251,7 @@
              </widget>
             </item>
             <item row="3" column="1">
-             <widget class="service::KeySequenceEdit" name="repeatEdit"/>
+             <widget class="service::KeySequenceEdit" name="repeatEdit" native="true"/>
             </item>
             <item row="0" column="0">
              <widget class="QLabel" name="label">
@@ -268,7 +268,7 @@
              </widget>
             </item>
             <item row="1" column="1">
-             <widget class="service::KeySequenceEdit" name="captureLockedEdit"/>
+             <widget class="service::KeySequenceEdit" name="captureLockedEdit" native="true"/>
             </item>
            </layout>
           </widget>
@@ -291,7 +291,23 @@
        </widget>
        <widget class="QWidget" name="pageRecognize">
         <layout class="QGridLayout" name="gridLayout_2">
-         <item row="3" column="2">
+         <item row="0" column="0">
+          <widget class="QLabel" name="label_2">
+           <property name="sizePolicy">
+            <sizepolicy hsizetype="Maximum" vsizetype="Preferred">
+             <horstretch>0</horstretch>
+             <verstretch>0</verstretch>
+            </sizepolicy>
+           </property>
+           <property name="text">
+            <string>Tessdata path:</string>
+           </property>
+          </widget>
+         </item>
+         <item row="1" column="2">
+          <widget class="QComboBox" name="tesseractLangCombo"/>
+         </item>
+         <item row="2" column="2">
           <spacer name="verticalSpacer_2">
            <property name="orientation">
             <enum>Qt::Vertical</enum>
@@ -317,19 +333,6 @@
            </property>
           </widget>
          </item>
-         <item row="0" column="0">
-          <widget class="QLabel" name="label_2">
-           <property name="sizePolicy">
-            <sizepolicy hsizetype="Maximum" vsizetype="Preferred">
-             <horstretch>0</horstretch>
-             <verstretch>0</verstretch>
-            </sizepolicy>
-           </property>
-           <property name="text">
-            <string>Tessdata path:</string>
-           </property>
-          </widget>
-         </item>
          <item row="1" column="0">
           <widget class="QLabel" name="label_4">
            <property name="sizePolicy">
@@ -346,19 +349,6 @@
            </property>
           </widget>
          </item>
-         <item row="1" column="2">
-          <widget class="QComboBox" name="tesseractLangCombo"/>
-         </item>
-         <item row="2" column="0">
-          <widget class="QLabel" name="label_24">
-           <property name="text">
-            <string>Library version</string>
-           </property>
-          </widget>
-         </item>
-         <item row="2" column="2">
-          <widget class="QComboBox" name="tesseractVersion"/>
-         </item>
         </layout>
        </widget>
        <widget class="QWidget" name="pageCorrect">