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 @@
 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(dependencies_dir + '/lib/libtesseract-*.so')
+    glob('/usr/lib/x86_64-linux-gnu/nss/*')
 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 @@
 c.print('>> Installing tesseract')
 
 install_dir = dependencies_dir
-url = 'https://github.com/tesseract-ocr/tesseract/archive/5.1.0.tar.gz'
-required_version = '5.1.0'
+required_version = '5.2.0'
+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
-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 = install_dir + '/tesseract.cache'
 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)
-        orig_lib = install_dir + '/bin/tesseract51.dll'
+        file_name_ver = required_version[0] + required_version[2]
+        dll = install_dir + '/bin/tesseract{}.dll'.format(file_name_ver)
+        lib = install_dir + '/lib/tesseract{}.lib'.format(file_name_ver)
+        if not os.path.exists(dll) or not os.path.exists(lib):
+            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(lib_suffix)
-        orig_lib = install_dir + '/lib/libtesseract.{}.dylib'.format(required_version)
+        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(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
+        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 False
+    return True
 
 
 if check_existing() and not 'FORCE' in os.environ:
@@ -71,60 +59,24 @@ def check_existing():
 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" \
--DBUILD_TRAINING_TOOLS=OFF -DBUILD_TESTS=OFF -DBUILD_SHARED_LIBS=ON -DSW_BUILD=OFF \
+cmake_args = '"{0}" \
+-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 @@ def check_existing():
 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')