PyMuPDF 1.24.13__tar.gz → 1.25.0__tar.gz

{PyMuPDF-1.24.13 → pymupdf-1.25.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PyMuPDF
-Version: 1.24.13
+Version: 1.25.0
 Summary: A high performance Python library for data extraction, analysis, conversion & manipulation of PDF (and other) documents.
 Description-Content-Type: text/markdown
 Author: Artifex

{PyMuPDF-1.24.13 → pymupdf-1.25.0}/changes.txt

@@ -2,6 +2,39 @@ Change Log
 ==========
 
 
+**Changes in version 1.25.0 (2024-12-05)**
+
+* Use MuPDF-1.25.1.
+
+* Fixed issues:
+
+  * **Fixed** `4026 <https://github.com/pymupdf/PyMuPDF/issues/4026>`_: page.get_text('blocks') output two piece of very similar text with different bbox
+  * **Fixed** `4004 <https://github.com/pymupdf/PyMuPDF/issues/4004>`_: Segmentation Fault When Updating PDF Form Field Value
+  * **Fixed** `3751 <https://github.com/pymupdf/PyMuPDF/issues/3751>`_: apply_redactions causes part of the page content to be hidden / transparent
+
+
+**Changes in version 1.24.14 (2024-11-19)**
+
+* Use MuPDF-1.24.11.
+
+* Fixed issues:
+
+  * **Fixed** `3448 <https://github.com/pymupdf/PyMuPDF/issues/3448>`_: get_pixmap function removes the table and leaves just the content behind
+  * **Fixed** `3758 <https://github.com/pymupdf/PyMuPDF/issues/3758>`_: Got "malloc(): unaligned tcache chunk detected Aborted (core dumped)" while using add_redact_annot/apply_redactions
+  * **Fixed** `3813 <https://github.com/pymupdf/PyMuPDF/issues/3813>`_: Stories: Ordered list count broken with nested unordered list
+  * **Fixed** `3933 <https://github.com/pymupdf/PyMuPDF/issues/3933>`_: font.valid_codepoints() - malfunction
+  * **Fixed** `4018 <https://github.com/pymupdf/PyMuPDF/issues/4018>`_: PyMuPDF hangs when iterating over zero page PDF pages backwards
+  * **Fixed** `4043 <https://github.com/pymupdf/PyMuPDF/issues/4043>`_: fullcopypage bug
+  * **Fixed** `4047 <https://github.com/pymupdf/PyMuPDF/issues/4047>`_: Segmentation Fault in add_redact_annot
+  * **Fixed** `4050 <https://github.com/pymupdf/PyMuPDF/issues/4050>`_: Content of dict returned by doc.embfile_info() does not fit to documentation
+
+* Other:
+
+  * Ensure that words from `Page.get_text()` never contain RTL/LTR char mixtures.
+  * Fix building with system MuPDF.
+  * Add dot product for points and vectors.
+
+
 **Changes in version 1.24.13 (2024-10-29)**
 
 * Fixed issues:

{PyMuPDF-1.24.13 → pymupdf-1.25.0}/pipcl.py

@@ -364,7 +364,7 @@ class Package:
                 * https://pypi.org/classifiers/
 
             requires_dist:
-                A string or list of strings. Also see PEP-508.
+                A string or list of strings. None items are ignored. Also see PEP-508.
             requires_python:
                 A string or list of strings.
             requires_external:
@@ -489,6 +489,7 @@ class Package:
                 Used as `zipfile.ZipFile()`'s `compresslevel` parameter when
                 creating wheels.
 
+        Occurrences of `None` in lists are ignored.
         '''
         assert name
         assert version
@@ -689,7 +690,7 @@ class Package:
             else:
                 items = self.fn_sdist()
 
-        prefix = f'{self.name}-{self.version}'
+        prefix = f'{_normalise(self.name)}-{self.version}'
         os.makedirs(sdist_directory, exist_ok=True)
         tarpath = f'{sdist_directory}/{prefix}.tar.gz'
         log2(f'Creating sdist: {tarpath}')
@@ -822,7 +823,7 @@ class Package:
         return ret
 
     def wheel_name(self):
-        return f'{self.name}-{self.version}-{self.tag_python()}-{self.tag_abi()}-{self.tag_platform()}.whl'
+        return f'{_normalise(self.name)}-{self.version}-{self.tag_python()}-{self.tag_abi()}-{self.tag_platform()}.whl'
 
     def wheel_name_match(self, wheel):
         '''
@@ -851,7 +852,7 @@ class Package:
                 log2(f'py_limited_api; {tag_python=} compatible with {self.tag_python()=}.')
                 py_limited_api_compatible = True
             
-        log2(f'{self.name == name=}')
+        log2(f'{_normalise(self.name) == name=}')
         log2(f'{self.version == version=}')
         log2(f'{self.tag_python() == tag_python=} {self.tag_python()=} {tag_python=}')
         log2(f'{py_limited_api_compatible=}')
@@ -860,7 +861,7 @@ class Package:
         log2(f'{self.tag_platform()=}')
         log2(f'{tag_platform.split(".")=}')
         ret = (1
-                and self.name == name
+                and _normalise(self.name) == name
                 and self.version == version
                 and (self.tag_python() == tag_python or py_limited_api_compatible)
                 and self.tag_abi() == tag_abi
@@ -1244,7 +1245,7 @@ class Package:
             )
 
     def _dist_info_dir( self):
-        return f'{self.name}-{self.version}.dist-info'
+        return f'{_normalise(self.name)}-{self.version}.dist-info'
 
     def _metainfo(self):
         '''
@@ -1261,7 +1262,8 @@ class Package:
                 return
             if isinstance( value, (tuple, list)):
                 for v in value:
-                    add( key, v)
+                    if v is not None:
+                        add( key, v)
                 return
             if key == 'License' and '\n' in value:
                 # This is ok because we write `self.license` into
@@ -1899,7 +1901,16 @@ def git_items( directory, submodules=False):
     return ret
 
 
-def run( command, capture=False, check=1, verbose=1, env_extra=None, caller=1):
+def run(
+        command,
+        *,
+        capture=False,
+        check=1,
+        verbose=1,
+        env_extra=None,
+        timeout=None,
+        caller=1,
+        ):
     '''
     Runs a command using `subprocess.run()`.
 
@@ -1923,6 +1934,10 @@ def run( command, capture=False, check=1, verbose=1, env_extra=None, caller=1):
             If true we show the command.
         env_extra:
             None or dict to add to environ.
+        timeout:
+            If not None, timeout in seconds; passed directly to
+            subprocess.run(). Note that on MacOS subprocess.run() seems to
+            leave processes running if timeout expires.
     Returns:
         check capture   Return
         --------------------------
@@ -1949,6 +1964,7 @@ def run( command, capture=False, check=1, verbose=1, env_extra=None, caller=1):
             check=check,
             encoding='utf8',
             env=env,
+            timeout=timeout,
             )
     if check:
         return cp.stdout if capture else None
@@ -2357,6 +2373,11 @@ def _fs_mtime( filename, default=0):
         return default
 
 
+def _normalise(name):
+    # https://packaging.python.org/en/latest/specifications/name-normalization/#name-normalization
+    return re.sub(r"[-_.]+", "-", name).lower()
+
+
 def _assert_version_pep_440(version):
     assert re.match(
                 r'^([1-9][0-9]*!)?(0|[1-9][0-9]*)(\.(0|[1-9][0-9]*))*((a|b|rc)(0|[1-9][0-9]*))?(\.post(0|[1-9][0-9]*))?(\.dev(0|[1-9][0-9]*))?$',

{PyMuPDF-1.24.13 → pymupdf-1.25.0}/scripts/gh_release.py

@@ -89,6 +89,14 @@ import test as test_py
 
 pymupdf_dir = os.path.abspath( f'{__file__}/../..')
 
+sys.path.insert(0, pymupdf_dir)
+import pipcl
+del sys.path[0]
+
+log = pipcl.log0
+run = pipcl.run
+
+
 def main():
 
     log( '### main():')
@@ -223,7 +231,7 @@ def build( platform_=None, valgrind=False):
     if platform.system() == 'Linux' and inputs_wheels_linux_pyodide:
         # Pyodide wheels are built by running scripts/test.py, not
         # cibuildwheel.
-        command = f'{sys.executable} scripts/test.py'
+        command = f'{sys.executable} scripts/test.py -P 1'
         if inputs_PYMUPDF_SETUP_MUPDF_BUILD:
             command += f' -m {shlex.quote(inputs_PYMUPDF_SETUP_MUPDF_BUILD)}'
         command += ' pyodide_wheel'
@@ -237,14 +245,14 @@ def build( platform_=None, valgrind=False):
             os.chdir( pymupdf_dir)
         # Create PyMuPDF sdist.
         run(f'{sys.executable} setup.py sdist')
-        assert glob.glob('dist/PyMuPDF-*.tar.gz')
+        assert glob.glob('dist/pymupdf-*.tar.gz')
         if inputs_flavours:
             # Create PyMuPDFb sdist.
             run(
                     f'{sys.executable} setup.py sdist',
                     env_extra=dict(PYMUPDF_SETUP_FLAVOUR='b'),
                     )
-            assert glob.glob('dist/PyMuPDFb-*.tar.gz')
+            assert glob.glob('dist/pymupdfb-*.tar.gz')
     
     # Build wheels.
     #
@@ -284,11 +292,11 @@ def build( platform_=None, valgrind=False):
                     ret.append(item)
             return ' '.join(ret)
     
-        cps = inputs_wheels_cps if inputs_wheels_cps else 'cp39* cp310* cp311* cp312* cp313'
+        cps = inputs_wheels_cps if inputs_wheels_cps else 'cp39* cp310* cp311* cp312* cp313*'
         set_if_unset( 'CIBW_BUILD', cps)
         for cp in cps.split():
             m = re.match('cp([0-9]+)[*]', cp)
-            assert m
+            assert m, f'{cps=} {cp=}'
             v = int(m.group(1))
             if v == 314:
                 # Need to set CIBW_PRERELEASE_PYTHONS, otherwise cibuildwheel
@@ -351,6 +359,8 @@ def build( platform_=None, valgrind=False):
             if pass_:
                 env_pass(name)
 
+        env_pass('PYMUPDF_SETUP_PY_LIMITED_API')
+        
         if os.environ.get('PYMUPDF_SETUP_LIBCLANG'):
             env_pass('PYMUPDF_SETUP_LIBCLANG')
     
@@ -389,7 +399,8 @@ def build( platform_=None, valgrind=False):
             env_pass('PYMUPDF_SETUP_PY_LIMITED_API')
             CIBW_BUILD_old = env_extra.get('CIBW_BUILD')
             assert CIBW_BUILD_old is not None
-            env_set('CIBW_BUILD', 'cp39*')
+            cp = cps.split()[0]
+            env_set('CIBW_BUILD', cp)
             log(f'Building single wheel.')
             run( f'cibuildwheel{platform_arg}', env_extra=env_extra)
             
@@ -413,7 +424,16 @@ def build( platform_=None, valgrind=False):
             #
             env_set('CIBW_REPAIR_WHEEL_COMMAND', '')
             
-            log(f'Testing on all python versions using wheels in wheelhouse/.')
+            if platform.system() == 'Linux' and env_extra.get('CIBW_ARCHS_LINUX') == 'aarch64':
+                log(f'Testing all Python versions on linux-aarch64 is too slow and is killed by github after 6h.')
+                log(f'Testing on restricted python versions using wheels in wheelhouse/.')
+                # Testing only on first and last python versions.
+                cp1 = cps.split()[0]
+                cp2 = cps.split()[-1]
+                cp = cp1 if cp1 == cp2 else f'{cp1} {cp2}'
+                env_set('CIBW_BUILD', cp)
+            else:
+                log(f'Testing on all python versions using wheels in wheelhouse/.')
             run( f'cibuildwheel{platform_arg}', env_extra=env_extra)
             
         elif inputs_flavours:
@@ -455,7 +475,7 @@ def build( platform_=None, valgrind=False):
             # careful to avoid incompatible wheels, e.g. 32 vs 64-bit wheels
             # coexist during Windows builds.
             #
-            env_set('CIBW_BEFORE_TEST', f'python scripts/gh_release.py pip_install wheelhouse/PyMuPDFb')
+            env_set('CIBW_BEFORE_TEST', f'python scripts/gh_release.py pip_install wheelhouse/pymupdfb')
         
             set_cibuild_test()
         
@@ -576,49 +596,6 @@ else:
         return os.path.relpath(path, start)
 
 
-def log(text, caller=0):
-    '''
-    Writes `text` to stdout with prefix showing caller path relative to
-    pymupdf_dir and fn name.
-    '''
-    frame_record = inspect.stack( context=0)[ caller+1]
-    filename    = frame_record.filename
-    line        = frame_record.lineno
-    function    = frame_record.function
-    prefix = f'{relpath(filename, pymupdf_dir)}:{line}:{function}(): '
-    print(textwrap.indent(text, prefix), flush=1)
-
-
-def run(command, env_extra=None, check=1, timeout=None):
-    '''
-    Runs a command using subprocess.run().
-    Args:
-        command:
-            The command to run.
-        env_extra:
-            None or dict containing extra environment variable settings to add
-            to os.environ.
-        check:
-            Whether to raise exception if command fails.
-        timeout:
-            If not None, timeout in seconds; passed directory to
-            subprocess.run(). Note that on MacOS subprocess.run() seems to
-            leave processes running if timeout expires.
-    '''
-    env = None
-    message = 'Running: '
-    if env_extra:
-        env = os.environ.copy()
-        env.update(env_extra)
-        message += '\n[environment:\n'
-        for n, v in env_extra.items():
-            message += f'    {n}={shlex.quote(v)}\n'
-        message += ']\n'
-    message += f'{command}'
-    log(message, caller=1)
-    return subprocess.run(command, check=check, shell=1, env=env, timeout=timeout)
-
-
 def platform_tag():
     bits = cpu_bits()
     if platform.system() == 'Windows':

{PyMuPDF-1.24.13 → pymupdf-1.25.0}/scripts/sysinstall.py

@@ -49,7 +49,7 @@ Args:
     --pymupdf-do 0|1
         Whether to build and install pymupdf.
     --root <root>
-        Root of install directory; default is `/`.
+        Root of install directory; default is 'pymupdf-sysinstall-test-root'.
     --tesseract5 0|1
         If 1 (the default), we force installation of libtesseract-dev version
         5 (which is not available as a default package in Ubuntu-22.04) from
@@ -86,6 +86,14 @@ import sysconfig
 
 import test as test_py
 
+pymupdf_dir = os.path.abspath( f'{__file__}/../..')
+
+sys.path.insert(0, pymupdf_dir)
+import pipcl
+del sys.path[0]
+
+log = pipcl.log0
+
 # Requirements for a system build and install:
 #
 # system packages (Debian names):
@@ -106,12 +114,12 @@ g_sys_packages = [
 def main():
     
     if 1:
-        print(f'## {__file__}: Starting.')
-        print(f'{sys.executable=}')
-        print(f'{platform.python_version()=}')
-        print(f'{__file__=}')
-        print(f'{sys.argv=}')
-        print(f'{sysconfig.get_path("platlib")=}')
+        log(f'## {__file__}: Starting.')
+        log(f'{sys.executable=}')
+        log(f'{platform.python_version()=}')
+        log(f'{__file__=}')
+        log(f'{sys.argv=}')
+        log(f'{sysconfig.get_path("platlib")=}')
         run_command(f'python -V', check=0)
         run_command(f'python3 -V', check=0)
         run_command(f'sudo python -V', check=0)
@@ -132,8 +140,7 @@ def main():
     packages = True
     prefix = '/usr/local'
     pymupdf_do = True
-    pymupdf_dir = os.path.abspath( f'{__file__}/../..')
-    root = 'sysinstall_test'
+    root = 'pymupdf-sysinstall-test-root'
     tesseract5 = True
     pytest_args = None
     pytest_do = True
@@ -152,7 +159,7 @@ def main():
         except StopIteration:
             break
         if arg in ('-h', '--help'):
-            print(__doc__)
+            log(__doc__)
             return
         elif arg == '--mupdf-do':       mupdf_do = int(next(args))
         elif arg == '--mupdf-dir':      mupdf_dir = next(args)
@@ -161,7 +168,6 @@ def main():
         elif arg == '--packages':       packages = int(next(args))
         elif arg == '--prefix':         prefix = next(args)
         elif arg == '--pymupdf-do':     pymupdf_do = int(next(args))
-        elif arg == '--pymupdf-dir':    pymupdf_dir = next(args)
         elif arg == '--root':           root = next(args)
         elif arg == '--tesseract5':     tesseract5 = int(next(args))
         elif arg == '--pytest-do':      pytest_do = int(next(args))
@@ -184,30 +190,30 @@ def main():
     sudo = ''
     if root == '/':
         sudo = f'sudo PATH={os.environ["PATH"]} '
-    def run(command):
-        return run_command(command, doit=mupdf_do)
+    def run(command, env_extra=None):
+        return run_command(command, doit=mupdf_do, env_extra=env_extra)
     # Get MuPDF from git if specified.
     #
     if mupdf_git:
         # Update existing checkout or do `git clone`.
         if os.path.exists(mupdf_dir):
-            print(f'## Update MuPDF checkout {mupdf_dir}.')
+            log(f'## Update MuPDF checkout {mupdf_dir}.')
             run(f'cd {mupdf_dir} && git pull && git submodule update --init')
         else:
             # No existing git checkout, so do a fresh clone.
-            print(f'## Clone MuPDF into {mupdf_dir}.')
+            log(f'## Clone MuPDF into {mupdf_dir}.')
             run(f'git clone --recursive --depth 1 --shallow-submodules {mupdf_git} {mupdf_dir}')
     
     if packages:
         # Install required system packages. We assume a Debian package system.
         #
-        print('## Install system packages required by MuPDF.')
+        log('## Install system packages required by MuPDF.')
         run(f'sudo apt update')
         run(f'sudo apt install {" ".join(g_sys_packages)}')
         # Ubuntu-22.04 has freeglut3-dev, not libglut-dev.
         run(f'sudo apt install libglut-dev | sudo apt install freeglut3-dev')
         if tesseract5:
-            print(f'## Force installation of libtesseract-dev version 5.')
+            log(f'## Force installation of libtesseract-dev version 5.')
             # https://stackoverflow.com/questions/76834972/how-can-i-run-pytesseract-python-library-in-ubuntu-22-04
             #
             run('sudo apt install -y software-properties-common')
@@ -220,12 +226,12 @@ def main():
     # Build+install MuPDF. We use mupd:Makefile's install-shared-python target.
     #
     if pip == 'sudo':
-        print('## Installing Python packages required for building MuPDF and PyMuPDF.')
+        log('## Installing Python packages required for building MuPDF and PyMuPDF.')
         run(f'sudo pip install --upgrade pip')
         names = test_py.wrap_get_requires_for_build_wheel(f'{__file__}/../..')
         run(f'sudo pip install {names}')
     
-    print('## Build and install MuPDF.')
+    log('## Build and install MuPDF.')
     command = f'cd {mupdf_dir}'
     command += f' && {sudo}make'
     command += f' -j {multiprocessing.cpu_count()}'
@@ -246,10 +252,10 @@ def main():
     
     # Build+install PyMuPDF.
     #
-    print('## Build and install PyMuPDF.')
+    log('## Build and install PyMuPDF.')
     def run(command):
         return run_command(command, doit=pymupdf_do)
-    flags_freetype2 = run_command('pkg-config --cflags freetype2', capture_output=1).stdout.strip()
+    flags_freetype2 = run_command('pkg-config --cflags freetype2', capture=1)
     compile_flags = f'-I {root_prefix}/include {flags_freetype2}'
     link_flags = f'-L {root_prefix}/lib'
     env = ''
@@ -258,7 +264,7 @@ def main():
     env += f'LDFLAGS="-L {root}/{prefix}/lib" '
     env += f'PYMUPDF_SETUP_MUPDF_BUILD= '       # Use system MuPDF.
     if use_installer:
-        print(f'## Building wheel.')
+        log(f'## Building wheel.')
         if pip == 'venv':
             venv_name = 'venv-pymupdf-sysinstall'
         run(f'pwd')
@@ -277,7 +283,7 @@ def main():
         wheel = glob.glob(f'dist/*')
         assert len(wheel) == 1, f'{wheel=}'
         wheel = wheel[0]
-        print(f'## Installing wheel using `installer`.')
+        log(f'## Installing wheel using `installer`.')
         pv = '.'.join(platform.python_version_tuple()[:2])
         p = f'{root_prefix}/lib/python{pv}'
         # `python -m installer` fails to overwrite existing files.
@@ -315,13 +321,10 @@ def main():
                     if leaf in dirnames:
                         pythonpath.append(os.path.join(dirpath, leaf))
         pythonpath = ':'.join(pythonpath)
-        print(f'{pythonpath=}')
+        log(f'{pythonpath=}')
     else:
         command = f'{env} pip install -vv --root {root} {os.path.abspath(pymupdf_dir)}'
         run( command)
-        sys.path.insert(0, pymupdf_dir)
-        import pipcl
-        del sys.path[0]
         pythonpath = pipcl.install_dir(root)
     
     # Show contents of installation directory. This is very slow on github,
@@ -330,9 +333,9 @@ def main():
         
     # Run pytest tests.
     #
-    print('## Run PyMuPDF pytest tests.')
-    def run(command):
-        return run_command(command, doit=pytest_do)
+    log('## Run PyMuPDF pytest tests.')
+    def run(command, env_extra=None):
+        return run_command(command, doit=pytest_do, env_extra=env_extra)
     import gh_release
     if pip == 'venv':
         # Create venv.
@@ -384,16 +387,14 @@ def main():
         command += f' -p {shlex.quote(pytest_args)}'
     if pytest_do:
         command += ' test'
-    run(command)
+    run(command, env_extra=dict(PYMUPDF_SYSINSTALL_TEST='1'))
 
 
-def run_command(command, capture_output=False, check=True, doit=True):
+def run_command(command, capture=False, check=True, doit=True, env_extra=None):
     if doit:
-        print(f'## Running: {command}')
-        sys.stdout.flush()
-        return subprocess.run(command, shell=1, check=check, text=1, capture_output=capture_output)
+        return pipcl.run(command, capture=capture, check=check, caller=2, env_extra=env_extra)
     else:
-        print(f'## Would have run: {command}')
+        log(f'## Would have run: {command}', caller=2)
 
 
 if __name__ == '__main__':