PyMuPDF 1.24.6__tar.gz → 1.24.8__tar.gz

{PyMuPDF-1.24.6 → PyMuPDF-1.24.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PyMuPDF
-Version: 1.24.6
+Version: 1.24.8
 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-1.24.6/READMErb.md → PyMuPDF-1.24.8/READMEb.md

@@ -1,4 +1,4 @@
-# PyMuPDF 1.24.6
+# PyMuPDFb
 
 This wheel contains [MuPDF](https://mupdf.readthedocs.io/) shared libraries for
 use by [PyMuPDF](https://pymupdf.readthedocs.io/).

PyMuPDF-1.24.8/READMEd.md

@@ -0,0 +1,4 @@
+# PyMuPDFd
+
+This wheel contains [MuPDF](https://mupdf.readthedocs.io/) build-time files
+that were used to build [PyMuPDF](https://pymupdf.readthedocs.io/).

{PyMuPDF-1.24.6 → PyMuPDF-1.24.8}/changes.txt

@@ -2,6 +2,30 @@ Change Log
 ==========
 
 
+**Changes in version 1.24.8 (2024-07-22)**
+
+* Fixed issues:
+
+  * **Fixed** `3636 <https://github.com/pymupdf/PyMuPDF/issues/3636>`_: API documentation for the open function is not obvious to find.
+  * **Fixed** `3654 <https://github.com/pymupdf/PyMuPDF/issues/3654>`_: docx parsing was broken in 1.24.7
+  * **Fixed** `3677 <https://github.com/pymupdf/PyMuPDF/issues/3677>`_: Unable to extract subset font name using the newer versions of PyMuPDF : 1.24.6 and 1.24.7.
+  * **Fixed** `3687 <https://github.com/pymupdf/PyMuPDF/issues/3687>`_: Page.get_text results in AssertionError for epub files
+
+Other:
+
+  * Fixed various spelling mistakes spotted by codespell.
+  * Improved how we modify MuPDF's default configuration on Windows.
+  * Make text search to work with ligatures.
+
+
+**Changes in version 1.24.7 (2024-06-26)**
+
+* Fixed issues:
+
+  * **Fixed** `3615 <https://github.com/pymupdf/PyMuPDF/issues/3615>`_: Document.pagemode or Document.pagelayout crashes for epub files
+  * **Fixed** `3616 <https://github.com/pymupdf/PyMuPDF/issues/3616>`_: not last version reported
+
+
 **Changes in version 1.24.6 (2024-06-25)**
 
 * Fixed issues:

{PyMuPDF-1.24.6 → PyMuPDF-1.24.8}/pipcl.py

@@ -516,13 +516,7 @@ class Package:
         assert re.match('([A-Z0-9]|[A-Z0-9][A-Z0-9._-]*[A-Z0-9])$', name, re.IGNORECASE), \
                 f'Bad name: {name!r}'
 
-
-        # PEP-440.
-        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]*))?$',
-                    version,
-                ), \
-                f'Bad version: {version!r}.'
+        _assert_version_pep_440(version)
 
         # https://packaging.python.org/en/latest/specifications/binary-distribution-format/
         if tag_python:
@@ -614,7 +608,10 @@ class Package:
             # pypi.org.
             #
             import setuptools
-            tag_platform = setuptools.distutils.util.get_platform().replace('-', '_').replace('.', '_')
+            tag_platform1 = setuptools.distutils.util.get_platform()
+            tag_platform2 = sysconfig.get_platform()
+            assert tag_platform1 == tag_platform2, f'{tag_platform1=} != {tag_platform2=}'
+            tag_platform = tag_platform2.replace('-', '_').replace('.', '_').lower()
 
             # We need to patch things on MacOS.
             #
@@ -1471,7 +1468,7 @@ def build_extension(
         debug2 = ''
         if debug:
             debug2 = '/Zi'  # Generate .pdb.
-            # debug2 = '/Z7'    # Embded debug info in .obj files.
+            # debug2 = '/Z7'    # Embed debug info in .obj files.
 
         # As of 2023-08-23, it looks like VS tools create slightly
         # .dll's each time, even with identical inputs.
@@ -1493,7 +1490,7 @@ def build_extension(
 
                     # Input/output files:
                     {T}{path_cpp}               # /Tp specifies C++ source file.
-                    /Fo{path_obj}               # Output file.
+                    /Fo{path_obj}               # Output file. codespell:ignore
 
                     # Include paths:
                     {includes_text}
@@ -1786,7 +1783,7 @@ def git_items( directory, submodules=False):
     ret = []
     for path in text.decode('utf8').strip().split( '\n'):
         path2 = os.path.join(directory, path)
-        # Sometimes git ls-files seems to list empty/non-existant directories
+        # Sometimes git ls-files seems to list empty/non-existent directories
         # within submodules.
         #
         if not os.path.exists(path2):
@@ -1832,7 +1829,7 @@ def run( command, capture=False, check=1, verbose=1):
     nl = '\n'
     if verbose:
         log1( f'Running: {nl.join(lines)}')
-    sep = ' ' if windows() else '\\\n'
+    sep = ' ' if windows() else ' \\\n'
     command2 = sep.join( lines)
     cp = subprocess.run(
             command2,
@@ -1884,8 +1881,8 @@ class PythonFlags:
 
         if windows():
             wp = wdev.WindowsPython()
-            self.includes = f'/I"{wp.root}\\include"'
-            self.ldflags = f'/LIBPATH:"{wp.root}\\libs"'
+            self.includes = f'/I"{wp.include}"'
+            self.ldflags = f'/LIBPATH:"{wp.libs}"'
 
         elif pyodide():
             _include_dir = os.environ[ 'PYO3_CROSS_INCLUDE_DIR']
@@ -2113,7 +2110,7 @@ def run_if( command, out, *prerequisites):
     '''
     doit = False
     cmd_path = f'{out}.cmd'
-    
+
     if not doit:
         out_mtime = _fs_mtime( out)
         if out_mtime == 0:
@@ -2247,6 +2244,15 @@ def _fs_mtime( filename, default=0):
     except OSError:
         return default
 
+
+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]*))?$',
+                version,
+            ), \
+            f'Bad version: {version!r}.'
+
+
 g_verbose = int(os.environ.get('PIPCL_VERBOSE', '1'))
 
 def verbose(level=None):

{PyMuPDF-1.24.6 → PyMuPDF-1.24.8}/scripts/gh_release.py

@@ -59,6 +59,10 @@ action inputs, which can't be easily translated into command-line arguments.
 
 Building for Pyodide
 
+    If GITHUB_EVENT_NAME is 'schedule' we assume we are running as a scheduled
+    Github workflow, and if inputs_PYMUPDF_SETUP_MUPDF_BUILD is unset we change
+    it to use MuPDF master branch.
+    
     If `inputs_wheels_linux_pyodide` is true and we are on Linux, we clone
     `emsdk.git`, set it up, and run `pyodide build`. This runs our setup.py
     with CC etc set up to create Pyodide binaries in a wheel called, for
@@ -73,6 +77,7 @@ import glob
 import inspect
 import os
 import platform
+import re
 import shlex
 import subprocess
 import sys
@@ -208,6 +213,14 @@ def build( platform_=None, valgrind=False):
     # Build Pyodide wheel if specified.
     #
     if platform.system() == 'Linux' and inputs_wheels_linux_pyodide:
+        # In scheduled runs (by .github/workflows/test_pyodide.yml), use MuPDF
+        # master.
+        GITHUB_EVENT_NAME = os.getenv('GITHUB_EVENT_NAME')
+        if GITHUB_EVENT_NAME == 'schedule':
+            if inputs_PYMUPDF_SETUP_MUPDF_BUILD in ('', None):
+                log(f'Overriding inputs_PYMUPDF_SETUP_MUPDF_BUILD because {GITHUB_EVENT_NAME=} {inputs_PYMUPDF_SETUP_MUPDF_BUILD=}.')
+                inputs_PYMUPDF_SETUP_MUPDF_BUILD = 'git:--branch master https://github.com/ArtifexSoftware/mupdf.git'
+                log(f'{inputs_PYMUPDF_SETUP_MUPDF_BUILD=}')
         build_pyodide_wheel(inputs_wheels_implementations, inputs_PYMUPDF_SETUP_MUPDF_BUILD)
     
     # Build sdist(s).
@@ -250,7 +263,14 @@ def build( platform_=None, valgrind=False):
         # external symbol PyUnicode_DecodeRawUnicodeEscape`.
         # 2024-06-05: musllinux on aarch64 fails because libclang cannot find
         # libclang.so.
-        set_if_unset( 'CIBW_SKIP', '"pp* *i686 cp36* cp37* *musllinux*aarch64*"')
+        #
+        # 2024-07-09: exclude Windows 32-bit Python-3.13 because when running
+        # under 32-bit Python-3.13, there is no 64-bit Python-3.13 available
+        # via `py -3.13`, which we need because libclang fails on 32-bit
+        # Windows. Hopefully this will work when Python-3.13 is officially
+        # released.
+        #
+        set_if_unset( 'CIBW_SKIP', 'pp* *i686 cp36* cp37* *musllinux*aarch64* cp313-win32')
     
         def make_string(*items):
             ret = list()
@@ -261,6 +281,15 @@ def build( platform_=None, valgrind=False):
     
         cps = inputs_wheels_cps if inputs_wheels_cps else 'cp38* cp39* cp310* cp311* cp312*'
         set_if_unset( 'CIBW_BUILD', cps)
+        for cp in cps.split():
+            m = re.match('cp([0-9]+)[*]', cp)
+            assert m
+            v = int(m.group(1))
+            if v == 313:
+                # Need to set CIBW_PRERELEASE_PYTHONS, otherwise cibuildwheel
+                # will refuse.
+                log(f'Setting CIBW_PRERELEASE_PYTHONS for Python version {cp=}.')
+                set_if_unset( 'CIBW_PRERELEASE_PYTHONS', '1')
     
         if platform.system() == 'Linux':
             set_if_unset(
@@ -346,6 +375,9 @@ def build( platform_=None, valgrind=False):
     
         run('pip install cibuildwheel')
     
+        # We include MuPDF build-time files on certain platforms.
+        flavour_d = platform.system() in ('Linux', 'Windows') and cpu_bits() == 64
+        
         if inputs_flavours:
             # Build and test PyMuPDF and PyMuPDFb wheels.
             #
@@ -353,9 +385,13 @@ def build( platform_=None, valgrind=False):
             # First build PyMuPDFb wheel. cibuildwheel will build a single wheel
             # here, which will work with any python version on current OS.
             #
-            env_set( 'PYMUPDF_SETUP_FLAVOUR', 'b', pass_=1)
+            flavour = 'b'
+            if flavour_d:
+                # Include MuPDF build-time files.
+                flavour += 'd'
+            env_set( 'PYMUPDF_SETUP_FLAVOUR', flavour, pass_=1)
             run( f'cibuildwheel{platform_arg}', env_extra)
-            run( 'echo after flavour=b')
+            run( 'echo after {flavour=}')
             run( 'ls -l wheelhouse')
 
             # Now set environment to build PyMuPDF wheels. cibuildwheel will build
@@ -385,15 +421,21 @@ def build( platform_=None, valgrind=False):
         
             set_cibuild_test()
         
-            env_set( 'PYMUPDF_SETUP_FLAVOUR', 'p', pass_=1)
+            # Build main PyMuPDF wheel.
+            flavour = 'p'
+            env_set( 'PYMUPDF_SETUP_FLAVOUR', flavour, pass_=1)
+            run( f'cibuildwheel{platform_arg}', env_extra=env_extra)
         
         else:
             # Build and test wheels which contain everything.
             #
+            flavour = 'pb'
+            if flavour_d:
+                flavour += 'd'
             set_cibuild_test()
-            env_set( 'PYMUPDF_SETUP_FLAVOUR', 'pb', pass_=1)
+            env_set( 'PYMUPDF_SETUP_FLAVOUR', flavour, pass_=1)
     
-        run( f'cibuildwheel{platform_arg}', env_extra=env_extra)
+            run( f'cibuildwheel{platform_arg}', env_extra=env_extra)
     
         run( 'ls -lt wheelhouse')
 
@@ -563,8 +605,11 @@ def pyodide_setup(clean=False):
     # Clone emsdk.
     #
     dir_emsdk = 'emsdk'
-    if clean and os.path.exists(dir_emsdk):
-        shutil.rmtree( dir_emsdk, ignore_errors=1)
+    if clean:
+        shutil.rmtree(dir_emsdk, ignore_errors=1)
+        # 2024-06-25: old `.pyodide-xbuildenv` directory was breaking build, so
+        # important to remove it here.
+        shutil.rmtree('.pyodide-xbuildenv', ignore_errors=1)
     if not os.path.exists(dir_emsdk):
         command += f' && echo "### cloning emsdk.git"'
         command += f' && git clone https://github.com/emscripten-core/emsdk.git {dir_emsdk}'
@@ -675,9 +720,8 @@ def log(text, caller=0):
     filename    = frame_record.filename
     line        = frame_record.lineno
     function    = frame_record.function
-    prefix = f'{relpath(filename, pymupdf_dir)}:{function}(): '
-    print(textwrap.indent(text, prefix))
-    sys.stdout.flush()
+    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):
@@ -697,10 +741,14 @@ def run(command, env_extra=None, check=1, timeout=None):
             leave processes running if timeout expires.
     '''
     env = None
+    message = 'Running:'
     if env_extra:
         env = os.environ.copy()
         env.update(env_extra)
-    log(f'Running: {command}')
+        for n, v in env_extra.items():
+            message += f' {n}={v}'
+    message += f' {command}'
+    log(message, caller=1)
     return subprocess.run(command, check=check, shell=1, env=env, timeout=timeout)
 
 
@@ -716,7 +764,7 @@ def platform_tag():
         assert 0, f'Unrecognised: {platform.system()=}'
 
 
-test_packages = 'pytest fontTools pymupdf-fonts flake8 pylint'
+test_packages = 'pytest fontTools pymupdf-fonts flake8 pylint codespell'
 if platform.system() == 'Windows' and cpu_bits() == 32:
     # No pillow wheel available, and doesn't build easily.
     pass

{PyMuPDF-1.24.6 → PyMuPDF-1.24.8}/scripts/sysinstall.py

@@ -15,7 +15,7 @@ We run install commands with `sudo` if `--root /` is used.
 Note that we run some commands with sudo; it's important that these use the
 same python as non-sudo, otherwise things can be build and installed for
 different python versions. For example when we are run from a github action, it
-should not do `- uses: actions/setup-python@v2` but instead use whatever system
+should not do `- uses: actions/setup-python@v5` but instead use whatever system
 python is already defined.
 
 Args:

{PyMuPDF-1.24.6 → PyMuPDF-1.24.8}/scripts/test.py

@@ -17,7 +17,7 @@ Examples:
         Build and test using internal checkout of mupdf 1.24.x branch from Github.
 
 Usage:
-    scripts/test.py <options> <commands>
+    scripts/test.py <options> <command(s)>
 
 * Commands are handled in order, so for example `build` should usually be
   before `test`.
@@ -60,7 +60,7 @@ Options:
         Location of local mupdf/ directory or 'git:...' to be used
         when building PyMuPDF. [This sets environment variable
         PYMUPDF_SETUP_MUPDF_BUILD, which is used by PyMuPDF/setup.py. If not
-        specifed PyMuPDF will download its default mupdf .tgz.]
+        specified PyMuPDF will download its default mupdf .tgz.]
     -p <pytest-options>
         Set pytest options; default is ''.
     -t <names>
@@ -81,6 +81,10 @@ Options:
         If true (the default on non-OpenBSD systems), we let pip create and use
         its own new venv to build PyMuPDF. Otherwise we force pip to use the
         current venv.
+    --build-flavour <build_flavour>
+        Combination of 'p', 'b', 'd'. See ../setup.py's description of
+        PYMUPDF_SETUP_FLAVOUR. Default is 'pb', i.e. self-contained PyMuPDF
+        wheels without MuPDF build-time files.
     --build-mupdf 0|1
         Whether to rebuild mupdf when we build PyMuPDF. Default is 1.
     --gdb 0|1
@@ -101,6 +105,8 @@ Commands:
     test
         Runs PyMuPDF's pytest tests in venv. Default is to test rebased and
         unoptimised rebased; use `-i` to change this.
+    wheel
+        Build wheel.
 
 Environment:
     PYMUDF_SCRIPTS_TEST_options
@@ -131,8 +137,10 @@ def main(argv):
     build_isolation = None
     valgrind = False
     s = True
+    build_do = 'i'
     build_type = None
     build_mupdf = True
+    build_flavour = 'pb'
     gdb = False
     test_fitz = True
     implementations = None
@@ -187,6 +195,8 @@ def main(argv):
             timeout = float(next(args))
         elif arg == '-v':
             venv = int(next(args))
+        elif arg == '--build-flavour':
+            build_flavour = next(args)
         elif arg == '--build-mupdf':
             build_mupdf = int(next(args))
         elif arg == '--gdb':
@@ -202,7 +212,7 @@ def main(argv):
     
     commands = list()
     while 1:
-        assert arg in ('build', 'buildtest', 'test'), \
+        assert arg in ('build', 'buildtest', 'test', 'wheel'), \
                 f'Unrecognised command: {arg=}.'
         commands.append(arg)
         try:
@@ -223,13 +233,15 @@ def main(argv):
                 )
         return
 
-    def do_build():
+    def do_build(wheel=False):
         build(
                 implementations=implementations,
                 build_type=build_type,
                 build_isolation=build_isolation,
                 venv_quick=venv_quick,
                 build_mupdf=build_mupdf,
+                build_flavour=build_flavour,
+                wheel=wheel,
                 )
     def do_test():
         test(
@@ -254,6 +266,8 @@ def main(argv):
         elif command == 'buildtest':
             do_build()
             do_test()
+        elif command == 'wheel':
+            do_build(wheel=True)
         else:
             assert 0
 
@@ -312,6 +326,8 @@ def build(
         build_isolation=None,
         venv_quick=False,
         build_mupdf=True,
+        build_flavour='pb',
+        wheel=False,
         ):
     '''
     Args:
@@ -363,7 +379,12 @@ def build(
         env_extra['PYMUPDF_SETUP_MUPDF_REBUILD'] = '0'
     if build_type:
         env_extra['PYMUPDF_SETUP_MUPDF_BUILD_TYPE'] = build_type
-    gh_release.run(f'pip install{build_isolation_text} -vv {pymupdf_dir}', env_extra=env_extra)
+    if build_flavour:
+        env_extra['PYMUPDF_SETUP_FLAVOUR'] = build_flavour
+    if wheel:
+        gh_release.run(f'pip wheel{build_isolation_text} -v {pymupdf_dir}', env_extra=env_extra)
+    else:
+        gh_release.run(f'pip install{build_isolation_text} -v {pymupdf_dir}', env_extra=env_extra)
 
 
 def test(
@@ -440,7 +461,7 @@ def test(
         elif gdb:
             command = f'{python} {pymupdf_dir_rel}/tests/run_compound.py{run_compound_args} gdb --args {python} -m pytest {pytest_options} {pytest_arg}'
         elif platform.system() == 'Windows':
-            # `python -m pytest` doesn' seem to work.
+            # `python -m pytest` doesn't seem to work.
            command = f'{python} {pymupdf_dir_rel}/tests/run_compound.py{run_compound_args} pytest {pytest_options} {pytest_arg}'
         else:
             # On OpenBSD `pip install pytest` doesn't seem to install the pytest