PyMuPDF 1.23.23__tar.gz → 1.23.25__tar.gz

{PyMuPDF-1.23.23 → PyMuPDF-1.23.25}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PyMuPDF
-Version: 1.23.23
+Version: 1.23.25
 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-1.23.23 → PyMuPDF-1.23.25}/changes.txt

@@ -2,6 +2,33 @@ Change Log
 ==========
 
 
+**Changes in version 1.23.25 (2024-02-20)**
+
+* Fixed issues:
+
+  * **Fixed** `3182 <https://github.com/pymupdf/PyMuPDF/issues/3182>`_: Pixmap.invert_irect argument type error
+  * **Fixed** `3186 <https://github.com/pymupdf/PyMuPDF/issues/3186>`_: extractText() extracts broken text from pdf
+  * **Fixed** `3191 <https://github.com/pymupdf/PyMuPDF/issues/3191>`_: Error on .find_tables()
+
+* Other:
+
+  * When building, be able to specify python-config directly, with environment
+    variable `PIPCL_PYTHON_CONFIG`.
+
+
+**Changes in version 1.23.24 (2024-02-19)**
+
+* Fixed issues:
+
+  * **Fixed** `3148 <https://github.com/pymupdf/PyMuPDF/issues/3148>`_: Table extraction - vertical text not handled correctly
+  * **Fixed** `3179 <https://github.com/pymupdf/PyMuPDF/issues/3179>`_: Table Detection: Incorrect Separation of Vector Graphics Clusters
+  * **Fixed** `3180 <https://github.com/pymupdf/PyMuPDF/issues/3180>`_: Cannot show optional content group: AttributeError: module 'fitz.mupdf' has no attribute 'pdf_array_push_drop'
+
+* Other:
+
+  * Be able to test system install using `sudo pip install` instead of a venv.
+
+
 **Changes in version 1.23.23 (2024-02-18)**
 
 * Fixed issues:

{PyMuPDF-1.23.23 → PyMuPDF-1.23.25}/pipcl.py

@@ -1769,6 +1769,9 @@ class PythonFlags:
     Compile/link flags for the current python, for example the include path
     needed to get `Python.h`.
 
+    The 'PIPCL_PYTHON_CONFIG' environment variable allows to override
+    the location of the python-config executable.
+
     Members:
         .includes:
             String containing compiler flags for include paths.
@@ -1789,43 +1792,45 @@ class PythonFlags:
             self.ldflags = f'-L {_lib_dir}'
 
         else:
-            # We use python-config which appears to work better than pkg-config
-            # because it copes with multiple installed python's, e.g.
-            # manylinux_2014's /opt/python/cp*-cp*/bin/python*.
-            #
-            # But... on non-macos it seems that we should not attempt to specify
-            # libpython on the link command. The manylinux docker containers
-            # don't actually contain libpython.so, and it seems that this
-            # deliberate. And the link command runs ok.
-            #
-            python_exe = os.path.realpath( sys.executable)
-            if darwin():
-                # Basic install of dev tools with `xcode-select --install` doesn't
-                # seem to provide a `python3-config` or similar, but there is a
-                # `python-config.py` accessible via sysconfig.
+            python_config = os.environ.get("PIPCL_PYTHON_CONFIG")
+            if not python_config:
+                # We use python-config which appears to work better than pkg-config
+                # because it copes with multiple installed python's, e.g.
+                # manylinux_2014's /opt/python/cp*-cp*/bin/python*.
                 #
-                # We try different possibilities and use the last one that
-                # works.
+                # But... on non-macos it seems that we should not attempt to specify
+                # libpython on the link command. The manylinux docker containers
+                # don't actually contain libpython.so, and it seems that this
+                # deliberate. And the link command runs ok.
                 #
-                python_config = None
-                for pc in (
-                        f'python3-config',
-                        f'{sys.executable} {sysconfig.get_config_var("srcdir")}/python-config.py',
-                        f'{python_exe}-config',
-                        ):
-                    e = subprocess.run(
-                            f'{pc} --includes',
-                            shell=1,
-                            stdout=subprocess.DEVNULL,
-                            stderr=subprocess.DEVNULL,
-                            check=0,
-                            ).returncode
-                    log1(f'{e=} from {pc!r}.')
-                    if e == 0:
-                        python_config = pc
-                assert python_config, f'Cannot find python-config'
-            else:
-                python_config = f'{python_exe}-config'
+                python_exe = os.path.realpath( sys.executable)
+                if darwin():
+                    # Basic install of dev tools with `xcode-select --install` doesn't
+                    # seem to provide a `python3-config` or similar, but there is a
+                    # `python-config.py` accessible via sysconfig.
+                    #
+                    # We try different possibilities and use the last one that
+                    # works.
+                    #
+                    python_config = None
+                    for pc in (
+                            f'python3-config',
+                            f'{sys.executable} {sysconfig.get_config_var("srcdir")}/python-config.py',
+                            f'{python_exe}-config',
+                            ):
+                        e = subprocess.run(
+                                f'{pc} --includes',
+                                shell=1,
+                                stdout=subprocess.DEVNULL,
+                                stderr=subprocess.DEVNULL,
+                                check=0,
+                                ).returncode
+                        log1(f'{e=} from {pc!r}.')
+                        if e == 0:
+                            python_config = pc
+                    assert python_config, f'Cannot find python-config'
+                else:
+                    python_config = f'{python_exe}-config'
             log1(f'Using {python_config=}.')
             try:
                 self.includes = run( f'{python_config} --includes', capture=1).strip()

{PyMuPDF-1.23.23 → PyMuPDF-1.23.25}/scripts/sysinstall.py

@@ -24,28 +24,39 @@ Args:
     --mupdf-git <git_args>
         Get or update `mupdf_dir` using git. If `mupdf_dir` already
         exists we run `git pull` in it; otherwise we run `git
-        clone` with `git_args`. For example:
+        clone` with `<git_args> <mupdf_dir>`. For example:
             --mupdf-git "--branch master https://github.com/ArtifexSoftware/mupdf.git"
-    --pymupdf-dir <pymupdf_dir>
-        Path of PyMuPDF checkout; default is 'PyMuPDF'.
+    --mupdf-so-mode <mode>
+        Used with `install -m <mode> ...` when installing MuPDF. For example
+        `--mupdf-so-mode 744`.
+    --packages 0|1
+        If 1 (the default) we install required system packages such as
+        `libfreetype-dev`.
+    --pip 0|venv|sudo
+        Whether/how to install Python packages.
+        If '0' we assume required packages are already available.
+        If 'sudo' we install required Python packages using `sudo pip install
+        ...`.
+        If 'venv' (the default) we install Python packages and run installer
+        and test commands inside venv's.
     --prefix:
         Directory within `root`; default is `/usr/local`. Must start with `/`.
+    --pymupdf-dir <pymupdf_dir>
+        Path of PyMuPDF checkout; default is 'PyMuPDF'.
     --root <root>
         Root of install directory; default is `/`.
     --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
         package repository ppa:alex-p/tesseract-ocr-devel.
-    --test-venv <venv_path>
-        Run tests in specified venv; the default is a hard-coded venv name. The
-        venv will be created and required packages installed using `pip`. If
-        `venv-path` is empty string, we do not create or use a venv, and will
-        instead attempt to install required packages with `pip` in the current
-        Python environment.
+    --test-venv <test_venv>
+        Set the name of the venv in which we run tests (only with `--pip
+        venv`); the default is a hard-coded venv name. The venv will be
+        created, and required packages installed using `pip`.
     --use-installer 0|1
         If 1 (the default), we use `python -m installer` to install PyMuPDF
-        from a generated wheel. Otherwise we use `pip install`, which refuses
-        to do a system install with `--root /`, referencing PEP-668.
+        from a generated wheel. [Otherwise we use `pip install`, which refuses
+        to do a system install with `--root /`, referencing PEP-668.]
     -m 0|1
         If 1 (the default) we build and install MuPDF, otherwise we just show
         what command we would have run.
@@ -61,12 +72,14 @@ To only show what commands would be run, but not actually run them, specify `-m
 '''
 
 import glob
+import multiprocessing
 import os
 import platform
 import subprocess
 import sys
 import sysconfig
 
+import test as test_py
 
 # Requirements for a system build and install:
 #
@@ -107,6 +120,8 @@ def main():
     mupdf = True
     mupdf_dir = 'mupdf'
     mupdf_git = None
+    mupdf_so_mode = None
+    packages = True
     prefix = '/usr/local'
     pymupdf = True
     pymupdf_dir = os.path.abspath( f'{__file__}/../..')
@@ -114,6 +129,7 @@ def main():
     tesseract5 = True
     test = True
     test_venv = 'venv-pymupdf-sysinstall-test'
+    pip = 'venv'
     
     # Parse command-line.
     #
@@ -128,12 +144,15 @@ def main():
             return
         elif arg == '--mupdf-dir':      mupdf_dir = next(args)
         elif arg == '--mupdf-git':      mupdf_git = next(args)
+        elif arg == '--mupdf-so-mode':  mupdf_so_mode = next(args)
+        elif arg == '--packages':       packages = int(next(args))
         elif arg == '--prefix':         prefix = 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 == '--test-venv':      test_venv = next(args)
         elif arg == '--use-installer':  use_installer = int(next(args))
+        elif arg == '--pip':            pip = next(args)
         elif arg == '-m':               mupdf = int(next(args))
         elif arg == '-p':               pymupdf = int(next(args))
         elif arg == '-t':               test = int(next(args))
@@ -141,6 +160,8 @@ def main():
             assert 0, f'Unrecognised arg: {arg!r}'
     
     assert prefix.startswith('/')
+    pip_values = ('0', 'sudo', 'venv')
+    assert pip in pip_values, f'Unrecognised --pip value {pip!r} should be one of: {pip_values!r}'
     root = os.path.abspath(root)
     root_prefix = f'{root}{prefix}'.replace('//', '/')
     
@@ -161,44 +182,54 @@ def main():
             print(f'## Clone MuPDF into {mupdf_dir}.')
             run(f'git clone --recursive --depth 1 --shallow-submodules {mupdf_git} {mupdf_dir}')
     
-    # Install required system packages. We assume a Debian package system.
-    #
-    print('## 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.')
-        # https://stackoverflow.com/questions/76834972/how-can-i-run-pytesseract-python-library-in-ubuntu-22-04
+    if packages:
+        # Install required system packages. We assume a Debian package system.
         #
-        run('sudo apt install -y software-properties-common')
-        run('sudo add-apt-repository ppa:alex-p/tesseract-ocr-devel')
-        run('sudo apt update')
-        run('sudo apt install -y libtesseract-dev')
-    else:
-        run('sudo apt install libtesseract-dev')
+        print('## 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.')
+            # https://stackoverflow.com/questions/76834972/how-can-i-run-pytesseract-python-library-in-ubuntu-22-04
+            #
+            run('sudo apt install -y software-properties-common')
+            run('sudo add-apt-repository ppa:alex-p/tesseract-ocr-devel')
+            run('sudo apt update')
+            run('sudo apt install -y libtesseract-dev')
+        else:
+            run('sudo apt install libtesseract-dev')
     
     # 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.')
+        run(f'sudo pip install --upgrade pip')
+        names = (''
+                + test_py.get_pyproject_required(os.path.abspath(f'{__file__}/../../pyproject.toml'))
+                + ' '
+                + test_py.get_pyproject_required(os.path.abspath(f'{mupdf_dir}/pyproject.toml'))
+                )
+        run(f'sudo pip install {names}')
+    
     print('## Build and install MuPDF.')
-    if 1:
-        # Current MuPDF creates softlinks with `ln -s` which breaks if there
-        # was a previous build; it should do `ln -sf`. We make things work by
-        # deleting any existing softlinks here.
-        run(f'rm {root_prefix}/lib/libmupdf.so || true')
-        run(f'rm {root_prefix}/lib/libmupdfcpp.so || true')
     command = f'cd {mupdf_dir}'
     command += f' && {sudo}make'
+    command += f' -j {multiprocessing.cpu_count()}'
     #command += f' EXE_LDFLAGS=-Wl,--trace' # Makes linker generate diagnostics as it runs.
     command += f' DESTDIR={root}'
     command += f' HAVE_LEPTONICA=yes'
     command += f' HAVE_TESSERACT=yes'
     command += f' USE_SYSTEM_LIBS=yes'
+    command += f' VENV_FLAG={"--venv" if pip == "venv" else ""}'
+    if mupdf_so_mode:
+        command += f' SO_INSTALL_MODE={mupdf_so_mode}'
     command += f' build_prefix=system-libs-'
     command += f' prefix={prefix}'
     command += f' verbose=yes'
     command += f' install-shared-python'
+    command += f' INSTALL_MODE=755'
     run( command)
     
     # Build+install PyMuPDF.
@@ -217,23 +248,34 @@ def main():
     env += f'PYMUPDF_SETUP_IMPLEMENTATIONS=b'   # Only build the rebased implementation.
     if use_installer:
         print(f'## Building wheel.')
+        if pip == 'venv':
+            venv_name = 'venv-pymupdf-sysinstall'
         run(f'pwd')
         run(f'rm dist/* || true')
-        run(f'{env} pip wheel -vv -w dist {os.path.abspath(pymupdf_dir)}')
+        if pip == 'venv':
+            run(f'{sys.executable} -m venv {venv_name}')
+            run(f'. {venv_name}/bin/activate && pip install --upgrade pip')
+            run(f'. {venv_name}/bin/activate && pip install --upgrade installer')
+            run(f'{env} {venv_name}/bin/python -m pip wheel -vv -w dist {os.path.abspath(pymupdf_dir)}')
+        elif pip == 'sudo':
+            run(f'sudo pip install --upgrade pip')
+            run(f'sudo pip install installer')
+            run(f'{env} pip wheel -vv -w dist {os.path.abspath(pymupdf_dir)}')
+        else:
+            log(f'Not installing "installer" because {pip=}.')
         wheel = glob.glob(f'dist/*')
         assert len(wheel) == 1, f'{wheel=}'
         wheel = wheel[0]
         print(f'## Installing wheel using `installer`.')
-        venv = 'venv-pymupdf-sysinstall'
-        run(f'{sys.executable} -m venv {venv}')
-        run(f'. {venv}/bin/activate && pip install --upgrade pip')
-        run(f'. {venv}/bin/activate && pip install --upgrade installer')
         pv = '.'.join(platform.python_version_tuple()[:2])
         p = f'{root_prefix}/lib/python{pv}'
         # `python -m installer` fails to overwrite existing files.
         run(f'{sudo}rm -r {p}/site-packages/fitz || true')
         run(f'{sudo}rm -r {p}/site-packages/PyMuPDF-*.dist-info || true')
-        run(f'{sudo}{venv}/bin/python -m installer --destdir {root} --prefix {prefix} {wheel}')
+        if pip == 'venv':
+            run(f'{sudo}{venv_name}/bin/python -m installer --destdir {root} --prefix {prefix} {wheel}')
+        else:
+            run(f'{sudo}{sys.executable} -m installer --destdir {root} --prefix {prefix} {wheel}')
         # It seems that MuPDF Python bindings are installed into
         # `.../dist-packages` (from mupdf:Mafile's call of `$(shell python3
         # -c "import sysconfig; print(sysconfig.get_path('platlib'))")` while
@@ -267,7 +309,7 @@ def main():
         del sys.path[0]
         pythonpath = pipcl.install_dir(root)
     
-    # Show contents of installation director. This is very slow on github,
+    # Show contents of installation directory. This is very slow on github,
     # where /usr/local contains lots of things.
     #run(f'find {root_prefix}|sort')
         
@@ -277,20 +319,27 @@ def main():
     def run(command):
         return run_command(command, doit=test)
     import gh_release
-    # Create venv.
-    run(f'{sys.executable} -m venv {test_venv}')
-    # Install required packages.
-    command = f'. {test_venv}/bin/activate'
-    command += f' && pip install --upgrade pip'
-    command += f' && pip install --upgrade {gh_release.test_packages}'
-    run(command)
+    if pip == 'venv':
+        # Create venv.
+        run(f'{sys.executable} -m venv {test_venv}')
+        # Install required packages.
+        command = f'. {test_venv}/bin/activate'
+        command += f' && pip install --upgrade pip'
+        command += f' && pip install --upgrade {gh_release.test_packages}'
+        run(command)
+    elif pip == 'sudo':
+        run(f'sudo pip install --upgrade {gh_release.test_packages}')
+    else:
+        log(f'Not installing packages for testing because {pip=}.')
     # Run pytest.
     #
     # We need to set PYTHONPATH and LD_LIBRARY_PATH. In particular we
     # use pipcl.install_dir() to find where pipcl will have installed
     # PyMuPDF.
-    command = f'. {test_venv}/bin/activate'
-    command += f' && LD_LIBRARY_PATH={root_prefix}/lib PYTHONPATH={pythonpath}'
+    command = ''
+    if pip == 'venv':
+        command += f'. {test_venv}/bin/activate &&'
+    command += f' LD_LIBRARY_PATH={root_prefix}/lib PYTHONPATH={pythonpath}'
     command += f' pytest -k "not test_color_count and not test_3050" {pymupdf_dir}'
     run(command)
 

{PyMuPDF-1.23.23 → PyMuPDF-1.23.25}/scripts/test.py

@@ -410,12 +410,13 @@ def test(
         log('\n' + venv_info(pytest_args=f'{pytest_options} {pytest_arg}'))
 
 
-def get_pyproject_required():
+def get_pyproject_required(ppt=None):
     '''
     Returns space-separated names of required packages in pyproject.toml.  We
     do not do a proper parse and rely on the packages being in a single line.
     '''
-    ppt = os.path.abspath(f'{__file__}/../../pyproject.toml')
+    if ppt is None:
+        ppt = os.path.abspath(f'{__file__}/../../pyproject.toml')
     with open(ppt) as f:
         for line in f:
             m = re.match('^requires = \\[(.*)\\]$', line)

{PyMuPDF-1.23.23 → PyMuPDF-1.23.25}/setup.py

@@ -1072,7 +1072,7 @@ with open( f'{g_root}/READMErb.md', encoding='utf-8') as f:
 # We generate different wheels depending on g_flavour.
 #
 
-version = '1.23.23'
+version = '1.23.25'
 version_b = '1.23.22'
 
 tag_python = None

{PyMuPDF-1.23.23 → PyMuPDF-1.23.25}/src/__init__.py

@@ -16259,7 +16259,7 @@ def JM_invert_pixmap_rect( dest, b):
 
 def JM_irect_from_py(r):
     '''
-    PySequence to fz_irect. Default: infinite irect
+    PySequence to mupdf.FzIrect. Default: infinite irect
     '''
     if isinstance(r, mupdf.FzIrect):
         return r
@@ -16267,7 +16267,11 @@ def JM_irect_from_py(r):
         r = mupdf.FzIrect( r.x0, r.y0, r.x1, r.y1)
         return r
     if isinstance(r, Rect):
-        ret = mupdf.fz_make_irect(r.x0, r.y0, r.x1, r.y1)
+        ret = mupdf.FzRect(r.x0, r.y0, r.x1, r.y1)
+        ret = mupdf.FzIrect(ret)  # Uses fz_irect_from_rect().
+        return ret
+    if isinstance(r, mupdf.FzRect):
+        ret = mupdf.FzIrect(r)  # Uses fz_irect_from_rect().
         return ret
     if not r or not PySequence_Check(r) or PySequence_Size(r) != 4:
         return mupdf.FzIrect(mupdf.fz_infinite_irect)
@@ -16275,7 +16279,7 @@ def JM_irect_from_py(r):
     for i in range(4):
         f[i] = r[i]
         if f[i] is None:
-            return mupdf.FzRect(mupdf.fz_infinite_irect)
+            return mupdf.FzIrect(mupdf.fz_infinite_irect)
         if f[i] < FZ_MIN_INF_RECT:
             f[i] = FZ_MIN_INF_RECT
         if f[i] > FZ_MAX_INF_RECT:
@@ -17437,7 +17441,7 @@ def JM_set_ocg_arrays_imp(arr, list_):
             # Not found.
             continue
         obj = mupdf.pdf_new_indirect(pdf, xref, 0)
-        mupdf.pdf_array_push_drop(arr, obj)
+        mupdf.pdf_array_push(arr, obj)
 
 
 def JM_set_resource_property(ref, name, xref):
@@ -17814,9 +17818,6 @@ def PyUnicode_DecodeRawUnicodeEscape(s, errors='strict'):
     elif isinstance(s, bytes):
         rc = s[:]
     ret = rc.decode('raw_unicode_escape', errors=errors)
-    z = ret.find(chr(0))
-    if z >= 0:
-        ret = ret[:z]
     return ret
 
 
@@ -21729,8 +21730,8 @@ def int_rc(text):
     return int(text)
 
 VersionFitz = "1.23.10" # MuPDF version.
-VersionBind = "1.23.23" # PyMuPDF version.
-VersionDate = "2024-02-18 00:00:01"
+VersionBind = "1.23.25" # PyMuPDF version.
+VersionDate = "2024-02-20 00:00:01"
 VersionDate2 = VersionDate.replace('-', '').replace(' ', '').replace(':', '')
 version = (VersionBind, VersionFitz, VersionDate2)
 pymupdf_version_tuple = tuple( [int_rc(i) for i in VersionBind.split('.')])