Jules - added tests for upstream fixes in mupdf etc.

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -46,7 +46,7 @@ body:
     attributes:
       label: PyMuPDF version
       options:
-        - 
+        - 1.26.0
         - 1.25.5
         - 1.25.4
         - 1.25.3

.github/workflows/build_wheels.yml

@@ -7,7 +7,7 @@ on:
       flavours:
         description: 'If set, we build separate PyMuPDF and PyMuPDFb wheels.'
         type: boolean
-        default: true
+        default: false
 
       sdist:
         type: boolean

changes.txt

@@ -9,13 +9,20 @@ Change Log
 * Fixed issues:
 
   * **Fixed** `4324 <https://github.com/pymupdf/PyMuPDF/issues/4324>`_: cluster_drawings() fails to cluster horizontal and vertical thin lines
+  * **Fixed** `4363 <https://github.com/pymupdf/PyMuPDF/issues/4363>`_: Trouble with searching
   * **Fixed** `4404 <https://github.com/pymupdf/PyMuPDF/issues/4404>`_: IndexError in page.get_links()
   * **Fixed** `4412 <https://github.com/pymupdf/PyMuPDF/issues/4412>`_: Regression? Spurious error? in insert_pdf in v1.25.4
+  * **Fixed** `4423 <https://github.com/pymupdf/PyMuPDF/issues/4423>`_: pymupdf.mupdf.FzErrorFormat: code=7: cannot find object in xref error encountered after version 1.25.3
+  * **Fixed** `4435 <https://github.com/pymupdf/PyMuPDF/issues/4435>`_: get_pixmap method stuck on one page
   * **Fixed** `4439 <https://github.com/pymupdf/PyMuPDF/issues/4439>`_: New Xml class from data does not work - bug in code
+  * **Fixed** `4445 <https://github.com/pymupdf/PyMuPDF/issues/4445>`_: Broken XREF table incorrectly repaired
   * **Fixed** `4447 <https://github.com/pymupdf/PyMuPDF/issues/4447>`_: Stroke color of annotations cannot be correctly set
+  * **Fixed** `4479 <https://github.com/pymupdf/PyMuPDF/issues/4479>`_: set_layer_ui_config() toggles all layers rather than just one
+  * **Fixed** `4505 <https://github.com/pymupdf/PyMuPDF/issues/4505>`_: Follow Widget flag values up its parent structure
 
 * Other:
 
+  * Partial fixed for `4457 <https://github.com/pymupdf/PyMuPDF/issues/4457>`_: Wrong characters displayed after font subsetting (w/ native method)
   * Support image stamp annotations.
   * Support recoloring pages.
   * Added example of using Django's file storage API to open files with pymupdf.

docs/version.rst

@@ -1,6 +1,6 @@
 ----
 
-This documentation covers **PyMuPDF v1.25.5** features as of **2025-03-31 00:00:01**.
+This documentation covers **PyMuPDF v1.26.0** features as of **2025-03-31 00:00:01**.
 
 The major and minor versions of |PyMuPDF| and |MuPDF| will always be the same. Only the third qualifier (patch level) may deviate from that of |MuPDF|.
 

scripts/gh_release.py

@@ -448,7 +448,7 @@ def set_cibuild_test():
                 # Include MuPDF build-time files.
                 flavour += 'd'
             env_set( 'PYMUPDF_SETUP_FLAVOUR', flavour, pass_=1)
-            run( f'cibuildwheel{platform_arg}', env_extra)
+            run( f'cibuildwheel{platform_arg}', env_extra=env_extra)
             run( 'echo after {flavour=}')
             run( 'ls -l wheelhouse')
 

scripts/test.py

@@ -13,8 +13,8 @@
     ./PyMuPDF/scripts/test.py --mupdf 'git:https://git.ghostscript.com/mupdf.git' buildtest
         Build and test with internal checkout of mupdf master.
 
-    ./PyMuPDF/scripts/test.py --mupdf 'git:--branch 1.25.x https://github.com/ArtifexSoftware/mupdf.git' buildtest
-        Build and test using internal checkout of mupdf 1.25.x branch from Github.
+    ./PyMuPDF/scripts/test.py --mupdf 'git:--branch 1.26.x https://github.com/ArtifexSoftware/mupdf.git' buildtest
+        Build and test using internal checkout of mupdf 1.26.x branch from Github.
 
 Usage:
     scripts/test.py <options> <command(s)>

setup.py

@@ -1229,15 +1229,15 @@ def sdist():
 #
 
 # PyMuPDF version.
-version_p = '1.25.5'
+version_p = '1.26.0'
 
 # PyMuPDFb version. This is the PyMuPDF version whose PyMuPDFb wheels we will
 # (re)use if generating separate PyMuPDFb wheels. Though as of PyMuPDF-1.24.11
 # (2024-10-03) we no longer use PyMuPDFb wheels so this is actually unused.
 #
-version_b = '1.25.3'
+version_b = '1.26.0'
 
-version_mupdf = '1.25.6'
+version_mupdf = '1.26.1'
 
 if os.path.exists(f'{g_root}/{g_pymupdfb_sdist_marker}'):
 

src/__init__.py

@@ -375,7 +375,7 @@ def _int_rc(text):
 
 # Basic version information.
 #
-pymupdf_version = "1.25.5"
+pymupdf_version = "1.26.0"
 mupdf_version = mupdf.FZ_VERSION
 pymupdf_date = "2025-03-31 00:00:01"
 
@@ -5809,7 +5809,7 @@ def update_object(self, xref, text, page=None):
         pdf = _as_pdf_document(self)
         xreflen = mupdf.pdf_xref_len(pdf)
         if not _INRANGE(xref, 1, xreflen-1):
-            RAISEPY("bad xref", MSG_BAD_XREF, PyExc_ValueError)
+            RAISEPY("bad xref", MSG_BAD_XREF)
         ENSURE_OPERATION(pdf)
         # create new object with passed-in string
         new_obj = JM_pdf_obj_from_str(pdf, text)

tests/test_font.py

@@ -1,8 +1,14 @@
 """
 Tests for the Font class.
 """
-import pymupdf
 import os
+import platform
+import pymupdf
+import subprocess
+import textwrap
+
+import util
+
 
 def test_font1():
     text = "PyMuPDF"
@@ -218,3 +224,98 @@ def test_3887():
         print(f'Have saved to: {path_pixmap=}')
         assert set(output)==set(text)
 
+
+def test_4457():
+    print()
+    files = (
+            ('https://arxiv.org/pdf/2504.13180', 'test_4457_a.pdf', None, 4),
+            ('https://arxiv.org/pdf/2504.13181', 'test_4457_b.pdf', None, 9),
+            )
+    for url, name, size, rms_after_max in files:
+        path = util.download(url, name, size)
+
+        with pymupdf.open(path) as document:
+            page = document[0]
+
+            pixmap = document[0].get_pixmap()
+            path_pixmap = f'{path}.png'
+            pixmap.save(path_pixmap)
+            print(f'Have created: {path_pixmap=}')
+
+            text = page.get_text()
+            path_before = f'{path}.before.pdf'
+            path_after = f'{path}.after.pdf'
+            document.ez_save(path_before, garbage=4)
+            print(f'Have created {path_before=}')
+
+            document.subset_fonts()
+            document.ez_save(path_after, garbage=4)
+            print(f'Have created {path_after=}')
+
+        with pymupdf.open(path_before) as document:
+            text_before = document[0].get_text()
+            pixmap_before = document[0].get_pixmap()
+            path_pixmap_before = f'{path_before}.png'
+            pixmap_before.save(path_pixmap_before)
+            print(f'Have created: {path_pixmap_before=}')
+
+        with pymupdf.open(path_after) as document:
+            text_after = document[0].get_text()
+            pixmap_after = document[0].get_pixmap()
+            path_pixmap_after = f'{path_after}.png'
+            pixmap_after.save(path_pixmap_after)
+            print(f'Have created: {path_pixmap_after=}')
+
+        import gentle_compare
+        rms_before = gentle_compare.pixmaps_rms(pixmap, pixmap_before)
+        rms_after = gentle_compare.pixmaps_rms(pixmap, pixmap_after)
+        print(f'{rms_before=}')
+        print(f'{rms_after=}')
+
+        # Create .png file showing differences between <path> and <path_after>.
+        path_pixmap_after_diff = f'{path_after}.diff.png'
+        pixmap_after_diff = gentle_compare.pixmaps_diff(pixmap, pixmap_after)
+        pixmap_after_diff.save(path_pixmap_after_diff)
+        print(f'Have created: {path_pixmap_after_diff}')
+
+        # Extract text from <path>, <path_before> and <path_after> and write to
+        # files so we can show differences with `diff`.
+        path_text = os.path.normpath(f'{__file__}/../../tests/test_4457.txt')
+        path_text_before = f'{path_text}.before.txt'
+        path_text_after = f'{path_text}.after.txt'
+        with open(path_text, 'w', encoding='utf8') as f:
+            f.write(text)
+        with open(path_text_before, 'w', encoding='utf8') as f:
+            f.write(text_before)
+        with open(path_text_after, 'w', encoding='utf8') as f:
+            f.write(text_after)
+
+        # Can't write text to stdout on Windows because of encoding errors.
+        if platform.system() != 'Windows':
+            print(f'text:\n{textwrap.indent(text, "    ")}')
+            print(f'text_before:\n{textwrap.indent(text_before, "    ")}')
+            print(f'text_after:\n{textwrap.indent(text_after, "    ")}')
+            print(f'{path_text=}')
+            print(f'{path_text_before=}')
+            print(f'{path_text_after=}')
+
+            command = f'diff -u {path_text} {path_text_before}'
+            print(f'Running: {command}', flush=1)
+            subprocess.run(command, shell=1)
+
+            command = f'diff -u {path_text} {path_text_after}'
+            print(f'Running: {command}', flush=1)
+            subprocess.run(command, shell=1)
+
+        assert text_before == text
+        assert rms_before == 0
+
+        # As of 2025-05-20 there are some differences in some characters, e.g.
+        # the non-ascii characters in `Philipp Krahenbuhl`.
+        # See <path_pixmap> and <path_pixmap_after>.
+        assert rms_after < rms_after_max
+
+    # Avoid test failure caused by mupdf warnings.
+    wt = pymupdf.TOOLS.mupdf_warnings()
+    print(f'{wt=}')
+    assert wt == 'bogus font ascent/descent values (0 / 0)\n... repeated 5 times...'

tests/test_general.py

@@ -1834,3 +1834,43 @@ def test_4466():
             pixmap = page.get_pixmap(clip=(0, 0, 10, 10))
             print(f'{pixmap.n=} {pixmap.size=} {pixmap.stride=} {pixmap.width=} {pixmap.height=} {pixmap.x=} {pixmap.y=}', flush=1)
             pixmap.is_unicolor  # Used to crash.
+
+
+def test_4479():
+    # This passes with pymupdf-1.24.14, fails with pymupdf==1.25.*, passes with
+    # pymupdf-1.26.0.
+    print()
+    path = os.path.normpath(f'{__file__}/../../tests/resources/test_4479.pdf')
+    with pymupdf.open(path) as document:
+
+        def show(items):
+            for item in items:
+                print(f'    {repr(item)}')
+
+        items = document.layer_ui_configs()
+        show(items)
+        assert items == [
+                {'depth': 0, 'locked': 0, 'number': 0, 'on': 1, 'text': 'layer_0', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 1, 'on': 1, 'text': 'layer_1', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 2, 'on': 0, 'text': 'layer_2', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 3, 'on': 1, 'text': 'layer_3', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 4, 'on': 1, 'text': 'layer_4', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 5, 'on': 1, 'text': 'layer_5', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 6, 'on': 1, 'text': 'layer_6', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 7, 'on': 1, 'text': 'layer_7', 'type': 'checkbox'},
+                ]
+
+        document.set_layer_ui_config(0, pymupdf.PDF_OC_OFF)
+        items = document.layer_ui_configs()
+        show(items)
+        assert items == [
+                {'depth': 0, 'locked': 0, 'number': 0, 'on': 0, 'text': 'layer_0', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 1, 'on': 1, 'text': 'layer_1', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 2, 'on': 0, 'text': 'layer_2', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 3, 'on': 1, 'text': 'layer_3', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 4, 'on': 1, 'text': 'layer_4', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 5, 'on': 1, 'text': 'layer_5', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 6, 'on': 1, 'text': 'layer_6', 'type': 'checkbox'},
+                {'depth': 0, 'locked': 0, 'number': 7, 'on': 1, 'text': 'layer_7', 'type': 'checkbox'},
+                ]
+

tests/test_pixmap.py

@@ -510,3 +510,72 @@ def test_4336():
 
     if cc_old:
         assert cc == cc_old
+
+
+def test_4435():
+    print(f'{pymupdf.version=}')
+    path = os.path.normpath(f'{__file__}/../../tests/resources/test_4435.pdf')
+    with pymupdf.open(path) as document:
+        page = document[2]
+        print(f'Calling page.get_pixmap().', flush=1)
+        pixmap = page.get_pixmap(alpha=False, dpi=120)
+        print(f'Called page.get_pixmap().', flush=1)
+    wt = pymupdf.TOOLS.mupdf_warnings()
+    assert wt == 'bogus font ascent/descent values (0 / 0)\n... repeated 9 times...'
+
+
+def test_4423():
+    path = os.path.normpath(f'{__file__}/../../tests/resources/test_4423.pdf')
+    with pymupdf.open(path) as document:
+        path2 = f'{path}.pdf'
+        ee = None
+        try:
+            document.save(
+                    path2, 
+                    garbage=4,
+                    expand=1,
+                    deflate=True,
+                    pretty=True,
+                    no_new_id=True,
+                    )
+        except Exception as e:
+            print(f'Exception: {e}')
+            ee = e
+
+        if (1, 25, 5) <= pymupdf.mupdf_version_tuple < (1, 26):
+            assert ee, f'Did not receive the expected exception.'
+            wt = pymupdf.TOOLS.mupdf_warnings()
+            assert wt == 'dropping unclosed output'
+        else:
+            assert not ee, f'Received unexpected exception: {e}'
+            wt = pymupdf.TOOLS.mupdf_warnings()
+            assert wt == 'format error: cannot find object in xref (56 0 R)\nformat error: cannot find object in xref (68 0 R)'
+
+
+def test_4445():
+    print()
+    # Test case is large so we download it instead of having it in PyMuPDF
+    # git. We put it in `cache/` directory do it is not removed by `git clean`
+    # (unless `-d` is specified).
+    import util
+    path = util.download(
+            'https://github.com/user-attachments/files/19738242/ss.pdf',
+            'test_4445.pdf',
+            size=2671185,
+            )
+    with pymupdf.open(path) as document:
+        page = document[0]
+        pixmap = page.get_pixmap()
+        print(f'{pixmap.width=}')
+        print(f'{pixmap.height=}')
+        if pymupdf.mupdf_version_tuple >= (1, 26):
+            assert (pixmap.width, pixmap.height) == (792, 612)
+        else:
+            assert (pixmap.width, pixmap.height) == (612, 792)
+        if 0:
+            path_pixmap = f'{path}.png'
+            pixmap.save(path_pixmap)
+            print(f'Have created {path_pixmap=}')
+    wt = pymupdf.TOOLS.mupdf_warnings()
+    print(f'{wt=}')
+    assert wt == 'broken xref subsection, proceeding anyway.\nTrailer Size is off-by-one. Ignoring.'

tests/util.py

@@ -0,0 +1,28 @@
+import os
+import subprocess
+
+
+def download(url, name, size=None):
+    '''
+    Downloads from <url> to a local file and returns its path. 
+
+    If file already exists and matches <size> we do not re-download it.
+
+    We put local files within a `cache/` directory so that it is not deleted by
+    `git clean` (unless `-d` is specified).
+    '''
+    path = os.path.normpath(f'{__file__}/../../tests/cache/{name}')
+    if os.path.isfile(path) and (not size or os.stat(path).st_size == size):
+        print(f'Using existing file {path=}.')
+    else:
+        print(f'Downloading from {url=}.')
+        subprocess.run(f'pip install -U requests', check=1, shell=1)
+        import requests
+        r = requests.get(url, path, timeout=10)
+        r.raise_for_status()
+        if size is not None:
+            assert len(r.content) == size
+        os.makedirs(os.path.dirname(path), exist_ok=1)
+        with open(path, 'wb') as f:
+            f.write(r.content)
+    return path