diffoscope/comparators/pdf.py

@@ -34,6 +34,8 @@ from .utils.command import Command
 
 logger = logging.getLogger(__name__)
 
+PYPDF_MAJOR_VERSION = None
+
 try:
     try:
         # PyPDF 3.x
@@ -43,12 +45,22 @@ try:
         import PyPDF2 as pypdf
 
     try:
-        # pyPDF 2.x-3.x
-        import pypdf.PdfReader as PdfReader
+        try:
+            # pyPDF 3.x
+            from pypdf import PdfReader
+
+            PYPDF_MAJOR_VERSION = 3
+        except ImportError:
+            # pyPDF 2.x
+            import pypdf.PdfReader as PdfReader
+
+            PYPDF_MAJOR_VERSION = 2
     except ImportError:
         # PyPDF2 1.x
         import pypdf.PdfFileReader as PdfReader
 
+        PYPDF_MAJOR_VERSION = 1
+
     try:
         # PyPDF 3.x
         from pypdf.errors import PdfReadError
@@ -84,37 +96,10 @@ class PdfFile(File):
     def compare_details(self, other, source=None):
         xs = []
 
-        if pypdf is None:
-            pkg = get_package_provider("pypdf")
-            infix = f" from the '{pkg}' package " if pkg else " "
-            self.add_comment(
-                f"Installing the 'pypdf' Python module{infix}"
-                "may produce a better output."
-            )
-        else:
-            difference = Difference.from_text(
-                self.dump_pypdf_metadata(self),
-                self.dump_pypdf_metadata(other),
-                self.name,
-                other.name,
-            )
-            if difference:
-                difference.add_comment("Document info")
-            xs.append(difference)
-
-            difference = Difference.from_text(
-                self.dump_pypdf_annotations(self),
-                self.dump_pypdf_annotations(other),
-                self.name,
-                other.name,
-            )
-            if difference:
-                difference.add_comment("Annotations")
-            xs.append(difference)
-
+        xs.extend(self.gen_metadata_differences(other))
         xs.append(Difference.from_operation(Pdftotext, self.path, other.path))
 
-        # Don't include verbose dumppdf output unless we won't see any any
+        # Don't include verbose dumppdf output unless we won't see any
         # differences without it.
         if not any(xs):
             xs.append(
@@ -123,6 +108,38 @@ class PdfFile(File):
 
         return xs
 
+    def gen_metadata_differences(self, other):
+        if PYPDF_MAJOR_VERSION is None:
+            pkg = get_package_provider("pypdf")
+            infix = f" from the '{pkg}' package " if pkg else " "
+            self.add_comment(
+                f"Installing the 'pypdf' Python module{infix}"
+                "may produce a better output."
+            )
+
+            return
+
+        difference = Difference.from_text(
+            self.dump_pypdf_metadata(self),
+            self.dump_pypdf_metadata(other),
+            self.name,
+            other.name,
+        )
+        if difference:
+            difference.add_comment("Document info")
+        yield difference
+
+        difference = Difference.from_text(
+            self.dump_pypdf_annotations(self),
+            self.dump_pypdf_annotations(other),
+            self.name,
+            other.name,
+        )
+        if difference:
+            difference.add_comment("Annotations")
+        yield difference
+
+
     def dump_pypdf_metadata(self, file):
         try:
             pdf = PdfReader(file.path)
@@ -154,9 +171,14 @@ class PdfFile(File):
 
                 try:
                     for annot in page["/Annots"]:
-                        subtype = annot.getObject()["/Subtype"]
-                        if subtype == "/Text":
-                            xs.append(annot.getObject()["/Contents"])
+                        if PYPDF_MAJOR_VERSION == 3:
+                            subtype = annot.get_object()["/Subtype"]
+                            if subtype == "/Text":
+                                xs.append(annot.get_object()["/Contents"])
+                        else:
+                            subtype = annot.getObject()["/Subtype"]
+                            if subtype == "/Text":
+                                xs.append(annot.getObject()["/Contents"])
                 except KeyError:
                     pass
 

tests/comparators/test_pdf.py

@@ -22,7 +22,11 @@ import pytest
 from diffoscope.comparators.pdf import PdfFile
 
 from ..utils.data import load_fixture, assert_diff
-from ..utils.tools import skip_unless_tools_exist, skip_unless_module_exists
+from ..utils.tools import (
+    skip_unless_tools_exist,
+    skip_unless_module_exists,
+    skipif,
+)
 from ..utils.nonexisting import assert_non_existing
 
 
@@ -34,6 +38,18 @@ pdf1a = load_fixture("test_weird_non_unicode_chars1.pdf")
 pdf2a = load_fixture("test_weird_non_unicode_chars2.pdf")
 
 
+def skip_unless_pypdf3():
+    def fn():
+        try:
+            import pypdf
+        except ImportError:
+            return True
+
+        return not pypdf.__version__.startswith("3.")
+
+    return skipif(fn(), reason="pypdf not installed or not version 3.x+")
+
+
 def test_identification(pdf1):
     assert isinstance(pdf1, PdfFile)
 
@@ -69,9 +85,11 @@ def differences_metadata(pdf1, pdf1a):
     return pdf1.compare(pdf1a).details
 
 
+@skip_unless_pypdf3()
 @skip_unless_tools_exist("pdftotext")
-@skip_unless_module_exists("pypdf")
 def test_metadata(differences_metadata):
+    assert len(differences_metadata) == 2
+
     assert_diff(differences_metadata[0], "pdf_metadata_expected_diff")
 
 
@@ -80,7 +98,7 @@ def differences_annotations(pdf3, pdf4):
     return pdf3.compare(pdf4).details
 
 
+@skip_unless_pypdf3()
 @skip_unless_tools_exist("pdftotext")
-@skip_unless_module_exists("pypdf")
 def test_annotations(differences_annotations):
     assert_diff(differences_annotations[0], "pdf_annotations_expected_diff")

tests/data/pdf_annotations_expected_diff

-@@ -1,3 +1,7 @@
- This is a test PDF.
- 
-+10: Typewriter annotation
-+2: Another type of
-+comment
-+
- 
+@@ -0,0 +1,2 @@
++1: A PDF comment created in Okular
++11: Comment created with evince