ROB: Handle missing destinations in reader (#840)

pubpub-zz · web-flow · commit 3fe9e6ef0e54 · 2022-04-30T20:13:24.000+02:00
If a destination is missing, getDestinationPageNumber now returns -1 If `strict=False`, the first page is used as a fallback. The code triggering the exception was ```python from PyPDF2 import PdfFileReader # https://github.com/mstamy2/PyPDF2/files/6045010/thyroid.pdf with open("thyroid.pdf", "rb") as f: reader = PdfFileReader(f) bookmarks = pdf.getOutlines() for b in bookmarks: print(reader.getDestinationPageNumber(b) + 1) # page count starts from 0 ``` The error message was: PyPDF2.utils.PdfReadError: Unknown Destination Type: 0 Closes #604 Closes #821
diff --git a/PyPDF2/_reader.py b/PyPDF2/_reader.py
@@ -557,6 +557,8 @@ def _getPageNumberByIndirect(self, indirectRef):
                 id2num[x.indirectRef.idnum] = i
             self._pageId2Num = id2num
 
+        if isinstance(indirectRef, NullObject):
+             return -1
         if isinstance(indirectRef, int):
             idnum = indirectRef
         else:
@@ -595,7 +597,17 @@ def getDestinationPageNumber(self, destination):
     def _buildDestination(self, title, array):
         page, typ = array[0:2]
         array = array[2:]
-        return Destination(title, page, typ, *array)
+        try:
+            return Destination(title, page, typ, *array)
+        except PdfReadError:
+            warnings.warn("Unknown destination : " + title + " " + str(array))
+            if self.strict:
+                raise
+            else:
+                #create a link to first Page
+                return Destination(title, self.getPage(0).indirectRef,
+                                   TextStringObject("/Fit"))
+
 
     def _buildOutline(self, node):
         dest, title, outline = None, None, None
diff --git a/Resources/issue-604.pdf b/Resources/issue-604.pdf
diff --git a/Tests/test_reader.py b/Tests/test_reader.py
@@ -547,6 +547,42 @@ def test_reader_properties():
     assert reader.pageMode is None
     assert reader.isEncrypted is False
 
+@pytest.mark.parametrize(
+    "strict",
+    [(True), (False)],
+)
+def test_issue604(strict):
+    """
+    Test with invalid destinations
+    """
+    with open(os.path.join(RESOURCE_ROOT, "issue-604.pdf"), "rb") as f:
+        pdf = None
+        bookmarks = None
+        if strict:
+            with pytest.raises(PdfReadError) as exc:
+                pdf = PdfFileReader(f, strict=strict)
+                bookmarks = pdf.getOutlines()
+            if "Unknown Destination" not in exc.value.args[0]:
+                raise Exception("Expected exception not raised")
+            return # bookmarks not correct
+        else:
+            pdf = PdfFileReader(f, strict=strict)
+            bookmarks = pdf.getOutlines()
+
+        def getDestPages(x):
+            # print(x)
+            if isinstance(x,list):
+                r = [getDestPages(y) for y in x]
+                return r
+            else:
+                return pdf.getDestinationPageNumber(x) + 1
+
+        out = []
+        for (
+            b
+        ) in bookmarks:  # b can be destination or a list:preferred to just print them
+            out.append(getDestPages(b))
+    #print(out)
 
 def test_decode_permissions():
     reader = PdfFileReader(os.path.join(RESOURCE_ROOT, "crazyones.pdf"))
