ENH: Add color and font_format to PdfReader.outlines[i] (py-pdf#1104)

PyPDF2/_reader.py

@@ -78,6 +78,7 @@
     DictionaryObject,
     EncodedStreamObject,
     Field,
+    FloatObject,
     IndirectObject,
     NameObject,
     NullObject,
@@ -797,74 +798,56 @@ def _build_destination(
         title: str,
         array: List[Union[NumberObject, IndirectObject, NullObject, DictionaryObject]],
     ) -> Destination:
-        page, typ = None, None
-
-        if isinstance(array, type(None)):
-            # handle outlines without destination
-            page = NullObject()
-            typ = TextStringObject("/Fit")
-            return Destination(title, page, typ)
-        else:
-            page, typ = array[0:2]
-            array = array[2:]
-            try:
-                return Destination(title, page, typ, *array)  # type: ignore
-            except PdfReadError:
-                warnings.warn(f"Unknown destination: {title} {array}", PdfReadWarning)
-                if self.strict:
-                    raise
-                else:
-                    # create a link to first Page
-                    tmp = self.pages[0].indirect_ref
-                    indirect_ref = NullObject() if tmp is None else tmp
-                    return Destination(
-                        title, indirect_ref, TextStringObject("/Fit")  # type: ignore
-                    )
+        page, typ = array[0:2]
+        array = array[2:]
+        try:
+            return Destination(title, page, typ, *array)  # type: ignore
+        except PdfReadError:
+            warnings.warn(f"Unknown destination: {title} {array}", PdfReadWarning)
+            if self.strict:
+                raise
+            else:
+                # create a link to first Page
+                tmp = self.pages[0].indirect_ref
+                indirect_ref = NullObject() if tmp is None else tmp
+                return Destination(
+                    title, indirect_ref, TextStringObject("/Fit")  # type: ignore
+                )
 
     def _build_outline(self, node: DictionaryObject) -> Optional[Destination]:
         dest, title, outline = None, None, None
 
-        # title required for valid outline
-        try:
-            title = node["/Title"]
-        except KeyError:
-            raise PdfReadError(f"Invalid Outline Entry, Missing /Title attribute: {node!r}")
-
-        if "/A" in node:
+        if "/A" in node and "/Title" in node:
             # Action, section 8.5 (only type GoTo supported)
+            title = node["/Title"]
             action = cast(DictionaryObject, node["/A"])
             action_type = cast(NameObject, action[GoToActionArguments.S])
             if action_type == "/GoTo":
                 dest = action[GoToActionArguments.D]
-        elif "/Dest" in node:
+        elif "/Dest" in node and "/Title" in node:
             # Destination, section 8.2.1
+            title = node["/Title"]
             dest = node["/Dest"]
 
-        if isinstance(dest, ArrayObject):
-            outline = self._build_destination(title, dest)  # type: ignore
-        elif isinstance(dest, str):
-            # named destination
-            try:
-                outline = self._get_named_destinations()[dest]
+        # if destination found, then create outline
+        if dest:
+            if isinstance(dest, ArrayObject):
+                outline = self._build_destination(title, dest)  # type: ignore
+            elif isinstance(dest, str) and dest in self._namedDests:
+                outline = self._namedDests[dest]
                 outline[NameObject("/Title")] = title  # type: ignore
-            except KeyError:
-                # named destination not found in Name Dict
-                outline = self._build_destination(title, None)
-        elif isinstance(dest, type(None)):
-            # outline not required to have destination or GoTo action
-            # Table 8.4
-            outline = self._build_destination(title, dest)  # type: ignore
-        else:
-            raise PdfReadError(f"Unexpected destination {dest!r}")
-
-        # if color or text format specifications present, add to outline
-        if "/C" in node:
-            # Color of outline in (R, G, B) with values ranging 0.0-1.0
-            outline[NameObject("/C")] = node["/C"]
-        if "/F" in node:
-            # specifies style characteristics bold and/or italic
-            # 1=italic, 2=bold, 3=both
-            outline[NameObject("/F")] = node["/F"]
+            else:
+                raise PdfReadError(f"Unexpected destination {dest!r}")
+
+        # if outline created, add color and format if present
+        if outline:
+            if "/C" in node:
+                # Color of outline in (R, G, B) with values ranging 0.0-1.0
+                outline[NameObject("/C")] = ArrayObject(FloatObject(c) for c in node["/C"])  # type: ignore
+            if "/F" in node:
+                # specifies style characteristics bold and/or italic
+                # 1=italic, 2=bold, 3=both
+                outline[NameObject("/F")] = node["/F"]
 
         return outline
 

PyPDF2/generic.py

@@ -49,7 +49,6 @@
     cast,
 )
 from enum import IntFlag
-
 from ._codecs import (  # noqa: rev_encoding
     _pdfdoc_encoding,
     _pdfdoc_encoding_rev,
@@ -1733,9 +1732,9 @@ def additionalActions(self) -> Optional[DictionaryObject]:  # pragma: no cover
         return self.additional_actions
 
 
-class OutlineItemFlag(IntFlag):
+class OutlineFontFlag(IntFlag):
     """
-    A class used as an enumerable flag for formatting outline font
+    A class used as an enumerable flag for formatting an outline font
     """
 
     italic = 1
@@ -1892,10 +1891,10 @@ def bottom(self) -> Optional[FloatObject]:
     @property
     def color(self) -> Optional[tuple]:
         """Read-only property accessing the color in (R, G, B) with values 0.0-1.0"""
-        return self.get("/C", (0.0, 0.0, 0.0))
+        return self.get("/C", [FloatObject(0), FloatObject(0), FloatObject(0)])
 
     @property
-    def fmt(self) -> Optional[OutlineItemFlag]:
+    def font_format(self) -> Optional[OutlineFontFlag]:
         """Read-only property accessing the font type. 1=italic, 2=bold, 3=both"""
         return self.get("/F", 0)
 

tests/test_reader.py

@@ -858,3 +858,17 @@ def test_header(src, pdf_header):
     reader = PdfReader(src)
 
     assert reader.pdf_header == pdf_header
+
+
+def test_outline_color():
+    url = "https://corpora.tika.apache.org/base/docs/govdocs1/924/924546.pdf"
+    name = "tika-924546.pdf"
+    reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
+    assert reader.outlines[0].color == [0, 0, 1]
+
+
+def test_outline_font_format():
+    url = "https://corpora.tika.apache.org/base/docs/govdocs1/924/924546.pdf"
+    name = "tika-924546.pdf"
+    reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
+    assert reader.outlines[0].font_format == 2