Add a helper for directly output pdf streams.

Also, such directly output streams don't need an explicitly set Length,
as the stream outputter will fill in that entry itself.  This is why
e.g. the invalid entry for toUnicodeMapObject (where Length was set to
unicode_cmap rather than len(unicode_cmap)) was not a problem: it was
overwritten in endStream().

Author: anntzer

lib/matplotlib/backends/backend_pdf.py

@@ -849,6 +849,11 @@ def endStream(self):
             self.currentstream.end()
             self.currentstream = None
 
+    def outputStream(self, ref, data, *, extra=None):
+        self.beginStream(ref.id, None, extra)
+        self.currentstream.write(data)
+        self.endStream()
+
     def _write_annotations(self):
         for annotsObject, annotations in self._annotations:
             self.writeObject(annotsObject, annotations)
@@ -1053,13 +1058,10 @@ def createType1Descriptor(self, t1font, fontfile):
 
         self.writeObject(fontdescObject, descriptor)
 
-        self.beginStream(fontfileObject.id, None,
-                         {'Length1': len(t1font.parts[0]),
-                          'Length2': len(t1font.parts[1]),
-                          'Length3': 0})
-        self.currentstream.write(t1font.parts[0])
-        self.currentstream.write(t1font.parts[1])
-        self.endStream()
+        self.outputStream(fontfileObject, b"".join(t1font.parts[:2]),
+                          extra={'Length1': len(t1font.parts[0]),
+                                 'Length2': len(t1font.parts[1]),
+                                 'Length3': 0})
 
         return fontdescObject
 
@@ -1182,13 +1184,13 @@ def get_char_width(charcode):
             charprocs = {}
             for charname in sorted(rawcharprocs):
                 stream = rawcharprocs[charname]
-                charprocDict = {'Length': len(stream)}
+                charprocDict = {}
                 # The 2-byte characters are used as XObjects, so they
                 # need extra info in their dictionary
                 if charname in multi_byte_chars:
-                    charprocDict['Type'] = Name('XObject')
-                    charprocDict['Subtype'] = Name('Form')
-                    charprocDict['BBox'] = bbox
+                    charprocDict = {'Type': Name('XObject'),
+                                    'Subtype': Name('Form'),
+                                    'BBox': bbox}
                     # Each glyph includes bounding box information,
                     # but xpdf and ghostscript can't handle it in a
                     # Form XObject (they segfault!!!), so we remove it
@@ -1197,9 +1199,7 @@ def get_char_width(charcode):
                     # value.
                     stream = stream[stream.find(b"d1") + 2:]
                 charprocObject = self.reserveObject('charProc')
-                self.beginStream(charprocObject.id, None, charprocDict)
-                self.currentstream.write(stream)
-                self.endStream()
+                self.outputStream(charprocObject, stream, extra=charprocDict)
 
                 # Send the glyphs with ccode > 255 to the XObject dictionary,
                 # and the others to the font itself
@@ -1267,12 +1267,9 @@ def embedTTFType42(font, characters, descriptor):
 
             # Make fontfile stream
             descriptor['FontFile2'] = fontfileObject
-            self.beginStream(
-                fontfileObject.id,
-                self.reserveObject('length of font stream'),
-                {'Length1': fontdata.getbuffer().nbytes})
-            self.currentstream.write(fontdata.getvalue())
-            self.endStream()
+            self.outputStream(
+                fontfileObject, fontdata.getvalue(),
+                extra={'Length1': fontdata.getbuffer().nbytes})
 
             # Make the 'W' (Widths) array, CidToGidMap and ToUnicode CMap
             # at the same time
@@ -1331,10 +1328,9 @@ def embedTTFType42(font, characters, descriptor):
             rawcharprocs = _get_pdf_charprocs(filename, glyph_ids)
             for charname in sorted(rawcharprocs):
                 stream = rawcharprocs[charname]
-                charprocDict = {'Length': len(stream)}
-                charprocDict['Type'] = Name('XObject')
-                charprocDict['Subtype'] = Name('Form')
-                charprocDict['BBox'] = bbox
+                charprocDict = {'Type': Name('XObject'),
+                                'Subtype': Name('Form'),
+                                'BBox': bbox}
                 # Each glyph includes bounding box information,
                 # but xpdf and ghostscript can't handle it in a
                 # Form XObject (they segfault!!!), so we remove it
@@ -1343,27 +1339,17 @@ def embedTTFType42(font, characters, descriptor):
                 # value.
                 stream = stream[stream.find(b"d1") + 2:]
                 charprocObject = self.reserveObject('charProc')
-                self.beginStream(charprocObject.id, None, charprocDict)
-                self.currentstream.write(stream)
-                self.endStream()
+                self.outputStream(charprocObject, stream, extra=charprocDict)
 
                 name = self._get_xobject_glyph_name(filename, charname)
                 self.multi_byte_charprocs[name] = charprocObject
 
             # CIDToGIDMap stream
             cid_to_gid_map = "".join(cid_to_gid_map).encode("utf-16be")
-            self.beginStream(cidToGidMapObject.id,
-                             None,
-                             {'Length': len(cid_to_gid_map)})
-            self.currentstream.write(cid_to_gid_map)
-            self.endStream()
+            self.outputStream(cidToGidMapObject, cid_to_gid_map)
 
             # ToUnicode CMap
-            self.beginStream(toUnicodeMapObject.id,
-                             None,
-                             {'Length': unicode_cmap})
-            self.currentstream.write(unicode_cmap)
-            self.endStream()
+            self.outputStream(toUnicodeMapObject, unicode_cmap)
 
             descriptor['MaxWidth'] = max_width