issue #2959: fix PdfDict attribute access, text decoding, tests

Tests/test_file_pdf.py

@@ -137,13 +137,13 @@ class TestFilePdf(PillowTestCase):
         pdf = pdfParser.PdfParser(pdf_filename)
         self.assertEqual(len(pdf.pages), 1)
         self.assertEqual(len(pdf.info), 1)
-        self.assertEqual(pdfParser.decode_text(pdf.info["Producer"]), "pdfParser")
+        self.assertEqual(pdf.info.Producer, "pdfParser")
         # append some info
-        pdf.info["Title"] = "abc"
-        pdf.info["Author"] = "def"
-        pdf.info["Subject"] = "ghi"
-        pdf.info["Keywords"] = "jkl"
-        pdf.info["Creator"] = "hopper()"
+        pdf.info.Title = "abc"
+        pdf.info.Author = "def"
+        pdf.info.Subject = u"ghi\uABCD"
+        pdf.info.Keywords = "qw)e\\r(ty"
+        pdf.info.Creator = "hopper()"
         with open(pdf_filename, "r+b") as f:
             f.seek(0, os.SEEK_END)
             pdf.write_xref_and_trailer(f)
@@ -151,7 +151,7 @@ class TestFilePdf(PillowTestCase):
         pdf = pdfParser.PdfParser(pdf_filename)
         self.assertEqual(len(pdf.pages), 1)
         self.assertEqual(len(pdf.info), 6)
-        self.assertEqual(pdfParser.decode_text(pdf.info["Title"]), "abc")
+        self.assertEqual(pdf.info.Title, "abc")
         # append two images
         mode_CMYK = hopper("CMYK")
         mode_P = hopper("P")
@@ -160,8 +160,11 @@ class TestFilePdf(PillowTestCase):
         pdf = pdfParser.PdfParser(pdf_filename)
         self.assertEqual(len(pdf.pages), 3)
         self.assertEqual(len(pdf.info), 6)
-        self.assertEqual(pdfParser.decode_text(pdf.info["Title"]), "abc")
-        self.assertEqual(pdfParser.decode_text(pdf.info["Producer"]), "pdfParser")
+        self.assertEqual(pdfParser.decode_text(pdf.info[b"Title"]), "abc")
+        self.assertEqual(pdf.info.Title, "abc")
+        self.assertEqual(pdf.info.Producer, "pdfParser")
+        self.assertEqual(pdf.info.Keywords, "qw)e\\r(ty")
+        self.assertEqual(pdf.info.Subject, u"ghi\uABCD")
 
     def test_pdf_append_to_bytesio(self):
         im = hopper("RGB")

src/PIL/PdfImagePlugin.py

@@ -62,17 +62,17 @@ def _save(im, fp, filename, save_all=False):
         existing_pdf = pdfParser.PdfParser()
 
     if title:
-        existing_pdf.info["Title"] = title
+        existing_pdf.info.Title = title
     if author:
-        existing_pdf.info["Author"] = author
+        existing_pdf.info.Author = author
     if subject:
-        existing_pdf.info["Subject"] = subject
+        existing_pdf.info.Subject = subject
     if keywords:
-        existing_pdf.info["Keywords"] = keywords
+        existing_pdf.info.Keywords = keywords
     if creator:
-        existing_pdf.info["Creator"] = creator
+        existing_pdf.info.Creator = creator
     if producer:
-        existing_pdf.info["Producer"] = producer
+        existing_pdf.info.Producer = producer
 
     #
     # make sure image data is available

src/PIL/pdfParser.py

@@ -186,7 +186,7 @@ class PdfName():
         elif isinstance(name, bytes):
             self.name = name
         else:
-            self.name = name.encode("utf-8")
+            self.name = name.encode("us-ascii")
 
     @classmethod
     def from_pdf_stream(klass, data):
@@ -224,6 +224,24 @@ class PdfDict(UserDict):
     def __init__(self, *args, **kwargs):
         UserDict.__init__(self, *args, **kwargs)
 
+    def __setattr__(self, key, value):
+        if key == "data":
+            UserDict.__setattr__(self, key, value)
+        else:
+            if isinstance(key, str):
+                key = key.encode("us-ascii")
+            self[key] = value
+
+    def __getattr__(self, key):
+        try:
+            value = self[key]
+        except KeyError:
+            value = self[key.encode("us-ascii")]
+        if isinstance(value, bytes):
+            return decode_text(value)
+        else:
+            return value
+
     def __bytes__(self):
         out = bytearray(b"<<")
         for key, value in self.items():
@@ -624,6 +642,7 @@ class PdfParser:
         b"f": b"\f",
         b"(": b"(",
         b")": b")",
+        b"\\": b"\\",
         ord(b"n"): b"\n",
         ord(b"r"): b"\r",
         ord(b"t"): b"\t",
@@ -631,6 +650,7 @@ class PdfParser:
         ord(b"f"): b"\f",
         ord(b"("): b"(",
         ord(b")"): b")",
+        ord(b"\\"): b"\\",
         }
     @classmethod
     def get_literal_string(klass, data, offset):