Improve test, fix an off-by-one and simplify the code

jkseppan · jkseppan · commit 66d974a65d77 · 2015-06-14T22:37:32.000+03:00
diff --git a/lib/matplotlib/tests/test_type1font.py b/lib/matplotlib/tests/test_type1font.py
@@ -7,30 +7,23 @@
 import matplotlib.type1font as t1f
 import os.path
 import difflib
-import hashlib
-
-
-def sha1(data):
-    hash = hashlib.sha1()
-    hash.update(data)
-    return hash.hexdigest()
 
 
 def test_Type1Font():
     filename = os.path.join(os.path.dirname(__file__), 'cmr10.pfb')
     font = t1f.Type1Font(filename)
     slanted = font.transform({'slant': 1})
     condensed = font.transform({'extend': 0.5})
-    assert_equal(map(sha1, font.parts),
-                 ['f4ce890d648e67d413a91b3109fe67a732ced96f',
-                  'af46adb6c528956580c125c6abf3b5eb9983bbc1',
-                  'e2538a88a810bc207cfa1194b658ee8967042db8'])
+    rawdata = open(filename, 'rb').read()
+    assert_equal(font.parts[0], rawdata[0x0006:0x10c5])
+    assert_equal(font.parts[1], rawdata[0x10cb:0x897f])
+    assert_equal(font.parts[2], rawdata[0x8985:0x8ba6])
     assert_equal(font.parts[1:], slanted.parts[1:])
     assert_equal(font.parts[1:], condensed.parts[1:])
 
     differ = difflib.Differ()
-    diff = set(differ.compare(font.parts[0].splitlines(),
-                              slanted.parts[0].splitlines()))
+    diff = set(differ.compare(font.parts[0].decode('latin-1').splitlines(),
+                              slanted.parts[0].decode('latin-1').splitlines()))
     for line in (
          # Removes UniqueID
          '- FontDirectory/CMR10 known{/CMR10 findfont dup/UniqueID known{dup',
diff --git a/lib/matplotlib/type1font.py b/lib/matplotlib/type1font.py
@@ -29,6 +29,7 @@
 from six.moves import filter
 from six import unichr
 
+import binascii
 import io
 import itertools
 import numpy as np
@@ -92,8 +93,7 @@ def _read(self, file):
             if type == 1:       # ASCII text: include verbatim
                 data += segment
             elif type == 2:     # binary data: encode in hexadecimal
-                data += b''.join([('%02x' % ord(char)).encode('ascii')
-                                  for char in segment])
+                data += binascii.hexlify(segment)
             elif type == 3:     # end of file
                 break
             else:
@@ -123,9 +123,8 @@ def _split(self, data):
         # zeros backward
         idx = data.rindex(b'cleartomark') - 1
         zeros = 512
-        while zeros and ord(data[idx]) in (
-            ord(b'0'[0]), ord(b'\n'[0]), ord(b'\r'[0])):
-            if ord(data[idx]) == ord(b'0'[0]):
+        while zeros and data[idx] in b'0\n\r':
+            if data[idx] in b'0':
                 zeros -= 1
             idx -= 1
         if zeros:
@@ -136,10 +135,9 @@ def _split(self, data):
         # but if we read a pfa file, this part is already in hex, and
         # I am not quite sure if even the pfb format guarantees that
         # it will be in binary).
-        binary = b''.join([unichr(int(data[i:i + 2], 16)).encode('latin-1')
-                           for i in range(len1, idx, 2)])
+        binary = binascii.unhexlify(data[len1:idx+1])
 
-        return data[:len1], binary, data[idx:]
+        return data[:len1], binary, data[idx+1:]
 
     _whitespace_re = re.compile(br'[\0\t\r\014\n ]+')
     _token_re = re.compile(br'/{0,2}[^]\0\t\r\v\n ()<>{}/%[]+')
