fix the html_error_template not handling tracebacks from normal .py files with

a magic encoding comment

CHANGES

+0.2.3 
+- fixed the html_error_template not handling tracebacks from
+normal .py files with a magic encoding comment [ticket:88]
+
 0.2.2
 - cached blocks now use the current context when rendering
 an expired section, instead of the original context

lib/mako/exceptions.py

@@ -7,6 +7,7 @@
 """exception classes"""
 
 import traceback, sys, re
+from mako import util
 
 class MakoException(Exception):
     pass
@@ -139,7 +140,14 @@ class RichTraceback(object):
                     break
             else:
                 try:
-                    self.source = file(new_trcback[-1][0]).read()
+                    # A normal .py file (not a Template)
+                    fp = open(new_trcback[-1][0])
+                    encoding = util.parse_encoding(fp)
+                    fp.seek(0)
+                    self.source = fp.read()
+                    fp.close()
+                    if encoding:
+                        self.source = self.source.decode(encoding)
                 except IOError:
                     self.source = ''
                 self.lineno = new_trcback[-1][1]

lib/mako/util.py

@@ -16,7 +16,7 @@ try:
 except:
     from StringIO import StringIO
 
-import weakref, os, time
+import codecs, re, weakref, os, time
 
 try:
     import threading
@@ -130,6 +130,56 @@ class LRUCache(dict):
                     # on us. loop around and try again
                     break
 
+# Regexp to match python magic encoding line
+_PYTHON_MAGIC_COMMENT_re = re.compile(
+    r'[ \t\f]* \# .* coding[=:][ \t]*([-\w.]+)',
+    re.VERBOSE)
+
+def parse_encoding(fp):
+    """Deduce the encoding of a source file from magic comment.
+
+    It does this in the same way as the `Python interpreter`__
+
+    .. __: http://docs.python.org/ref/encodings.html
+
+    The ``fp`` argument should be a seekable file object.
+    """
+    pos = fp.tell()
+    fp.seek(0)
+    try:
+        line1 = fp.readline()
+        has_bom = line1.startswith(codecs.BOM_UTF8)
+        if has_bom:
+            line1 = line1[len(codecs.BOM_UTF8):]
+
+        m = _PYTHON_MAGIC_COMMENT_re.match(line1)
+        if not m:
+            try:
+                import parser
+                parser.suite(line1)
+            except (ImportError, SyntaxError):
+                # Either it's a real syntax error, in which case the source
+                # is not valid python source, or line2 is a continuation of
+                # line1, in which case we don't want to scan line2 for a magic
+                # comment.
+                pass
+            else:
+                line2 = fp.readline()
+                m = _PYTHON_MAGIC_COMMENT_re.match(line2)
+
+        if has_bom:
+            if m:
+                raise SyntaxError, \
+                      "python refuses to compile code with both a UTF8" \
+                      " byte-order-mark and a magic encoding comment"
+            return 'utf_8'
+        elif m:
+            return m.group(1)
+        else:
+            return None
+    finally:
+        fp.seek(pos)
+
 def restore__ast(_ast):
     """Attempt to restore the required classes to the _ast module if it
     appears to be missing them

test/exceptions_.py

@@ -60,6 +60,15 @@ ${u'привет'}
             assert False, ("This function should trigger a CompileException, "
                            "but didn't")
     
+    def test_py_utf8_html_error_template(self):
+        try:
+            foo = u'日本'
+            raise RuntimeError('test')
+        except:
+            html_error = exceptions.html_error_template().render()
+            assert 'RuntimeError: test' in html_error
+            assert "foo = u'日本'" in html_error
+
     def test_format_exceptions(self):
         l = TemplateLookup(format_exceptions=True)