From: Éric Araujo <merwok@netwok.org>
Date: Sat, 10 Sep 2011 16:10:58 +0000 (+0200)
Subject: Use bytes regex instead of decoding whole pages
X-Git-Tag: v3.3.0a1~1544^2~4
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=030cfe26a336826d08362cd60c8ea4be7775844b;p=thirdparty%2FPython%2Fcpython.git

Use bytes regex instead of decoding whole pages
---

diff --git a/Lib/packaging/pypi/simple.py b/Lib/packaging/pypi/simple.py
index 710355d6480e..76aad02416d0 100644
--- a/Lib/packaging/pypi/simple.py
+++ b/Lib/packaging/pypi/simple.py
@@ -159,22 +159,20 @@ class Crawler(BaseClient):
 
         Return a list of names.
         """
-        with self._open_url(self.index_url) as index:
-            if '*' in name:
-                name.replace('*', '.*')
-            else:
-                name = "%s%s%s" % ('*.?', name, '*.?')
-            name = name.replace('*', '[^<]*')  # avoid matching end tag
-            projectname = re.compile('<a[^>]*>(%s)</a>' % name, re.I)
-            matching_projects = []
+        if '*' in name:
+            name.replace('*', '.*')
+        else:
+            name = "%s%s%s" % ('*.?', name, '*.?')
+        name = name.replace('*', '[^<]*')  # avoid matching end tag
+        pattern = ('<a[^>]*>(%s)</a>' % name).encode('utf-8')
+        projectname = re.compile(pattern, re.I)
+        matching_projects = []
 
+        with self._open_url(self.index_url) as index:
             index_content = index.read()
 
-        # FIXME should use bytes I/O and regexes instead of decoding
-        index_content = index_content.decode()
-
         for match in projectname.finditer(index_content):
-            project_name = match.group(1)
+            project_name = match.group(1).decode('utf-8')
             matching_projects.append(self._get_project(project_name))
         return matching_projects