[pypy-commit] pypy urlparse-unquote-faster: fix quadratic slowness of unquote (caused by buliding string with inplace addition)

[kostialopuhin]

Author: Konstantin Lopuhin <kostia.lopuhin at gmail.com>
Branch: urlparse-unquote-faster
Changeset: r61124:c7ea99e3b4db
Date: 2013-02-12 00:22 +0400
http://bitbucket.org/pypy/pypy/changeset/c7ea99e3b4db/

Log:	fix quadratic slowness of unquote (caused by buliding string with
	inplace addition)

diff --git a/lib-python/2.7/urllib.py b/lib-python/2.7/urllib.py
--- a/lib-python/2.7/urllib.py
+++ b/lib-python/2.7/urllib.py
@@ -1205,15 +1205,16 @@
     # fastpath
     if len(res) == 1:
         return s
-    s = res[0]
+    res_list = [res[0]]
     for item in res[1:]:
         try:
-            s += _hextochr[item[:2]] + item[2:]
+            x = _hextochr[item[:2]] + item[2:]
         except KeyError:
-            s += '%' + item
+            x = '%' + item
         except UnicodeDecodeError:
-            s += unichr(int(item[:2], 16)) + item[2:]
-    return s
+            x = unichr(int(item[:2], 16)) + item[2:]
+        res_list.append(x)
+    return ''.join(res_list)
 
 def unquote_plus(s):
     """unquote('%7e/abc+def') -> '~/abc def'"""
diff --git a/lib-python/2.7/urlparse.py b/lib-python/2.7/urlparse.py
--- a/lib-python/2.7/urlparse.py
+++ b/lib-python/2.7/urlparse.py
@@ -321,15 +321,16 @@
     # fastpath
     if len(res) == 1:
         return s
-    s = res[0]
+    res_list = [res[0]]
     for item in res[1:]:
         try:
-            s += _hextochr[item[:2]] + item[2:]
+            x = _hextochr[item[:2]] + item[2:]
         except KeyError:
-            s += '%' + item
+            x = '%' + item
         except UnicodeDecodeError:
-            s += unichr(int(item[:2], 16)) + item[2:]
-    return s
+            x = unichr(int(item[:2], 16)) + item[2:]
+        res_list.append(x)
+    return ''.join(res_list)
 
 def parse_qs(qs, keep_blank_values=0, strict_parsing=0):
     """Parse a query given as a string argument.