From a41c64bfb97a7bba8ec138345d48b94d8734d27e Mon Sep 17 00:00:00 2001
From: Leonid Amirov <leonid.amirov@gmail.com>
Date: Mon, 2 Nov 2015 16:06:21 +0300
Subject: [PATCH] issue GH #1550 - fixed bugs in
 scrapy.utils.url.add_http_if_no_scheme(): when given URI where scheme is
 present, but not 'http' the function gave bad result

---
 scrapy/utils/url.py | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/scrapy/utils/url.py b/scrapy/utils/url.py
index 398407a64..3eac5fb3d 100644
--- a/scrapy/utils/url.py
+++ b/scrapy/utils/url.py
@@ -6,6 +6,7 @@ Some of the functions that used to be imported from this module have been moved
 to the w3lib.url module. Always import those from there instead.
 """
 import posixpath
+from urlparse import urlsplit, urlunsplit
 from six.moves.urllib.parse import (ParseResult, urlunparse, urldefrag,
                                     urlparse, parse_qsl, urlencode,
                                     unquote)
@@ -114,10 +115,16 @@ def escape_ajax(url):
 
 def add_http_if_no_scheme(url):
     """Add http as the default scheme if it is missing from the url."""
-    if url.startswith('//'):
-        url = 'http:' + url
-        return url
-    parser = parse_url(url)
-    if not parser.scheme or not parser.netloc:
-        url = 'http://' + url
-    return url
+    parts = urlsplit(url)
+    scheme = parts.scheme or "http"
+    if parts.netloc:
+        netloc = parts.netloc
+        path = parts.path
+    else:
+        path_parts = url.split("/", 1)
+        netloc = path_parts[0]
+        path = path_parts[1] if len(path_parts) > 1 else "/"
+
+    return urlunsplit((
+        scheme, netloc, path, parts.query, parts.fragment
+    ))