From 834723beda7859252a1eca2a8eaaa2a54dcd072c Mon Sep 17 00:00:00 2001
From: rastasheep <rastasheep3@gmail.com>
Date: Sun, 25 Oct 2015 20:47:38 +0100
Subject: [PATCH] Fallback to 'http' as default url schema if needed

---
 goose/__init__.py      | 6 ++++++
 goose/configuration.py | 4 ++++
 2 files changed, 10 insertions(+)

diff --git a/goose/__init__.py b/goose/__init__.py
index 409b5732..5565f978 100644
--- a/goose/__init__.py
+++ b/goose/__init__.py
@@ -22,6 +22,7 @@
 """
 import os
 import platform
+import urllib2
 from tempfile import mkstemp
 
 from goose.version import version_info, __version__
@@ -52,6 +53,11 @@ def extract(self, url=None, raw_html=None):
         Main method to extract an article object from a URL,
         pass in a url and get back a Article
         """
+        scheme, address = urllib2.splittype(url)
+
+        if not scheme:
+            url = self.config.default_scheme + url
+
         cc = CrawlCandidate(self.config, url, raw_html)
         return self.crawl(cc)
 
diff --git a/goose/configuration.py b/goose/configuration.py
index fcfa5b9a..827a3476 100644
--- a/goose/configuration.py
+++ b/goose/configuration.py
@@ -99,6 +99,10 @@ def __init__(self):
         # http timeout
         self.http_timeout = HTTP_DEFAULT_TIMEOUT
 
+        # default url scheme
+        # it will be use as fallback if url doesnt conain one
+        self.default_scheme = 'http://'
+
     def get_parser(self):
         return AVAILABLE_PARSERS[self.parser_class]