# HG changeset patch
# User souliane <souliane@mailoo.org>
# Date 1452857254 -3600
# Node ID 07390a9d1c098f0a54c9aa0a0ac49587d6911e19
# Parent  5f4d688d8b6eb5ad9788de2c92e36967fe155df4
tools (strings): add method fixXHTMLLinks to add a scheme if missing, and force opening in new tab

diff -r 5f4d688d8b6e -r 07390a9d1c09 frontends/src/tools/strings.py
--- a/frontends/src/tools/strings.py	Wed Jan 13 22:10:06 2016 +0100
+++ b/frontends/src/tools/strings.py	Fri Jan 15 12:27:34 2016 +0100
@@ -18,6 +18,7 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 import re
+from feed.atom import Content
 
 
 def getURLParams(url):
@@ -72,3 +73,24 @@
         return '<a href="%s" target="_blank">%s</a>' % (url, match.group(0))
     pattern = r"""<img[^>]* src="([^"]+)"[^>]*>"""
     return re.sub(pattern, repl, string)
+
+
+def fixXHTMLLinks(xhtml):
+    """Add http:// if the scheme is missing and force opening in a new window.
+
+    @param string (unicode): XHTML Content
+    """
+    subs = []
+    for match in re.finditer(r'<a( \w+="[^"]*")* ?/?>', xhtml):
+        tag = match.group(0)
+        url = re.search(r'href="([^"]*)"', tag)
+        if url:
+            if not re.search(r'target="([^"]*)"', tag):  # no target
+                subs.append((tag, '<a target="_blank"%s' % tag[2:]))
+            if not re.match(r"^\w+://", url.group(1)):  # no scheme
+                subs.append((url.group(0), 'href="http://%s"' % url.group(1)))
+
+    for url, new_url in subs:
+        xhtml = xhtml.replace(url, new_url)
+    return xhtml
+