Ensure makeSafeAbsoluteURI() catches ValueError

`urlparse` in Python 2.7 and up throws a ValueError for URIs like:
    http://bad]uri/

Fixes issue 321. Thanks to Google user kristaps.rats for reporting this!

git-svn-id: http://feedparser.googlecode.com/svn/trunk@679 73d2b349-402e-0410-baf4-070fd12ab5b7

NEWS

@@ -5,6 +5,7 @@ Changes coming in the next release:
     * Issue 310 (pubDate should map to `published`, not `updated`)
     * Issue 313 (include the compression test files in MANIFEST.in)
     * Issue 315 (HTTP server for unit tests runs on 0.0.0.0)
+    * Issue 321 (malformed URIs can cause ValueError to be thrown)
     * Issue 325 (map `description_detail` to `summary_detail`)
 
 5.1 - December 2, 2011

feedparser/feedparser.py

@@ -2542,15 +2542,24 @@ def _resolveRelativeURIs(htmlSource, baseURI, encoding, _type):
 def _makeSafeAbsoluteURI(base, rel=None):
     # bail if ACCEPTABLE_URI_SCHEMES is empty
     if not ACCEPTABLE_URI_SCHEMES:
-        return _urljoin(base, rel or u'')
+        try:
+            return _urljoin(base, rel or u'')
+        except ValueError:
+            return u''
     if not base:
         return rel or u''
     if not rel:
-        scheme = urlparse.urlparse(base)[0]
+        try:
+            scheme = urlparse.urlparse(base)[0]
+        except ValueError:
+            return u''
         if not scheme or scheme in ACCEPTABLE_URI_SCHEMES:
             return base
         return u''
-    uri = _urljoin(base, rel)
+    try:
+        uri = _urljoin(base, rel)
+    except ValueError:
+        return u''
     if uri.strip().split(':', 1)[0] not in ACCEPTABLE_URI_SCHEMES:
         return u''
     return uri

feedparser/feedparsertest.py

@@ -319,6 +319,20 @@ def fn(self):
     test_rel = _mktest(u'/new', u'http://d.test/new', 'relative uri')
     test_bad = _mktest(u'x://bad.test/', u'', 'unacceptable uri protocol')
 
+    def test_catch_ValueError(self):
+        'catch ValueError in Python 2.7 and up'
+        uri = u'http://bad]test/'
+        value1 = feedparser._makeSafeAbsoluteURI(uri)
+        value2 = feedparser._makeSafeAbsoluteURI(self.base, uri)
+        swap = feedparser.ACCEPTABLE_URI_SCHEMES
+        feedparser.ACCEPTABLE_URI_SCHEMES = ()
+        value3 = feedparser._makeSafeAbsoluteURI(self.base, uri)
+        feedparser.ACCEPTABLE_URI_SCHEMES = swap
+        # Only Python 2.7 and up throw a ValueError, otherwise uri is returned
+        self.assertTrue(value1 in (uri, u''))
+        self.assertTrue(value2 in (uri, u''))
+        self.assertTrue(value3 in (uri, u''))
+
 class TestConvertToIdn(unittest.TestCase):
     "Test IDN support (unavailable in Jython as of Jython 2.5.2)"
     # this is the greek test domain