Relative URIs are resolved incorrectly after redirects

[ghost]

vfaronov, 2010-09-10T23:23:28.000Z

What steps will reproduce the problem?

1. Prepare a resource http://example.org/foo serving up an RDF description that contains relative URIs, for example <#frag1>.
2. Prepare a resource http://example.org/bar that redirects (for example, HTTP 301) to http://example.org/foo.
3. Use RDFLib's Graph.parse() to parse http://example.org/bar.

What is the expected output? What do you see instead?

I expect the "real" URI http://example.org/foo to be used as the base URI, giving absolute URIs of the form http://example.org/foo#frag1. Instead, RDFLib uses the original requested URI http://example.org/bar as the base, giving http://example.org/bar#frag1.

What version of the product are you using? On what operating system?

RDFLib trunk (r1895) on GNU/Linux.

Please provide any additional information below.

RFC 3986 Uniform Resource Identifier (URI): Generic Syntax
http://tools.ietf.org/html/rfc3986#section-5.1.3

"Note that if the retrieval was the result of a redirected request, the last URI used (i.e., the URI that resulted in the actual retrieval of the representation) is the base URI."

Comment 1 by vfaronov

For a working example, see

 <http://linked-data.ru/example>

which 301s to (RDFa).

<http://linked-data.ru/example/>

Comment 2 by vfaronov

First attempt at a patch.
This changes the base URI resolution logic a bit, and I'm not 100% sure it doesn't break anything.

Index: rdflib/parser.py
===================================================================
--- rdflib/parser.py    (revision 1895)
+++ rdflib/parser.py    (working copy)
@@ -94,9 +94,11 @@
         except HTTPError, e:
             # TODO:
             raise Exception('"%s" while trying to open "%s"' % (e, self.url))
+        self.url = file.geturl()    # in case redirections took place
         self.content_type = file.info().get('content-type')
         self.content_type = self.content_type.split(";", 1)[0]
         self.setByteStream(file)
+        self.setPublicId(self.url)
         # TODO: self.setEncoding(encoding)

     def __repr__(self):
@@ -147,6 +149,8 @@
             else:
                 raise Exception("Unexpected type '%s' for source '%s'" % (type(source), source))

+    absolute_location = None
+    
     if location is not None:
         base = urljoin("file:", "%s/" % pathname2url(os.getcwd()))
         absolute_location = URIRef(location, base=base).defrag()
@@ -155,7 +159,6 @@
             file = __builtin__.file(filename, "rb")
         else:
             input_source = URLInputSource(absolute_location, format)
-        publicID = publicID or absolute_location

     if file is not None:
         input_source = FileInputSource(file)
@@ -168,13 +171,11 @@
     if input_source is None:
         raise Exception("could not create InputSource")
     else:
-        if publicID:
+        if publicID is not None:
             input_source.setPublicId(publicID)
-
-        # TODO: what motivated this bit?
-        id = input_source.getPublicId()
-        if id is None:
-            input_source.setPublicId("")
+        elif input_source.getPublicId() is None:
+            input_source.setPublicId(absolute_location or "")
+        
         return input_source

[ghost]

Currently working:

>>> graph.parse(location="http://linked-data.ru/example")
<Graph identifier=xNAYeDDA0 (<class 'rdflib.graph.Graph'>)>

[ghost]

Test created, fix applied, test passes.