Forward changes from JulienPalard's digitalbazaar#141

Author: meritasix

lib/pyld/documentloader/requests.py

@@ -1,40 +1,34 @@
 """
 Remote document loader using Requests.
-
 .. module:: jsonld.documentloader.requests
   :synopsis: Remote document loader using Requests
-
 .. moduleauthor:: Dave Longley
 .. moduleauthor:: Mike Johnson
 .. moduleauthor:: Tim McNamara <tim.mcnamara@okfn.org>
 .. moduleauthor:: Olaf Conradi <olaf@conradi.org>
 """
 import string
+import re
 import urllib.parse as urllib_parse
 
-from pyld.jsonld import (JsonLdError, parse_link_header, LINK_HEADER_REL)
+from pyld.jsonld import (JsonLdError, parse_link_header, LINK_HEADER_REL, prepend_base)
 
 
 def requests_document_loader(secure=False, **kwargs):
     """
     Create a Requests document loader.
-
     Can be used to setup extra Requests args such as verify, cert, timeout,
     or others.
-
     :param secure: require all requests to use HTTPS (default: False).
     :param **kwargs: extra keyword args for Requests get() call.
-
     :return: the RemoteDocument loader function.
     """
     import requests
 
     def loader(url, options={}):
         """
         Retrieves JSON-LD at the given URL.
-
         :param url: the URL to retrieve.
-
         :return: the RemoteDocument.
         """
         try:
@@ -69,30 +63,31 @@ def loader(url, options={}):
                 'contentType': content_type,
                 'contextUrl': None,
                 'documentUrl': response.url,
-                'document': response.json()
             }
             link_header = response.headers.get('link')
             if link_header:
                 linked_context = parse_link_header(link_header).get(
                     LINK_HEADER_REL)
                 # only 1 related link header permitted
                 if linked_context and content_type != 'application/ld+json':
-                  if isinstance(linked_context, list):
-                      raise JsonLdError(
-                          'URL could not be dereferenced, '
-                          'it has more than one '
-                          'associated HTTP Link Header.',
-                          'jsonld.LoadDocumentError',
-                          {'url': url},
-                          code='multiple context link headers')
-                  doc['contextUrl'] = linked_context['target']
+                    if isinstance(linked_context, list):
+                        raise JsonLdError(
+                            "URL could not be dereferenced, "
+                            "it has more than one "
+                            "associated HTTP Link Header.",
+                            "jsonld.LoadDocumentError",
+                            {"url": url},
+                            code="multiple context link headers")
+                    doc["contextUrl"] = linked_context["target"]
                 linked_alternate = parse_link_header(link_header).get('alternate')
                 # if not JSON-LD, alternate may point there
                 if (linked_alternate and
                         linked_alternate.get('type') == 'application/ld+json' and
                         not re.match(r'^application\/(\w*\+)?json$', content_type)):
                     doc['contentType'] = 'application/ld+json'
-                    doc['documentUrl'] = jsonld.prepend_base(url, linked_alternate['target'])
+                    doc['documentUrl'] = prepend_base(url, linked_alternate['target'])
+                    return loader(doc['documentUrl'], options=options)
+            doc["document"] = response.json()
             return doc
         except JsonLdError as e:
             raise e
@@ -102,4 +97,4 @@ def loader(url, options={}):
                 'jsonld.LoadDocumentError', code='loading document failed',
                 cause=cause)
 
-    return loader
+    return loader