From d0adbcbcf372a5568e64db96a472c8ccafa9f508 Mon Sep 17 00:00:00 2001
From: Julien Palard <julien@palard.fr>
Date: Fri, 9 Oct 2020 13:12:28 +0200
Subject: [PATCH] Always follow link header if content-type is not json.

---
 lib/pyld/documentloader/requests.py | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/lib/pyld/documentloader/requests.py b/lib/pyld/documentloader/requests.py
index 77f42e1d..40a34939 100644
--- a/lib/pyld/documentloader/requests.py
+++ b/lib/pyld/documentloader/requests.py
@@ -10,9 +10,10 @@
 .. 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):
@@ -69,7 +70,6 @@ def loader(url, options={}):
                 'contentType': content_type,
                 'contextUrl': None,
                 'documentUrl': response.url,
-                'document': response.json()
             }
             link_header = response.headers.get('link')
             if link_header:
@@ -77,22 +77,24 @@ def loader(url, options={}):
                     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