Return original document if present when requesting an object

ckanext/harvest/controllers/view.py

@@ -252,7 +252,15 @@ class ViewController(BaseController):
 
             # Check content type. It will probably be either XML or JSON
             try:
-                content = re.sub('<\?xml(.*)\?>','', obj['content'])
+
+                if obj['content']:
+                    content = obj['content']
+                elif 'original_document' in obj['extras']:
+                    content = obj['extras']['original_document']
+                else:
+                    abort(404,_('No content found'))
+
+                content = re.sub('<\?xml(.*)\?>','',content)
                 etree.fromstring(content)
                 response.content_type = 'application/xml'
             except XMLSyntaxError:
@@ -262,8 +270,8 @@ class ViewController(BaseController):
                 except ValueError:
                     pass
 
-            response.headers['Content-Length'] = len(obj['content'])
+            response.headers['Content-Length'] = len(content)
-            return obj['content']
+            return content
         except NotFound:
             abort(404,_('Harvest object not found'))
         except NotAuthorized,e:

ckanext/harvest/logic/dictization.py

@@ -44,10 +44,13 @@ def harvest_object_dictize(obj, context):
         out['package'] = obj.package.id
 
     out['errors'] = []
-
     for error in obj.errors:
         out['errors'].append(error.as_dict())
 
+    out['extras'] = {}
+    for extra in obj.extras:
+        out['extras'][extra.key] = extra.value
+
     return out
 
 def _get_source_status(source, context):