[#5] Add method for returning the original URL for a document

Harvesters implementing IHarvester can define a `get_original_url`
method that should return a URL pointing to the original location of a
document in the remote server. If present, this URL will be used on the
job reports.

Examples:
* For a CKAN record: http://{ckan-instance}/api/rest/{guid}
* For a WAF record: http://{waf-root}/{file-name}
* For a CSW record: http://{csw-server}/?Request=GetElementById&Id={guid}&...

ckanext/harvest/logic/action/get.py

@@ -161,22 +161,38 @@ def harvest_job_report(context, data_dict):
     if not job:
         raise NotFound
 
-    q = model.Session.query(harvest_model.HarvestObjectError) \
+    # Check if the harvester for this job's source has a method for returning
+    # the URL to the original document
+    original_url_builder = None
+    for harvester in PluginImplementations(IHarvester):
+        if harvester.info()['name'] == job.source.type:
+             if hasattr(harvester, 'get_original_url'):
+                original_url_builder = harvester.get_original_url
+
+    q = model.Session.query(harvest_model.HarvestObjectError, harvest_model.HarvestObject.guid) \
                       .join(harvest_model.HarvestObject) \
                       .filter(harvest_model.HarvestObject.harvest_job_id==job.id) \
                       .order_by(harvest_model.HarvestObjectError.harvest_object_id)
 
-    errors= {}
+    report = {}
-    for error in q.all():
+    for error, guid in q.all():
-        if not error.harvest_object_id in errors:
+        if not error.harvest_object_id in report:
-            errors[error.harvest_object_id] = []
+            report[error.harvest_object_id] = {
-        errors[error.harvest_object_id].append({
+                'guid': guid,
+                'errors': []
+            }
+            if original_url_builder:
+                url = original_url_builder(error.harvest_object_id)
+                if url:
+                    report[error.harvest_object_id]['original_url'] = url
+
+        report[error.harvest_object_id]['errors'].append({
             'message': error.message,
             'line': error.line,
             'type': error.stage
          })
 
-    return {'errors': errors}
+    return report
 
 def harvest_job_list(context,data_dict):
 

ckanext/harvest/templates_new/job/report.html

@@ -33,13 +33,23 @@
 
 
     <div style='font-size: 1.5em; margin: 1em 0;'>
-      {{ c.job_report.errors.keys()|length}} documents with errors
+      {{ c.job_report.keys()|length}} documents with errors
     </div>
 
-    {% for harvest_object_id in c.job_report.errors.keys() %}
+    {% for harvest_object_id in c.job_report.keys() %}
     <div>
-      <div>{{ harvest_object_id }}</div>
+        <div>
-      {% for error in c.job_report.errors[harvest_object_id] %}
+            {{ c.job_report[harvest_object_id].guid }}
+
+            {% if 'original_url' in c.job_report[harvest_object_id] %}
+            (<a href="{{ c.job_report[harvest_object_id].original_url }}">{{ _('Remote content') }}</a>)
+            {% endif %}
+
+            ({{ h.link_for_harvest_object(harvest_object_id,text=_('Local content')) }})
+
+
+        </div>
+      {% for error in c.job_report[harvest_object_id].errors %}
       <div style="margin-left: 2em">{{ error.message }}
         {% if error.line %}
         <span>(line {{error.line}})</span>