Update Automaton to add the method process_recjson.

modules/harvest_tools/automaton.py

@@ -551,49 +551,30 @@ class Automaton(object):
         # process the remaining identifiers
         [self.process_recid(rec_id) for rec_id in rec_ids]
 
-    def process_recid(self, rec_id):
-        """Process the publication:
+    def process_recjson(self, recjson):
+        """Process the publication provided as a JSON record:
 
-            * get the publication data from the store using its identifier
             * instantiate the record (RecordPubli, REcordConf, RecordThesis)
-            * process OAI data
             * check the record
             * insert new record in the database
 
-        Note:
-            * Design to never stop although exception are raised
-            * Have a look to the collection_logs and logs in order to
-              understand what happen.
-
         Args:
-            rec_id (int):
-                identifier of the publication in the store.
+            recjson (dict):
+                record provided by the store.
 
         """
         if self.dbg:
-            print("\nprocessing record", rec_id)
+            print("\nprocessing record")
 
         collection_logs = self.collection_logs
         harvester = self.harvester
         logs = self.logs
 
         # instantiate the record
-        try:
-            recjson = self.store.get_record(rec_id)
-            record = build_record(recjson)
-
-            if self.dbg:
-                print("\t", record.title())
+        record = build_record(recjson)
 
-        except Exception as e:
-            print(traceback.format_exc())
-            url = OAI_URL % (harvester.host, rec_id)
-            logs.append(Msg(harvester=harvester,
-                            collection=collection_logs[-1].title,
-                            record_id=rec_id,
-                            title=url))
-            logs[-1].reject(e)
-            return
+        if self.dbg:
+            print("\t", record.title().encode("utf-8"))
 
         # start the log for the record
         logs.append(Msg(harvester=harvester,
@@ -620,6 +601,46 @@ class Automaton(object):
             action = (action.upper() if isinstance(action, str) else action)
             print("\tlog:", action, log.txt)
 
+    def process_recid(self, rec_id):
+        """Process the publication identified by its record identifier:
+
+            * get the publication data from the store using its identifier
+            * instantiate the record (RecordPubli, REcordConf, RecordThesis)
+            * process OAI data
+            * check the record
+            * insert new record in the database
+
+        Note:
+            * Design to never stop although exception are raised
+            * Have a look to the collection_logs and logs in order to
+              understand what happen.
+
+        Args:
+            rec_id (int):
+                identifier of the publication in the store.
+
+        """
+        if self.dbg:
+            print("\nprocessing record", rec_id)
+
+        collection_logs = self.collection_logs
+        harvester = self.harvester
+        logs = self.logs
+
+        try:
+            recjson = self.store.get_record(rec_id)
+            self.process_recjson(recjson)
+
+        except Exception as e:
+            print(traceback.format_exc())
+            url = OAI_URL % (harvester.host, rec_id)
+            logs.append(Msg(harvester=harvester,
+                            collection=collection_logs[-1].title,
+                            record_id=rec_id,
+                            title=url))
+            logs[-1].reject(e)
+            return
+
     def process_url(self, host, collections):
         """Retrieve JSON objects from the invenio store and
         insert corresponding records in the database.