cif/parse.go

@@ -72,7 +72,6 @@ func parseCifDataStream(dataStream io.ReadCloser) (*parsedData, error) {
 		return nil, errors.New("unable to parse nil pointer")
 	}
 
-	// Initialise data structures
 	var parsed parsedData
 	parsed.assoc = make([]upstreamApi.JsonAssociationV1, 0)
 	parsed.sched = make([]upstreamApi.JsonScheduleV1, 0)
@@ -99,12 +98,9 @@ func parseCifDataStream(dataStream io.ReadCloser) (*parsedData, error) {
 				}
 				parsed.header = timetable
 			case "JsonAssociationV1":
-				var association upstreamApi.JsonAssociationV1
-				if err := json.Unmarshal(value, &association); err != nil {
-					log.Msg.Error("Error decoding JSONAssociationV1 object", zap.Error(err))
-					continue
-				}
-				parsed.assoc = append(parsed.assoc, association)
+				// Association data is not currently used
+				// but may be used in the future
+				continue
 			case "JsonScheduleV1":
 				var schedule upstreamApi.JsonScheduleV1
 				if err := json.Unmarshal(value, &schedule); err != nil {

cif/readme.md

@@ -0,0 +1,11 @@
+# package cif
+
+This package follows a similar pattern to `package corpus`.
+
+First, CheckCorpus() retreived cif metadata from the database and determines whether an update is required and what type of update.
+
+Then, one of the update functions is called which run through the update process.  There are two update types, 'full' and 'update'.  A 'full' update drops the entire timetable collection and rebuilds with a full CIF download.  'update' downloads CIF updates from specified days and applies updates.
+
+Downloads are handled by `package nrod` which returns an io.ReadCloser which is passed to the parsing function.
+
+Currently the parsing function returns a parsedCif pointer, however this is using significant memory due to the size of a full CIF download (Often around 4.5GB).  The intention is to instead use a worker pool to handle the data.