Building out timetable logic

src/timetable.py

@@ -47,12 +47,30 @@ def getTimetable(full = False):
   download_url = full_data_url if full else update_data_url
   response = requests.get(downloadUrl, auth=(CORPUS_USER, CORPUS_PASS))
   mongo.incrementCounter("schedule_api")
-  return response
+  return zlib.decompress(response.content, 16+zlib.MAX_WBITS)
 
-def parseTimetable(data):
-    ## Parse the update file
-      ## Check how the downloaded data is displayed
-  return
+def loopTimetable(data):
+  listify = data.splitlines()
+  documents = []
+  for item in listify:
+    dic = json.loads(item)
+    if ('JsonTimetableV1' in dic):
+      timestamp = dic['JsonTimetableV1']['timestamp']
+      sequence  = dic['JsonTimetableV1']['Metadata']['sequence']
+      detail = {timestamp: timestamp, sequence: sequence}
+      # Do something with this data here
+      # Check if timestamp and sequence are correct, if not trigger a full download
+    elif ('TiplocV1' in dic):
+      print("Disregarding TIPLOC Data")
+      # Not used as TIPLOCs etc. are sourced from CORPUS
+    elif ('JsonAssociationV1' in dic):
+      print("JsonAssociationData")
+      # Associates trains with eachother - not planning to use yet.
+    elif ('JsonScheduleV1' in dic):
+      print("JsonScheduleData")
+      document = insertSchedule(dic)
+      documents.append(document)
+  return documents
 
 def runUpdate():
   required = isUpdateRequired()
@@ -66,3 +84,41 @@ def runUpdate():
     log.out("timetable.runUpdate: timetable update is not needed")
     return "done"
   ## Check what happens if there is no update
+
+def insertSchedule(sch_record):
+  schedule = sch_record['JsonScheduleV1']
+  scheduleId = schedule['CIF_train_uid']
+  transactionType = schedule['transaction_type']
+  document = {
+    'stpIndicator': schedule['CIF_stp_indicator'],
+    'trainUid': scheduleId,
+    'headcode': schedule['schedule_segment']['signalling_id'],
+    'powerType': schedule['schedule_segment']['CIF_power_type'],
+    'planSpeed': schedule['schedule_segment']['CIF_speed'],
+    'scheduleEndDate': schedule['schedule_end_date']
+  }
+  return document
+
+#  Proposed Document Schema:
+#  {
+#  stp_indicator: "O",
+#  train_uid: "C07284"
+#  atoc_code: "GW"
+#  schedule_days_runs: [0,1,2,3,4,5,6] # Sunday-Saturday
+#  schedule_end_date: "2023-06-02"
+#  headcode: "5G30"
+#  power_type: "DMU"
+#  speed: "090"
+#  catering_code: null
+#  service_branding: ""
+#  passenger_stops: [
+#                     {
+#                       'tiploc': "TIPLOC",
+#                       'pb_arr': "PublicArrival",
+#                       'pb_dep': "PublicDepartr"
+#                     }
+#                   ]
+
+### CURRENT STATE: loopTimetable and insertSchedule builds the data into
+### a suitable format to send to Mongo, passengerStops are not yet
+### parsed and there needs to be logic around the transaction_type