erp5_wendelin_telecom_ingestion: handle ingestion of modified files

bt5/erp5_wendelin_telecom_ingestion/SkinTemplateItem/portal_skins/erp5_wendelin_telecom/IngestionPolicy_getIngestionOperationAndParameterDict.py

@@ -31,14 +31,23 @@ try:
 
   if data_ingestion is not None:
     if data_ingestion.getSimulationState() != 'started':
-      # CHECK HASH
-      #if hash is equal:
+      modified = False
+      if size != None and size != "":
+        data_stream = portal_catalog.getResultValue(
+          portal_type = 'Data Stream',
+          validation_state = 'validated',
+          reference = data_ingestion_reference)
+        if size != data_stream.getSize():
+          modified = True
+        elif hash_value != None and hash_value != "" and hash_value != data_stream.getVersion():
+          modified = True
+      if not modified:
         context.logEntry("An older ingestion for reference %s was already done." % data_ingestion_reference)
         raise ValueError("An older ingestion for reference %s was already done." % data_ingestion_reference)
-      #else:
-      #portal.ERP5Site_invalidateIngestionObjects(data_ingestion.getReference())
+      else:
+        context.logEntry("Ingestion of modified file. Old version will be invalidated.")
+        portal.ERP5Site_invalidateIngestionObjects(data_ingestion_reference)
 
-  #if data_ingestion is None:
   specialise_value_list = [x.getObject() for x in portal_catalog.searchResults(
     portal_type = 'Data Supply',
     reference = 'embulk',

bt5/erp5_wendelin_telecom_ingestion/SkinTemplateItem/portal_skins/erp5_wendelin_telecom/ingestionReferenceExists.py

@@ -16,6 +16,7 @@ try:
   # remove supplier and eof from reference
   data_ingestion_reference = '/'.join(reference.split('/')[1:-3])
   EOF = reference.split('/')[-3]
+  size = reference.split('/')[-2]
 
   if data_ingestion_reference is "":
     context.logEntry("[ERROR] Data Ingestion reference is not well formated")
@@ -39,6 +40,9 @@ try:
       pass
     return FALSE
 
+  if size != "" and size != None:
+    # this is a modified file
+    return FALSE
   context.logEntry("[ERROR] Data Ingestion reference %s already exists" % data_ingestion_reference)
   return TRUE
 except Exception as e: