print("Searching through entities")
export_records(old_client, folder_path, list_of_types_to_consider)
# Discover the Relationship Guids that will be uploaded later
print("Discovering guids to remap from disk")
relationship_guids = discover_guids(folder_path, ["relationshipGuid"])
with open(relationships_guid_path, 'a+') as fp:
for relationship in relationship_guids:
fp.write(relationship)
fp.write("\n")
# Get Guids and replace them
print("Remapping guids from disk and into memory")
old_entity_guids = discover_guids(folder_path, indicators)
orig_guid_to_temp_guid = {g: str(gt.get_guid()) for g in old_entity_guids}
remapped_entities = remap_guids(orig_guid_to_temp_guid, folder_path,
output_path)
print("Processing entities in memory")
for entity in remapped_entities:
# Strip the relationshipAttributes, they will be added later.
entity["relationshipAttributes"] = {}
entity.pop("lastModifiedTS")
entity.pop("createdBy")
entity.pop("updatedBy")
entity.pop("createTime")
entity.pop("updateTime")
input("Ready to upload entities. Continue? Ctrl + C to back out now.")
# Upload and get results["guidAssignments"] for new guids
results = new_client.upload_entities(batch=remapped_entities)
print("Writing glossary terms in prep to be re-mapped to temporary guids")
with open(glossary_prep_path, 'w') as fp:
json.dump(glossary_terms_copy, fp)
print("Discovering guids and remapping guids")
# Remap the guids and write them back out to the output_path
old_glossary_guids = discover_guids([glossary_prep_path],
["guid", "termGuid"])
# Provide a new guid (temp guid) for the upload (must be a negative number)
orig_guid_to_temp_guid = {
g: str(gt.get_guid())
for g in old_glossary_guids
}
# Execute the find and replace of old guid with temp guid
remapped_glossary = remap_guids(orig_guid_to_temp_guid,
[glossary_prep_path], output_path)
print("Processing the glossary terms in memory")
headers = [
"antonyms", "classifies", "isA", "preferredTerms", "preferredToTerms",
"replacedBy", "replacementTerms", "seeAlso", "synonyms",
"translatedTerms", "translationTerms", "validValues", "validValuesFor"
]
for term in remapped_glossary:
# Remove everything that will be created automatically
try:
term["anchor"].pop("relationGuid")
term.pop("lastModifiedTS")
term.pop("createdBy")
term.pop("qualifiedName")
with open(
os.path.join(relationships_path,
f"batch-{counter}.json"), 'w') as fp:
json.dump(buffer, fp)
buffer = []
counter = counter + 1
if len(buffer) > 0:
with open(os.path.join(relationships_path, f"batch-last.json"),
'w') as fp:
json.dump(buffer, fp)
# Now we can read load the remapping files and
# Load in the remapping files
print('Remapping guids')
remapped_relationships = remap_guids(old_to_new_guids, relationships_path,
output_path)
# Clean up the remapped relationship and upload one by one...
# This will take a while...
gt = GuidTracker()
counter = 0
skipped = 0
total_relationships = len(remapped_relationships)
for relationship in remapped_relationships:
inner_relationship = relationship["relationship"]
inner_relationship["guid"] = str(gt.get_guid())
# Pop attributes that break the upload
inner_relationship.pop("updateTime")
inner_relationship.pop("lastModifiedTS")
inner_relationship.pop("updatedBy")
inner_relationship.pop("createTime")