def rename(docs, target=None, **kwargs):
assert target, "To run this migration you need a target."
for doc in docs:
raw = RawDocument({
'doc': doc.doc,
'docID': doc.docID,
'source': target,
'filetype': doc.filetype,
'timestamps': doc.timestamps,
'versions': doc.versions
})
assert doc.source != target, "Can't rename {} to {}, names are the same.".format(
doc.source, target)
if not kwargs.get('dry'):
tasks.process_raw(raw)
tasks.process_normalized(tasks.normalize(raw, raw['source']), raw)
logger.info('Processed document from {} with id {}'.format(
doc.source, raw['docID']))
es.delete(index=settings.ELASTIC_INDEX,
doc_type=doc.source,
id=raw['docID'],
ignore=[404])
es.delete(index='share_v1',
doc_type=doc.source,
id=raw['docID'],
ignore=[404])
logger.info('Deleted document from {} with id {}'.format(
doc.source, raw['docID']))
def delete(doc, sources=None, **kwargs):
assert sources, "To run this migration you need a source."
doc.timeout(5).delete()
es.delete(index=settings.ELASTIC_INDEX, doc_type=sources, id=doc.docID, ignore=[404])
es.delete(index='share_v1', doc_type=sources, id=doc.docID, ignore=[404])
logger.info('Deleted document from {} with id {}'.format(sources, doc.docID))
def rename(source, target, dry=True):
assert source != target, "Can't rename {} to {}, names are the same".format(source, target)
count = 0
exceptions = []
for doc in documents(source):
count += 1
try:
raw = RawDocument({
'doc': doc.doc,
'docID': doc.docID,
'source': target,
'filetype': doc.filetype,
'timestamps': doc.timestamps,
'versions': doc.versions
})
if not dry:
process_raw(raw)
process_normalized(normalize(raw, raw['source']), raw)
logger.info('Processed document from {} with id {}'.format(source, raw['docID']))
except Exception as e:
logger.exception(e)
exceptions.append(e)
else:
if not dry:
es.delete(index=settings.ELASTIC_INDEX, doc_type=source, id=raw['docID'], ignore=[404])
es.delete(index='share_v1', doc_type=source, id=raw['docID'], ignore=[404])
logger.info('Deleted document from {} with id {}'.format(source, raw['docID']))
if dry:
logger.info('Dry run complete')
for ex in exceptions:
logger.exception(e)
logger.info('{} documents processed, with {} exceptions'.format(count, len(exceptions)))
def delete_by_source(source):
count = 0
exceptions = []
for doc in documents(source):
count += 1
try:
doc.delete()
es.delete(index=settings.ELASTIC_INDEX, doc_type=source, id=doc.docID, ignore=[404])
es.delete(index='share_v1', doc_type=source, id=doc.docID, ignore=[404])
except Exception as e:
logger.exception(e)
exceptions.append(e)
for ex in exceptions:
logger.exception(e)
logger.info('{} documents processed, with {} exceptions'.format(count, len(exceptions)))
def rename(doc, target=None, **kwargs):
assert target, "To run this migration you need a target."
raw = RawDocument({
'doc': doc.doc,
'docID': doc.docID,
'source': target,
'filetype': doc.filetype,
'timestamps': doc.timestamps,
'versions': doc.versions
})
assert doc.source != target, "Can't rename {} to {}, names are the same.".format(doc.source, target)
if not kwargs.get('dry'):
tasks.process_raw(raw)
tasks.process_normalized(tasks.normalize(raw, raw['source']), raw)
logger.info('Processed document from {} with id {}'.format(doc.source, raw['docID']))
es.delete(index=settings.ELASTIC_INDEX, doc_type=doc.source, id=raw['docID'], ignore=[404])
es.delete(index='share_v1', doc_type=doc.source, id=raw['docID'], ignore=[404])
logger.info('Deleted document from {} with id {}'.format(doc.source, raw['docID']))
