force garbage collection.

This commit is contained in:
Nathan TeBlunthuis 2025-07-22 13:13:18 -07:00
parent 6557e25af7
commit 076df15740

View File

@ -3,6 +3,7 @@
# original wikiq headers are: title articleid revid date_time anon
# editor editor_id minor text_size text_entropy text_md5 reversion
# additions_size deletions_size
import gc
import argparse
import json
import os.path
@ -675,7 +676,7 @@ class WikiqParser:
if self.partition_namespaces is True:
writer = pq_writers[page.mwpage.namespace]
writer.write(pa.record_batch(row_buffer, schema=schema))
gc.collect()
page_count += 1
print(