This commit is contained in:
2023-12-02 03:50:05 +08:00
parent 655fc8c1c0
commit 0dbd957454
2 changed files with 12 additions and 6 deletions

View File

@@ -55,9 +55,9 @@ def process_images(conn, ocr, es):
item = [x for x in ocr.ocr(image) if x['text'] and not x['text'].isdigit() and len(x['text']) > 1]
text = ' '.join([x['text'] for x in item])
print(id, text)
save_text(conn, id, json.dumps(item, ensure_ascii=False, cls=MyEncoder))
es.index(index='web_images', id=id, body={'content': text})
conn.commit()
#save_text(conn, id, json.dumps(item, ensure_ascii=False, cls=MyEncoder))
#es.index(index='web_images', id=id, body={'content': text})
#conn.commit()
def main():
es = Elasticsearch(config['ELASTICSEARCH_HOST'], basic_auth=(config['ELASTICSEARCH_USERNAME'], config['ELASTICSEARCH_PASSWORD']), verify_certs=False)