|
@@ -1166,7 +1166,7 @@ class KbmService:
|
|
logger.info(f"开始处理 PDF,document_id: {document_id}, max_tokens: {max_tokens}")
|
|
logger.info(f"开始处理 PDF,document_id: {document_id}, max_tokens: {max_tokens}")
|
|
text_chunks = []
|
|
text_chunks = []
|
|
try:
|
|
try:
|
|
- sentence_chunks, page_images = pdf_utils.PDFProcessor(1).get_results(file_content, document_id, max_tokens, bucket_name, KbmService)
|
|
|
|
|
|
+ sentence_chunks, page_images = pdf_utils.PDFProcessor(2).get_results(file_content, document_id, max_tokens, bucket_name, KbmService)
|
|
# 一张页面 , 一张页面的完这
|
|
# 一张页面 , 一张页面的完这
|
|
for sentence_chunk in sentence_chunks:
|
|
for sentence_chunk in sentence_chunks:
|
|
text_chunks.append( KbmService.split_text_by_semantic(sentence_chunk, max_tokens, bucket_name))
|
|
text_chunks.append( KbmService.split_text_by_semantic(sentence_chunk, max_tokens, bucket_name))
|