Update ppt_chunker.py

ppt_chunker.py

@@ -12,7 +12,7 @@
 #
 import re
 import copy
-from io import BytesIO
+from io import StringIO
 
 from PIL import Image
 
@@ -22,7 +22,7 @@ from nlp import rag_tokenizer, tokenize, is_english
 from pptx import Presentation
 from ordered_multimap import OrderedMultiIndexMapWeakRef
 
-def ppt_chunk(filename, binary=None, from_page=0, to_page=100000,
+def ppt_chunk(file_like, from_page=0, to_page=100000,
           lang="English", **kwargs):
     """
     The supported file formats are pptx.
@@ -37,7 +37,9 @@ def ppt_chunk(filename, binary=None, from_page=0, to_page=100000,
     doc["title_sm_tks"] = rag_tokenizer.fine_grained_tokenize(doc["title_tks"])
     weakDict = OrderedMultiIndexMapWeakRef()
 
-    ppt = Presentation(filename if not binary else BytesIO(binary))
+    source_stream = StringIO(file_like.read())
+    ppt = Presentation(source_stream)
+    source_stream.close()
     total_pages = len(ppt.slides)
 
     ppt_parser = RAGFlowPptParser()