Fix office file indexing (#230)

1 year ago · 4f5cacca46
parent 0aab6a071a
commit 4f5cacca46
4 changed files with 11 additions and 3 deletions
--- a/models/index_model.py
+++ b/models/index_model.py
@ -381,7 +381,12 @@ class Index_handler:
            "application/json": ".json",
            "image/png": ".png",
            "image/": ".jpg",
-            "vnd.": ".pptx",
+            "ms-powerpoint": ".ppt",
+            "presentationml.presentation": ".pptx",
+            "ms-excel": ".xls",
+            "spreadsheetml.sheet": ".xlsx",
+            "msword": ".doc",
+            "wordprocessingml.document": ".docx",
            "audio/": ".mp3",
            "video/": ".mp4",
            "epub": ".epub",
--- a/pyproject.toml
+++ b/pyproject.toml
@ -40,7 +40,8 @@ dependencies = [
 "python-pptx==0.6.21",
 "langchain==0.0.105",
 "unidecode==1.3.6",
-"tqdm==4.64.1"
+"tqdm==4.64.1",
+"docx2txt==0.8"
 ]

 dynamic = ["version"]
--- a/requirements.txt
+++ b/requirements.txt
@ -23,3 +23,4 @@ langchain==0.0.105
 openai-whisper
 unidecode==1.3.6
 tqdm==4.64.1
+docx2txt==0.8
--- a/requirements_base.txt
+++ b/requirements_base.txt
@ -20,4 +20,5 @@ protobuf==3.20.2
 python-pptx==0.6.21
 langchain==0.0.105
 unidecode==1.3.6
-tqdm==4.64.1
+tqdm==4.64.1
+docx2txt==0.8