Spaces:
Running
Running
Update requirements.txt
Browse files- requirements.txt +19 -21
requirements.txt
CHANGED
@@ -14,34 +14,32 @@ torch==2.1.2
|
|
14 |
sentencepiece
|
15 |
|
16 |
# PDF, DOCX, PPTX, XLSX processing
|
17 |
-
pymupdf # PDF
|
18 |
-
python-docx
|
19 |
-
python-pptx
|
20 |
-
openpyxl
|
21 |
|
22 |
-
# OCR & Image
|
23 |
-
Pillow
|
24 |
easyocr
|
25 |
-
|
|
|
26 |
|
27 |
-
#
|
28 |
gTTS
|
29 |
|
30 |
-
#
|
31 |
pandas
|
32 |
-
pdfplumber # for
|
33 |
-
|
34 |
-
# Optional: chart/table structure parsing (custom logic to be added)
|
35 |
-
layoutparser==0.3.4 # CPU-compatible version
|
36 |
|
37 |
-
# Retrieval + QA
|
38 |
-
chromadb==0.4.15
|
39 |
|
40 |
-
# Session Memory
|
41 |
-
redis
|
42 |
|
43 |
-
#
|
44 |
-
opencv-python-headless
|
45 |
-
|
46 |
-
# General
|
47 |
tqdm
|
|
|
|
|
|
|
|
14 |
sentencepiece
|
15 |
|
16 |
# PDF, DOCX, PPTX, XLSX processing
|
17 |
+
pymupdf # PyMuPDF for PDF
|
18 |
+
python-docx
|
19 |
+
python-pptx
|
20 |
+
openpyxl
|
21 |
|
22 |
+
# OCR & Image Processing
|
|
|
23 |
easyocr
|
24 |
+
opencv-python-headless # required for EasyOCR and layoutparser
|
25 |
+
Pillow
|
26 |
|
27 |
+
# Voice Output
|
28 |
gTTS
|
29 |
|
30 |
+
# Table & Chart QA from documents
|
31 |
pandas
|
32 |
+
pdfplumber # for tables in PDFs
|
33 |
+
layoutparser==0.3.4 # CPU-compatible layout parser (no detectron2)
|
|
|
|
|
34 |
|
35 |
+
# Retrieval + Long QA
|
36 |
+
chromadb==0.4.15 # for retrieval-based QA
|
37 |
|
38 |
+
# Session Memory (optional Redis-based or in-memory)
|
39 |
+
redis
|
40 |
|
41 |
+
# General Utilities
|
|
|
|
|
|
|
42 |
tqdm
|
43 |
+
|
44 |
+
# Compatibility (avoid Pydantic v2 issues)
|
45 |
+
pydantic==1.10.13
|