Spaces:
Build error
Build error
Upload 34 files
Browse files- .gitattributes +30 -0
- app.py +129 -0
- outputs/Screenshot (3).png +0 -0
- packages.txt +2 -0
- requirements.txt +18 -0
- uploads/Adani_Ports.pdf +3 -0
- uploads/Asian_Paints.pdf +3 -0
- uploads/Axis_Bank.pdf +3 -0
- uploads/Bajaj_Finance.pdf +3 -0
- uploads/Bajaj_Finserv.pdf +3 -0
- uploads/Bharti_Airtel.pdf +3 -0
- uploads/HCL_Tech.pdf +3 -0
- uploads/HDFC_Bank.pdf +3 -0
- uploads/Hindustan_Unilever.pdf +3 -0
- uploads/ICICI_Bank.pdf +3 -0
- uploads/ITC.pdf +3 -0
- uploads/IndusInd_Bank.pdf +3 -0
- uploads/Infosys.pdf +3 -0
- uploads/JSW_Steel.pdf +3 -0
- uploads/Kotak_Bank.pdf +3 -0
- uploads/L&T.pdf +3 -0
- uploads/Mahindra&Mahindra.pdf +3 -0
- uploads/Maruti_Suzuki.pdf +3 -0
- uploads/NTPC.pdf +3 -0
- uploads/Nestle.pdf +3 -0
- uploads/Powergrid.pdf +3 -0
- uploads/Reliance.pdf +3 -0
- uploads/SBI_Bank.pdf +3 -0
- uploads/Sun_Pharma.pdf +3 -0
- uploads/TCS.pdf +3 -0
- uploads/Tata_Motors.pdf +3 -0
- uploads/Tata_Steel.pdf +3 -0
- uploads/Tech_Mahindra.pdf +3 -0
- uploads/Titan.pdf +3 -0
- uploads/Ultratech.pdf +3 -0
.gitattributes
CHANGED
@@ -33,3 +33,33 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
uploads/Adani_Ports.pdf filter=lfs diff=lfs merge=lfs -text
|
37 |
+
uploads/Asian_Paints.pdf filter=lfs diff=lfs merge=lfs -text
|
38 |
+
uploads/Axis_Bank.pdf filter=lfs diff=lfs merge=lfs -text
|
39 |
+
uploads/Bajaj_Finance.pdf filter=lfs diff=lfs merge=lfs -text
|
40 |
+
uploads/Bajaj_Finserv.pdf filter=lfs diff=lfs merge=lfs -text
|
41 |
+
uploads/Bharti_Airtel.pdf filter=lfs diff=lfs merge=lfs -text
|
42 |
+
uploads/HCL_Tech.pdf filter=lfs diff=lfs merge=lfs -text
|
43 |
+
uploads/HDFC_Bank.pdf filter=lfs diff=lfs merge=lfs -text
|
44 |
+
uploads/Hindustan_Unilever.pdf filter=lfs diff=lfs merge=lfs -text
|
45 |
+
uploads/ICICI_Bank.pdf filter=lfs diff=lfs merge=lfs -text
|
46 |
+
uploads/IndusInd_Bank.pdf filter=lfs diff=lfs merge=lfs -text
|
47 |
+
uploads/Infosys.pdf filter=lfs diff=lfs merge=lfs -text
|
48 |
+
uploads/ITC.pdf filter=lfs diff=lfs merge=lfs -text
|
49 |
+
uploads/JSW_Steel.pdf filter=lfs diff=lfs merge=lfs -text
|
50 |
+
uploads/Kotak_Bank.pdf filter=lfs diff=lfs merge=lfs -text
|
51 |
+
uploads/L&T.pdf filter=lfs diff=lfs merge=lfs -text
|
52 |
+
uploads/Mahindra&Mahindra.pdf filter=lfs diff=lfs merge=lfs -text
|
53 |
+
uploads/Maruti_Suzuki.pdf filter=lfs diff=lfs merge=lfs -text
|
54 |
+
uploads/Nestle.pdf filter=lfs diff=lfs merge=lfs -text
|
55 |
+
uploads/NTPC.pdf filter=lfs diff=lfs merge=lfs -text
|
56 |
+
uploads/Powergrid.pdf filter=lfs diff=lfs merge=lfs -text
|
57 |
+
uploads/Reliance.pdf filter=lfs diff=lfs merge=lfs -text
|
58 |
+
uploads/SBI_Bank.pdf filter=lfs diff=lfs merge=lfs -text
|
59 |
+
uploads/Sun_Pharma.pdf filter=lfs diff=lfs merge=lfs -text
|
60 |
+
uploads/Tata_Motors.pdf filter=lfs diff=lfs merge=lfs -text
|
61 |
+
uploads/Tata_Steel.pdf filter=lfs diff=lfs merge=lfs -text
|
62 |
+
uploads/TCS.pdf filter=lfs diff=lfs merge=lfs -text
|
63 |
+
uploads/Tech_Mahindra.pdf filter=lfs diff=lfs merge=lfs -text
|
64 |
+
uploads/Titan.pdf filter=lfs diff=lfs merge=lfs -text
|
65 |
+
uploads/Ultratech.pdf filter=lfs diff=lfs merge=lfs -text
|
app.py
ADDED
@@ -0,0 +1,129 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import os
|
3 |
+
from tqdm import tqdm
|
4 |
+
|
5 |
+
from unstructured.partition.pdf import partition_pdf
|
6 |
+
from langchain.schema.document import Document
|
7 |
+
import google.generativeai as genai
|
8 |
+
from langchain_huggingface import HuggingFaceEmbeddings
|
9 |
+
from langchain_community.vectorstores import FAISS
|
10 |
+
import shutil
|
11 |
+
|
12 |
+
from wasabi import msg
|
13 |
+
|
14 |
+
from PIL import Image
|
15 |
+
|
16 |
+
genai.configure(api_key = 'AIzaSyB342Fh-nkRaO38BshbyI4-s0T9orVpsMw')
|
17 |
+
model = genai.GenerativeModel('gemini-1.5-flash')
|
18 |
+
|
19 |
+
files = os.listdir('uploads')
|
20 |
+
files = [f'uploads/{file}' for file in files]
|
21 |
+
|
22 |
+
documents = []
|
23 |
+
|
24 |
+
for file in tqdm(files , total = len(files) , leave = False) :
|
25 |
+
|
26 |
+
elements = partition_pdf(
|
27 |
+
filename = file ,
|
28 |
+
extract_images_in_pdf = True ,
|
29 |
+
infer_table_structure = True ,
|
30 |
+
chunking_strategy = 'by_title' ,
|
31 |
+
max_characters = 4000 ,
|
32 |
+
new_after_n_chars = 3800 ,
|
33 |
+
combine_text_under_n_chars = 2000 ,
|
34 |
+
extract_image_block_output_dir = 'outputs'
|
35 |
+
)
|
36 |
+
|
37 |
+
for element in elements :
|
38 |
+
|
39 |
+
element = element.to_dict()
|
40 |
+
|
41 |
+
metadata = element['metadata']
|
42 |
+
|
43 |
+
if 'text_as_html' in metadata : documents.append(
|
44 |
+
Document(
|
45 |
+
page_content = metadata['text_as_html'] ,
|
46 |
+
metadata = {
|
47 |
+
'type' : 'text' ,
|
48 |
+
'metadata' : element
|
49 |
+
}
|
50 |
+
)
|
51 |
+
)
|
52 |
+
|
53 |
+
else : documents.append(
|
54 |
+
Document(
|
55 |
+
page_content = element['text'] ,
|
56 |
+
metadata = {
|
57 |
+
'type' : 'text' ,
|
58 |
+
'metadata' : element
|
59 |
+
}
|
60 |
+
)
|
61 |
+
)
|
62 |
+
|
63 |
+
images = os.listdir('outputs')
|
64 |
+
images = [f'outputs/{image}' for image in images]
|
65 |
+
|
66 |
+
for image in tqdm(images , total = len(images) , leave = False) :
|
67 |
+
|
68 |
+
image = Image.open(image)
|
69 |
+
|
70 |
+
try :
|
71 |
+
|
72 |
+
response = model.generate_content([
|
73 |
+
image ,
|
74 |
+
'Explain the Image'
|
75 |
+
])
|
76 |
+
|
77 |
+
response = response.text
|
78 |
+
|
79 |
+
except Exception as e : msg.fail(f'----| FAIL : COULDNT CALL THE IMAGE DESCRIPTION API : {e}') ; response = 'COuldnt Call Model for this'
|
80 |
+
|
81 |
+
documents.append(
|
82 |
+
Document(
|
83 |
+
page_content = response ,
|
84 |
+
metadata = {
|
85 |
+
'type' : 'image' ,
|
86 |
+
'metadata' : {
|
87 |
+
'image' : image
|
88 |
+
}
|
89 |
+
}
|
90 |
+
)
|
91 |
+
)
|
92 |
+
|
93 |
+
shutil.rmtree('uploads')
|
94 |
+
|
95 |
+
vc = FAISS.from_documents(
|
96 |
+
documents = documents ,
|
97 |
+
embedding = HuggingFaceEmbeddings(model_name = 'all-MiniLM-L6-v2')
|
98 |
+
)
|
99 |
+
|
100 |
+
def run_rag(query) :
|
101 |
+
|
102 |
+
similar_docs = vc.similarity_search(query , k = 4)
|
103 |
+
context = [doc.page_content for doc in similar_docs]
|
104 |
+
|
105 |
+
prompt = f'''
|
106 |
+
You are a Helpfull Chatbot that helps users with their queries
|
107 |
+
|
108 |
+
- You will be provided with a query
|
109 |
+
- You will be provided with a context as well
|
110 |
+
|
111 |
+
Your task is to generate a response to the query based on the context provided
|
112 |
+
|
113 |
+
Context : {context}
|
114 |
+
|
115 |
+
Query : {query}
|
116 |
+
|
117 |
+
'''
|
118 |
+
|
119 |
+
response = model.generate_content(prompt)
|
120 |
+
|
121 |
+
return response.text
|
122 |
+
|
123 |
+
demo = gr.Interface(
|
124 |
+
fn = run_rag ,
|
125 |
+
inputs = 'text' ,
|
126 |
+
outputs = 'text'
|
127 |
+
)
|
128 |
+
|
129 |
+
demo.launch()
|
outputs/Screenshot (3).png
ADDED
![]() |
packages.txt
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
poppler-utils
|
2 |
+
tesseract-ocr
|
requirements.txt
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
unstructured
|
2 |
+
unstructured_inference
|
3 |
+
unstructured_pytesseract
|
4 |
+
pytesseract
|
5 |
+
pi_heif
|
6 |
+
pdfminer.six
|
7 |
+
langchain
|
8 |
+
langchain_community
|
9 |
+
langchain_huggingface
|
10 |
+
sentence-transformers
|
11 |
+
faiss-cpu
|
12 |
+
groq
|
13 |
+
google
|
14 |
+
google-generativeai
|
15 |
+
gradio
|
16 |
+
tqdm
|
17 |
+
Pillow
|
18 |
+
wasabi
|
uploads/Adani_Ports.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e14479f84d48f530cf11f37b8b45f327b98908fe4c714c09c042f44e1b40e15d
|
3 |
+
size 23441567
|
uploads/Asian_Paints.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9a1375324a65d21104e91597c2a8c974f29da03c6e275c47456eb9ee3f317858
|
3 |
+
size 26576588
|
uploads/Axis_Bank.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:da21de351524368d87dd801a2c97b86de799bd3d6fde21c94a57686e41e8196d
|
3 |
+
size 11263914
|
uploads/Bajaj_Finance.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:49071340391999e6863619b33ceb8a52c5c3a6d19cf94a7f5577fd6ea2304075
|
3 |
+
size 19813034
|
uploads/Bajaj_Finserv.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8743975cc9720e2db0f861da7ddc45b97c85ae5bea2202a91a9e0eb230604edf
|
3 |
+
size 22926656
|
uploads/Bharti_Airtel.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ab401f9504fd5fd7cf840235ecd411cf69eb82d4a853a2e6a9d70e4724801e89
|
3 |
+
size 25376207
|
uploads/HCL_Tech.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bf28874294660cb882e56c84021ce910fea0a1d5a150649604d051e72174d21a
|
3 |
+
size 8340206
|
uploads/HDFC_Bank.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fcd7ef37de26997fed3bcf2b7cfb9eb90f08d4e2617f5f3f967a1ac621c2a73e
|
3 |
+
size 17350949
|
uploads/Hindustan_Unilever.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ffb3d9f6bde2d6fa5882011ad4d394f1d6ddbdfcd3d6080617d652a90fa10595
|
3 |
+
size 21616344
|
uploads/ICICI_Bank.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4a62858135454fc5f6e09b018102cba98813928d2b2e72023e86194c0087cd09
|
3 |
+
size 11777065
|
uploads/ITC.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cef9cf09cefa71094985018a88b936733ff2c3a02b7cc6cd772fdeb8b75a9c97
|
3 |
+
size 24199409
|
uploads/IndusInd_Bank.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:24102ada097cfc42f46cc2fb4d3f520fae76e01749075491cca4e6d0be91ea93
|
3 |
+
size 19735093
|
uploads/Infosys.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b0a9bb9e802aff5f09733b8c78c88e9878732ac46e0fb29754c6da87ad47326a
|
3 |
+
size 11441269
|
uploads/JSW_Steel.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5946803f37a1c099be7b58fb4a8746f57ebd475ef7b673994e93746ffc62b508
|
3 |
+
size 17148166
|
uploads/Kotak_Bank.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea21e25405bc8c9badb4085b42eade96559f4b915a2f383b4fee33a151d47602
|
3 |
+
size 18338231
|
uploads/L&T.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a86677b95602b7ebafcd16295deafbd7a2cba4a90c502aed37f05dee9dafcf8a
|
3 |
+
size 18910726
|
uploads/Mahindra&Mahindra.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d46740998cf312d94a1dd6a4d6f9ee1bc73cbd8b9efa4f974687f1017fb736e7
|
3 |
+
size 24944665
|
uploads/Maruti_Suzuki.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e298354886131e8e253df60ac25a725e140f89452ad6a2f4c0e857074107548c
|
3 |
+
size 23558415
|
uploads/NTPC.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fb943a9cb4db67822a4f1657454cd476e08d73259a18e990a6a36e29c78ec8aa
|
3 |
+
size 23736915
|
uploads/Nestle.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a2bd86b483d58de5cb6fac36113ff1851d39d14d666c33f24c8cac382f180dfa
|
3 |
+
size 8982580
|
uploads/Powergrid.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fb943a9cb4db67822a4f1657454cd476e08d73259a18e990a6a36e29c78ec8aa
|
3 |
+
size 21240959
|
uploads/Reliance.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4d7d6038b4d6957981fb422bc501ea893765c24611ea1fa8a5d3cd9425541eae
|
3 |
+
size 12354677
|
uploads/SBI_Bank.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4aacda3744fdcc9aea893c0c0cd35c22e5f30a93513e37f86da29f30770961dc
|
3 |
+
size 16162498
|
uploads/Sun_Pharma.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dbb9d8021b365cc095ef9e5c6917e4ec4fe7cca079672ece3beb6284128c9343
|
3 |
+
size 10123693
|
uploads/TCS.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:992b3cb2a47a9ce84b127d23d7dc1fb22967d7592d1e8af2c2120b6764da8651
|
3 |
+
size 10182737
|
uploads/Tata_Motors.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7c7610ebbea4ff3dabcb1c63d673b8b2e3c6c0f45d17daf1a88908fbf8f04a25
|
3 |
+
size 20573259
|
uploads/Tata_Steel.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:870be0ca867c1c562337a5319d09cac353a85dbb7907dce7c13ee47ad7af3e21
|
3 |
+
size 26006738
|
uploads/Tech_Mahindra.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:304878c8f4fba07d1eb1d120556c067b46d56d23993c56b5a3f552b4f109b477
|
3 |
+
size 17536967
|
uploads/Titan.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1979b0e5973d1c07ebf1a76d5580554c45acd67242746c6673a0ab276d1de6e3
|
3 |
+
size 21978349
|
uploads/Ultratech.pdf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e250887b207b20483cf040589ab9723a935dca317ce9bccd050cfdef059c0a8b
|
3 |
+
size 29394403
|