analyticsbyte
commited on
Commit
•
045d75e
1
Parent(s):
ad76dfc
Upload 3 files
Browse filesThe invoice data extraction code
- app.py +39 -0
- pipeline.py +94 -0
- requirements.txt +6 -0
app.py
ADDED
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
from dotenv import load_dotenv
|
3 |
+
|
4 |
+
from pipeline import create_docs
|
5 |
+
|
6 |
+
|
7 |
+
|
8 |
+
def main():
|
9 |
+
load_dotenv()
|
10 |
+
|
11 |
+
st.set_page_config(page_title="Invoice Extraction Bot")
|
12 |
+
st.title("Invoice Extraction Bot...💁 ")
|
13 |
+
st.subheader("I can help you in extracting invoice data")
|
14 |
+
|
15 |
+
|
16 |
+
# Upload the Invoices (pdf files)...
|
17 |
+
pdf = st.file_uploader("Upload invoices here, only PDF files allowed", type=["pdf"],accept_multiple_files=True)
|
18 |
+
|
19 |
+
submit=st.button("Extract Data")
|
20 |
+
|
21 |
+
if submit:
|
22 |
+
with st.spinner('Wait for it...'):
|
23 |
+
df=create_docs(pdf)
|
24 |
+
st.write(df.head())
|
25 |
+
|
26 |
+
data_as_csv= df.to_csv(index=False).encode("utf-8")
|
27 |
+
st.download_button(
|
28 |
+
"Download data as CSV",
|
29 |
+
data_as_csv,
|
30 |
+
"benchmark-tools.csv",
|
31 |
+
"text/csv",
|
32 |
+
key="download-tools-csv",
|
33 |
+
)
|
34 |
+
st.success("Hope I was able to save your time❤️")
|
35 |
+
|
36 |
+
|
37 |
+
#Invoking main function
|
38 |
+
if __name__ == '__main__':
|
39 |
+
main()
|
pipeline.py
ADDED
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# from langchain.llms import OpenAI
|
2 |
+
from langchain_openai import OpenAI
|
3 |
+
from pypdf import PdfReader
|
4 |
+
from langchain.llms.openai import OpenAI
|
5 |
+
import pandas as pd
|
6 |
+
import re
|
7 |
+
import replicate
|
8 |
+
from langchain.prompts import PromptTemplate
|
9 |
+
|
10 |
+
#Extract Information from PDF file
|
11 |
+
def get_pdf_text(pdf_doc):
|
12 |
+
text = ""
|
13 |
+
pdf_reader = PdfReader(pdf_doc)
|
14 |
+
for page in pdf_reader.pages:
|
15 |
+
text += page.extract_text()
|
16 |
+
return text
|
17 |
+
|
18 |
+
|
19 |
+
|
20 |
+
#Function to extract data from text
|
21 |
+
def extracted_data(pages_data):
|
22 |
+
|
23 |
+
template = """Extract all the following values : invoice no., Description, Quantity, date,
|
24 |
+
Unit price , Amount, Total, email, phone number and address from this data: {pages}
|
25 |
+
|
26 |
+
Expected output: remove any dollar symbols {{'Invoice no.': '1001329','Description': 'Office Chair','Quantity': '2','Date': '5/4/2023','Unit price': '1100.00','Amount': '2200.00','Total': '2200.00','Email': 'Santoshvarma0988@gmail.com','Phone number': '9999999999','Address': 'Mumbai, India'}}
|
27 |
+
"""
|
28 |
+
prompt_template = PromptTemplate(input_variables=["pages"], template=template)
|
29 |
+
|
30 |
+
llm = OpenAI(temperature=.7)
|
31 |
+
full_response=llm(prompt_template.format(pages=pages_data))
|
32 |
+
|
33 |
+
|
34 |
+
#The below code will be used when we want to use LLAMA 2 model, we will use Replicate for hosting our model....
|
35 |
+
|
36 |
+
#output = replicate.run('replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1',
|
37 |
+
#input={"prompt":prompt_template.format(pages=pages_data) ,
|
38 |
+
#"temperature":0.1, "top_p":0.9, "max_length":512, "repetition_penalty":1})
|
39 |
+
|
40 |
+
#full_response = ''
|
41 |
+
#for item in output:
|
42 |
+
#full_response += item
|
43 |
+
|
44 |
+
|
45 |
+
#print(full_response)
|
46 |
+
return full_response
|
47 |
+
|
48 |
+
|
49 |
+
# iterate over files in
|
50 |
+
# that user uploaded PDF files, one by one
|
51 |
+
def create_docs(user_pdf_list):
|
52 |
+
|
53 |
+
df = pd.DataFrame({'Invoice no.': pd.Series(dtype='str'),
|
54 |
+
'Description': pd.Series(dtype='str'),
|
55 |
+
'Quantity': pd.Series(dtype='str'),
|
56 |
+
'Date': pd.Series(dtype='str'),
|
57 |
+
'Unit price': pd.Series(dtype='str'),
|
58 |
+
'Amount': pd.Series(dtype='int'),
|
59 |
+
'Total': pd.Series(dtype='str'),
|
60 |
+
'Email': pd.Series(dtype='str'),
|
61 |
+
'Phone number': pd.Series(dtype='str'),
|
62 |
+
'Address': pd.Series(dtype='str')
|
63 |
+
})
|
64 |
+
|
65 |
+
for filename in user_pdf_list:
|
66 |
+
|
67 |
+
print(filename)
|
68 |
+
raw_data=get_pdf_text(filename)
|
69 |
+
print(raw_data)
|
70 |
+
print("extracted raw data")
|
71 |
+
|
72 |
+
llm_extracted_data=extracted_data(raw_data)
|
73 |
+
print("llm extracted data")
|
74 |
+
#Adding items to our list - Adding data & its metadata
|
75 |
+
|
76 |
+
pattern = r'{(.+)}'
|
77 |
+
match = re.search(pattern, llm_extracted_data, re.DOTALL)
|
78 |
+
|
79 |
+
data_dict = {}
|
80 |
+
|
81 |
+
if match:
|
82 |
+
extracted_text = match.group(1)
|
83 |
+
# Converting the extracted text to a dictionary
|
84 |
+
data_dict = eval('{' + extracted_text + '}')
|
85 |
+
print(data_dict)
|
86 |
+
else:
|
87 |
+
print("No match found.")
|
88 |
+
|
89 |
+
df=df._append([data_dict], ignore_index=True)
|
90 |
+
print("********************DONE***************")
|
91 |
+
#df=df.append(save_to_dataframe(llm_extracted_data), ignore_index=True)
|
92 |
+
|
93 |
+
df.head()
|
94 |
+
return df
|
requirements.txt
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
langchain==0.0.351
|
2 |
+
streamlit==1.29.0
|
3 |
+
openai==1.5.0
|
4 |
+
python-dotenv==1.0.0
|
5 |
+
pypdf==3.17.3
|
6 |
+
replicate==0.9.0
|