Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,35 +1,65 @@
|
|
| 1 |
-
import
|
| 2 |
-
import
|
| 3 |
-
import
|
|
|
|
| 4 |
from transformers import pipeline
|
| 5 |
import PyPDF2
|
| 6 |
import docx
|
| 7 |
import matplotlib.pyplot as plt
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
from simple_salesforce import Salesforce
|
| 9 |
|
| 10 |
# Initialize the BERT-based NLP pipeline
|
| 11 |
model_name = "dbmdz/bert-large-cased-finetuned-conll03-english" # Example, replace with your model
|
| 12 |
nlp_pipeline = pipeline("ner", model=model_name)
|
| 13 |
|
| 14 |
-
#
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
# Function to analyze contract text
|
| 18 |
def analyze_contract(contract_text):
|
| 19 |
try:
|
| 20 |
# Run the contract through the NLP pipeline
|
| 21 |
results = nlp_pipeline(contract_text)
|
| 22 |
-
|
| 23 |
# Parse and score clauses (this is a simplified version)
|
| 24 |
risk_score = 0
|
| 25 |
high_risk_clauses = []
|
| 26 |
-
|
| 27 |
for result in results:
|
| 28 |
# This assumes 'labels' are risk-related; adjust as per model output
|
| 29 |
if result['label'] in ["PENALTY", "OBLIGATION", "DELAY"]: # Customize as per your model's tags
|
| 30 |
high_risk_clauses.append(result['word'])
|
| 31 |
risk_score += 10 # Example scoring logic, modify as needed
|
| 32 |
-
|
| 33 |
return {
|
| 34 |
"high_risk_clauses": high_risk_clauses,
|
| 35 |
"risk_score": risk_score
|
|
@@ -37,72 +67,84 @@ def analyze_contract(contract_text):
|
|
| 37 |
except Exception as e:
|
| 38 |
return {"error": str(e)}
|
| 39 |
|
| 40 |
-
# Streamlit UI
|
| 41 |
-
st.title("Contract Risk Analyzer")
|
| 42 |
|
| 43 |
-
#
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
if contract_file is not None:
|
| 47 |
contract_text = ""
|
| 48 |
-
if
|
| 49 |
-
# Read PDF
|
| 50 |
try:
|
| 51 |
-
pdf_reader = PyPDF2.PdfReader(
|
| 52 |
for page in pdf_reader.pages:
|
| 53 |
contract_text += page.extract_text()
|
| 54 |
except Exception as e:
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
elif
|
| 58 |
-
# Read DOCX
|
| 59 |
try:
|
| 60 |
-
doc = docx.Document(
|
| 61 |
for para in doc.paragraphs:
|
| 62 |
contract_text += para.text
|
| 63 |
except Exception as e:
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
elif
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 84 |
# Generate the risk heatmap (simplified here)
|
| 85 |
-
st.subheader("Risk Heatmap")
|
| 86 |
-
|
| 87 |
-
# Create a simple bar chart to visualize the risk score
|
| 88 |
fig, ax = plt.subplots()
|
| 89 |
-
ax.barh(['Contract'], [
|
| 90 |
ax.set_xlim(0, 100) # Assuming risk score ranges from 0 to 100
|
| 91 |
ax.set_xlabel("Risk Score")
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
'
|
| 100 |
-
'
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
'
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import dash
|
| 2 |
+
from dash import dcc, html
|
| 3 |
+
import dash_bootstrap_components as dbc
|
| 4 |
+
import gradio as gr
|
| 5 |
from transformers import pipeline
|
| 6 |
import PyPDF2
|
| 7 |
import docx
|
| 8 |
import matplotlib.pyplot as plt
|
| 9 |
+
import numpy as np
|
| 10 |
+
import pandas as pd
|
| 11 |
+
import requests
|
| 12 |
+
import json
|
| 13 |
from simple_salesforce import Salesforce
|
| 14 |
|
| 15 |
# Initialize the BERT-based NLP pipeline
|
| 16 |
model_name = "dbmdz/bert-large-cased-finetuned-conll03-english" # Example, replace with your model
|
| 17 |
nlp_pipeline = pipeline("ner", model=model_name)
|
| 18 |
|
| 19 |
+
# Initialize Dash App
|
| 20 |
+
app = dash.Dash(__name__, external_stylesheets=[dbc.themes.BOOTSTRAP])
|
| 21 |
+
|
| 22 |
+
# Define app layout
|
| 23 |
+
app.layout = html.Div([
|
| 24 |
+
dbc.Row([
|
| 25 |
+
dbc.Col(html.H1("Contract Risk Analyzer", style={'textAlign': 'center'})),
|
| 26 |
+
]),
|
| 27 |
+
dbc.Row([
|
| 28 |
+
dbc.Col(html.Div([
|
| 29 |
+
html.Label("Upload Contract"),
|
| 30 |
+
dcc.Upload(
|
| 31 |
+
id='upload-data',
|
| 32 |
+
children=html.Button('Upload File'),
|
| 33 |
+
multiple=False
|
| 34 |
+
),
|
| 35 |
+
html.Div(id='file-upload-status'),
|
| 36 |
+
]), width=12),
|
| 37 |
+
]),
|
| 38 |
+
dbc.Row([
|
| 39 |
+
dbc.Col(html.Div(id='output-text'), width=12),
|
| 40 |
+
]),
|
| 41 |
+
dbc.Row([
|
| 42 |
+
dbc.Col(dcc.Graph(id='risk-heatmap'), width=12),
|
| 43 |
+
]),
|
| 44 |
+
])
|
| 45 |
+
|
| 46 |
|
| 47 |
# Function to analyze contract text
|
| 48 |
def analyze_contract(contract_text):
|
| 49 |
try:
|
| 50 |
# Run the contract through the NLP pipeline
|
| 51 |
results = nlp_pipeline(contract_text)
|
| 52 |
+
|
| 53 |
# Parse and score clauses (this is a simplified version)
|
| 54 |
risk_score = 0
|
| 55 |
high_risk_clauses = []
|
| 56 |
+
|
| 57 |
for result in results:
|
| 58 |
# This assumes 'labels' are risk-related; adjust as per model output
|
| 59 |
if result['label'] in ["PENALTY", "OBLIGATION", "DELAY"]: # Customize as per your model's tags
|
| 60 |
high_risk_clauses.append(result['word'])
|
| 61 |
risk_score += 10 # Example scoring logic, modify as needed
|
| 62 |
+
|
| 63 |
return {
|
| 64 |
"high_risk_clauses": high_risk_clauses,
|
| 65 |
"risk_score": risk_score
|
|
|
|
| 67 |
except Exception as e:
|
| 68 |
return {"error": str(e)}
|
| 69 |
|
|
|
|
|
|
|
| 70 |
|
| 71 |
+
# Function to parse uploaded contract
|
| 72 |
+
def parse_contract(file_content, file_type):
|
|
|
|
|
|
|
| 73 |
contract_text = ""
|
| 74 |
+
if file_type == "application/pdf":
|
|
|
|
| 75 |
try:
|
| 76 |
+
pdf_reader = PyPDF2.PdfReader(file_content)
|
| 77 |
for page in pdf_reader.pages:
|
| 78 |
contract_text += page.extract_text()
|
| 79 |
except Exception as e:
|
| 80 |
+
return f"Error reading PDF: {str(e)}"
|
| 81 |
+
|
| 82 |
+
elif file_type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
|
|
|
|
| 83 |
try:
|
| 84 |
+
doc = docx.Document(file_content)
|
| 85 |
for para in doc.paragraphs:
|
| 86 |
contract_text += para.text
|
| 87 |
except Exception as e:
|
| 88 |
+
return f"Error reading DOCX: {str(e)}"
|
| 89 |
+
|
| 90 |
+
elif file_type == "text/plain":
|
| 91 |
+
contract_text = file_content.decode("utf-8")
|
| 92 |
+
|
| 93 |
+
return contract_text
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
# Callback to handle file upload
|
| 97 |
+
@app.callback(
|
| 98 |
+
[dash.dependencies.Output('file-upload-status', 'children'),
|
| 99 |
+
dash.dependencies.Output('output-text', 'children'),
|
| 100 |
+
dash.dependencies.Output('risk-heatmap', 'figure')],
|
| 101 |
+
[dash.dependencies.Input('upload-data', 'contents'),
|
| 102 |
+
dash.dependencies.State('upload-data', 'filename'),
|
| 103 |
+
dash.dependencies.State('upload-data', 'type')]
|
| 104 |
+
)
|
| 105 |
+
def update_output(file_contents, filename, file_type):
|
| 106 |
+
if file_contents is not None:
|
| 107 |
+
# Parse the contract
|
| 108 |
+
contract_text = parse_contract(file_contents, file_type)
|
| 109 |
+
|
| 110 |
+
if contract_text:
|
| 111 |
+
# Analyze the contract
|
| 112 |
+
analysis_results = analyze_contract(contract_text)
|
| 113 |
+
|
| 114 |
+
if "error" in analysis_results:
|
| 115 |
+
return "Error", f"An error occurred during analysis: {analysis_results['error']}", {}
|
| 116 |
+
|
| 117 |
+
# Display high-risk clauses and overall risk score
|
| 118 |
+
high_risk_clauses = analysis_results["high_risk_clauses"]
|
| 119 |
+
risk_score = analysis_results["risk_score"]
|
| 120 |
+
|
| 121 |
+
high_risk_text = f"High Risk Clauses: {', '.join(high_risk_clauses)}"
|
| 122 |
+
risk_score_text = f"Overall Risk Score: {risk_score}"
|
| 123 |
+
|
| 124 |
# Generate the risk heatmap (simplified here)
|
|
|
|
|
|
|
|
|
|
| 125 |
fig, ax = plt.subplots()
|
| 126 |
+
ax.barh(['Contract'], [risk_score], color='red')
|
| 127 |
ax.set_xlim(0, 100) # Assuming risk score ranges from 0 to 100
|
| 128 |
ax.set_xlabel("Risk Score")
|
| 129 |
+
|
| 130 |
+
# Returning results for display
|
| 131 |
+
return "File Uploaded Successfully", [high_risk_text, risk_score_text], {
|
| 132 |
+
'data': [{
|
| 133 |
+
'x': ['Contract'],
|
| 134 |
+
'y': [risk_score],
|
| 135 |
+
'type': 'bar',
|
| 136 |
+
'name': 'Risk Score',
|
| 137 |
+
'marker': {'color': 'red'}
|
| 138 |
+
}],
|
| 139 |
+
'layout': {
|
| 140 |
+
'title': 'Risk Heatmap',
|
| 141 |
+
'xaxis': {'title': 'Risk Score'},
|
| 142 |
+
'yaxis': {'title': 'Contract'}
|
| 143 |
+
}
|
| 144 |
+
}
|
| 145 |
+
|
| 146 |
+
return "No File Uploaded", "", {}
|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
if __name__ == '__main__':
|
| 150 |
+
app.run_server(debug=True)
|