Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,12 +1,21 @@
|
|
| 1 |
-
from flask import Flask, request, jsonify
|
| 2 |
import PyPDF2
|
| 3 |
from transformers import pipeline
|
| 4 |
import psycopg2
|
| 5 |
import json
|
| 6 |
-
import
|
|
|
|
|
|
|
| 7 |
|
| 8 |
app = Flask(__name__)
|
| 9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
# Database connection
|
| 11 |
conn = psycopg2.connect(
|
| 12 |
dbname="contracts_db",
|
|
@@ -21,6 +30,48 @@ cursor = conn.cursor()
|
|
| 21 |
ner_pipeline = pipeline("ner", model="dslim/bert-base-NER", aggregation_strategy="simple")
|
| 22 |
compliance_pipeline = pipeline("text-classification", model="facebook/bart-large-mnli")
|
| 23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
def extract_text_from_pdf(file):
|
| 25 |
try:
|
| 26 |
pdf_reader = PyPDF2.PdfReader(file)
|
|
@@ -33,6 +84,10 @@ def extract_text_from_pdf(file):
|
|
| 33 |
|
| 34 |
@app.route('/upload_contract', methods=['POST'])
|
| 35 |
def upload_contract():
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
if 'contract' not in request.files:
|
| 37 |
return jsonify({"error": "No file uploaded"}), 400
|
| 38 |
|
|
@@ -56,7 +111,7 @@ def upload_contract():
|
|
| 56 |
"score": result['scores'][0]
|
| 57 |
})
|
| 58 |
|
| 59 |
-
# Store in
|
| 60 |
try:
|
| 61 |
cursor.execute(
|
| 62 |
"INSERT INTO contracts (content, entities, compliance) VALUES (%s, %s, %s) RETURNING id",
|
|
@@ -68,6 +123,17 @@ def upload_contract():
|
|
| 68 |
conn.rollback()
|
| 69 |
return jsonify({"error": str(e)}), 500
|
| 70 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
return jsonify({
|
| 72 |
"contract_id": contract_id,
|
| 73 |
"entities": extracted_data,
|
|
|
|
| 1 |
+
from flask import Flask, request, jsonify, redirect, url_for
|
| 2 |
import PyPDF2
|
| 3 |
from transformers import pipeline
|
| 4 |
import psycopg2
|
| 5 |
import json
|
| 6 |
+
import requests
|
| 7 |
+
from simple_salesforce import Salesforce
|
| 8 |
+
import urllib.parse
|
| 9 |
|
| 10 |
app = Flask(__name__)
|
| 11 |
|
| 12 |
+
# Salesforce Connected App credentials
|
| 13 |
+
SALESFORCE_CONSUMER_KEY = 'your_consumer_key'
|
| 14 |
+
SALESFORCE_CONSUMER_SECRET = 'your_consumer_secret'
|
| 15 |
+
SALESFORCE_REDIRECT_URI = 'http://localhost:5000/oauth/callback'
|
| 16 |
+
SALESFORCE_AUTH_URL = 'https://login.salesforce.com/services/oauth2/authorize'
|
| 17 |
+
SALESFORCE_TOKEN_URL = 'https://login.salesforce.com/services/oauth2/token'
|
| 18 |
+
|
| 19 |
# Database connection
|
| 20 |
conn = psycopg2.connect(
|
| 21 |
dbname="contracts_db",
|
|
|
|
| 30 |
ner_pipeline = pipeline("ner", model="dslim/bert-base-NER", aggregation_strategy="simple")
|
| 31 |
compliance_pipeline = pipeline("text-classification", model="facebook/bart-large-mnli")
|
| 32 |
|
| 33 |
+
# Store OAuth tokens
|
| 34 |
+
access_token = None
|
| 35 |
+
instance_url = None
|
| 36 |
+
|
| 37 |
+
@app.route('/login')
|
| 38 |
+
def login():
|
| 39 |
+
params = {
|
| 40 |
+
'client_id': SALESFORCE_CONSUMER_KEY,
|
| 41 |
+
'redirect_uri': SALESFORCE_REDIRECT_URI,
|
| 42 |
+
'response_type': 'code',
|
| 43 |
+
'scope': 'api refresh_token'
|
| 44 |
+
}
|
| 45 |
+
auth_url = f"{SALESFORCE_AUTH_URL}?{urllib.parse.urlencode(params)}"
|
| 46 |
+
return redirect(auth_url)
|
| 47 |
+
|
| 48 |
+
@app.route('/oauth/callback')
|
| 49 |
+
def oauth_callback():
|
| 50 |
+
global access_token, instance_url
|
| 51 |
+
code = request.args.get('code')
|
| 52 |
+
if not code:
|
| 53 |
+
return jsonify({"error": "Authorization code not found"}), 400
|
| 54 |
+
|
| 55 |
+
data = {
|
| 56 |
+
'grant_type': 'authorization_code',
|
| 57 |
+
'client_id': SALESFORCE_CONSUMER_KEY,
|
| 58 |
+
'client_secret': SALESFORCE_CONSUMER_SECRET,
|
| 59 |
+
'redirect_uri': SALESFORCE_REDIRECT_URI,
|
| 60 |
+
'code': code
|
| 61 |
+
}
|
| 62 |
+
response = requests.post(SALESFORCE_TOKEN_URL, data=data)
|
| 63 |
+
if response.status_code != 200:
|
| 64 |
+
return jsonify({"error": "Failed to obtain access token"}), 500
|
| 65 |
+
|
| 66 |
+
token_data = response.json()
|
| 67 |
+
access_token = token_data['access_token']
|
| 68 |
+
instance_url = token_data['instance_url']
|
| 69 |
+
return redirect(url_for('index'))
|
| 70 |
+
|
| 71 |
+
@app.route('/')
|
| 72 |
+
def index():
|
| 73 |
+
return jsonify({"message": "Welcome to Contract Analyzer. Please log in to Salesforce."})
|
| 74 |
+
|
| 75 |
def extract_text_from_pdf(file):
|
| 76 |
try:
|
| 77 |
pdf_reader = PyPDF2.PdfReader(file)
|
|
|
|
| 84 |
|
| 85 |
@app.route('/upload_contract', methods=['POST'])
|
| 86 |
def upload_contract():
|
| 87 |
+
global access_token, instance_url
|
| 88 |
+
if not access_token or not instance_url:
|
| 89 |
+
return jsonify({"error": "Please authenticate with Salesforce first"}), 401
|
| 90 |
+
|
| 91 |
if 'contract' not in request.files:
|
| 92 |
return jsonify({"error": "No file uploaded"}), 400
|
| 93 |
|
|
|
|
| 111 |
"score": result['scores'][0]
|
| 112 |
})
|
| 113 |
|
| 114 |
+
# Store in PostgreSQL
|
| 115 |
try:
|
| 116 |
cursor.execute(
|
| 117 |
"INSERT INTO contracts (content, entities, compliance) VALUES (%s, %s, %s) RETURNING id",
|
|
|
|
| 123 |
conn.rollback()
|
| 124 |
return jsonify({"error": str(e)}), 500
|
| 125 |
|
| 126 |
+
# Push to Salesforce
|
| 127 |
+
try:
|
| 128 |
+
sf = Salesforce(instance_url=instance_url, session_id=access_token)
|
| 129 |
+
sf.Contract__c.create({
|
| 130 |
+
'Content__c': text[:1000], # Truncate for Salesforce field limit
|
| 131 |
+
'Entities__c': json.dumps(extracted_data),
|
| 132 |
+
'Compliance__c': json.dumps(compliance_results)
|
| 133 |
+
})
|
| 134 |
+
except Exception as e:
|
| 135 |
+
return jsonify({"error": f"Failed to save to Salesforce: {str(e)}"}), 500
|
| 136 |
+
|
| 137 |
return jsonify({
|
| 138 |
"contract_id": contract_id,
|
| 139 |
"entities": extracted_data,
|