Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,65 +1,153 @@
|
|
| 1 |
-
import
|
| 2 |
-
from salesforce_utils import get_salesforce_objects, get_token, create_record, attach_pdf
|
| 3 |
from ai_mapping import run_ai_mapping
|
| 4 |
from ocr_utils import extract_text_from_pdf
|
| 5 |
-
import tempfile
|
| 6 |
import os
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
-
def
|
| 9 |
-
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
-
# Get token and Salesforce objects
|
| 13 |
try:
|
| 14 |
token, instance_url = get_token()
|
| 15 |
except Exception as e:
|
| 16 |
return f"❌ Salesforce authentication failed: {str(e)}", None
|
| 17 |
|
| 18 |
-
#
|
| 19 |
try:
|
| 20 |
-
|
| 21 |
-
if
|
| 22 |
-
return "⚠️ No text extracted from PDF. It may be empty or require OCR.", None
|
| 23 |
except Exception as e:
|
| 24 |
-
return f"❌ Failed to
|
| 25 |
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
|
|
|
| 31 |
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
return f"❌ Failed to create record: {record_response}", mappings
|
| 40 |
-
except Exception as e:
|
| 41 |
-
return f"❌ Failed to create Salesforce record: {str(e)}", mappings
|
| 42 |
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
|
| 51 |
-
|
| 52 |
-
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
object_list = gr.Dropdown(choices=get_salesforce_object_names(), label="Salesforce Object")
|
| 57 |
|
| 58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
|
| 60 |
-
|
| 61 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
|
| 63 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
|
| 65 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
from salesforce_utils import get_salesforce_objects, get_salesforce_object_fields, get_token, create_record, attach_pdf
|
| 3 |
from ai_mapping import run_ai_mapping
|
| 4 |
from ocr_utils import extract_text_from_pdf
|
|
|
|
| 5 |
import os
|
| 6 |
+
import json
|
| 7 |
+
import tempfile
|
| 8 |
+
|
| 9 |
+
# Initialize session state for failed records
|
| 10 |
+
if 'failed_records' not in st.session_state:
|
| 11 |
+
st.session_state.failed_records = []
|
| 12 |
|
| 13 |
+
def save_failed_record(pdf_name, object_name, error, mappings):
|
| 14 |
+
"""Log failed records for reconciliation."""
|
| 15 |
+
st.session_state.failed_records.append({
|
| 16 |
+
"pdf_name": pdf_name,
|
| 17 |
+
"object_name": object_name,
|
| 18 |
+
"error": error,
|
| 19 |
+
"mappings": mappings
|
| 20 |
+
})
|
| 21 |
+
|
| 22 |
+
def process_contract(uploaded_files, object_name, manual_mappings):
|
| 23 |
+
"""Process uploaded PDFs and create Salesforce records."""
|
| 24 |
+
if not uploaded_files:
|
| 25 |
+
return "❌ No files uploaded.", None
|
| 26 |
|
|
|
|
| 27 |
try:
|
| 28 |
token, instance_url = get_token()
|
| 29 |
except Exception as e:
|
| 30 |
return f"❌ Salesforce authentication failed: {str(e)}", None
|
| 31 |
|
| 32 |
+
# Get object fields for mapping
|
| 33 |
try:
|
| 34 |
+
object_fields = get_salesforce_object_fields(token, instance_url, object_name)
|
| 35 |
+
object_field_names = [field['name'] for field in object_fields if field.get('createable')]
|
|
|
|
| 36 |
except Exception as e:
|
| 37 |
+
return f"❌ Failed to fetch object fields: {str(e)}", None
|
| 38 |
|
| 39 |
+
results = []
|
| 40 |
+
for pdf_file in uploaded_files:
|
| 41 |
+
pdf_name = pdf_file.name
|
| 42 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
|
| 43 |
+
tmp.write(pdf_file.read())
|
| 44 |
+
tmp_path = tmp.name
|
| 45 |
|
| 46 |
+
try:
|
| 47 |
+
# Step 1: OCR
|
| 48 |
+
text_data = extract_text_from_pdf(tmp_path)
|
| 49 |
+
if not text_data:
|
| 50 |
+
save_failed_record(pdf_name, object_name, "No text extracted from PDF", {})
|
| 51 |
+
results.append(f"⚠️ {pdf_name}: No text extracted")
|
| 52 |
+
continue
|
|
|
|
|
|
|
|
|
|
| 53 |
|
| 54 |
+
# Step 2: AI Mapping
|
| 55 |
+
ai_result = run_ai_mapping(text_data, tmp_path, object_field_names)
|
| 56 |
+
if ai_result['status'] == 'failed':
|
| 57 |
+
save_failed_record(pdf_name, object_name, ai_result['error'], ai_result['mappings'])
|
| 58 |
+
results.append(f"❌ {pdf_name}: {ai_result['error']}")
|
| 59 |
+
continue
|
| 60 |
+
|
| 61 |
+
# Apply manual mappings (if provided)
|
| 62 |
+
mappings = {k: v['value'] for k, v in ai_result['mappings'].items()}
|
| 63 |
+
for field, value in manual_mappings.items():
|
| 64 |
+
if value and field in object_field_names:
|
| 65 |
+
mappings[field] = value
|
| 66 |
+
|
| 67 |
+
# Step 3: Create Salesforce record
|
| 68 |
+
record_response = create_record(object_name, mappings, token, instance_url)
|
| 69 |
+
if 'id' in record_response:
|
| 70 |
+
attach_pdf(record_response['id'], tmp_path, token, instance_url)
|
| 71 |
+
results.append(f"✅ {pdf_name}: Record created (ID: {record_response['id']})")
|
| 72 |
+
else:
|
| 73 |
+
save_failed_record(pdf_name, object_name, f"Failed to create record: {record_response}", mappings)
|
| 74 |
+
results.append(f"❌ {pdf_name}: Failed to create record: {record_response}")
|
| 75 |
+
except Exception as e:
|
| 76 |
+
save_failed_record(pdf_name, object_name, str(e), {})
|
| 77 |
+
results.append(f"❌ {pdf_name}: {str(e)}")
|
| 78 |
+
finally:
|
| 79 |
+
os.unlink(tmp_path)
|
| 80 |
+
|
| 81 |
+
return "\n".join(results), ai_result
|
| 82 |
+
|
| 83 |
+
def retry_failed_record(index, object_name, manual_mappings):
|
| 84 |
+
"""Retry a failed record with manual corrections."""
|
| 85 |
+
failed_record = st.session_state.failed_records.pop(index)
|
| 86 |
+
pdf_name = failed_record['pdf_name']
|
| 87 |
+
with open(failed_record['pdf_path'], 'rb') as f:
|
| 88 |
+
result, ai_result = process_contract([f], object_name, manual_mappings)
|
| 89 |
+
return result
|
| 90 |
|
| 91 |
+
# Streamlit UI
|
| 92 |
+
st.title("📄 Smart Contract Migrator (Streamlit)")
|
| 93 |
|
| 94 |
+
# Epic 1: PDF Upload
|
| 95 |
+
uploaded_files = st.file_uploader("Upload Contract PDFs", type="pdf", accept_multiple_files=True)
|
|
|
|
| 96 |
|
| 97 |
+
# Epic 2: Salesforce Object Selection
|
| 98 |
+
try:
|
| 99 |
+
token, instance_url = get_token()
|
| 100 |
+
objects = get_salesforce_objects(token, instance_url)
|
| 101 |
+
object_names = [obj['name'] for obj in objects if obj.get('createable')]
|
| 102 |
+
except Exception as e:
|
| 103 |
+
st.error(f"❌ Failed to fetch Salesforce objects: {str(e)}")
|
| 104 |
+
object_names = []
|
| 105 |
+
|
| 106 |
+
object_name = st.selectbox("Select Salesforce Object", object_names)
|
| 107 |
+
|
| 108 |
+
# Display object fields
|
| 109 |
+
if object_name:
|
| 110 |
+
try:
|
| 111 |
+
object_fields = get_salesforce_object_fields(token, instance_url, object_name)
|
| 112 |
+
field_names = [field['name'] for field in object_fields if field.get('createable')]
|
| 113 |
+
st.write("Available Fields:")
|
| 114 |
+
st.write(field_names)
|
| 115 |
+
except Exception as e:
|
| 116 |
+
st.error(f"❌ Failed to fetch fields for {object_name}: {str(e)}")
|
| 117 |
+
field_names = []
|
| 118 |
|
| 119 |
+
# Manual mapping input
|
| 120 |
+
manual_mappings = {}
|
| 121 |
+
if field_names:
|
| 122 |
+
st.subheader("Manual Field Mappings (Optional)")
|
| 123 |
+
for field in field_names:
|
| 124 |
+
manual_mappings[field] = st.text_input(f"{field}", "")
|
| 125 |
|
| 126 |
+
# Process button
|
| 127 |
+
if st.button("Extract, Map, and Upload"):
|
| 128 |
+
if uploaded_files and object_name:
|
| 129 |
+
status, ai_result = process_contract(uploaded_files, object_name, manual_mappings)
|
| 130 |
+
st.text_area("Status", status)
|
| 131 |
+
if ai_result:
|
| 132 |
+
st.json({
|
| 133 |
+
"mappings": ai_result['mappings'],
|
| 134 |
+
"unmapped_fields": ai_result['unmapped_fields'],
|
| 135 |
+
"status": ai_result['status']
|
| 136 |
+
})
|
| 137 |
|
| 138 |
+
# Epic 6: Reconciliation Dashboard
|
| 139 |
+
st.subheader("Reconciliation Dashboard")
|
| 140 |
+
if st.session_state.failed_records:
|
| 141 |
+
for i, record in enumerate(st.session_state.failed_records):
|
| 142 |
+
with st.expander(f"Failed Record: {record['pdf_name']}"):
|
| 143 |
+
st.write(f"Object: {record['object_name']}")
|
| 144 |
+
st.write(f"Error: {record['error']}")
|
| 145 |
+
st.write("Mappings:", record['mappings'])
|
| 146 |
+
manual_mappings_retry = {}
|
| 147 |
+
for field in field_names:
|
| 148 |
+
manual_mappings_retry[field] = st.text_input(f"{field} (Retry)", key=f"retry_{i}_{field}")
|
| 149 |
+
if st.button("Retry", key=f"retry_btn_{i}"):
|
| 150 |
+
result = retry_failed_record(i, record['object_name'], manual_mappings_retry)
|
| 151 |
+
st.write(result)
|
| 152 |
+
else:
|
| 153 |
+
st.write("No failed records.")
|