Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,47 +1,179 @@
|
|
1 |
import os
|
2 |
os.environ["OMP_NUM_THREADS"] = "1"
|
3 |
|
|
|
|
|
4 |
import gradio as gr
|
5 |
from utils import extract_kyc_fields
|
6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
"""
|
10 |
-
|
11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
"""
|
13 |
results = {"aadhaar": None, "pan": None}
|
14 |
|
|
|
|
|
|
|
|
|
15 |
if aadhaar_file:
|
16 |
try:
|
17 |
res = extract_kyc_fields(aadhaar_file.name)
|
18 |
res["source_file"] = os.path.basename(aadhaar_file.name)
|
19 |
results["aadhaar"] = res
|
20 |
except Exception as e:
|
21 |
-
results["aadhaar"] = {"error": f"Aadhaar OCR failed: {str(e)}"}
|
22 |
|
|
|
23 |
if pan_file:
|
24 |
try:
|
25 |
res = extract_kyc_fields(pan_file.name)
|
26 |
res["source_file"] = os.path.basename(pan_file.name)
|
27 |
results["pan"] = res
|
28 |
except Exception as e:
|
29 |
-
results["pan"] = {"error": f"PAN OCR failed: {str(e)}"}
|
30 |
|
31 |
-
|
32 |
-
return {"error": "Please upload both Aadhaar and PAN files."}
|
33 |
|
34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
35 |
|
|
|
36 |
|
37 |
-
|
|
|
|
|
38 |
gr.Markdown(
|
39 |
"""
|
40 |
-
# π§Ύ Smart KYC OCR
|
41 |
-
Upload
|
42 |
-
Click **Extract KYC Info** to get structured output.
|
43 |
|
44 |
-
|
|
|
|
|
45 |
"""
|
46 |
)
|
47 |
|
@@ -57,20 +189,21 @@ with gr.Blocks(title="Smart KYC OCR") as demo:
|
|
57 |
file_types=[".jpg", ".jpeg", ".png"]
|
58 |
)
|
59 |
|
|
|
|
|
60 |
submit_btn = gr.Button("π Extract KYC Info", variant="primary")
|
61 |
-
output_json = gr.JSON(label="π
|
62 |
|
63 |
submit_btn.click(
|
64 |
fn=process_documents,
|
65 |
-
inputs=[aadhaar_uploader, pan_uploader],
|
66 |
outputs=output_json,
|
67 |
)
|
68 |
|
69 |
gr.Markdown("---")
|
70 |
gr.Markdown(
|
71 |
"""
|
72 |
-
π **
|
73 |
-
No data is stored or shared.
|
74 |
"""
|
75 |
)
|
76 |
|
|
|
1 |
import os
|
2 |
os.environ["OMP_NUM_THREADS"] = "1"
|
3 |
|
4 |
+
import re
|
5 |
+
import datetime as dt
|
6 |
import gradio as gr
|
7 |
from utils import extract_kyc_fields
|
8 |
|
9 |
+
# ------------------------------------------------------------------
|
10 |
+
# π CREDENTIALS STRATEGY
|
11 |
+
# 1) Preferred: read from environment (HF Space Secrets / .env).
|
12 |
+
# 2) Fallback (NOT RECOMMENDED): hardcoded defaults below.
|
13 |
+
# If you insist, replace the REPLACE_ME values locally.
|
14 |
+
# ------------------------------------------------------------------
|
15 |
+
SF_DEFAULT_USERNAME = "REPLACE_ME_USERNAME" # e.g., "[email protected]"
|
16 |
+
SF_DEFAULT_PASSWORD = "REPLACE_ME_PASSWORD" # e.g., "Lic@2025"
|
17 |
+
SF_DEFAULT_TOKEN = "REPLACE_ME_SECURITY_TOKEN" # e.g., "AmmfRcd6IiYaRtSGntBnzNMQU"
|
18 |
+
SF_DEFAULT_DOMAIN = "login" # "login" (prod) or "test" (sandbox)
|
19 |
+
# ------------------------------------------------------------------
|
20 |
|
21 |
+
# simple-salesforce is required for SF push
|
22 |
+
try:
|
23 |
+
from simple_salesforce import Salesforce
|
24 |
+
SF_AVAILABLE = True
|
25 |
+
except Exception:
|
26 |
+
SF_AVAILABLE = False
|
27 |
+
|
28 |
+
|
29 |
+
# ---------- helpers ----------
|
30 |
+
def _parse_birthdate(dob_text: str):
|
31 |
+
"""
|
32 |
+
Normalize DOB to YYYY-MM-DD for Salesforce Date fields.
|
33 |
+
Supports dd/mm/yyyy, dd-mm-yyyy, dd.mm.yyyy, yyyy-mm-dd, or just YYYY (mapped to mid-year).
|
34 |
+
"""
|
35 |
+
if not dob_text or dob_text == "Not found":
|
36 |
+
return None
|
37 |
+
s = dob_text.strip()
|
38 |
+
|
39 |
+
m = re.fullmatch(r"(\d{4})-(\d{2})-(\d{2})", s)
|
40 |
+
if m:
|
41 |
+
y, mo, d = map(int, m.groups())
|
42 |
+
try:
|
43 |
+
return dt.date(y, mo, d).isoformat()
|
44 |
+
except ValueError:
|
45 |
+
return None
|
46 |
+
|
47 |
+
m = re.fullmatch(r"(\d{2})[./-](\d{2})[./-](\d{4})", s)
|
48 |
+
if m:
|
49 |
+
d, mo, y = map(int, m.groups())
|
50 |
+
try:
|
51 |
+
return dt.date(y, mo, d).isoformat()
|
52 |
+
except ValueError:
|
53 |
+
return None
|
54 |
+
|
55 |
+
m = re.fullmatch(r"(19|20)\d{2}", s)
|
56 |
+
if m:
|
57 |
+
y = int(s)
|
58 |
+
try:
|
59 |
+
return dt.date(y, 6, 15).isoformat()
|
60 |
+
except ValueError:
|
61 |
+
return None
|
62 |
+
|
63 |
+
return None
|
64 |
+
|
65 |
+
|
66 |
+
def sf_connect_from_env_or_defaults():
|
67 |
"""
|
68 |
+
Tries env vars first (HF Space Secrets / .env), then hardcoded defaults.
|
69 |
+
Env vars: SF_USERNAME, SF_PASSWORD, SF_SECURITY_TOKEN, SF_DOMAIN
|
70 |
+
"""
|
71 |
+
if not SF_AVAILABLE:
|
72 |
+
raise RuntimeError("simple-salesforce is not installed.")
|
73 |
+
|
74 |
+
username = os.getenv("SF_USERNAME", SF_DEFAULT_USERNAME)
|
75 |
+
password = os.getenv("SF_PASSWORD", SF_DEFAULT_PASSWORD)
|
76 |
+
token = os.getenv("SF_SECURITY_TOKEN", SF_DEFAULT_TOKEN)
|
77 |
+
domain = os.getenv("SF_DOMAIN", SF_DEFAULT_DOMAIN or "login")
|
78 |
+
|
79 |
+
# Basic validation: block if still placeholders
|
80 |
+
if any(v.startswith("REPLACE_ME") for v in [username, password, token]):
|
81 |
+
raise ValueError(
|
82 |
+
"Salesforce credentials missing. "
|
83 |
+
"Either set Space Secrets/ENV (SF_USERNAME, SF_PASSWORD, SF_SECURITY_TOKEN, SF_DOMAIN) "
|
84 |
+
"or replace the REPLACE_ME_* constants in app.py."
|
85 |
+
)
|
86 |
+
|
87 |
+
return Salesforce(username=username, password=password, security_token=token, domain=domain)
|
88 |
+
|
89 |
+
|
90 |
+
def sf_push_kyc_record(sf, ocr_results):
|
91 |
+
"""
|
92 |
+
Create one KYC_Record__c combining Aadhaar + PAN.
|
93 |
+
Fields:
|
94 |
+
Aadhaar_Number__c, Aadhaar_Name__c, Aadhaar_DOB__c (Date)
|
95 |
+
PAN_Number__c, Pan_Name__c, Pan_DOB__c (Date)
|
96 |
+
"""
|
97 |
+
a = ocr_results.get("aadhaar") or {}
|
98 |
+
p = ocr_results.get("pan") or {}
|
99 |
+
|
100 |
+
aadhaar_number = a.get("aadhaar_number") if (a.get("card_type") == "AADHAAR") else None
|
101 |
+
aadhaar_name = a.get("name") if (a.get("card_type") == "AADHAAR") else None
|
102 |
+
aadhaar_dob = _parse_birthdate(a.get("dob")) if (a.get("card_type") == "AADHAAR") else None
|
103 |
+
|
104 |
+
pan_number = p.get("pan_number") if (p.get("card_type") == "PAN") else None
|
105 |
+
pan_name = p.get("name") if (p.get("card_type") == "PAN") else None
|
106 |
+
pan_dob = _parse_birthdate(p.get("dob")) if (p.get("card_type") == "PAN") else None
|
107 |
+
|
108 |
+
payload = {
|
109 |
+
"Aadhaar_Number__c": aadhaar_number,
|
110 |
+
"Aadhaar_Name__c": aadhaar_name,
|
111 |
+
"Aadhaar_DOB__c": aadhaar_dob,
|
112 |
+
"PAN_Number__c": pan_number,
|
113 |
+
"Pan_Name__c": pan_name,
|
114 |
+
"Pan_DOB__c": pan_dob,
|
115 |
+
}
|
116 |
+
payload = {k: v for k, v in payload.items() if v is not None}
|
117 |
+
|
118 |
+
try:
|
119 |
+
result = sf.KYC_Record__c.create(payload)
|
120 |
+
return {"success": True, "id": result.get("id"), "payload": payload}
|
121 |
+
except Exception as e:
|
122 |
+
return {"success": False, "error": str(e), "payload": payload}
|
123 |
+
|
124 |
+
|
125 |
+
# ---------- gradio callback ----------
|
126 |
+
def process_documents(aadhaar_file, pan_file, push_to_sf):
|
127 |
+
"""
|
128 |
+
- Runs OCR on Aadhaar and PAN separately.
|
129 |
+
- Optionally pushes a single KYC_Record__c to Salesforce.
|
130 |
"""
|
131 |
results = {"aadhaar": None, "pan": None}
|
132 |
|
133 |
+
if not aadhaar_file and not pan_file:
|
134 |
+
return {"error": "Please upload at least one file (Aadhaar and/or PAN)."}
|
135 |
+
|
136 |
+
# OCR Aadhaar
|
137 |
if aadhaar_file:
|
138 |
try:
|
139 |
res = extract_kyc_fields(aadhaar_file.name)
|
140 |
res["source_file"] = os.path.basename(aadhaar_file.name)
|
141 |
results["aadhaar"] = res
|
142 |
except Exception as e:
|
143 |
+
results["aadhaar"] = {"error": f"Aadhaar OCR failed: {str(e)}", "card_type": "UNKNOWN"}
|
144 |
|
145 |
+
# OCR PAN
|
146 |
if pan_file:
|
147 |
try:
|
148 |
res = extract_kyc_fields(pan_file.name)
|
149 |
res["source_file"] = os.path.basename(pan_file.name)
|
150 |
results["pan"] = res
|
151 |
except Exception as e:
|
152 |
+
results["pan"] = {"error": f"PAN OCR failed: {str(e)}", "card_type": "UNKNOWN"}
|
153 |
|
154 |
+
output = {"ocr": results}
|
|
|
155 |
|
156 |
+
if push_to_sf:
|
157 |
+
try:
|
158 |
+
sf = sf_connect_from_env_or_defaults()
|
159 |
+
created = sf_push_kyc_record(sf, results)
|
160 |
+
output["salesforce"] = {"pushed": created.get("success", False), **created}
|
161 |
+
except Exception as e:
|
162 |
+
output["salesforce"] = {"pushed": False, "error": str(e)}
|
163 |
|
164 |
+
return output
|
165 |
|
166 |
+
|
167 |
+
# ---------- UI ----------
|
168 |
+
with gr.Blocks(title="Smart KYC OCR β Salesforce (KYC_Record__c)") as demo:
|
169 |
gr.Markdown(
|
170 |
"""
|
171 |
+
# π§Ύ Smart KYC OCR β Salesforce
|
172 |
+
Upload **Aadhaar** and **PAN** in separate boxes, then (optional) push a single **KYC_Record__c**.
|
|
|
173 |
|
174 |
+
- Preferred: configure `SF_USERNAME`, `SF_PASSWORD`, `SF_SECURITY_TOKEN`, `SF_DOMAIN` as environment variables
|
175 |
+
(Hugging Face **Space Secrets**).
|
176 |
+
- Emergency fallback: edit the `SF_DEFAULT_*` constants in `app.py` (NOT recommended).
|
177 |
"""
|
178 |
)
|
179 |
|
|
|
189 |
file_types=[".jpg", ".jpeg", ".png"]
|
190 |
)
|
191 |
|
192 |
+
push_to_sf = gr.Checkbox(label="Push to Salesforce (create KYC_Record__c)", value=False)
|
193 |
+
|
194 |
submit_btn = gr.Button("π Extract KYC Info", variant="primary")
|
195 |
+
output_json = gr.JSON(label="π Output (OCR + Salesforce)")
|
196 |
|
197 |
submit_btn.click(
|
198 |
fn=process_documents,
|
199 |
+
inputs=[aadhaar_uploader, pan_uploader, push_to_sf],
|
200 |
outputs=output_json,
|
201 |
)
|
202 |
|
203 |
gr.Markdown("---")
|
204 |
gr.Markdown(
|
205 |
"""
|
206 |
+
π **Security note:** Avoid hardcoding secrets. Prefer Space Secrets / environment variables.
|
|
|
207 |
"""
|
208 |
)
|
209 |
|