Spaces:
Sleeping
Sleeping
Update utils.py
Browse files
utils.py
CHANGED
|
@@ -5,6 +5,7 @@ import re
|
|
| 5 |
import difflib
|
| 6 |
from simple_salesforce import Salesforce
|
| 7 |
from dotenv import load_dotenv
|
|
|
|
| 8 |
|
| 9 |
# Load .env variables
|
| 10 |
load_dotenv()
|
|
@@ -51,6 +52,16 @@ def extract_text_from_pdf(pdf_path):
|
|
| 51 |
except Exception as e:
|
| 52 |
return None, str(e)
|
| 53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
# ✅ Extract key-value pairs smartly from the PDF text
|
| 55 |
def extract_key_value_pairs(pdf_path):
|
| 56 |
try:
|
|
@@ -95,9 +106,9 @@ def extract_key_value_pairs(pdf_path):
|
|
| 95 |
)
|
| 96 |
if match_dates:
|
| 97 |
keys.append("Agreement Start Date")
|
| 98 |
-
values.append(match_dates.group(1).strip())
|
| 99 |
keys.append("Agreement End Date")
|
| 100 |
-
values.append(match_dates.group(2).strip())
|
| 101 |
|
| 102 |
return [{"keys": keys, "values": values}], None
|
| 103 |
except Exception as e:
|
|
@@ -163,4 +174,4 @@ def attach_pdf(sf, record_id, pdf_path):
|
|
| 163 |
def log_failure(pdf_path, object_name, error):
|
| 164 |
with open("failures.json", "a") as f:
|
| 165 |
json.dump({"pdf": pdf_path, "object": object_name, "error": error}, f)
|
| 166 |
-
f.write("\n")
|
|
|
|
| 5 |
import difflib
|
| 6 |
from simple_salesforce import Salesforce
|
| 7 |
from dotenv import load_dotenv
|
| 8 |
+
from datetime import datetime
|
| 9 |
|
| 10 |
# Load .env variables
|
| 11 |
load_dotenv()
|
|
|
|
| 52 |
except Exception as e:
|
| 53 |
return None, str(e)
|
| 54 |
|
| 55 |
+
# ✅ Helper: Convert date to YYYY-MM-DD
|
| 56 |
+
def convert_to_salesforce_date(date_str):
|
| 57 |
+
for fmt in ("%m/%d/%Y", "%m/%d/%y"):
|
| 58 |
+
try:
|
| 59 |
+
dt = datetime.strptime(date_str, fmt)
|
| 60 |
+
return dt.strftime("%Y-%m-%d")
|
| 61 |
+
except ValueError:
|
| 62 |
+
continue
|
| 63 |
+
return date_str # Return original if no match
|
| 64 |
+
|
| 65 |
# ✅ Extract key-value pairs smartly from the PDF text
|
| 66 |
def extract_key_value_pairs(pdf_path):
|
| 67 |
try:
|
|
|
|
| 106 |
)
|
| 107 |
if match_dates:
|
| 108 |
keys.append("Agreement Start Date")
|
| 109 |
+
values.append(convert_to_salesforce_date(match_dates.group(1).strip()))
|
| 110 |
keys.append("Agreement End Date")
|
| 111 |
+
values.append(convert_to_salesforce_date(match_dates.group(2).strip()))
|
| 112 |
|
| 113 |
return [{"keys": keys, "values": values}], None
|
| 114 |
except Exception as e:
|
|
|
|
| 174 |
def log_failure(pdf_path, object_name, error):
|
| 175 |
with open("failures.json", "a") as f:
|
| 176 |
json.dump({"pdf": pdf_path, "object": object_name, "error": error}, f)
|
| 177 |
+
f.write("\n")
|