Update backend/agents/normalizer.py

backend/agents/normalizer.py

@@ -1,6 +1,4 @@
 from datetime import datetime
-from .normalizer import normalize_job_data
-
 import re
 
 def clean_html(raw_html):
@@ -25,12 +23,11 @@ def normalize_job_data(job_data, source):
     if source == "RemoteOK":
         date_posted_str = 'N/A'
         try:
-            # FIX: Safely convert timestamp, handling potential strings or errors.
+            # Safely convert timestamp
             timestamp = int(job_data.get('date', 0))
             if timestamp > 0:
                 date_posted_str = datetime.fromtimestamp(timestamp).strftime('%Y-%m-%d')
         except (ValueError, TypeError, OSError):
-            # If the date is not a valid number or causes an error, it will default to 'N/A'.
             pass
 
         return {
@@ -40,9 +37,10 @@ def normalize_job_data(job_data, source):
             'location': job_data.get('location', 'Remote'),
             'description': clean_html(job_data.get('description', '')),
             'url': job_data.get('url', ''),
-            'date_posted': date_posted_str, # Use the safely handled string
+            'date_posted': date_posted_str,
             'source': source
         }
+
     elif source == "LinkedIn (Stub)":
         return {
             'id': f"linkedin_{job_data.get('title', '').replace(' ', '_')}",
@@ -54,17 +52,19 @@ def normalize_job_data(job_data, source):
             'date_posted': job_data.get('posted_at', 'N/A'),
             'source': source
         }
+
     elif source == "Upwork (Stub)":
         return {
             'id': f"upwork_{job_data.get('title', '').replace(' ', '_')}",
             'title': job_data.get('title', 'N/A'),
-            'company': 'Upwork Client', # Upwork is client-based
+            'company': 'Upwork Client',
             'location': job_data.get('client', {}).get('country', 'Remote'),
             'description': clean_html(job_data.get('snippet', '')),
             'url': job_data.get('url', ''),
             'date_posted': job_data.get('published_on', 'N/A').split('T')[0],
             'source': source
         }
+
     elif source == "Freelancer (Stub)":
         return {
             'id': f"freelancer_{job_data.get('title', '').replace(' ', '_')}",
@@ -76,6 +76,7 @@ def normalize_job_data(job_data, source):
             'date_posted': datetime.fromtimestamp(job_data.get('submitdate', 0)).strftime('%Y-%m-%d'),
             'source': source
         }
+
     else:
         # Generic fallback
         return {
@@ -88,4 +89,3 @@ def normalize_job_data(job_data, source):
             'date_posted': 'N/A',
             'source': source
         }
-