debug_data.py

"""
Debug script to verify grant data is loading correctly.
Run with: python debug_data.py
"""

from pathlib import Path
from src.analyzer.data_loader import load_current_grants
from src.analyzer.search.service import get_grant_by_id

print("=" * 60)
print("GRANT DATA DEBUG")
print("=" * 60)

# Load all grants
print("\n1. Loading grants from snapshots...")
grants = load_current_grants(Path('data/snapshots'))
print(f"   ✓ Loaded {len(grants)} grants")

# Check grant 2313 specifically
print("\n2. Checking competition-2313 (Battery Feasibility)...")
grant_2313 = next((g for g in grants if '2313' in g.id), None)

if grant_2313:
    print(f"   ✓ Found grant: {grant_2313.title}")
    print(f"   Status: {grant_2313.status}")
    print(f"   Has funding: {grant_2313.funding is not None}")

    if grant_2313.funding:
        print(f"   Funding min: £{grant_2313.funding.min:,.0f}" if grant_2313.funding.min else "   Funding min: None")
        print(f"   Funding max: £{grant_2313.funding.max:,.0f}" if grant_2313.funding.max else "   Funding max: None")
        print(f"   Total pot: £{grant_2313.funding.total_pot:,.0f}" if grant_2313.funding.total_pot else "   Total pot: None")
    else:
        print("   ❌ NO FUNDING DATA")

    print(f"   Has summary: {grant_2313.summary is not None and len(grant_2313.summary) > 0}")
    print(f"   Close date: {grant_2313.close_date}")
    print(f"   URL: {grant_2313.url}")
else:
    print("   ❌ Grant 2313 not found in loaded grants!")

# Check grant 2314
print("\n3. Checking competition-2314 (Battery Concept)...")
grant_2314 = next((g for g in grants if '2314' in g.id), None)

if grant_2314:
    print(f"   ✓ Found grant: {grant_2314.title}")
    print(f"   Status: {grant_2314.status}")

    if grant_2314.funding:
        print(f"   Funding min: £{grant_2314.funding.min:,.0f}" if grant_2314.funding.min else "   Funding min: None")
        print(f"   Funding max: £{grant_2314.funding.max:,.0f}" if grant_2314.funding.max else "   Funding max: None")
    else:
        print("   ❌ NO FUNDING DATA")
else:
    print("   ❌ Grant 2314 not found!")

# Check search index
print("\n4. Checking search index (get_grant_by_id)...")
try:
    indexed_grant = get_grant_by_id("competition-2313")
    if indexed_grant:
        print(f"   ✓ Found via search index: {indexed_grant.title}")
        print(f"   Status via index: {indexed_grant.status}")
        print(f"   Has funding via index: {indexed_grant.funding is not None}")

        if indexed_grant.funding:
            print(f"   Funding via index: £{indexed_grant.funding.min:,.0f} - £{indexed_grant.funding.max:,.0f}")
        else:
            print("   ❌ NO FUNDING DATA VIA INDEX")
    else:
        print("   ❌ Not found via search index")
except Exception as e:
    print(f"   ❌ Error accessing index: {e}")

# Summary
print("\n" + "=" * 60)
print("DIAGNOSIS:")
print("=" * 60)

has_data = grant_2313 and grant_2313.funding and grant_2313.funding.min
has_indexed_data = False

try:
    indexed_grant = get_grant_by_id("competition-2313")
    has_indexed_data = indexed_grant and indexed_grant.funding and indexed_grant.funding.min
except:
    pass

if has_data and has_indexed_data:
    print("✓ Data is present and index is working")
    print("  Issue is likely in context building or LLM prompt")
elif has_data and not has_indexed_data:
    print("✓ Data exists in snapshots")
    print("❌ Search index is not preserving the data")
    print("  ACTION: Rebuild search index")
elif not has_data:
    print("❌ Data missing from snapshot files")
    print("  ACTION: Re-run scraper to get funding data")
else:
    print("? Unclear - check logs above")

print("=" * 60)