import json
import sys
from pathlib import Path

sys.stdout.reconfigure(encoding='utf-8')

# Load 1st audit results
with open('resources/knowledge/technical_info/01_임도/05_원가정보/품셈_원문md_마스터데이터_대조_검증보고서.json', 'r', encoding='utf-8') as f:
    audit_v1 = json.load(f)

# Load updated master
with open('resources/data_work_item_master/work_item_master_2026-01-01.json', 'r', encoding='utf-8') as f:
    master_v2 = json.load(f)

tables_v2 = {}
for wi in master_v2['work_items']:
    for t in wi.get('tables', []):
        tid = t.get('pum_table_id')
        if tid:
            tables_v2[tid] = t
for t in master_v2.get('orphan_tables', []):
    tid = t.get('pum_table_id')
    if tid:
        tables_v2[tid] = t

print("=== 1. 밑수 60건 추적 ===")
basis_v1 = audit_v1['item3_basis']
resolved_basis = []
remaining_basis = []
changed_basis = []

for item in basis_v1:
    tid = item['table_id']
    t_v2 = tables_v2.get(tid)
    if not t_v2:
        continue
    q2 = t_v2.get('basis_quantity')
    u2 = t_v2.get('basis_unit')
    
    # compare with v1 master state
    old_q = item['master_qty']
    old_u = item['master_unit']
    md_q = item['md_qty']
    md_u = item['md_unit']
    
    # Check if resolved
    # resolved means q2 == md_q and u2 == md_u (or unit alias matched)
    u2_norm = str(u2).replace('㎡','m2').replace('㎥','m3').replace('인/','').replace('당','')
    md_u_norm = str(md_u).replace('㎡','m2').replace('㎥','m3').replace('인/','').replace('당','')
    
    is_resolved = False
    if q2 is not None and u2 is not None:
        if abs(float(q2) - float(md_q)) < 1e-4 and u2_norm == md_u_norm:
            is_resolved = True
            
    if is_resolved:
        resolved_basis.append({
            'table_id': tid,
            'section': item['section'],
            'v1_master': f"{old_q} {old_u}",
            'v2_master': f"{q2} {u2}",
            'md': f"{md_q} {md_u}"
        })
    else:
        if q2 != old_q or u2 != old_u:
            changed_basis.append({
                'table_id': tid,
                'section': item['section'],
                'v1_master': f"{old_q} {old_u}",
                'v2_master': f"{q2} {u2}",
                'md': f"{md_q} {md_u}"
            })
        else:
            remaining_basis.append({
                'table_id': tid,
                'section': item['section'],
                'v1_master': f"{old_q} {old_u}",
                'v2_master': f"{q2} {u2}",
                'md': f"{md_q} {md_u}"
            })

print(f"Total Basis v1 issues: {len(basis_v1)}")
print(f"  - Resolved (일치 해결됨): {len(resolved_basis)}")
print(f"  - Changed but still mismatch (변경되었으나 여전히 불일치): {len(changed_basis)}")
print(f"  - Unchanged (미해결/그대로): {len(remaining_basis)}")

print("\n--- Resolved Basis Sample (해결된 16건) ---")
for r in resolved_basis:
    print(f"  [{r['table_id']}] {r['section']}: {r['v1_master']} -> {r['v2_master']} (MD: {r['md']})")

if changed_basis:
    print("\n--- Changed Basis ---")
    for c in changed_basis:
        print(f"  [{c['table_id']}] {c['section']}: {c['v1_master']} -> {c['v2_master']} (MD: {c['md']})")

print("\n=== 2. [주] 계수 누락 9건 추적 ===")
notes_v1 = audit_v1['item4_notes']
resolved_notes = []
remaining_notes = []

for item in notes_v1:
    tid = item['table_id']
    t_v2 = tables_v2.get(tid)
    if not t_v2:
        continue
    notes_v2 = t_v2.get('notes', [])
    # Check if notes_v2 has content now
    if notes_v2:
        resolved_notes.append({
            'table_id': tid,
            'section': item['section'],
            'notes_count_v2': len(notes_v2),
            'notes_preview': notes_v2[:2]
        })
    else:
        remaining_notes.append({
            'table_id': tid,
            'section': item['section']
        })

print(f"Total [주] v1 missing: {len(notes_v1)}")
print(f"  - Resolved (notes 채워짐): {len(resolved_notes)}")
print(f"  - Remaining (여전히 notes 빈칸): {len(remaining_notes)}")
for r in resolved_notes:
    print(f"  [{r['table_id']}] {r['section']}: {r['notes_count_v2']} notes added -> sample: {r['notes_preview'][0][:60]}")
for rm in remaining_notes:
    print(f"  [{rm['table_id']}] {rm['section']}: STILL EMPTY")

print("\n=== 3. reference 27건 추적 ===")
refs_v1 = audit_v1['item5_references']
changed_refs = []
rem_refs = []
for item in refs_v1:
    tid = item['table_id']
    t_v2 = tables_v2.get(tid)
    if not t_v2:
        continue
    pform_v2 = t_v2.get('pum_form')
    if pform_v2 != 'reference':
        changed_refs.append({'table_id': tid, 'section': item['section'], 'form_v2': pform_v2})
    else:
        rem_refs.append({'table_id': tid, 'section': item['section']})

print(f"Total reference v1 suspects: {len(refs_v1)}")
print(f"  - Changed form: {len(changed_refs)}")
print(f"  - Kept reference: {len(rem_refs)}")