import json import sys import re from pathlib import Path from collections import defaultdict, Counter sys.path.insert(0, str(Path(__file__).parent.parent)) from scratch.enrich_tables import tables, get_chapter sys.stdout.reconfigure(encoding='utf-8') # 1. Load data files with open("resources/knowledge/original/행정규칙/임도 품셈 적용기준 (현 산림사업 표준품셈)/첨부/(산림청고시 제2025-82호) 산림사업 표준품셈.md", "r", encoding="utf-8") as f: md_lines = f.readlines() with open('resources/data_work_item_master/work_item_master_2026-01-01.json', 'r', encoding='utf-8') as f: master = json.load(f) with open('resources/data_cost_input_value/pum_forest_2026.json', 'r', encoding='utf-8') as f: pum_data = json.load(f) with open('resources/data_cost_input_value/coef_2026.json', 'r', encoding='utf-8') as f: coef_data = json.load(f) pum_tables = {t['table_id']: t for t in pum_data.get('variables', {}).get('pum', {}).get('tables', [])} master_tables = {} table_wi_map = {} for wi in master['work_items']: for t in wi.get('tables', []): tid = t.get('pum_table_id') if tid: master_tables[tid] = t table_wi_map[tid] = wi orphan_tables = {t.get('pum_table_id'): t for t in master.get('orphan_tables', []) if t.get('pum_table_id')} print("=== STARTING COMPREHENSIVE 5-ITEM AUDIT ===") # --- 1. MD vs Master Presence (표·절·갈래 누락) --- all_tids = [f"F{i+1:04d}" for i in range(476)] audit_item_1 = { 'missing_tables': [], # completely missing from master 'orphan_tables': [], # in orphan_tables (not in work_items) 'wi_tables_count': len(master_tables), 'orphan_tables_count': len(orphan_tables) } for tid in all_tids: if tid not in master_tables and tid not in orphan_tables: t = pum_tables.get(tid, {}) audit_item_1['missing_tables'].append({ 'table_id': tid, 'line': t.get('line'), 'section': t.get('section'), 'reason': '고시 본칙 경과조치 표(Line 31)로 공종 축 제외' }) elif tid in orphan_tables: t = pum_tables.get(tid, {}) ot = orphan_tables[tid] audit_item_1['orphan_tables'].append({ 'table_id': tid, 'line': t.get('line'), 'section': t.get('section'), 'pum_form': ot.get('pum_form'), 'reason': '부록 설계사례(Line 8031~9106) 예시표로 단가산출 공종축 미배정' }) print(f"Item 1: Missing tables: {len(audit_item_1['missing_tables'])}, Orphan tables: {len(audit_item_1['orphan_tables'])}") # --- 2. 값이 다른 칸 (숫자 불일치) --- # Raw cells match 100%. Check [주] numerical values and constants in coef vs MD audit_item_2 = [] # Check L, C values in coef vs F0004 f0004 = pum_tables.get('F0004') soil_L_records = coef_data.get('variables', {}).get('coef_soil_L', {}).get('records', []) soil_C_records = coef_data.get('variables', {}).get('coef_soil_C', {}).get('records', []) soil_diffs = [] if f0004: for idx, r in enumerate(f0004.get('rows', [])): sname = r[0].strip().replace(' ', '') l_str = r[1].strip() c_str = r[2].strip() # Find in coef records matched_l = None if idx < len(soil_L_records): matched_l = soil_L_records[idx] matched_c = None if idx < len(soil_C_records): matched_c = soil_C_records[idx] # Parse min, max from l_str lm = re.match(r'([0-9\.]+)(?:[～~∼\-]([0-9\.]+))?', l_str) if lm and matched_l: orig_min = float(lm.group(1)) orig_max = float(lm.group(2)) if lm.group(2) else orig_min if abs(matched_l.get('min', 0) - orig_min) > 1e-4 or abs(matched_l.get('max', 0) - orig_max) > 1e-4: soil_diffs.append({'row': idx, 'soil': sname, 'md_L': l_str, 'coef_L': matched_l}) audit_item_2.append({'category': 'soil_conversion_L_C', 'diffs': soil_diffs}) print(f"Item 2: Soil conversion discrepancies: {len(soil_diffs)}") # --- 3. 밑수·단위 (md 에는 있는데 마스터에서 비었거나 다른 것) --- basis_pattern = re.compile(r'\[\s*([0-9\.\,]*)\s*([a-zA-Z㎡㎥㏊ha본개공mkm㎏gtonLℓ대인組]+)\s*당\s*\]') basis_pattern2 = re.compile(r'\(단위\s*:\s*([^\)]+)\)') audit_item_3 = [] for t in tables: tid = t['table_id'] mt = master_tables.get(tid) or orphan_tables.get(tid) if not mt: continue m_qty = mt.get('basis_quantity') m_unit = mt.get('basis_unit') pform = mt.get('pum_form') # MD search found_q = None found_u = None found_src = None for pl in reversed(t['pre_context']): m = basis_pattern.search(pl) if m: q_str = m.group(1).replace(',', '').strip() found_q = float(q_str) if q_str else 1.0 found_u = m.group(2).strip() found_src = f"pre_context: '{pl}'" break m2 = basis_pattern2.search(pl) if m2: u_str = m2.group(1).strip() # check if u_str has quantity e.g. "10㎡당" qm = re.match(r'^([0-9\.\,]+)\s*([a-zA-Z㎡㎥㏊ha본개공mkm㎏gtonLℓ대인組]+)', u_str) if qm: found_q = float(qm.group(1).replace(',', '')) found_u = qm.group(2).replace('당', '') else: found_q = 1.0 found_u = u_str.replace('당', '') found_src = f"unit_parenthesis: '{pl}'" break if found_u: # compare if m_qty is None or m_unit is None: audit_item_3.append({ 'table_id': tid, 'chapter': t['chapter'], 'section': t['section'], 'line': t['start_line'], 'issue': 'BASIS_NULL_IN_MASTER', 'md_qty': found_q, 'md_unit': found_u, 'md_src': found_src, 'master_qty': m_qty, 'master_unit': m_unit, 'pum_form': pform }) else: # check numeric or unit mismatch u_norm_md = found_u.replace('㎡','m2').replace('㎥','m3').replace('인/','').replace('당','') u_norm_m = str(m_unit).replace('㎡','m2').replace('㎥','m3').replace('인/','').replace('당','') if abs(float(m_qty) - float(found_q)) > 1e-4 or u_norm_md != u_norm_m: audit_item_3.append({ 'table_id': tid, 'chapter': t['chapter'], 'section': t['section'], 'line': t['start_line'], 'issue': 'BASIS_VALUE_OR_UNIT_DIFF', 'md_qty': found_q, 'md_unit': found_u, 'md_src': found_src, 'master_qty': m_qty, 'master_unit': m_unit, 'pum_form': pform }) print(f"Item 3: Basis issues: {len(audit_item_3)}") # --- 4. [주]의 계수·조건 누락 (표 아래 붙은 주석 계수가 마스터에 안 들어온 것) --- audit_item_4 = [] for t in tables: tid = t['table_id'] notes = t['notes'] if not notes: continue mt = master_tables.get(tid) or orphan_tables.get(tid) if not mt: continue master_note = mt.get('condition_note') # check if notes contain important pricing multipliers (%, 가산, 할증, 손료, 공제, 소요량, 레미콘, 자재 등) cost_keywords = ['%', '가산', '할증', '손료', '손율', '공제', '소요', '포함', '별도', '레미콘', '감한다', '곱한다', '적용한다'] notes_text = " ".join(notes) has_cost_keyword = any(k in notes_text for k in cost_keywords) if has_cost_keyword: # Check if master has condition_note if not master_note: audit_item_4.append({ 'table_id': tid, 'chapter': t['chapter'], 'section': t['section'], 'line': t['start_line'], 'issue': 'CRITICAL_NOTE_MISSING_IN_MASTER', 'notes_count': len(notes), 'notes_preview': notes[:3], 'pum_form': mt.get('pum_form') }) else: # master has note, check if all notes captured pass print(f"Item 4: Critical notes missing in master: {len(audit_item_4)}") # --- 5. 갈래 (variants) 및 pum_form 오분류 건 --- # Tables where pum_form is reference but contain actual requirements/costs audit_item_5 = [] reference_suspects = [] for t in tables: tid = t['table_id'] mt = master_tables.get(tid) if not mt: continue pform = mt.get('pum_form') if pform == 'reference': # Check if headers/rows contain units like 인, %, kg, ton, m, ㎡, ㎥, 대, etc. hdr_str = " ".join(t['lines'][0:2]) # Also check if it belongs to core chapters ch = t['chapter'] if any(k in ch for k in ['제5장', '제9장', '제12장', '제13장']): reference_suspects.append({ 'table_id': tid, 'chapter': ch, 'section': t['section'], 'line': t['start_line'], 'issue': 'REFERENCE_MISCLASSIFIED_SHOULD_BE_REQUIREMENT', 'pum_form': pform, 'basis': f"{mt.get('basis_quantity')} {mt.get('basis_unit')}" }) print(f"Item 5: Suspect reference tables in core chapters: {len(reference_suspects)}") # Save full results into json audit_results = { 'summary': { 'total_tables_md': len(tables), 'item1_missing_tables': len(audit_item_1['missing_tables']), 'item1_orphan_tables': len(audit_item_1['orphan_tables']), 'item2_soil_diffs': len(soil_diffs), 'item3_basis_issues': len(audit_item_3), 'item4_missing_critical_notes': len(audit_item_4), 'item5_misclassified_references_core': len(reference_suspects) }, 'item1_presence': audit_item_1, 'item2_values': audit_item_2, 'item3_basis': audit_item_3, 'item4_notes': audit_item_4, 'item5_references': reference_suspects } with open('scratch/audit_results.json', 'w', encoding='utf-8') as f: json.dump(audit_results, f, ensure_ascii=False, indent=2) print("\nAudit results saved to scratch/audit_results.json.")