import os, sys, re, json

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

base_dir = "resources/knowledge/original/원가계산/건설공사_표준품셈"

all_md_files = []
for root, dirs, files in os.walk(base_dir):
    for f in files:
        if f.endswith('.md') and not f.startswith('_') and '개정사항' not in f and '2026년_건설공사_표준품셈.md' not in f:
            if '01_공통부문\\제8장' in os.path.join(root, f) or '01_공통부문/제8장' in os.path.join(root, f):
                continue
            all_md_files.append(os.path.join(root, f))

# Also add merged file
all_md_files.append(os.path.join(base_dir, "2026년_건설공사_표준품셈.md"))

print(f"Auditing {len(all_md_files)} files for 2 flaws...")

header_divider_mismatches = []
spec_3row_shift_candidates = []

for fpath in all_md_files:
    rel_path = os.path.relpath(fpath, base_dir)
    with open(fpath, 'r', encoding='utf-8') as f:
        lines = f.readlines()
        
    in_table = False
    header_rows = []
    divider_cols = 0
    divider_line = 0
    
    for idx, line in enumerate(lines):
        line_num = idx + 1
        line_str = line.strip()
        
        if line_str.startswith('|') and line_str.endswith('|'):
            cells = [c.strip() for c in line_str.split('|')[1:-1]]
            
            # Check if this is a divider row
            if all(re.match(r'^:?-+:?$', c) for c in cells if c):
                divider_cols = len(cells)
                divider_line = line_num
                # Check preceding header row
                if header_rows:
                    last_hdr = header_rows[-1]
                    if len(last_hdr['cells']) != divider_cols:
                        header_divider_mismatches.append({
                            "file": rel_path,
                            "line": divider_line,
                            "header_line": last_hdr['line'],
                            "header_cols": len(last_hdr['cells']),
                            "divider_cols": divider_cols,
                            "header_snippet": last_hdr['snippet']
                        })
                in_table = True
            elif not in_table:
                header_rows.append({
                    "line": line_num,
                    "cells": cells,
                    "snippet": line_str[:80]
                })
            else:
                # Table body row
                # Check Flaw 1: 3-row spec shifted (e.g. output ㎾ on a row alone or pushed to next line)
                # Looking for orphaned spec line with ㎾, ton, HP etc. where rest of row is empty or mismatched
                if len(cells) > 2:
                    c0 = cells[0]
                    c1 = cells[1] if len(cells) > 1 else ""
                    # If first cell is just a unit/spec like "출력 ㎾" or "㎾" and following data columns are misaligned
                    if re.match(r'^(출력|규격|용량)?\s*(\(?㎾\)?|\(?ton\)?|\(?HP\)?|\(?PS\)?)$', c0):
                        spec_3row_shift_candidates.append({
                            "file": rel_path,
                            "line": line_num,
                            "snippet": line_str[:100]
                        })
        else:
            in_table = False
            header_rows = []

print(f"\n1. Header vs Divider column mismatches: {len(header_divider_mismatches)}")
for m in header_divider_mismatches:
    print(f"   {m['file']} L{m['line']}: header={m['header_cols']} vs divider={m['divider_cols']} ({m['header_snippet']})")

print(f"\n2. Spec 3-row shifted candidates: {len(spec_3row_shift_candidates)}")
for s in spec_3row_shift_candidates:
    print(f"   {s['file']} L{s['line']}: {s['snippet']}")

with open('scratch/laptop_sub_flaws_audit.json', 'w', encoding='utf-8') as f:
    json.dump({
        "header_divider_mismatches": header_divider_mismatches,
        "spec_3row_shift_candidates": spec_3row_shift_candidates
    }, f, ensure_ascii=False, indent=2)