import glob, re, os, sys, fitz

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

pdf_path = 'resources/knowledge/original/원가계산/건설공사_표준품셈/2026년_건설공사_표준품셈.pdf'
doc = fitz.open(pdf_path)

target_files = []
target_files.extend(glob.glob('resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/*.md'))
target_files.extend(glob.glob('resources/knowledge/original/원가계산/건설공사_표준품셈/05_유지관리부문/*.md'))
target_files.append('resources/knowledge/original/원가계산/건설공사_표준품셈/04_기계설비부문/제13장_플랜트설비공사.md')

suspicious_cases = []

# Scan markdown tables for suspicious remarks or 조건 cells
for fpath in target_files:
    fname = os.path.basename(fpath)
    with open(fpath, 'r', encoding='utf-8') as f:
        lines = f.readlines()
        
    current_sec = 'UNKNOWN'
    in_table = False
    headers = []
    table_rows = []
    header_idx = 0
    
    for idx, line in enumerate(lines):
        line_s = line.strip()
        sec_m = re.search(r'(\d+-\d+(?:-\d+)?(?:\s+[^\n(]+)?)', line_s)
        if sec_m and not line_s.startswith('|'):
            current_sec = sec_m.group(1).strip()
            
        if line_s.startswith('|') and not line_s.startswith('|---'):
            cols = [c.strip() for c in line_s.split('|')[1:-1]]
            if not in_table:
                # Header row
                if any('비고' in c or '비 고' in c or '구분' in c or '직종' in c or '규격' in c for c in cols):
                    headers = cols
                    in_table = True
                    header_idx = idx + 1
                    table_rows = []
                    continue
            if in_table:
                table_rows.append((idx + 1, cols))
        else:
            if in_table and (line_s == '' or not line_s.startswith('|')):
                # Check table_rows for suspicious patterns
                # Find remarks col index if any
                remarks_idx = -1
                for ci, h in enumerate(headers):
                    if '비고' in h or '비 고' in h:
                        remarks_idx = ci
                        break
                        
                for ri, (ln, rcols) in enumerate(table_rows):
                    # Check remarks cell
                    if remarks_idx != -1 and remarks_idx < len(rcols):
                        rval = rcols[remarks_idx]
                        # Patterns: starts with %, contains lonely number/percentage, or short fragment
                        if re.match(r'^[\d\.]+\s*[%％]', rval) or rval in ['45%', '50%', '30%'] or re.match(r'^[0-9]+[가-힣a-zA-Z]*$', rval) and len(rval) <= 4:
                            suspicious_cases.append({
                                'type': 'Suspicious Remarks Value',
                                'file': fname,
                                'sec': current_sec,
                                'line': ln,
                                'row': rcols,
                                'remarks': rval,
                                'reason': f'Remarks starts with percentage/number: {rval}'
                            })
                    # Check if rcols has ditto mark following an empty cell or strange transition
                    for ci, cval in enumerate(rcols):
                        if cval in ['〃', '"', '″'] and ri == 0:
                            suspicious_cases.append({
                                'type': 'Ditto on First Row',
                                'file': fname,
                                'sec': current_sec,
                                'line': ln,
                                'row': rcols,
                                'col_idx': ci,
                                'reason': f'Ditto mark on first row of table: {cval}'
                            })
                            
                in_table = False
                headers = []
                table_rows = []

print(f"Total suspicious cases found in MD: {len(suspicious_cases)}")
for sc in suspicious_cases:
    print(f"[{sc['file']}] {sc['sec']} (L{sc['line']}): {sc['reason']}")
    print(f"   Row: {sc['row']}")