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))

text_tables = []

for fpath in sorted(all_md_files):
    rel_path = os.path.relpath(fpath, base_dir)
    with open(fpath, 'r', encoding='utf-8') as f:
        lines = f.readlines()
        
    for idx, line in enumerate(lines):
        line_num = idx + 1
        line_str = line.strip()
        # Non-table line containing table keywords
        if not line_str.startswith('|'):
            if ('구 분' in line_str or '구분' in line_str) and ('단위' in line_str or '단 위' in line_str or '수량' in line_str or '수 량' in line_str):
                # Filter out obvious false positives
                if len(line_str) > 30:
                    text_tables.append({
                        "file": rel_path,
                        "line": line_num,
                        "snippet": line_str[:120]
                    })

print(f"Total non-table lines with embedded tables: {len(text_tables)}")
for tt in text_tables:
    print(f"{tt['file']} L{tt['line']}: {tt['snippet']}")

with open('scratch/text_inline_tables.json', 'w', encoding='utf-8') as f:
    json.dump(text_tables, f, ensure_ascii=False, indent=2)