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)

# Ditot / remarks targets
target_sections = [
    ('제4장_궤도공사.md', '4-5-2', 205),
    ('제7장_항만공사.md', '7-1-2', 29),
    ('제7장_항만공사.md', '7-1-4', 86),
    ('제8장_지반조사.md', '8-4-2', 271),
    ('제9장_측량.md', '9-3-1', 703),
    ('제9장_측량.md', '9-5-4', 2701),
    ('제13장_플랜트설비공사.md', '13-4-3', 1544),
    ('제13장_플랜트설비공사.md', '13-5-1', 1703),
    ('제13장_플랜트설비공사.md', '13-5-2', 1734),
    ('제13장_플랜트설비공사.md', '13-7-14', 3012),
    ('제13장_플랜트설비공사.md', '13-7-16', 3040),
    ('제13장_플랜트설비공사.md', '13-7-17', 3053),
    ('제13장_플랜트설비공사.md', '13-7-18', 3067),
    ('제13장_플랜트설비공사.md', '13-7-19', 3079),
    ('제13장_플랜트설비공사.md', '13-7-20', 3096),
    ('제13장_플랜트설비공사.md', '13-7-21', 3110),
    ('제13장_플랜트설비공사.md', '13-10-2', 3513),
    ('제13장_플랜트설비공사.md', '13-10-4', 3586)
]

print("=== Checking Ditot Sections in PDF ===")
for fname, sec, line in target_sections:
    # Find page in PDF
    found_pages = []
    for p_num in range(len(doc)):
        text = doc[p_num].get_text()
        if sec in text:
            found_pages.append(p_num)
    print(f"\n[{fname}] Section {sec} (md line {line}) -> Found in PDF page(s): {[p+1 for p in found_pages]}")
    for p in found_pages[:2]:
        page = doc[p]
        tabs = page.find_tables()
        for t_idx, tab in enumerate(tabs):
            df = tab.extract()
            # print first 3 and last 2 rows
            print(f"  --- Table {t_idx} on Page {p+1} (rows: {len(df)}) ---")
            for r in df[:5]:
                print(f"    {r}")
            if len(df) > 5:
                print("    ...")
                for r in df[-2:]:
                    print(f"    {r}")