import os, sys, re

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

target_files = [
    # 02_토목부문
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제1장_도로포장공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제2장_하천공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제3장_터널공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제4장_궤도공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제5장_강구조공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제6장_관부설및접합공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제7장_항만공사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제8장_지반조사.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/02_토목부문/제9장_측량.md',
    # 05_유지관리부문
    'resources/knowledge/original/원가계산/건설공사_표준품셈/05_유지관리부문/제1장_공통.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/05_유지관리부문/제2장_토목.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/05_유지관리부문/제3장_건축.md',
    'resources/knowledge/original/원가계산/건설공사_표준품셈/05_유지관리부문/제4장_기계설비.md',
    # 04_기계설비부문 제13장
    'resources/knowledge/original/원가계산/건설공사_표준품셈/04_기계설비부문/제13장_플랜트설비공사.md',
]

print("=== Scanning for Overlapped / Duplicated Squashed Rows in Same Table ===")

overlap_cases = []

for fpath in target_files:
    fname = os.path.basename(fpath)
    with open(fpath, 'r', encoding='utf-8') as f:
        lines = f.readlines()
        
    tables = []
    cur_t = []
    cur_start = 0
    for idx, l in enumerate(lines):
        line_str = l.strip()
        if line_str.startswith('|') and line_str.endswith('|'):
            if not cur_t:
                cur_start = idx + 1
            cur_t.append((idx + 1, line_str))
        else:
            if cur_t:
                tables.append((cur_start, cur_t))
                cur_t = []
    if cur_t:
        tables.append((cur_start, cur_t))
        
    for t_start, t_rows in tables:
        # Separate header and data rows
        data_rows = []
        for lno, r_str in t_rows:
            cells = [c.strip() for c in r_str.split('|')[1:-1]]
            if all(re.match(r'^:?-+:?$', c) for c in cells if c):
                continue
            if cells and cells[0] in ["비고", "비 고", "주", "[주]"]:
                continue
            data_rows.append((lno, r_str, cells))
            
        if len(data_rows) < 3:
            continue
            
        # For each row in the table, check if it is a "squashed row"
        # (contains multiple words or numbers in a single cell)
        for r_idx, (lno, r_str, cells) in enumerate(data_rows):
            # Check if any cell has multi-values
            is_squashed = False
            multi_words = []
            for c in cells:
                # E.g. 2 or more distinct words (length >= 2) in the cell
                w_list = [w for w in c.split() if len(w) >= 2 and w not in ["-", "·", "ㆍ"]]
                if len(w_list) >= 2:
                    is_squashed = True
                    multi_words.extend(w_list)
                    
            if not is_squashed:
                continue
                
            # Now, check if these multi_words appear in PRECEDING rows of the SAME table as separate entries!
            matched_preceding_rows = []
            for prev_idx in range(r_idx):
                prev_lno, prev_str, prev_cells = data_rows[prev_idx]
                prev_words = set(w for c in prev_cells for w in c.split() if len(w) >= 2)
                # Does prev_words share words with multi_words?
                overlap = prev_words.intersection(set(multi_words))
                # Filter out generic words like '인부', '보통인부'
                overlap = {w for w in overlap if w not in ["인부", "보통인부", "특별인부", "플랜트", "설비공", "용접공", "배관공"]}
                if overlap:
                    matched_preceding_rows.append((prev_lno, list(overlap), prev_str[:60]))
                    
            if len(matched_preceding_rows) >= 2:
                # This squashed row matches 2 or more preceding distinct rows in the same table!
                overlap_cases.append({
                    'file': fname,
                    'line': lno,
                    'table_start': t_start,
                    'squashed_row': r_str[:90],
                    'matched_preceding': matched_preceding_rows
                })

print(f"Total squashed rows overlapping with preceding rows in same table: {len(overlap_cases)}")
for oc in overlap_cases:
    print(f"\n[{oc['file']}:L{oc['line']}] (Table starting at L{oc['table_start']})")
    print(f"   Squashed row: {oc['squashed_row']}")
    print(f"   Matches {len(oc['matched_preceding'])} preceding rows:")
    for plno, ow, pstr in oc['matched_preceding'][:5]:
        print(f"      L{plno} (overlap: {ow}): {pstr}")