Update wehago matching logic and exclude reports

scripts/analyze_wehago_case_report.py

@@ -0,0 +1,688 @@
+from __future__ import annotations
+
+import csv
+import re
+import sqlite3
+import zipfile
+from collections import Counter, defaultdict
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+from xml.sax.saxutils import escape
+
+import sys
+
+ROOT = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(ROOT))
+
+from runtime_config import DB_PATH as CONFIG_DB_PATH  # noqa: E402
+
+
+YEAR = 2025
+WEHAGO_STATUSES = ("voucher_matched", "voucher_recheck", "voucher_unmatched")
+MATCHED_STATUSES = ("voucher_matched", "voucher_recheck")
+REPORT_DIR = ROOT / "reports"
+DB_PATH = ROOT / "data.db" if (ROOT / "data.db").exists() else CONFIG_DB_PATH
+
+
+def clean(value: Any) -> str:
+    return "" if value is None else str(value).strip()
+
+
+def amount(value: Any) -> float:
+    try:
+        return float(value or 0)
+    except (TypeError, ValueError):
+        return 0.0
+
+
+def pct(n: int, d: int) -> str:
+    return f"{(n / d * 100):.2f}%" if d else "0.00%"
+
+
+def norm_text(value: str) -> str:
+    text = clean(value)
+    text = text.replace("(주)", "").replace("㈜", "").replace("（주）", "")
+    text = text.replace("주식회사", "").replace("유한회사", "").replace("재단법인", "")
+    text = text.replace("사단법인", "").replace("(재)", "").replace("(사)", "")
+    text = re.sub(r"\b외\s*\d+\s*명\b", "", text)
+    text = re.sub(r"[\s()/,_\-.·]+", "", text)
+    return text
+
+
+def account_core(value: str) -> str:
+    text = clean(value)
+    text = text.replace("원가)", "").replace("판관)", "")
+    text = re.sub(r"\([^)]*\)", "", text)
+    text = re.sub(r"[\s/]+", "", text)
+    return text
+
+
+def account_family(value: str) -> str:
+    text = clean(value)
+    core = account_core(text)
+    pairs = [
+        ("복리후생", "welfare"),
+        ("접대", "entertainment"),
+        ("부서비", "department"),
+        ("여비교통", "travel"),
+        ("시내교통", "travel"),
+        ("차량유지", "vehicle"),
+        ("차량렌탈", "vehicle"),
+        ("지급임차료", "rent"),
+        ("임차료", "rent"),
+        ("통신", "communication"),
+        ("수도광열", "utility"),
+        ("전력", "utility"),
+        ("전기요금", "utility"),
+        ("가스수도", "utility"),
+        ("소모품", "supplies"),
+        ("사무용품", "supplies"),
+        ("전산용품", "supplies"),
+        ("교육훈련", "training"),
+        ("행사비용", "event"),
+        ("합사경비", "site_office"),
+        ("감리현장운영비", "site_operation"),
+        ("관리현장운영비", "site_operation"),
+        ("외주비", "outsourcing"),
+        ("기술협력비", "outsourcing"),
+        ("설계외주비", "outsourcing"),
+        ("세금과공과", "tax_dues"),
+        ("수수료", "fee"),
+        ("선급금", "advance"),
+        ("전도금", "advance"),
+        ("보통예금", "cash"),
+        ("외상매입금", "payable"),
+        ("미지급금", "payable"),
+        ("외상매출금", "receivable"),
+        ("미수금", "receivable"),
+        ("용역미수금", "receivable"),
+        ("부가세", "tax"),
+        ("매입세액", "tax"),
+        ("매출세액", "tax"),
+        ("예수", "withholding"),
+        ("용역수입", "revenue_service"),
+        ("설계용역수입", "revenue_service"),
+        ("임대수입", "revenue_rent"),
+        ("이자수익", "revenue_interest"),
+        ("잡이익", "other_income"),
+        ("잡손실", "other_loss"),
+    ]
+    for token, family in pairs:
+        if token in text or token in core:
+            return family
+    return core
+
+
+def is_substantive_reclass(left: str, right: str) -> bool:
+    left_family = account_family(left)
+    right_family = account_family(right)
+    if not left_family or not right_family or left_family == right_family:
+        return False
+    if "tax" in {left_family, right_family}:
+        return False
+    if {left_family, right_family} in (
+        {"cash", "payable"},
+        {"cash", "receivable"},
+        {"payable", "receivable"},
+    ):
+        return False
+    return True
+
+
+def is_tax_account(account: str) -> bool:
+    return any(token in clean(account) for token in ("부가세", "매입세액", "매출세액", "불공제"))
+
+
+def is_clearing_account(account: str) -> bool:
+    return any(
+        token in clean(account)
+        for token in (
+            "보통예금",
+            "외상매입금",
+            "외상매출금",
+            "미지급금",
+            "미수금",
+            "선급금",
+            "전도금",
+            "예수금",
+            "예수",
+            "부가세",
+            "세액",
+        )
+    )
+
+
+def row_side_amount(row: dict[str, Any], prefix: str) -> tuple[str, float]:
+    debit = amount(row[f"{prefix}_debit"])
+    credit = amount(row[f"{prefix}_credit"])
+    if abs(debit) >= abs(credit):
+        return "debit", debit
+    return "credit", credit
+
+
+def has_ledger(row: dict[str, Any]) -> bool:
+    return bool(
+        clean(row["ledger_account_name"])
+        or clean(row["ledger_desc"])
+        or abs(amount(row["ledger_debit"])) >= 0.5
+        or abs(amount(row["ledger_credit"])) >= 0.5
+    )
+
+
+def has_erp(row: dict[str, Any]) -> bool:
+    return bool(
+        clean(row["voucher_account_name"])
+        or clean(row["voucher_desc"])
+        or abs(amount(row["voucher_debit"])) >= 0.5
+        or abs(amount(row["voucher_credit"])) >= 0.5
+    )
+
+
+def latest_signature(conn: sqlite3.Connection) -> str:
+    row = conn.execute(
+        """
+        SELECT signature
+        FROM wehago_compare_query_groups
+        WHERE start_year = ? AND end_year = ? AND signature LIKE 'compare-query-v5|%'
+        GROUP BY signature
+        ORDER BY MAX(updated_at) DESC
+        LIMIT 1
+        """,
+        (YEAR, YEAR),
+    ).fetchone()
+    if not row:
+        raise RuntimeError("latest compare signature not found")
+    return clean(row[0])
+
+
+def group_key(group: dict[str, Any]) -> tuple[str, int]:
+    return (clean(group["status_key"]), int(group["group_index"]))
+
+
+def display_key(group: dict[str, Any]) -> str:
+    return f"{group['fiscal_year']} {clean(group['ledger_date'])} {clean(group['voucher_no'])}".strip()
+
+
+def summarize_group(group: dict[str, Any], rows: list[dict[str, Any]]) -> str:
+    erp = clean(group["draft_no"]) or "ERP 전표 없음"
+    ledger_accounts = clean(group["ledger_accounts"]) or "-"
+    erp_accounts = clean(group["voucher_accounts"]) or "-"
+    ledger_vendors = clean(group["ledger_vendors"]) or "-"
+    erp_vendors = clean(group["voucher_vendors"]) or "-"
+    descs = []
+    for row in rows:
+        for key in ("ledger_desc", "voucher_desc"):
+            text = clean(row.get(key))
+            if text and text not in descs:
+                descs.append(text)
+        if len(descs) >= 2:
+            break
+    desc = " / ".join(descs[:2])
+    return (
+        f"{display_key(group)} | ERP {erp} | "
+        f"더존 계정 {ledger_accounts} / ERP 계정 {erp_accounts} | "
+        f"더존 거래처 {ledger_vendors} / ERP 거래처 {erp_vendors}"
+        + (f" | 적요 {desc}" if desc else "")
+    )
+
+
+def account_reclass(rows: list[dict[str, Any]]) -> bool:
+    for row in rows:
+        if not (has_ledger(row) and has_erp(row)):
+            continue
+        ledger_account = clean(row["ledger_account_name"])
+        erp_account = clean(row["voucher_account_name"])
+        if not ledger_account or not erp_account:
+            continue
+        if is_tax_account(ledger_account) or is_tax_account(erp_account):
+            continue
+        ledger_side, ledger_amount = row_side_amount(row, "ledger")
+        erp_side, erp_amount = row_side_amount(row, "voucher")
+        if ledger_side == erp_side and abs(ledger_amount - erp_amount) < 0.5:
+            if is_substantive_reclass(ledger_account, erp_account):
+                return True
+    return False
+
+
+def broad_account_difference(rows: list[dict[str, Any]]) -> bool:
+    for row in rows:
+        if not (has_ledger(row) and has_erp(row)):
+            continue
+        if clean(row["ledger_account_name"]) and clean(row["voucher_account_name"]):
+            if account_core(row["ledger_account_name"]) != account_core(row["voucher_account_name"]):
+                return True
+    return False
+
+
+def vendor_change(rows: list[dict[str, Any]]) -> bool:
+    for row in rows:
+        if not (has_ledger(row) and has_erp(row)):
+            continue
+        ledger_vendor = clean(row["ledger_vendor"])
+        erp_vendor = clean(row["voucher_vendor"])
+        if not ledger_vendor or not erp_vendor:
+            continue
+        if "국민" in ledger_vendor and "국민" in erp_vendor:
+            continue
+        left = norm_text(ledger_vendor)
+        right = norm_text(erp_vendor)
+        if not left or not right:
+            continue
+        if left in right or right in left:
+            continue
+        common_len = 0
+        for i in range(len(left)):
+            for j in range(i + 3, len(left) + 1):
+                if left[i:j] in right:
+                    common_len = max(common_len, j - i)
+        if common_len >= 4:
+            continue
+        if left != right:
+            ledger_side, ledger_amount = row_side_amount(row, "ledger")
+            erp_side, erp_amount = row_side_amount(row, "voucher")
+            if ledger_side == erp_side and abs(ledger_amount - erp_amount) < 0.5:
+                return True
+    return False
+
+
+def amount_display_difference(rows: list[dict[str, Any]]) -> bool:
+    ledger_only = [r for r in rows if has_ledger(r) and not has_erp(r) and not is_tax_account(r["ledger_account_name"])]
+    erp_only = [r for r in rows if has_erp(r) and not has_ledger(r) and not is_tax_account(r["voucher_account_name"])]
+    if not ledger_only or not erp_only:
+        return False
+    tax_like = [
+        r
+        for r in rows
+        if is_tax_account(r["ledger_account_name"]) or is_tax_account(r["voucher_account_name"])
+    ]
+    for lrow in ledger_only:
+        lside, lamt = row_side_amount(lrow, "ledger")
+        if lside != "debit" or lamt <= 0:
+            continue
+        for erow in erp_only:
+            eside, eamt = row_side_amount(erow, "voucher")
+            if eside != "debit" or eamt <= 0 or lamt <= eamt:
+                continue
+            diff = lamt - eamt
+            looks_vat = abs(diff - round(eamt * 0.1)) <= 2 or any(
+                abs(diff - max(amount(t["ledger_debit"]), amount(t["ledger_credit"]), amount(t["voucher_debit"]), amount(t["voucher_credit"]))) <= 2
+                for t in tax_like
+            )
+            if looks_vat:
+                return True
+    return False
+
+
+def advance_unsettled(group: dict[str, Any], rows: list[dict[str, Any]]) -> bool:
+    text = " ".join(
+        [clean(group["ledger_accounts"]), clean(group["ledger_vendors"])]
+        + [clean(r["ledger_desc"]) + " " + clean(r["ledger_account_name"]) + " " + clean(r["ledger_vendor"]) for r in rows]
+    )
+    return bool(
+        ("주재비" in text)
+        or ("전도금" in text and ("운영" in text or "정산" in text or "감리현장" in text))
+        or ("관리현장운영비" in text)
+    )
+
+
+def management_scope_difference(group: dict[str, Any], rows: list[dict[str, Any]]) -> bool:
+    text = " ".join(
+        [clean(group["ledger_accounts"]), clean(group["ledger_vendors"])]
+        + [clean(r["ledger_desc"]) + " " + clean(r["ledger_account_name"]) + " " + clean(r["ledger_vendor"]) for r in rows]
+    )
+    return any(token in text for token in ("RP 매수", "RP 매도", "기타예금", "투자자산", "유가증권", "증권", "CMA", "HMC투자"))
+
+
+def writing_method_difference(group: dict[str, Any], rows: list[dict[str, Any]]) -> bool:
+    if group["status_key"] not in MATCHED_STATUSES:
+        return False
+    has_l_only = any(has_ledger(r) and not has_erp(r) for r in rows)
+    has_e_only = any(has_erp(r) and not has_ledger(r) for r in rows)
+    if not (has_l_only and has_e_only):
+        return False
+    return True
+
+
+def primary_vendor(group: dict[str, Any]) -> str:
+    vendors = clean(group.get("ledger_vendors"))
+    if not vendors:
+        vendors = clean(group.get("voucher_vendors"))
+    return norm_text(vendors.split(",")[0])
+
+
+def group_month(group: dict[str, Any]) -> str:
+    date_text = clean(group.get("ledger_date"))
+    match = re.match(r"(\d{1,2})[-./]", date_text)
+    return match.group(1).zfill(2) if match else ""
+
+
+def build_diverse_samples(
+    case_keys: set[tuple[str, int]],
+    group_by_key: dict[tuple[str, int], dict[str, Any]],
+    rows_by_group: dict[tuple[str, int], list[dict[str, Any]]],
+    excluded: set[tuple[str, int]],
+    limit: int = 5,
+) -> list[str]:
+    candidates = [
+        key
+        for key in sorted(case_keys, key=lambda k: (group_by_key[k]["ledger_date"], group_by_key[k]["voucher_no"], k[1]))
+        if key not in excluded
+    ]
+    selected: list[tuple[str, int]] = []
+    used_months: set[str] = set()
+    used_vendors: set[str] = set()
+
+    def try_pick(require_new_month: bool, require_new_vendor: bool) -> None:
+        if len(selected) >= limit:
+            return
+        for key in candidates:
+            if key in selected:
+                continue
+            group = group_by_key[key]
+            month = group_month(group)
+            vendor = primary_vendor(group)
+            if require_new_month and month and month in used_months:
+                continue
+            if require_new_vendor and vendor and vendor in used_vendors:
+                continue
+            selected.append(key)
+            if month:
+                used_months.add(month)
+            if vendor:
+                used_vendors.add(vendor)
+            if len(selected) >= limit:
+                return
+
+    try_pick(require_new_month=True, require_new_vendor=True)
+    try_pick(require_new_month=False, require_new_vendor=True)
+    try_pick(require_new_month=False, require_new_vendor=False)
+    return [summarize_group(group_by_key[key], rows_by_group[key]) for key in selected[:limit]]
+
+
+def docx_paragraph(text: str, style: str | None = None) -> str:
+    style_xml = f'<w:pPr><w:pStyle w:val="{style}"/></w:pPr>' if style else ""
+    return f"<w:p>{style_xml}<w:r><w:t>{escape(text)}</w:t></w:r></w:p>"
+
+
+def docx_table(headers: list[str], rows: list[list[str]]) -> str:
+    def cell(value: str, bold: bool = False) -> str:
+        run_pr = "<w:rPr><w:b/></w:rPr>" if bold else ""
+        return (
+            "<w:tc><w:tcPr><w:tcW w:w=\"2400\" w:type=\"dxa\"/></w:tcPr>"
+            f"<w:p><w:r>{run_pr}<w:t>{escape(str(value))}</w:t></w:r></w:p></w:tc>"
+        )
+
+    table_rows = ["<w:tr>" + "".join(cell(header, True) for header in headers) + "</w:tr>"]
+    for row in rows:
+        table_rows.append("<w:tr>" + "".join(cell(str(value)) for value in row) + "</w:tr>")
+    return (
+        "<w:tbl><w:tblPr><w:tblW w:w=\"0\" w:type=\"auto\"/>"
+        "<w:tblBorders><w:top w:val=\"single\" w:sz=\"4\" w:space=\"0\" w:color=\"808080\"/>"
+        "<w:left w:val=\"single\" w:sz=\"4\" w:space=\"0\" w:color=\"808080\"/>"
+        "<w:bottom w:val=\"single\" w:sz=\"4\" w:space=\"0\" w:color=\"808080\"/>"
+        "<w:right w:val=\"single\" w:sz=\"4\" w:space=\"0\" w:color=\"808080\"/>"
+        "<w:insideH w:val=\"single\" w:sz=\"4\" w:space=\"0\" w:color=\"808080\"/>"
+        "<w:insideV w:val=\"single\" w:sz=\"4\" w:space=\"0\" w:color=\"808080\"/></w:tblBorders>"
+        "</w:tblPr>"
+        + "".join(table_rows)
+        + "</w:tbl>"
+    )
+
+
+def write_docx(
+    path: Path,
+    title: str,
+    meta_lines: list[str],
+    summary_rows: list[list[str]],
+    detail_rows_for_doc: list[list[str]],
+    sample_rows_by_case: dict[str, list[str]],
+) -> None:
+    body: list[str] = [docx_paragraph(title, "Title")]
+    body.extend(docx_paragraph(line) for line in meta_lines)
+    body.append(docx_paragraph("요약", "Heading1"))
+    body.append(docx_table(["구분", "전표 수", "전체 대비", "비고"], summary_rows))
+    body.append(docx_paragraph("세부 유형별 수치", "Heading1"))
+    body.append(docx_table(["대분류", "세부 유형", "전표 수", "전체 대비", "산정 기준"], detail_rows_for_doc))
+    body.append(docx_paragraph("추가 사례", "Heading1"))
+    for case_name, sample_rows in sample_rows_by_case.items():
+        body.append(docx_paragraph(case_name, "Heading2"))
+        for sample in sample_rows:
+            body.append(docx_paragraph(sample))
+
+    document_xml = (
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
+        '<w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">'
+        "<w:body>"
+        + "".join(body)
+        + '<w:sectPr><w:pgSz w:w="11906" w:h="16838"/>'
+        '<w:pgMar w:top="1440" w:right="1000" w:bottom="1440" w:left="1000" w:header="720" w:footer="720" w:gutter="0"/>'
+        "</w:sectPr></w:body></w:document>"
+    )
+    styles_xml = (
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
+        '<w:styles xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">'
+        '<w:style w:type="paragraph" w:styleId="Title"><w:name w:val="Title"/>'
+        '<w:rPr><w:b/><w:sz w:val="32"/></w:rPr></w:style>'
+        '<w:style w:type="paragraph" w:styleId="Heading1"><w:name w:val="heading 1"/>'
+        '<w:rPr><w:b/><w:sz w:val="26"/></w:rPr></w:style>'
+        '<w:style w:type="paragraph" w:styleId="Heading2"><w:name w:val="heading 2"/>'
+        '<w:rPr><w:b/><w:sz w:val="22"/></w:rPr></w:style>'
+        "</w:styles>"
+    )
+    content_types = (
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
+        '<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types">'
+        '<Default Extension="rels" ContentType="application/vnd.openxmlformats-package.relationships+xml"/>'
+        '<Default Extension="xml" ContentType="application/xml"/>'
+        '<Override PartName="/word/document.xml" ContentType="application/vnd.openxmlformats-officedocument.wordprocessingml.document.main+xml"/>'
+        '<Override PartName="/word/styles.xml" ContentType="application/vnd.openxmlformats-officedocument.wordprocessingml.styles+xml"/>'
+        "</Types>"
+    )
+    rels = (
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
+        '<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">'
+        '<Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument" Target="word/document.xml"/>'
+        "</Relationships>"
+    )
+    doc_rels = (
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
+        '<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">'
+        '<Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles" Target="styles.xml"/>'
+        "</Relationships>"
+    )
+    with zipfile.ZipFile(path, "w", compression=zipfile.ZIP_DEFLATED) as docx:
+        docx.writestr("[Content_Types].xml", content_types)
+        docx.writestr("_rels/.rels", rels)
+        docx.writestr("word/_rels/document.xml.rels", doc_rels)
+        docx.writestr("word/document.xml", document_xml)
+        docx.writestr("word/styles.xml", styles_xml)
+
+
+def main() -> None:
+    conn = sqlite3.connect(DB_PATH)
+    conn.row_factory = sqlite3.Row
+    signature = latest_signature(conn)
+
+    groups = [
+        dict(row)
+        for row in conn.execute(
+            f"""
+            SELECT *
+            FROM wehago_compare_query_groups
+            WHERE start_year = ? AND end_year = ? AND signature = ?
+              AND status_key IN ({','.join('?' for _ in WEHAGO_STATUSES)})
+            ORDER BY status_key, group_index
+            """,
+            (YEAR, YEAR, signature, *WEHAGO_STATUSES),
+        )
+    ]
+    rows_by_group: dict[tuple[str, int], list[dict[str, Any]]] = defaultdict(list)
+    for row in conn.execute(
+        f"""
+        SELECT *
+        FROM wehago_compare_query_rows
+        WHERE start_year = ? AND end_year = ? AND signature = ?
+          AND status_key IN ({','.join('?' for _ in WEHAGO_STATUSES)})
+        ORDER BY status_key, group_index, row_index
+        """,
+        (YEAR, YEAR, signature, *WEHAGO_STATUSES),
+    ):
+        record = dict(row)
+        rows_by_group[(record["status_key"], int(record["group_index"]))].append(record)
+
+    total = len(groups)
+    status_counts = Counter(group["status_key"] for group in groups)
+    cases: dict[str, set[tuple[str, int]]] = defaultdict(set)
+
+    for group in groups:
+        key = group_key(group)
+        rows = rows_by_group[key]
+        if group["status_key"] in MATCHED_STATUSES:
+            if broad_account_difference(rows):
+                cases["광의 계정명 차이"].add(key)
+            if account_reclass(rows):
+                cases["계정 재분류 후보"].add(key)
+            if vendor_change(rows):
+                cases["거래처/대상자 변경 및 보완"].add(key)
+            if amount_display_difference(rows):
+                cases["금액 표시 기준 차이"].add(key)
+            if writing_method_difference(group, rows):
+                cases["전표 작성 방식 차이"].add(key)
+        if group["status_key"] == "voucher_unmatched":
+            if advance_unsettled(group, rows):
+                cases["ERP상 주재비/전도금 미정산"].add(key)
+            if management_scope_difference(group, rows):
+                cases["관리 대상 차이"].add(key)
+
+    info_union = (
+        cases["계정 재분류 후보"]
+        | cases["거래처/대상자 변경 및 보완"]
+        | cases["금액 표시 기준 차이"]
+    )
+    unmatched_union = (
+        set(group_key(g) for g in groups if g["status_key"] == "voucher_unmatched")
+        | cases["전표 작성 방식 차이"]
+    )
+    voucher_unmatched_keys = set(group_key(g) for g in groups if g["status_key"] == "voucher_unmatched")
+    residual_unentered = voucher_unmatched_keys - cases["ERP상 주재비/전도금 미정산"] - cases["관리 대상 차이"]
+    cases["전표 미입력/ERP 직접 대응 없음"].update(residual_unentered)
+
+    group_by_key = {group_key(group): group for group in groups}
+
+    given_examples = {
+        ("voucher_unmatched", 71),
+        ("voucher_matched", 2833),
+        ("voucher_matched", 2016),
+        ("voucher_unmatched", 174),
+        ("voucher_matched", 51),
+        ("voucher_unmatched", 431),
+        ("voucher_unmatched", 1042),
+    }
+
+    def samples(case_name: str, limit: int = 5) -> list[str]:
+        return build_diverse_samples(cases[case_name], group_by_key, rows_by_group, given_examples, limit)
+
+    ts = datetime.now().strftime("%Y%m%d_%H%M%S")
+    md_path = REPORT_DIR / f"wehago_case_distribution_report_{YEAR}_{ts}.md"
+    csv_path = REPORT_DIR / f"wehago_case_distribution_samples_{YEAR}_{ts}.csv"
+    docx_path = REPORT_DIR / f"wehago_case_distribution_report_{YEAR}_{ts}.docx"
+
+    summary_rows = [
+        ["① 매칭 전표의 정보 차이(중복 제거)", f"{len(info_union):,}", pct(len(info_union), total), "계정 재분류 후보, 거래처 변경, 금액 표시 차이 중 하나 이상"],
+        ["② 미매칭/부분미매칭 발생 유형(중복 제거)", f"{len(unmatched_union):,}", pct(len(unmatched_union), total), "더존-only 미매칭 + 전표 작성 방식 차이"],
+        ["더존-only 미매칭", f"{status_counts['voucher_unmatched']:,}", pct(status_counts["voucher_unmatched"], total), "ERP 직접 대응 행이 없는 더존 전표"],
+        ["재검토", f"{status_counts['voucher_recheck']:,}", pct(status_counts["voucher_recheck"], total), "매칭 후보이나 검토 필요"],
+    ]
+    detail_rows = [
+        ("①", "계정 재분류 후보", "매칭/재검토 중 금액·차대 방향은 같고, 계정의 핵심 의미군이 달라진 행 포함. 단순 계정명·보조명 차이는 제외"),
+        ("①", "광의 계정명 차이", "세액·채권채무·예금 등 시스템 계정명 차이를 포함한 계정명 차이. 참고 지표로만 사용"),
+        ("①", "거래처/대상자 변경 및 보완", "매칭/재검토 중 금액·차대 방향은 같고, 거래처 핵심 명칭이 서로 겹치지 않는 행 포함"),
+        ("①", "금액 표시 기준 차이", "ERP 공급가/부가세 분리 금액이 더존 비용 합산 표시로 나타난 후보"),
+        ("②", "ERP상 주재비/전도금 미정산", "더존-only 중 주재비, 전도금, 관리현장운영비 정산 문구/계정 포함"),
+        ("②", "전표 작성 방식 차이", "매칭/재검토 중 한쪽 행만 남는 분리·합산 전표 구조 포함"),
+        ("②", "관리 대상 차이", "더존-only 중 RP, 기타예금, 증권, 투자자산 등 관리 대상 거래"),
+        ("②", "전표 미입력/ERP 직접 대응 없음", "더존-only 중 위 주재비/투자관리 유형으로 분류되지 않은 잔여"),
+    ]
+    detail_rows_for_doc = [
+        [major, name, f"{len(cases[name]):,}", pct(len(cases[name]), total), note]
+        for major, name, note in detail_rows
+    ]
+    sample_rows_by_case = {
+        name: samples(name)
+        for _, name, _ in detail_rows
+        if name != "광의 계정명 차이"
+    }
+
+    lines: list[str] = []
+    lines.append(f"# 더존 전표 기준 케이스별 분포 분석 ({YEAR})")
+    lines.append("")
+    lines.append(f"- 기준 DB: `{DB_PATH}`")
+    lines.append(f"- 기준 projection: `{signature}`")
+    lines.append(f"- 분모: WEHAGO/더존 전표 그룹 {total:,}건 = 매칭 {status_counts['voucher_matched']:,}건 + 재검토 {status_counts['voucher_recheck']:,}건 + 더존-only 미매칭 {status_counts['voucher_unmatched']:,}건")
+    lines.append("- 보완 기준: 계정 재분류는 계정의 핵심 의미군이 달라진 경우만 포함하고, 단순 계정명·보조명 차이는 제외했다.")
+    lines.append("- 보완 기준: 거래처/대상자 변경은 핵심 회사명·조직명 문자열이 서로 겹치지 않는 경우만 포함하고, 약칭·법인격·부서명 차이는 제외했다.")
+    lines.append("- 사례 선정: 2025년 1월부터 순차 추출하되, 월과 거래처가 과도하게 겹치지 않도록 우선 분산 추출했다.")
+    lines.append("- 주의: 세부유형은 한 전표가 둘 이상의 유형에 동시에 해당될 수 있어 단순 합산하면 대분류 중복이 발생한다.")
+    lines.append("")
+    lines.append("## 요약")
+    lines.append("")
+    lines.append("| 구분 | 전표 수 | 전체 대비 | 비고 |")
+    lines.append("|---|---:|---:|---|")
+    for row in summary_rows:
+        lines.append("| " + " | ".join(row) + " |")
+    lines.append("")
+    lines.append("## 세부 유형별 수치")
+    lines.append("")
+    lines.append("| 대분류 | 세부 유형 | 전표 수 | 전체 대비 | 산정 기준 |")
+    lines.append("|---|---|---:|---:|---|")
+    for major, name, note in detail_rows:
+        lines.append(f"| {major} | {name} | {len(cases[name]):,} | {pct(len(cases[name]), total)} | {note} |")
+    lines.append("")
+    lines.append("## 추가 사례")
+    lines.append("")
+    for _, name, _ in detail_rows:
+        if name == "광의 계정명 차이":
+            continue
+        lines.append(f"### {name}")
+        sample_list = sample_rows_by_case[name]
+        if not sample_list:
+            lines.append("- 추가 사례 없음")
+        else:
+            for item in sample_list:
+                lines.append(f"- {item}")
+        lines.append("")
+
+    md_path.write_text("\n".join(lines), encoding="utf-8")
+
+    with csv_path.open("w", newline="", encoding="utf-8-sig") as f:
+        writer = csv.writer(f)
+        writer.writerow(["case_name", "sample_no", "sample"])
+        for _, name, _ in detail_rows:
+            if name == "광의 계정명 차이":
+                continue
+            for idx, item in enumerate(sample_rows_by_case[name], start=1):
+                writer.writerow([name, idx, item])
+
+    meta_lines = [
+        f"기준 DB: {DB_PATH}",
+        f"분모: WEHAGO/더존 전표 그룹 {total:,}건 = 매칭 {status_counts['voucher_matched']:,}건 + 재검토 {status_counts['voucher_recheck']:,}건 + 더존-only 미매칭 {status_counts['voucher_unmatched']:,}건",
+        "보완 기준: 계정 재분류는 계정의 핵심 의미군이 달라진 경우만 포함하고, 거래처 변경은 핵심 회사명·조직명이 서로 겹치지 않는 경우만 포함했다.",
+        "사례 선정: 월과 거래처가 과도하게 겹치지 않도록 우선 분산 추출했다.",
+    ]
+    write_docx(
+        docx_path,
+        f"더존 전표 기준 케이스별 분포 분석 ({YEAR})",
+        meta_lines,
+        summary_rows,
+        detail_rows_for_doc,
+        sample_rows_by_case,
+    )
+
+    print(md_path)
+    print(csv_path)
+    print(docx_path)
+
+
+if __name__ == "__main__":
+    main()