# -*- coding: utf-8 -*-
import json
import re
from pathlib import Path
from collections import defaultdict

SQL_PATH = Path(r"C:\dev\LLTQ\ETL\feiqiu-ETL\etl_billiards\database\schema_ODS_doc.sql")
DOC_DIR = Path(r"C:\dev\LLTQ\export\test-json-doc")

TABLE_CN = {
    "member_profiles": "会员档案/会员账户信息",
    "member_balance_changes": "会员余额变更流水",
    "member_stored_value_cards": "会员储值/卡券账户列表",
    "recharge_settlements": "充值结算记录",
    "settlement_records": "结账/结算记录",
    "assistant_cancellation_records": "助教作废/取消记录",
    "assistant_accounts_master": "助教档案主数据",
    "assistant_service_records": "助教服务流水",
    "site_tables_master": "门店桌台主数据",
    "table_fee_discount_records": "台费折扣记录",
    "table_fee_transactions": "台费流水",
    "goods_stock_movements": "商品库存变动流水",
    "stock_goods_category_tree": "商品分类树",
    "goods_stock_summary": "商品库存汇总",
    "payment_transactions": "支付流水",
    "refund_transactions": "退款流水",
    "platform_coupon_redemption_records": "平台券核销/使用记录",
    "tenant_goods_master": "租户商品主数据",
    "group_buy_packages": "团购套餐主数据",
    "group_buy_redemption_records": "团购核销记录",
    "settlement_ticket_details": "结算小票明细",
    "store_goods_master": "门店商品主数据",
    "store_goods_sales_records": "门店商品销售流水",
}

COMMON_FIELD_PURPOSE = {
    "tenant_id": "租户/品牌 ID，用于商户维度过滤与关联。",
    "site_id": "门店 ID，用于门店维度过滤与关联。",
    "register_site_id": "会员注册门店 ID，用于归属门店维度关联。",
    "site_name": "门店名称快照，用于直接展示。",
    "id": "本表主键 ID，用于唯一标识一条记录。",
    "system_member_id": "系统级会员 ID（跨门店/跨卡种统一到‘人’的维度）。",
    "order_trade_no": "订单交易号，用于串联同一订单下的各类消费明细。",
    "order_settle_id": "订单结算/结账主键，用于关联结算记录与小票明细。",
    "order_pay_id": "关联支付流水的主键 ID，用于追溯支付明细。",
    "point": "积分余额，用于记录会员积分取值。",
    "growth_value": "成长值/成长积分，用于会员成长与等级评估。",
    "referrer_member_id": "推荐人会员 ID，用于记录会员推荐/拉新关系。",
    "create_time": "记录创建时间（业务侧产生时间）。",
    "status": "状态枚举，用于标识记录当前业务状态。",
    "user_status": "用户状态枚举，用于标识会员账户/用户可用状态。",
    "is_delete": "逻辑删除标记（0=否，1=是）。",
    "payload": "完整原始 JSON 记录快照，用于回溯与二次解析。",
    "source_file": "ETL 元数据：原始导出文件名，用于数据追溯。",
    "source_endpoint": "ETL 元数据：采集来源（接口/文件路径），用于数据追溯。",
    "fetched_at": "ETL 元数据：采集/入库时间戳，用于口径对齐与增量处理。",
}

ETL_META_FIELDS = {"source_file", "source_endpoint", "fetched_at"}


def _first_sentence(text: str, max_len: int = 120) -> str:
    s = re.sub(r"\s+", " ", (text or "").strip())
    if not s:
        return ""
    parts = re.split(r"[。；;]\s*", s)
    s = parts[0].strip() if parts else s
    if len(s) > max_len:
        s = s[: max_len - 1] + "…"
    return s


def _escape_sql(s: str) -> str:
    return (s or "").replace("'", "''")

def normalize_key(s: str) -> str:
    return re.sub(r"[_\-\s]", "", (s or "").lower())


def snake_to_lower_camel(s: str) -> str:
    parts = re.split(r"[_\-\s]+", s)
    if not parts:
        return s
    first = parts[0].lower()
    rest = "".join(p[:1].upper() + p[1:] for p in parts[1:] if p)
    return first + rest


def snake_to_upper_camel(s: str) -> str:
    parts = re.split(r"[_\-\s]+", s)
    return "".join(p[:1].upper() + p[1:] for p in parts if p)


def find_key_in_record(record: dict, token: str) -> str | None:
    if not isinstance(record, dict) or not token:
        return None
    if token in record:
        return token
    norm_to_key = {normalize_key(k): k for k in record.keys()}
    candidates = [
        token,
        token.lower(),
        token.upper(),
        snake_to_lower_camel(token),
        snake_to_upper_camel(token),
    ]
    for c in candidates:
        nk = normalize_key(c)
        if nk in norm_to_key:
            return norm_to_key[nk]
    return None


def _infer_purpose(_table: str, col: str) -> str:
    if col in COMMON_FIELD_PURPOSE:
        return COMMON_FIELD_PURPOSE[col]

    lower = col.lower()
    if lower.endswith("_id"):
        return "标识类 ID 字段，用于关联/定位相关实体。"
    if lower.endswith("_time") or lower.endswith("time"):
        return "时间字段，用于记录业务时间点/发生时间。"
    if any(k in lower for k in ["amount", "money", "fee", "price", "deduct", "cost"]):
        return "金额字段，用于计费/结算/分摊等金额计算。"
    if any(k in lower for k in ["count", "num", "number", "seconds", "qty"]):
        return "数量/时长字段，用于统计与计量。"
    if lower.endswith("_name") or lower.endswith("name"):
        return "名称字段，用于展示与辅助识别。"
    if lower.endswith("_code") or lower.endswith("code"):
        return "编码/枚举字段，用于表示类型、等级或业务枚举。"
    if lower.startswith("is_") or lower.startswith("able_") or lower.startswith("can_"):
        return "布尔/开关字段，用于表示权限、可用性或状态开关。"

    return "来自 JSON 导出的原始字段，用于保留业务取值。"


def _format_example(value, max_len: int = 120) -> str:
    if value is None:
        return "NULL"
    if isinstance(value, bool):
        return "true" if value else "false"
    if isinstance(value, (int, float)):
        return str(value)
    if isinstance(value, str):
        s = value.strip()
        if len(s) > max_len:
            s = s[: max_len - 1] + "…"
        return s
    if isinstance(value, list):
        if not value:
            return "[]"
        sample = value[0]
        rendered = json.dumps(sample, ensure_ascii=False)
        if len(value) > 1:
            rendered = f"[{rendered}, …] (len={len(value)})"
        else:
            rendered = f"[{rendered}]"
        if len(rendered) > max_len:
            rendered = rendered[: max_len - 1] + "…"
        return rendered
    if isinstance(value, dict):
        keys = list(value)[:6]
        mini = {k: value.get(k) for k in keys}
        rendered = json.dumps(mini, ensure_ascii=False)
        if len(value) > len(keys):
            rendered = rendered[:-1] + ", …}"
        if len(rendered) > max_len:
            rendered = rendered[: max_len - 1] + "…"
        return rendered
    rendered = str(value)
    if len(rendered) > max_len:
        rendered = rendered[: max_len - 1] + "…"
    return rendered


def _find_best_record_list(data, columns):
    cols = set(columns)
    best = None
    best_score = -1

    queue = [(data, 0)]
    visited = 0
    while queue and visited < 20000:
        node, depth = queue.pop(0)
        visited += 1
        if depth > 8:
            continue
        if isinstance(node, list):
            if node and all(isinstance(x, dict) for x in node[:3]):
                scores = []
                for x in node[:5]:
                    scores.append(len(set(x.keys()) & cols))
                score = sum(scores) / max(1, len(scores))
                if score > best_score:
                    best_score = score
                    best = node
                for x in node[:10]:
                    queue.append((x, depth + 1))
            else:
                for x in node[:50]:
                    queue.append((x, depth + 1))
        elif isinstance(node, dict):
            for v in list(node.values())[:80]:
                queue.append((v, depth + 1))

    return best


def _find_best_record_list_and_node(data, columns):
    cols = set(columns)
    best = None
    best_score = -1
    best_path = []

    queue = [(data, 0, [])]
    visited = 0
    while queue and visited < 25000:
        node, depth, path = queue.pop(0)
        visited += 1
        if depth > 10:
            continue

        if isinstance(node, list):
            if node and all(isinstance(x, dict) for x in node[:3]):
                scores = []
                for x in node[:5]:
                    scores.append(len(set(x.keys()) & cols))
                score = sum(scores) / max(1, len(scores))
                if score > best_score:
                    best_score = score
                    best = node
                    best_path = path
                for x in node[:10]:
                    queue.append((x, depth + 1, path))
            else:
                for x in node[:80]:
                    queue.append((x, depth + 1, path))
        elif isinstance(node, dict):
            for k, v in list(node.items())[:120]:
                queue.append((v, depth + 1, path + [str(k)]))

    node_str = ".".join(best_path) if best_path else "$"
    return best or [], node_str


def _choose_examples(records, columns):
    examples = {}
    if not records:
        return examples
    for col in columns:
        val = None
        for r in records[:120]:
            if isinstance(r, dict) and col in r and r[col] not in (None, ""):
                val = r[col]
                break
        examples[col] = val
    return examples


def _extract_header_fields(line: str, columns_set):
    s = line.strip()
    if not s:
        return []

    # 支持 1. id / 1.1 siteProfile / 8. tenant_id
    m = re.match(r"^\d+(?:\.\d+)*[\.)]?\s+(.+)$", s)
    if m:
        s = m.group(1).strip()

    parts = re.split(r"\s*[/、,，]\s*", s)
    fields = [p.strip() for p in parts if p.strip() in columns_set]

    if not fields and s in columns_set:
        fields = [s]

    if fields and len(line) <= 120:
        return fields
    return []


def _parse_field_purpose_from_block(block_lines):
    lines = [l.rstrip() for l in block_lines]

    def pick_after_label(labels):
        for i, l in enumerate(lines):
            for lab in labels:
                if lab in l:
                    after = l.split(lab, 1)[1].strip()
                    if after:
                        return after
                    buf = []
                    j = i + 1
                    while j < len(lines) and not lines[j].strip():
                        j += 1
                    for k in range(j, len(lines)):
                        if not lines[k].strip():
                            break
                        if re.match(r"^[\w\u4e00-\u9fff]+[:：]", lines[k].strip()):
                            break
                        buf.append(lines[k].strip())
                    if buf:
                        return " ".join(buf)
        return ""

    # 兼容「含义（结合其它文件）：」「含义（推测）：」等变体
    picked = pick_after_label(["含义：", "含义:"])
    if not picked:
        for i, l in enumerate(lines):
            s = l.strip()
            m = re.match(r"^含义.*[:：]\s*(.*)$", s)
            if m:
                after = m.group(1).strip()
                if after:
                    picked = after
                else:
                    buf = []
                    j = i + 1
                    while j < len(lines) and not lines[j].strip():
                        j += 1
                    for k in range(j, len(lines)):
                        if not lines[k].strip():
                            break
                        if re.match(r"^[\w\u4e00-\u9fff]+[:：]", lines[k].strip()):
                            break
                        buf.append(lines[k].strip())
                    if buf:
                        picked = " ".join(buf)
                break

    if not picked:
        picked = pick_after_label(["作用：", "作用:"])
    if not picked:
        for i, l in enumerate(lines):
            s = l.strip()
            m = re.match(r"^作用.*[:：]\s*(.*)$", s)
            if m:
                after = m.group(1).strip()
                if after:
                    picked = after
                break

    if not picked:
        # 兜底：尽量避开“类型：/唯一值个数：”这类描述
        for l in lines:
            s = l.strip()
            if not s:
                continue
            if any(
                s.startswith(prefix)
                for prefix in [
                    "类型：",
                    "非空：",
                    "唯一值",
                    "观测",
                    "特征",
                    "统计",
                    "分布",
                    "说明：",
                    "关联：",
                    "结构关系",
                    "和其它表",
                    "重复记录",
                    "全部为",
                ]
            ):
                continue
            picked = s
            break

    return _first_sentence(picked, 160)


def _is_poor_purpose(purpose: str) -> bool:
    s = (purpose or "").strip()
    if not s:
        return True
    if s.endswith("：") or s.endswith(":"):
        return True
    if s.startswith("全部为"):
        return True
    if s.startswith("含义") and ("：" in s or ":" in s) and len(s) <= 12:
        return True
    return False


def parse_analysis(analysis_text: str, columns):
    columns_set = set(columns)
    blocks = defaultdict(list)

    current_fields = []
    buf = []

    for raw in analysis_text.splitlines():
        fields = _extract_header_fields(raw, columns_set)
        if fields:
            if current_fields and buf:
                for f in current_fields:
                    blocks[f].extend(buf)
            current_fields = fields
            buf = []
        else:
            if current_fields:
                buf.append(raw)

    if current_fields and buf:
        for f in current_fields:
            blocks[f].extend(buf)

    purposes = {}
    for col in columns:
        if col in blocks and blocks[col]:
            p = _parse_field_purpose_from_block(blocks[col])
            if p:
                purposes[col] = p
    return purposes


def parse_columns_from_ddl(create_sql: str):
    start = create_sql.find("(")
    end = create_sql.rfind(")")
    body = create_sql[start + 1 : end]

    cols = []
    for line in body.splitlines():
        s = line.strip().rstrip(",")
        if not s:
            continue
        if s.startswith(")"):
            continue
        if s.upper().startswith("CONSTRAINT "):
            continue
        m = re.match(r"^([A-Za-z_][A-Za-z0-9_]*)\s+", s)
        if not m:
            continue
        name = m.group(1)
        if name.upper() in {"PRIMARY", "UNIQUE", "FOREIGN", "CHECK"}:
            continue
        cols.append(name)
    return cols


def build_comment_block(table: str, columns, analysis_text: str, records):
    # records_node: 由外部确定，避免这里重复遍历 JSON
    records, records_node = records
    purposes = parse_analysis(analysis_text, columns)
    examples = _choose_examples(records, columns)

    table_cn = TABLE_CN.get(table, table)

    table_comment = (
        f"ODS 原始明细表：{table_cn}。"
        f"来源：C:/dev/LLTQ/export/test-json-doc/{table}.json；分析：{table}-Analysis.md。"
        f"字段以导出原样为主；ETL 补充 source_file/source_endpoint/fetched_at，并保留 payload 为原始记录快照。"
    )

    lines = []
    lines.append(f"COMMENT ON TABLE billiards_ods.{table} IS '{_escape_sql(table_comment)}';")

    for col in columns:
        json_file = f"{table}.json"
        if col in ETL_META_FIELDS:
            json_field = f"{json_file} - ETL元数据 - 无"
        elif col == "payload":
            json_field = f"{json_file} - {records_node} - $"
        else:
            actual = None
            for r in records[:50]:
                if isinstance(r, dict):
                    actual = find_key_in_record(r, col)
                    if actual:
                        break
            field_name = actual or col
            json_field = f"{json_file} - {records_node} - {field_name}"

        purpose = purposes.get(col) or _infer_purpose(table, col)
        purpose = _first_sentence(purpose, 140) or _infer_purpose(table, col)
        if _is_poor_purpose(purpose):
            purpose = COMMON_FIELD_PURPOSE.get(col) or _infer_purpose(table, col)

        if col in ETL_META_FIELDS:
            if col == "source_file":
                ex = f"{table}.json"
            elif col == "source_endpoint":
                ex = f"C:/dev/LLTQ/export/test-json-doc/{table}.json"
            else:
                ex = "2025-11-10T00:00:00+08:00"
        elif col == "payload":
            ex = "{...}"
        else:
            ex = _format_example(examples.get(col))

        func = purpose
        if "用于" not in func:
            func = "用于" + func.rstrip("。")

        # ODS来源：表名-字段名（ODS自身字段）；ETL补充字段标记
        if col in ETL_META_FIELDS:
            ods_src = f"{table} - {col}（ETL补充）"
        else:
            ods_src = f"{table} - {col}"

        comment = (
            f"【说明】{purpose}。"
            f"     【示例】{ex}（{func}）。"
            f"     【ODS来源】{ods_src}。"
            f"     【JSON字段】{json_field}。"
        )
        lines.append(
            f"COMMENT ON COLUMN billiards_ods.{table}.{col} IS '{_escape_sql(comment)}';"
        )

    return "\n".join(lines)


text = SQL_PATH.read_text(encoding="utf-8")
newline = "\r\n" if "\r\n" in text else "\n"

kept = []
for raw_line in text.splitlines(True):
    stripped = raw_line.lstrip()
    if stripped.startswith("--"):
        continue
    if re.match(r"^\s*COMMENT ON\s+(TABLE|COLUMN)\s+", raw_line):
        continue
    kept.append(raw_line)

clean = "".join(kept)

create_re = re.compile(
    r"(CREATE TABLE IF NOT EXISTS\s+billiards_ods\.(?P<table>[A-Za-z0-9_]+)\s*\([\s\S]*?\)\s*;)" ,
    re.M,
)

out_parts = []
last = 0
count = 0
for m in create_re.finditer(clean):
    out_parts.append(clean[last : m.end()])
    table = m.group("table")
    create_sql = m.group(1)

    cols = parse_columns_from_ddl(create_sql)
    analysis_text = (DOC_DIR / f"{table}-Analysis.md").read_text(encoding="utf-8")
    data = json.loads((DOC_DIR / f"{table}.json").read_text(encoding="utf-8"))
    record_list, record_node = _find_best_record_list_and_node(data, cols)

    out_parts.append(newline + newline + build_comment_block(table, cols, analysis_text, (record_list, record_node)) + newline + newline)
    last = m.end()
    count += 1

out_parts.append(clean[last:])

result = "".join(out_parts)
result = re.sub(r"(?:\r?\n){4,}", newline * 3, result)

backup = SQL_PATH.with_suffix(SQL_PATH.suffix + ".rewrite2.bak")
backup.write_text(text, encoding="utf-8")
SQL_PATH.write_text(result, encoding="utf-8")

print(f"Rewrote comments for {count} tables. Backup: {backup}")