MetonaSqlark/tests/v080-case-expression.test.ts

/**
 * v0.8.0 回归套件 —— B-4 结构化 CASE 表达式
 * ============================================================================
 * 修复前 `parseCaseExpression` 用**正则**在 SQL 文本上切分 WHEN/THEN/ELSE：
 *
 * ```ts
 * /WHEN\s+([\s\S]*?)\s+THEN\s+([\s\S]*?)(?=\s+WHEN\s+|\s+ELSE\s+|\s*$)/gi
 * ```
 *
 * 它不认字符串字面量、不认嵌套结构，于是产生三类错误结果（全部实测确认）：
 *
 * | 输入 | 修复前实测 | 应有 |
 * |---|---|---|
 * | 嵌套 CASE（`THEN CASE ... END ELSE ...`） | `"big' END ELSE 'small"` / `null` | huge/big/small |
 * | 条件引用不存在的列 | 每行静默变成 ELSE 值，**无报错** | COLUMN_NOT_FOUND |
 * | `GROUP BY CASE ... END` | `COLUMN_NOT_FOUND Unknown column "CASE WHEN ..."` | 正常分组 |
 *
 * 第三项尤其反直觉：错误信息说"未知列 CASE WHEN ..."，因为分组把 CASE 原文
 * 当成了列名 —— 而"按条件分组"是 SQL 里最常见的分析写法之一。
 *
 * 新实现（`src/query/expression.ts`）复用 `sql/lexer` 的 token 流做递归下降，
 * 并把条件交给与 WHERE **完全相同**的 `parseWhereCondition`。
 *
 * 顺带修掉的词法层缺陷：`Token.position` 的语义**按 token 类型不一致** ——
 * `readString` 用 `position + 1`（指向引号之内），`readIdentifier`/`readNumber`
 * 用 `position - len`（指向首字符）。任何"按 position 切片"的调用方都会对字符串
 * 多切/少切一个字符（实测：`'big'` 被切成 `"big'"`）。现统一为"token 首字符下标"。
 */
import { describe, it, expect, beforeEach } from '@jest/globals';
import { MetonaSqlark } from '../src/core';
import { parseCaseExpression, evaluateCase, assertCaseColumnsExist } from '../src/query/expression';
import { tokenize } from '../src/sql/lexer';
import { TokenType } from '../src/sql/tokens';
import { rows as rowsOf } from './helpers/assertions';
import type { DatabaseConfig } from '../src/constants';

const ENGINES: Array<[string, DatabaseConfig['mode'], Partial<DatabaseConfig>]> = [
  ['memory', 'memory', {}],
  ['disk', 'disk', {}],
  ['hybrid', 'hybrid', {}],
  ['aria', 'aria', { diskEngine: 'memory' }],
];

// ---------------------------------------------------------------------------
//  词法层：position 必须统一为"token 首字符下标"
// ---------------------------------------------------------------------------

describe('[v0.8.0] B-4 词法层：Token.position 语义统一', () => {
  it('所有 token 的 position 都指向自己的第一个字符', () => {
    const sql = `SELECT 'abc', "col", ident, 123 FROM t`;
    for (const tok of tokenize(sql)) {
      if (tok.type === TokenType.EOF) continue;
      // 从 position 开始重新词法化，必须得到同一个 token —— 这是 position
      // "指向首字符"的可判定判据（字符串/分隔标识符此前会差 1）。
      const relexed = tokenize(sql.slice(tok.position));
      expect(relexed[0].value).toBe(tok.value);
      expect(relexed[0].type).toBe(tok.type);
    }
  });

  it('字符串字面量含引号时切片完整（回归护栏）', () => {
    const sql = `THEN 'big' ELSE`;
    const tokens = tokenize(sql).filter((t) => t.type === TokenType.STRING);
    expect(tokens).toHaveLength(1);
    const start = tokens[0].position;
    // 'big' 共 5 个字符（含两个引号）
    expect(sql.slice(start, start + 5)).toBe("'big'");
  });
});

// ---------------------------------------------------------------------------
//  解析层：结构化结果
// ---------------------------------------------------------------------------

describe('[v0.8.0] B-4 CASE 解析', () => {
  it('基本结构：分支顺序、ELSE、别名', () => {
    const expr = parseCaseExpression("CASE WHEN n > 10 THEN 'big' ELSE 'small' END AS band")!;
    expect(expr.alias).toBe('band');
    expect(expr.elseText).toBe("'small'");
    expect(expr.branches).toHaveLength(1);
    expect(expr.branches[0].conditionText).toBe('n > 10');
    expect(expr.branches[0].resultText).toBe("'big'");
  });

  it('多分支 + 无 ELSE', () => {
    const expr = parseCaseExpression("CASE WHEN a = 1 THEN 'x' WHEN a = 2 THEN 'y' END")!;
    expect(expr.branches).toHaveLength(2);
    expect(expr.elseText).toBeNull();
  });

  it('嵌套 CASE 被完整切分（修复前会被切成残片）', () => {
    const expr = parseCaseExpression(
      "CASE WHEN n > 10 THEN CASE WHEN n > 25 THEN 'huge' ELSE 'big' END ELSE 'small' END",
    )!;
    expect(expr.branches).toHaveLength(1);
    expect(expr.branches[0].resultText)
      .toBe("CASE WHEN n > 25 THEN 'huge' ELSE 'big' END");
    expect(expr.elseText).toBe("'small'");
  });

  it('字符串字面量内的 WHEN/ELSE/THEN 不参与切分', () => {
    for (const word of ['WHEN', 'ELSE', 'THEN', 'END']) {
      const expr = parseCaseExpression(`CASE WHEN s = '${word}' THEN 'hit' ELSE 'miss' END`)!;
      expect(expr.branches).toHaveLength(1);
      expect(expr.branches[0].conditionText).toBe(`s = '${word}'`);
      expect(expr.branches[0].resultText).toBe("'hit'");
      expect(expr.elseText).toBe("'miss'");
    }
  });

  it('条件里的括号不影响切分', () => {
    const expr = parseCaseExpression("CASE WHEN (a > 1 AND b < 2) THEN 'x' ELSE 'y' END")!;
    expect(expr.branches[0].conditionText).toBe('(a > 1 AND b < 2)');
  });

  it('非 CASE 文本返回 null（供调用方分流）', () => {
    expect(parseCaseExpression('n > 10')).toBeNull();
    expect(parseCaseExpression('COUNT(*)')).toBeNull();
  });

  it('结构不完整显式报错', () => {
    expect(() => parseCaseExpression('CASE WHEN a > 1 END')).toThrow(/THEN/);
    expect(() => parseCaseExpression('CASE END')).toThrow(/no WHEN branch/);
  });

  it('条件无法解析时报 PARSE_ERROR（不静默降级）', () => {
    expect(() => parseCaseExpression('CASE WHEN THEN 1 ELSE 2 END')).toThrow();
  });

  it('解析结果被缓存（同文本多次解析返回同一对象）', () => {
    const text = "CASE WHEN z > 1 THEN 'a' ELSE 'b' END";
    expect(parseCaseExpression(text)).toBe(parseCaseExpression(text));
  });
});

// ---------------------------------------------------------------------------
//  求值层：与行数据结合
// ---------------------------------------------------------------------------

describe('[v0.8.0] B-4 CASE 求值', () => {
  it('第一个 TRUE 分支胜出；UNKNOWN 不算命中', () => {
    const expr = parseCaseExpression("CASE WHEN a > 1 THEN 'big' WHEN a > 0 THEN 'small' ELSE 'none' END")!;
    expect(evaluateCase(expr, { a: 5 })).toBe('big');
    expect(evaluateCase(expr, { a: 2 })).toBe('big');
    expect(evaluateCase(expr, { a: 0 })).toBe('none');
    // NULL 参与比较 → UNKNOWN → 落到 ELSE（与 WHERE 只保留 TRUE 一致）
    expect(evaluateCase(expr, { a: null })).toBe('none');
  });

  it('THEN/ELSE 支持字面量、列引用与 NULL', () => {
    expect(evaluateCase(parseCaseExpression('CASE WHEN a > 0 THEN b ELSE NULL END')!, { a: 1, b: 'col-value' }))
      .toBe('col-value');
    expect(evaluateCase(parseCaseExpression('CASE WHEN a > 0 THEN 42 ELSE NULL END')!, { a: 1 })).toBe(42);
    expect(evaluateCase(parseCaseExpression('CASE WHEN a > 0 THEN TRUE ELSE FALSE END')!, { a: 1 })).toBe(true);
    expect(evaluateCase(parseCaseExpression("CASE WHEN a > 0 THEN 'it''s' ELSE 'no' END")!, { a: 1 })).toBe("it's");
  });

  it('嵌套 CASE 逐层求值', () => {
    const expr = parseCaseExpression(
      "CASE WHEN n > 10 THEN CASE WHEN n > 25 THEN 'huge' ELSE 'big' END ELSE 'small' END",
    )!;
    expect(evaluateCase(expr, { n: 30 })).toBe('huge');
    expect(evaluateCase(expr, { n: 20 })).toBe('big');
    expect(evaluateCase(expr, { n: 1 })).toBe('small');
  });

  it('无法识别的结果表达式抛 NOT_SUPPORTED（不把原文当字符串返回）', () => {
    // 修复前：正则残片 `"big' END ELSE 'small"` 会被原样当成字符串返回给用户
    const expr = parseCaseExpression('CASE WHEN a > 0 THEN a + 1 ELSE 0 END')!;
    expect(() => evaluateCase(expr, { a: 1 })).toThrow(/Unsupported expression/);
  });

  it('结果引用不存在的列抛 COLUMN_NOT_FOUND', () => {
    const expr = parseCaseExpression('CASE WHEN a > 0 THEN nope ELSE 0 END')!;
    expect(() => evaluateCase(expr, { a: 1 })).toThrow(/Unknown column "nope"/);
  });
});

// ---------------------------------------------------------------------------
//  列存在性校验
// ---------------------------------------------------------------------------

describe('[v0.8.0] B-4 CASE 列存在性校验', () => {
  const available = new Set(['n', 's', 't.n', 't.s']);

  it('条件与结果里的列都校验', () => {
    expect(() => assertCaseColumnsExist(
      parseCaseExpression('CASE WHEN nope > 1 THEN s ELSE s END')!, available, 'CASE',
    )).toThrow(/Unknown column "nope"/);
    expect(() => assertCaseColumnsExist(
      parseCaseExpression('CASE WHEN n > 1 THEN nope ELSE s END')!, available, 'CASE',
    )).toThrow(/Unknown column "nope"/);
  });

  it('合格引用不报错（含表别名与嵌套）', () => {
    expect(() => assertCaseColumnsExist(
      parseCaseExpression('CASE WHEN t.n > 1 THEN t.s ELSE n END')!, available, 'CASE',
    )).not.toThrow();
    expect(() => assertCaseColumnsExist(
      parseCaseExpression("CASE WHEN n > 1 THEN CASE WHEN s = 'x' THEN 1 ELSE 0 END ELSE 2 END")!,
      available, 'CASE',
    )).not.toThrow();
  });
});

// ---------------------------------------------------------------------------
//  端到端：四个引擎
// ---------------------------------------------------------------------------

describe('[v0.8.0] B-4 CASE 端到端（四引擎一致）', () => {
  describe.each(ENGINES)('%s 引擎', (label, mode, extra) => {
    let db: MetonaSqlark;
    let seq = 0;

    beforeEach(async () => {
      seq += 1;
      db = await MetonaSqlark.create({
        name: `b4-${label}-${seq}-${Math.random().toString(36).slice(2)}`,
        mode,
        ...extra,
      });
      await db.defineTable('t', {
        id: { type: 'string', primaryKey: true },
        n: { type: 'number' },
        s: { type: 'string' },
      });
      await db.query("INSERT INTO t VALUES ('1',1,'a'),('2',20,'WHEN'),('3',30,'ELSE'),('4',NULL,'THEN')");
    });

    afterEach(async () => {
      await db.close();
    });

    it('搜索式 CASE：分支与 ELSE 正确', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query("SELECT id, CASE WHEN n > 10 THEN 'big' ELSE 'small' END AS band FROM t ORDER BY id"),
      );
      expect(rows).toEqual([
        { id: '1', band: 'small' },
        { id: '2', band: 'big' },
        { id: '3', band: 'big' },
        { id: '4', band: 'small' },   // n 为 NULL → UNKNOWN → ELSE
      ]);
    });

    it('嵌套 CASE（修复前返回字符串残片）', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query(
          "SELECT id, CASE WHEN n > 10 THEN CASE WHEN n > 25 THEN 'huge' ELSE 'big' END ELSE 'small' END AS r "
          + 'FROM t ORDER BY id',
        ),
      );
      expect(rows.map((r) => r.r)).toEqual(['small', 'big', 'huge', 'small']);
    });

    it('字符串字面量含关键字（修复前可能错切）', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query("SELECT id, CASE WHEN s = 'WHEN' THEN 'hit' ELSE 'miss' END AS r FROM t ORDER BY id"),
      );
      expect(rows.map((r) => r.r)).toEqual(['miss', 'hit', 'miss', 'miss']);
    });

    it('无 ELSE 时未命中为 NULL', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query("SELECT id, CASE WHEN n > 25 THEN 'only' END AS r FROM t ORDER BY id"),
      );
      expect(rows.map((r) => r.r)).toEqual([null, null, 'only', null]);
    });

    it('聚合 CASE（条件计数）', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query('SELECT SUM(CASE WHEN n > 10 THEN 1 ELSE 0 END) AS c FROM t'),
      );
      expect(rows).toEqual([{ c: 2 }]);
    });

    it('GROUP BY CASE 表达式（修复前抛"未知列 CASE WHEN ..."）', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query(
          "SELECT CASE WHEN n > 10 THEN 'big' ELSE 'small' END AS band, COUNT(*) AS c "
          + 'FROM t GROUP BY band ORDER BY band',
        ),
      );
      expect(rows).toEqual([{ band: 'big', c: 2 }, { band: 'small', c: 2 }]);
    });

    it('WHERE 里的 CASE 条件', async () => {
      const rows = rowsOf<Record<string, unknown>>(
        await db.query('SELECT id FROM t WHERE CASE WHEN n > 10 THEN 1 ELSE 0 END = 1 ORDER BY id'),
      );
      expect(rows.map((r) => r.id)).toEqual(['2', '3']);
    });

    it('未知列：条件里报 COLUMN_NOT_FOUND（修复前静默全为 ELSE 值）', async () => {
      await expect(
        db.query("SELECT CASE WHEN nope > 1 THEN 'x' ELSE 'y' END AS r FROM t"),
      ).rejects.toMatchObject({ code: 'COLUMN_NOT_FOUND' });
      await expect(
        db.query("SELECT CASE WHEN n > 1 THEN nope ELSE 'y' END AS r FROM t"),
      ).rejects.toMatchObject({ code: 'COLUMN_NOT_FOUND' });
    });

    it('CASE 别名列可被 ORDER BY 引用', async () => {
      // 数据：id=1/4 → small（n=1 / NULL），id=2/3 → big（n=20/30）
      // `band DESC` → 'small' > 'big'（字典序），故 small 组在前，组内按 id 升序
      const rows = rowsOf<Record<string, unknown>>(
        await db.query(
          "SELECT id, CASE WHEN n > 10 THEN 'big' ELSE 'small' END AS band FROM t ORDER BY band DESC, id",
        ),
      );
      expect(rows.map((r) => r.id)).toEqual(['1', '4', '2', '3']);
      // 升序则反过来
      const asc = rowsOf<Record<string, unknown>>(
        await db.query(
          "SELECT id, CASE WHEN n > 10 THEN 'big' ELSE 'small' END AS band FROM t ORDER BY band, id",
        ),
      );
      expect(asc.map((r) => r.id)).toEqual(['2', '3', '1', '4']);
    });
  });
});