docs: add syntax rework plan

2026-08-06 09:05:09 +08:00
parent a4e17a3c0b
commit 7db0742f50
1 changed files with 109 additions and 0 deletions
@@ -0,0 +1,109 @@
+# Implementation Plan: Syntax Clarity & Parser Robustness
+
+Status: **Draft — not yet applied**
+
+## Goals
+
+1. Eliminate the backtracking heuristics in the value parser by making the value grammar context-free.
+2. Remove redundant/ambiguous syntax forms.
+3. Make union resolution deterministic and structural (not error-message-driven).
+4. Address the `csv-parse` quote conflict.
+5. Keep the `;` separator (forced by CSV constraints — not the source of ambiguity).
+
+All changes are **breaking** to the DSL → bump to `2.0.0`, update README + `csv-loader.md`, add a migration note.
+
+---
+
+## Phase 1 — Mandatory brackets for composite values
+
+**Files:** `src/value-parser.ts`, `src/index.test.ts`, `src/csv-loader/reference-resolver.ts`
+
+- Remove the `allowOmitBrackets` parameter from `parseValue`, `parseTupleValue`, `parseArrayValue`.
+- Delete the `elementIsTupleOrArray` disambiguation block and all `savedPos` restore logic in `parseArrayValue`.
+- In `parseValue` (top-level), drop the `allowOmitBrackets = schema.type === "tuple" || "array"` special case — brackets are always required.
+- Values now must be fully bracketed: `[a; 1]; [b; 2]` (no more `[a; 1]; [b; 2]` without outer brackets).
+
+**Tests to update** (currently assert bracket-optional behavior):
+- `index.test.ts`: "should parse tuple without brackets" [L177], "should parse array without brackets" [L215], "should parse array of tuples without outer brackets" [L249].
+- `reference-resolver.ts` / `index.test.ts` reference tests: "should parse array reference IDs without brackets" [L685].
+
+**Decision point:** Full mandatory brackets, or a *top-level-only* carve-out (brackets optional only when the value is the whole cell and the element is non-composite)? Recommend **full mandatory** for a clean grammar; note the carve-out as a fallback if ergonomics matter more.
+
+---
+
+## Phase 2 — Single array form `Type[]`
+
+**Files:** `src/parser.ts`, `src/index.test.ts`
+
+- Remove the `[Type][]` array syntax from `parseSchemaInternal`. Keep only `Type[]`.
+- This makes `[string]` a **1-tuple** (currently it collapses to an array). Update the tuple branch: `[t]` → `{ type: "tuple", elements: [t] }`; the special case `elements.length === 1 && !elements[0].name` becomes a real tuple.
+- Update `schemaToTypeString` in `src/type-utils.ts` — the `array` case currently special-cases tuple elements; with `[Type][]` gone, arrays are always `elementType[]` (keep the `(union)[]` paren wrapping).
+
+**Tests to update:**
+- `index.test.ts`: "should parse array of tuples" [L239] uses `[string; number][]` → becomes `[string; number][]` (unchanged) but "array of tuples without outer brackets" [L249] changes per Phase 1.
+- Any test using `[number][]`, `[string][]` schema strings (e.g. README examples, `csv-loader.md`).
+
+---
+
+## Phase 3 — Deterministic union resolution
+
+**Files:** `src/csv-loader/reference-resolver.ts`, `src/validator.ts` (via `type-utils.ts`), `src/csv-loader/module-gen.ts`
+
+- **Rule:** In a union, always try **non-reference members before reference members**, regardless of author order. This makes fallback structural instead of error-message-driven.
+- In `parseValueWithReferences` and `resolveNestedReferences`, replace the `/not found|Circular reference|Failed to load/` error-message inspection with a fixed ordering: partition members into `nonRef` / `ref`, try `nonRef` first, then `ref`.
+- In `module-gen.ts` `generateSchemaResolutionCode` union case, apply the same ordering so generated code matches runtime behavior.
+- **Document the rule** in README (union member ordering is currently "first match wins" per AGENTS.md).
+
+**Note:** This changes behavior for `@users[] | string` — `string` would now win for plain strings. Flag as a deliberate semantic change.
+
+---
+
+## Phase 4 — Quote conflict resolution
+
+**Files:** `src/parser.ts`, `src/csv-loader/loader.ts`, README, `csv-loader.md`
+
+**Options (pick one):**
+- **(a) Standardize on single-quoted literals only** — minimal change, but leaves a footgun.
+- **(b) Quote the entire schema cell** so `csv-parse` treats it as one field — requires loader-side handling to strip the outer quotes before parsing.
+- **(c) Drop quotes entirely** — use bare-token literals (e.g. `on` / `off` as identifiers). Most robust, largest change.
+
+**Recommendation:** (b) — it's the only option that fixes the root cause (schema cells containing `"` break `csv-parse`) without redesigning the literal syntax. Requires: in `loader.ts`, detect schema-row cells that are fully quoted and unwrap before `parseSchema`.
+
+---
+
+## Phase 5 — Cleanup & consistency
+
+**Files:** `src/csv-loader/reference-resolver.ts`, `src/csv-loader/loader.ts`, `src/csv-loader/module-gen.ts`
+
+- **Reverse-reference performance:** `resolveReverseReference` does `refTable.filter(...)` per row. Build a `Map<fk, rows[]>` lookup once per referenced table (mirroring what `module-gen.ts` already generates) and reuse it. Cache the lookup alongside the parsed table in `referenceTableCache`.
+- **Remove the `,` stop-character in `parseReferenceValue`** (value parser has a `,` stop char the schema parser doesn't — drift). Make reference ID parsing consistent with the rest of the value grammar.
+- **`int`/`float`/`number` → `number`:** document the collapse in README (type-level collapse is intentional; parse-time distinction remains).
+
+---
+
+## Phase 6 — Docs & migration
+
+**Files:** `README.md`, `csv-loader.md`, `AGENTS.md`
+
+- Update all syntax tables and examples for mandatory brackets + single array form.
+- Add a **Migration section** listing the breaking changes:
+  1. Composite values must be fully bracketed.
+  2. `[Type][]` array form removed — use `Type[]`.
+  3. `[single]` is now a 1-tuple, not an array.
+  4. Union resolution now prefers non-reference members.
+  5. (If Phase 4b) schema cells may be fully quoted.
+- Update `AGENTS.md` gotchas: union ordering rule, quote handling.
+
+---
+
+## Validation
+
+- `npm run typecheck`
+- `npm run test` (update `src/index.test.ts`, `src/csv-loader/*.test.ts` first)
+- Manually verify the integration fixture (`user_rev.csv` / `order_rev.csv`) still resolves.
+
+---
+
+## Suggested execution order
+
+Phases 1 → 2 are tightly coupled (both touch bracket parsing) — do them together. Phase 3 is independent. Phase 4 is independent. Phases 5–6 are cleanup/docs and can go last.