Files
WRNexusJS/packages/compiler/src/parser.ts
T

799 lines
24 KiB
TypeScript

/**
* Recursive-descent parser for `.wrn`, producing a small AST.
*
* Grammar (subset of the vision, but real):
*
* page <Name> {
* types { <TypeScript declarations> }
* props { <ident>: <type> [= <expr>] } // no default means required
* state <ident>: <type> = <expr> // type annotation is optional
* view { <html> } // plain HTML (see parseHtmlView)
* seo { title = "Home" description = "..." }
* ssr { api <name> <METHOD> <path> { <render js> } functions { <raw js> } }
* client { api <name> <METHOD> <path> { <render js> } functions { <raw js> } }
* style { <raw css> } // zero or more, inlined with the page
* functions { <raw js> } // zero or more, shared helpers
* api <METHOD> <path> { <raw js> } // zero or more
* realtime <name> { on <evt>(<args>) { <raw js> } * } // zero or more
* }
*
* The `view` block is written as ordinary HTML — nothing new to learn. Text may
* contain `{expr}` interpolation, attributes may be hyphenated (`data-*`), and
* `@event="..."` declares a client event binding. See `parseHtmlView`.
*/
import { Lexer, LexError, type Token } from "./tokenizer.ts";
import { validateTypedInitializer } from "./types.ts";
export interface StateDecl {
name: string;
/** Explicit TypeScript-style type annotation, when supplied. */
valueType?: string;
/** Raw JS initializer expression, e.g. `0` or `'x'`. */
expr: string;
}
export interface Attr {
name: string;
value: string;
/** True for `@event` bindings (vs. plain HTML attributes). */
event: boolean;
/** True for a valueless boolean attribute, e.g. `<button disabled>`. */
boolean?: boolean;
}
/**
* HTML void elements: they have no children and no closing tag.
* @see https://html.spec.whatwg.org/multipage/syntax.html#void-elements
*/
export const VOID_ELEMENTS = new Set([
"area",
"base",
"br",
"col",
"embed",
"hr",
"img",
"input",
"link",
"meta",
"param",
"source",
"track",
"wbr",
]);
export type ViewNode =
| { type: "text"; value: string }
| { type: "element"; tag: string; attrs: Attr[]; children: ViewNode[] }
/**
* A server-side loop: `{#each <list> as <item>[, <index>]} …body… {:empty} …empty… {/each}`.
* `list` is a JS expression (evaluated on the server, may reference an `ssr` data
* binding). The `body` is rendered once per item with `{item.field}` interpolation;
* `empty` renders when the list is empty. See codegen `compileEach`.
*/
| {
type: "each";
list: string;
item: string;
index?: string;
body: ViewNode[];
empty: ViewNode[];
}
/**
* A server-side conditional: `{#if <expr>} … {:else if <expr>} … {:else} … {/if}`.
* Rendered branches are chosen on the server. Each branch's `cond` is a JS expression
* (`null` for the final `{:else}`); the first truthy branch renders. See `compileIfExpr`.
*/
| { type: "if"; branches: { cond: string | null; body: ViewNode[] }[] };
export interface ApiBlock {
method: string;
path: string;
body: string;
}
export type SeoBlock = Record<string, string>;
export type DataMode = "ssr" | "client";
export interface DataApiBlock {
mode: DataMode;
name: string;
method: string;
path: string;
body: string;
}
export interface ModeFunctionsBlock {
mode: DataMode;
body: string;
}
export type LifecycleHookName = "mount" | "update" | "unmount";
export interface LifecycleBlock {
mount?: string;
update?: string;
unmount?: string;
}
export interface WatchBlock {
state: string;
body: string;
}
export interface RealtimeHandler {
event: string;
args: string[];
body: string;
}
export interface RealtimeBlock {
name: string;
handlers: RealtimeHandler[];
}
export interface PropDecl {
name: string;
/** Explicit TypeScript-style type annotation, when supplied. */
valueType?: string;
/** Props without a default are required. */
required: boolean;
/** Raw JS default expression, e.g. `0` or `'Count'`. Its type drives coercion. */
default: string;
}
export interface PageAst {
type: "page";
/**
* `page` is a route, `component` is a reusable fragment,
* and `layout` is a reusable page wrapper.
*/
kind: "page" | "component" | "layout";
name: string;
/** Name of the page layout (`app/layouts/<layout>.wrn`), if the page sets one. */
layout?: string;
/** Declared component props (empty for pages). */
props: PropDecl[];
/** Raw declarations from `types { ... }`, emitted as TypeScript. */
types: string[];
states: StateDecl[];
seo: SeoBlock;
view: ViewNode[];
styles: string[];
functions: string[];
dataApis: DataApiBlock[];
modeFunctions: ModeFunctionsBlock[];
lifecycle: LifecycleBlock;
watches: WatchBlock[];
apis: ApiBlock[];
realtimes: RealtimeBlock[];
}
export class ParseError extends Error {}
function parseSeoBlock(body: string): SeoBlock {
const out: SeoBlock = {};
const pair =
/([A-Za-z][A-Za-z0-9_-]*)\s*=\s*(?:"((?:\\.|[^"\\])*)"|'((?:\\.|[^'\\])*)'|([^\n;]+))/g;
for (const match of body.matchAll(pair)) {
const key = match[1]!;
const rawValue = match[2] ?? match[3] ?? match[4] ?? "";
out[key] = unescapeSeoValue(rawValue.trim());
}
return out;
}
function unescapeSeoValue(value: string): string {
return value.replace(/\\(["'\\nrt])/g, (_match, ch: string) => {
if (ch === "n") return "\n";
if (ch === "r") return "\r";
if (ch === "t") return "\t";
return ch;
});
}
export function parse(source: string): PageAst {
const lx = new Lexer(source);
const expect = (type: Token["type"]): Token => {
const t = lx.next();
if (t.type !== type) {
throw new ParseError(`Expected ${type} but got '${t.value || t.type}' at offset ${t.pos}`);
}
return t;
};
const expectKeyword = (kw: string): void => {
const t = lx.next();
if (t.type !== "ident" || t.value !== kw) {
throw new ParseError(`Expected '${kw}' but got '${t.value || t.type}' at offset ${t.pos}`);
}
};
try {
// A file may contain a page, reusable component, or reusable layout.
const opener = lx.next();
if (opener.type !== "ident" || !["page", "component", "layout"].includes(opener.value)) {
throw new ParseError(
`Expected 'page', 'component', or 'layout' but got '${
opener.value || opener.type
}' at offset ${opener.pos}`,
);
}
const kind = opener.value as "page" | "component" | "layout";
const name = expect("ident").value;
expect("lbrace");
let layout: string | undefined;
const props: PropDecl[] = [];
const types: string[] = [];
const states: StateDecl[] = [];
const seo: SeoBlock = {};
const view: ViewNode[] = [];
const styles: string[] = [];
const functions: string[] = [];
const dataApis: DataApiBlock[] = [];
const modeFunctions: ModeFunctionsBlock[] = [];
const lifecycle: LifecycleBlock = {};
const watches: WatchBlock[] = [];
const apis: ApiBlock[] = [];
const realtimes: RealtimeBlock[] = [];
while (lx.peek().type !== "rbrace") {
const kw = lx.peek();
if (kw.type === "eof") throw new ParseError(`Unexpected end of input inside ${kind}`);
if (kw.type !== "ident") {
throw new ParseError(`Expected a ${kind} member keyword at offset ${kw.pos}`);
}
switch (kw.value) {
case "layout": {
// layout = "public" — selects app/layouts/<name>.wrn for this page.
lx.next();
expect("eq");
layout = expect("string").value;
break;
}
case "props": {
// props { name: Type = <default> } — omit the default for required props.
lx.next();
expect("lbrace");
while (lx.peek().type !== "rbrace") {
const t = lx.peek();
if (t.type === "eof") throw new ParseError("Unexpected end of input inside props");
if (t.type !== "ident") {
throw new ParseError(`Expected a prop name at offset ${t.pos}`);
}
const pName = expect("ident").value;
let valueType: string | undefined;
let hasDefault = false;
if (lx.peek().type === "colon") {
lx.next();
const annotation = lx.readTypeAnnotation();
valueType = annotation.type;
hasDefault = annotation.hasDefault;
} else {
expect("eq");
hasDefault = true;
}
const defaultValue = hasDefault ? lx.readToLineEnd() : "undefined";
props.push({ name: pName, valueType, required: !hasDefault, default: defaultValue });
}
expect("rbrace");
break;
}
case "state": {
lx.next();
const sName = expect("ident").value;
let valueType: string | undefined;
if (lx.peek().type === "colon") {
lx.next();
const annotation = lx.readTypeAnnotation();
valueType = annotation.type;
if (!annotation.hasDefault) {
throw new ParseError(`State '${sName}' requires an initializer`);
}
} else {
expect("eq");
}
states.push({ name: sName, valueType, expr: lx.readToLineEnd() });
break;
}
case "types": {
lx.next();
types.push(lx.readBalancedBraces());
break;
}
case "view": {
lx.next();
expect("lbrace");
// The view body is plain HTML. Parse it straight off the source
// (the token lexer isn't used for markup), then resume after the
// block's closing `}`.
const { nodes, endPos } = parseHtmlView(lx.src, lx.pos);
view.push(...nodes);
lx.pos = endPos;
expect("rbrace");
break;
}
case "seo": {
lx.next();
Object.assign(seo, parseSeoBlock(lx.readBalancedBraces()));
break;
}
case "api": {
lx.next();
const method = expect("ident").value.toUpperCase();
const path = lx.readPath();
const body = lx.readBalancedBraces();
apis.push({ method, path, body });
break;
}
case "ssr":
case "client": {
const mode: DataMode = kw.value === "ssr" ? "ssr" : "client";
lx.next();
expect("lbrace");
while (lx.peek().type !== "rbrace") {
const member = lx.peek();
if (member.type === "eof") {
throw new ParseError(`Unexpected end of input inside ${mode} block`);
}
if (member.type !== "ident") {
throw new ParseError(`Expected a ${mode} member keyword at offset ${member.pos}`);
}
switch (member.value) {
case "api": {
lx.next();
const name = expect("ident").value;
const method = expect("ident").value.toUpperCase();
const path = lx.readPath();
const body = lx.readBalancedBraces();
dataApis.push({ mode, name, method, path, body });
break;
}
case "functions": {
lx.next();
modeFunctions.push({ mode, body: lx.readBalancedBraces() });
break;
}
default:
throw new ParseError(
`Unknown ${mode} member '${member.value}' at offset ${member.pos}`,
);
}
}
expect("rbrace");
break;
}
case "realtime": {
lx.next();
const rName = expect("ident").value;
expect("lbrace");
const handlers: RealtimeHandler[] = [];
while (lx.peek().type !== "rbrace") {
expectKeyword("on");
const event = expect("ident").value;
expect("lparen");
const args: string[] = [];
while (lx.peek().type !== "rparen") {
args.push(expect("ident").value);
if (lx.peek().type === "comma") lx.next();
}
expect("rparen");
handlers.push({ event, args, body: lx.readBalancedBraces() });
}
expect("rbrace");
realtimes.push({ name: rName, handlers });
break;
}
case "style": {
lx.next();
styles.push(lx.readBalancedBraces());
break;
}
case "lifecycle": {
lx.next();
expect("lbrace");
while (lx.peek().type !== "rbrace") {
const hook = lx.peek();
if (hook.type === "eof") {
throw new ParseError("Unexpected end of input inside lifecycle block");
}
if (hook.type !== "ident") {
throw new ParseError(`Expected a lifecycle hook at offset ${hook.pos}`);
}
if (hook.value !== "mount" && hook.value !== "update" && hook.value !== "unmount") {
throw new ParseError(`Unknown lifecycle hook '${hook.value}' at offset ${hook.pos}`);
}
const hookName = hook.value as LifecycleHookName;
lx.next();
if (lifecycle[hookName] !== undefined) {
throw new ParseError(`Duplicate lifecycle hook '${hookName}' at offset ${hook.pos}`);
}
lifecycle[hookName] = lx.readBalancedBraces();
}
expect("rbrace");
break;
}
case "watch": {
lx.next();
const stateName = expect("ident").value;
const body = lx.readBalancedBraces();
watches.push({
state: stateName,
body,
});
break;
}
case "functions": {
lx.next();
functions.push(lx.readBalancedBraces());
break;
}
default:
throw new ParseError(`Unknown page member '${kw.value}' at offset ${kw.pos}`);
}
}
expect("rbrace");
const declaredStates = new Set(states.map((state) => state.name));
for (const watcher of watches) {
if (!declaredStates.has(watcher.state)) {
throw new ParseError(`Cannot watch undeclared state '${watcher.state}'`);
}
}
for (const prop of props) {
const problem = validateTypedInitializer(`Prop '${prop.name}'`, prop.valueType, prop.default);
if (problem) throw new ParseError(problem);
}
for (const state of states) {
const problem = validateTypedInitializer(
`State '${state.name}'`,
state.valueType,
state.expr,
);
if (problem) throw new ParseError(problem);
}
return {
type: "page",
kind,
name,
layout,
props,
types,
states,
seo,
view,
styles,
functions,
dataApis,
modeFunctions,
lifecycle,
watches,
apis,
realtimes,
};
} catch (err) {
if (err instanceof LexError) throw new ParseError(err.message);
throw err;
}
}
/**
* Parse the body of a `view { ... }` block as plain HTML.
*
* `src` is the whole `.wrn` source; `pos` points just past the view block's
* opening `{`. Returns the parsed nodes plus the index of the block's closing
* `}` (left for the caller to consume). It is intentionally lenient — you write
* markup the way you already know:
*
* - `<tag attr="v" @event="expr">children</tag>` — elements with attributes
* - `<tag/>` and HTML void elements (`<br>`, `<img>`, …) — no closing tag
* - text may contain `{expr}` interpolation, kept verbatim for the runtime
* - `@event="..."` becomes a client event binding; hyphenated names are fine
* - `<!-- comments -->` are dropped
*
* `{` and `}` in text are reserved for interpolation; a lone `<` that isn't a
* tag is treated as literal text.
*/
export function parseHtmlView(src: string, pos: number): { nodes: ViewNode[]; endPos: number } {
let i = pos;
const isNameStart = (c: string): boolean => /[A-Za-z_]/.test(c);
const isTagNamePart = (c: string): boolean => /[A-Za-z0-9_$:.-]/.test(c);
const isWs = (c: string): boolean => c === " " || c === "\t" || c === "\n" || c === "\r";
const fail = (msg: string): never => {
throw new ParseError(`${msg} at offset ${i}`);
};
const skipWs = (): void => {
while (i < src.length && isWs(src[i]!)) i++;
};
/** Read a `{...}` interpolation (brace-balanced), braces included. */
const readInterpolation = (): string => {
const start = i;
let depth = 0;
for (; i < src.length; i++) {
if (src[i] === "{") depth++;
else if (src[i] === "}" && --depth === 0) {
i++;
return src.slice(start, i);
}
}
return fail("Unterminated `{` interpolation in view");
};
const readQuoted = (): string => {
const quote = src[i];
if (quote !== '"' && quote !== "'") return fail("Expected a quoted attribute value");
i++;
const start = i;
while (i < src.length && src[i] !== quote) i++;
if (i >= src.length) return fail("Unterminated attribute value");
const value = src.slice(start, i);
i++; // closing quote
return value;
};
const readTagName = (): string => {
if (i >= src.length || !isNameStart(src[i]!)) {
return fail("Expected a tag name");
}
const start = i++;
while (i < src.length && isTagNamePart(src[i]!)) {
i++;
}
return src.slice(start, i);
};
const readAttributeName = (): string => {
if (i >= src.length) {
return fail("Expected an attribute name");
}
const start = i;
while (i < src.length) {
const char = src[i];
const next = src[i + 1];
if (
char === "=" ||
char === ">" ||
char === '"' ||
char === "'" ||
char === " " ||
char === "\t" ||
char === "\n" ||
char === "\r" ||
(char === "/" && next === ">")
) {
break;
}
i++;
}
if (i === start) {
return fail("Expected an attribute name");
}
return src.slice(start, i);
};
const parseTag = (): ViewNode => {
i++; // consume '<'
const tag = readTagName();
const attrs: Attr[] = [];
for (;;) {
skipWs();
const c = src[i];
if (c === undefined) return fail(`Unterminated <${tag}> tag`);
if (c === ">") {
i++;
break;
}
if (c === "/" && src[i + 1] === ">") {
i += 2;
return { type: "element", tag, attrs, children: [] };
}
if (c === "@") {
i++;
const name = readAttributeName();
skipWs();
if (src[i] !== "=") return fail(`Expected '=' after @${name}`);
i++;
skipWs();
attrs.push({ name, value: readQuoted(), event: true });
continue;
}
const name = readAttributeName();
skipWs();
if (src[i] === "=") {
i++;
skipWs();
attrs.push({ name, value: readQuoted(), event: false });
} else {
attrs.push({ name, value: "", event: false, boolean: true });
}
}
if (VOID_ELEMENTS.has(tag.toLowerCase())) {
return { type: "element", tag, attrs, children: [] };
}
const children = parseNodeList("element");
// parseNodeList stops at the parent's closing tag `</`.
if (src[i] !== "<" || src[i + 1] !== "/") return fail(`Expected </${tag}>`);
i += 2;
skipWs();
const close = readTagName();
if (close !== tag) return fail(`Mismatched </${close}>, expected </${tag}>`);
skipWs();
if (src[i] !== ">") return fail(`Expected '>' to close </${tag}>`);
i++;
return { type: "element", tag, attrs, children };
};
const EACH_HEADER =
/^\{#each\s+([\s\S]+?)\s+as\s+([A-Za-z_$][\w$]*)\s*(?:,\s*([A-Za-z_$][\w$]*)\s*)?\}$/;
/** Parse `{#each <list> as <item>[, <index>]} …body… {:empty} …empty… {/each}`. */
function parseEach(): ViewNode {
const header = readInterpolation(); // reads the full `{#each …}`
const m = EACH_HEADER.exec(header);
if (!m) return fail(`Invalid {#each …} header: ${header}`);
const list = m[1]!.trim();
const item = m[2]!;
const index = m[3];
const body = parseNodeList("each"); // stops at {:empty} or {/each}
let empty: ViewNode[] = [];
if (src.startsWith("{:empty}", i)) {
i += "{:empty}".length;
empty = parseNodeList("each"); // stops at {/each}
}
if (!src.startsWith("{/each}", i)) return fail("Expected `{/each}` to close `{#each}`");
i += "{/each}".length;
return { type: "each", list, item, index, body, empty };
}
/** Parse `{#if <expr>} … {:else if <expr>} … {:else} … {/if}`. */
function parseIf(): ViewNode {
const header = readInterpolation(); // reads the full `{#if …}`
const m = /^\{#if\s+([\s\S]+?)\s*\}$/.exec(header);
if (!m) return fail(`Invalid {#if …} header: ${header}`);
const branches: { cond: string | null; body: ViewNode[] }[] = [
{ cond: m[1]!.trim(), body: parseNodeList("if") },
];
for (;;) {
if (src.startsWith("{:else if", i)) {
const h = readInterpolation();
const mm = /^\{:else if\s+([\s\S]+?)\s*\}$/.exec(h);
if (!mm) return fail(`Invalid {:else if …}: ${h}`);
branches.push({ cond: mm[1]!.trim(), body: parseNodeList("if") });
continue;
}
if (src.startsWith("{:else}", i)) {
i += "{:else}".length;
branches.push({ cond: null, body: parseNodeList("if") });
continue;
}
break;
}
if (!src.startsWith("{/if}", i)) return fail("Expected `{/if}` to close `{#if}`");
i += "{/if}".length;
return { type: "if", branches };
}
/**
* Parse a run of nodes. `mode` sets the terminator:
* - "root": stops at the view block's closing `}`
* - "element": stops at the parent element's closing tag (`</`)
* - "each": stops (without consuming) at `{:empty}` or `{/each}`
* - "if": stops (without consuming) at `{:else …}` or `{/if}`
* `{#each …}` and `{#if …}` start nested blocks in any mode.
*/
function parseNodeList(mode: "root" | "element" | "each" | "if"): ViewNode[] {
const nodes: ViewNode[] = [];
let text = "";
const flush = (): void => {
if (text.length > 0) {
nodes.push({ type: "text", value: text });
text = "";
}
};
for (;;) {
if (i >= src.length) {
return mode === "root"
? fail("Unexpected end of view (missing `}`)")
: fail("Unclosed block");
}
const c = src[i]!;
if (c === "<") {
const next = src[i + 1];
if (next === "/") {
flush();
break; // parent's closing tag
}
if (src.startsWith("<!--", i)) {
const end = src.indexOf("-->", i + 4);
i = end === -1 ? src.length : end + 3;
continue;
}
if (next !== undefined && (isNameStart(next) || next === "!")) {
flush();
nodes.push(parseTag());
continue;
}
// A lone `<` that doesn't start a tag: treat as literal text.
text += c;
i++;
continue;
}
if (c === "{") {
if (src.startsWith("{#each", i)) {
flush();
nodes.push(parseEach());
continue;
}
if (src.startsWith("{#if", i)) {
flush();
nodes.push(parseIf());
continue;
}
if (mode === "each" && (src.startsWith("{:empty}", i) || src.startsWith("{/each}", i))) {
flush();
break; // loop-section terminator; left for parseEach
}
if (mode === "if" && (src.startsWith("{:else", i) || src.startsWith("{/if}", i))) {
flush();
break; // conditional-section terminator; left for parseIf
}
text += readInterpolation();
continue;
}
if (c === "}" && mode === "root") {
flush();
break; // view terminator; leave `}` for the caller
}
text += c;
i++;
}
return nodes;
}
const nodes = parseNodeList("root");
return { nodes, endPos: i };
}