gofront 0.0.8 → 0.0.9

package/src/templ-lexer.js

@@ -0,0 +1,737 @@
+// Templ lexer: tokenizes .templ files into a stream of tokens.
+//
+// A .templ file is a Go-like file where `templ Name(params) { ... }` declarations
+// contain HTML bodies instead of Go statements. All other constructs (package,
+// import, regular func, var, const, type) are lexed as normal Go.
+//
+// Token types produced (in addition to the standard Go token types from lexer.js):
+//   TEMPL_KW       — the `templ` keyword
+//   HTML_OPEN      — <tag attr...> (includes the `>`)
+//   HTML_CLOSE     — </tag>
+//   HTML_SELF      — <tag .../>
+//   HTML_TEXT      — literal text between tags
+//   TEMPL_EXPR     — { goExpr } inside HTML (the goExpr tokens are inlined)
+//   TEMPL_COMP     — @Call(...) inside HTML
+//   TEMPL_CHILDREN — { children... }
+//
+// The approach: scan the file character-by-character. Outside templ bodies, use
+// the existing Lexer class to tokenise Go. Inside templ bodies, scan HTML.
+
+import { Lexer, T, Token } from "./lexer.js";
+
+export const TT = {
+	TEMPL_KW: "templ",
+	HTML_OPEN: "HTML_OPEN",
+	HTML_CLOSE: "HTML_CLOSE",
+	HTML_SELF: "HTML_SELF",
+	HTML_TEXT: "HTML_TEXT",
+	TEMPL_EXPR: "TEMPL_EXPR", // inline Go expression token stream
+	TEMPL_COMP: "TEMPL_COMP", // @Component(args) call
+	TEMPL_CHILDREN: "TEMPL_CHILDREN",
+	TEMPL_IF: "TEMPL_IF",
+	TEMPL_ELSE: "TEMPL_ELSE",
+	TEMPL_FOR: "TEMPL_FOR",
+	TEMPL_SWITCH: "TEMPL_SWITCH", // switch expr {
+	TEMPL_CASE: "TEMPL_CASE", // case expr:
+	TEMPL_DEFAULT: "TEMPL_DEFAULT", // default:
+	TEMPL_END: "TEMPL_END", // closing } of if/for/switch/templ body
+};
+
+// Describes a parsed HTML attribute
+// { kind: "static", name, value } | { kind: "expr", name, tokens } | { kind: "bool", name }
+
+export class TemplLexer {
+	constructor(source, filename = "<templ>") {
+		this.src = source;
+		this.pos = 0;
+		this.line = 1;
+		this.col = 1;
+		this.filename = filename;
+		this.tokens = [];
+	}
+
+	// ── Character primitives ────────────────────────────────────
+
+	peek(n = 0) {
+		return this.src[this.pos + n];
+	}
+
+	advance() {
+		const ch = this.src[this.pos++];
+		if (ch === "\n") {
+			this.line++;
+			this.col = 1;
+		} else {
+			this.col++;
+		}
+		return ch;
+	}
+
+	eof() {
+		return this.pos >= this.src.length;
+	}
+
+	skipWhitespace() {
+		while (!this.eof() && /[ \t\r\n]/.test(this.peek())) this.advance();
+	}
+
+	skipLineComment() {
+		while (!this.eof() && this.peek() !== "\n") this.advance();
+	}
+
+	skipBlockComment() {
+		while (!this.eof()) {
+			if (this.peek() === "*" && this.peek(1) === "/") {
+				this.advance();
+				this.advance();
+				return;
+			}
+			this.advance();
+		}
+		this.err("unterminated block comment");
+	}
+
+	err(msg) {
+		throw new Error(`${this.filename}:${this.line}:${this.col}: ${msg}`);
+	}
+
+	// ── Main tokenise entry point ────────────────────────────────
+
+	tokenize() {
+		// Scan the entire file. We identify `templ Name(params) {` blocks
+		// and handle their bodies in HTML mode; everything else is Go.
+		while (!this.eof()) {
+			this.skipWhitespace();
+			if (this.eof()) break;
+
+			// Skip Go-style comments
+			if (this.peek() === "/" && this.peek(1) === "/") {
+				this.skipLineComment();
+				continue;
+			}
+			if (this.peek() === "/" && this.peek(1) === "*") {
+				this.advance();
+				this.advance();
+				this.skipBlockComment();
+				continue;
+			}
+
+			// Detect `templ` keyword at identifier boundaries
+			if (this._matchKeyword("templ")) {
+				this.tokens.push(new Token(TT.TEMPL_KW, "templ", this.line, this.col));
+				this._lexTemplDecl();
+				continue;
+			}
+
+			// Everything else: scan to end of top-level declaration using the
+			// raw Go lexer, accumulating tokens until the next `templ` keyword
+			// or EOF.
+			this._lexGoChunk();
+		}
+
+		this.tokens.push(new Token(T.EOF, "", this.line, this.col));
+		return this.tokens;
+	}
+
+	// ── Go chunk lexing ──────────────────────────────────────────
+
+	// Scan ahead to find the next `templ ` keyword at top level (not inside
+	// braces/parens/strings/comments). Collect all source text up to that
+	// point, lex it as normal Go, and append the tokens.
+	_lexGoChunk() {
+		const start = this.pos;
+		let depth = 0; // brace depth
+
+		while (!this.eof()) {
+			// Check for `templ` keyword at top level (depth === 0)
+			if (depth === 0 && this._peekKeyword("templ")) break;
+
+			const ch = this.peek();
+
+			// Strings (raw or interpreted)
+			if (ch === '"' || ch === "'") {
+				this._skipGoString(ch);
+				continue;
+			}
+			if (ch === "`") {
+				this._skipRawString();
+				continue;
+			}
+
+			// Comments
+			if (ch === "/" && this.peek(1) === "/") {
+				this.skipLineComment();
+				continue;
+			}
+			if (ch === "/" && this.peek(1) === "*") {
+				this.advance();
+				this.advance();
+				this.skipBlockComment();
+				continue;
+			}
+
+			if (ch === "{") depth++;
+			else if (ch === "}") {
+				if (depth === 0) break; // should not happen at top level, stop
+				depth--;
+			}
+			this.advance();
+		}
+
+		if (this.pos === start) {
+			// Nothing to lex; skip one char to avoid infinite loop
+			this.advance();
+			return;
+		}
+
+		const chunk = this.src.slice(start, this.pos);
+		for (const t of this._goTokens(chunk)) this.tokens.push(t);
+	}
+
+	// ── Templ declaration lexing ─────────────────────────────────
+
+	// Called right after consuming `templ` keyword.
+	// Lexes: Name(params) { <HTML body> }
+	_lexTemplDecl() {
+		this.skipWhitespace();
+
+		// Name
+		const nameStart = this.pos;
+		while (!this.eof() && /\w/.test(this.peek())) this.advance();
+		const name = this.src.slice(nameStart, this.pos);
+		if (!name) this.err("expected component name after 'templ'");
+		this.tokens.push(new Token(T.IDENT, name, this.line, this.col));
+
+		this.skipWhitespace();
+
+		// Parameter list ( ... ) — lex as Go
+		if (this.peek() !== "(") this.err("expected '(' after templ name");
+		const paramStart = this.pos;
+		this._skipBalanced("(", ")");
+		const paramSrc = this.src.slice(paramStart, this.pos);
+		for (const t of this._goTokens(paramSrc)) this.tokens.push(t);
+
+		this.skipWhitespace();
+
+		// Opening { of body
+		if (this.peek() !== "{") this.err("expected '{' after templ params");
+		this.advance(); // consume {
+		this.tokens.push(new Token(T.LBRACE, "{", this.line, this.col));
+
+		// HTML body
+		this._lexHtmlBody(1);
+	}
+
+	// Lex the HTML body of a templ declaration.
+	// depth: current brace depth (starts at 1 for the opening { of templ)
+	_lexHtmlBody(depth) {
+		while (!this.eof()) {
+			this.skipWhitespace();
+			if (this.eof()) break;
+
+			const ch = this.peek();
+
+			// Closing brace: if depth becomes 0, we're done
+			if (ch === "}") {
+				if (depth === 1) {
+					this.advance(); // consume }
+					this.tokens.push(new Token(T.RBRACE, "}", this.line, this.col));
+					return;
+				}
+				// depth > 1 should not happen at the top of a templ body; treat as end
+				this.advance();
+				this.tokens.push(new Token(TT.TEMPL_END, "}", this.line, this.col));
+				return;
+			}
+
+			this._lexHtmlNode();
+		}
+	}
+
+	// Lex HTML nodes for a switch-case body until case/default/} without consuming
+	// the terminator (the caller handles it).
+	_lexCaseBody() {
+		while (!this.eof()) {
+			this.skipWhitespace();
+			if (this.eof()) break;
+			const ch = this.peek();
+			if (ch === "}") break;
+			if (this._peekKeyword("case") || this._peekKeyword("default")) break;
+			this._lexHtmlNode();
+		}
+	}
+
+	// Lex a single HTML node at the current position.
+	_lexHtmlNode() {
+		const ch = this.peek();
+
+		// HTML comment <!-- ... -->
+		if (ch === "<" && this.src.startsWith("<!--", this.pos)) {
+			this._skipHtmlComment();
+			return;
+		}
+		// Closing tag </tag>
+		if (ch === "<" && this.peek(1) === "/") {
+			this._lexHtmlCloseTag();
+			return;
+		}
+		// Opening tag <tag ...> or <tag .../>
+		if (ch === "<") {
+			this._lexHtmlOpenTag();
+			return;
+		}
+		// @Component(args) call
+		if (ch === "@") {
+			this._lexTemplComponent();
+			return;
+		}
+		// { children... } or { goExpr }
+		if (ch === "{") {
+			this._lexBraceExprOrControl();
+			return;
+		}
+		// Control flow keywords
+		if (this._peekKeyword("if")) {
+			this._lexTemplIf();
+			return;
+		}
+		if (this._peekKeyword("for")) {
+			this._lexTemplFor();
+			return;
+		}
+		if (this._peekKeyword("switch")) {
+			this._lexTemplSwitch();
+			return;
+		}
+		// Plain text content
+		this._lexHtmlText();
+	}
+
+	// ── HTML element lexing ───────────────────────────────────────
+
+	_lexHtmlOpenTag() {
+		const line = this.line;
+		const col = this.col;
+		this.advance(); // consume <
+		this.skipWhitespace();
+
+		// Tag name
+		const tagName = this._readHtmlName();
+		if (!tagName) this.err("expected tag name");
+
+		const attrs = [];
+		let selfClose = false;
+
+		// Attributes
+		while (!this.eof()) {
+			this.skipWhitespace();
+			const c = this.peek();
+
+			if (c === "/" && this.peek(1) === ">") {
+				this.advance();
+				this.advance(); // />
+				selfClose = true;
+				break;
+			}
+			if (c === ">") {
+				this.advance(); // >
+				break;
+			}
+
+			// Attribute name
+			const attrName = this._readHtmlAttrName();
+			if (!attrName) break;
+
+			this.skipWhitespace();
+
+			// Conditional boolean attribute: name?={expr}
+			if (this.peek() === "?" && this.peek(1) === "=") {
+				this.advance();
+				this.advance(); // ?=
+				this.skipWhitespace();
+				if (this.peek() !== "{") this.err("expected '{' after ?=");
+				const exprTokens = this._readGoExprInBraces();
+				attrs.push({ kind: "cond-bool", name: attrName, tokens: exprTokens });
+				continue;
+			}
+
+			if (this.peek() !== "=") {
+				// Boolean attribute: no value
+				attrs.push({ kind: "bool", name: attrName });
+				continue;
+			}
+			this.advance(); // =
+			this.skipWhitespace();
+
+			if (this.peek() === "{") {
+				// Expression attribute: name={ goExpr }
+				const exprTokens = this._readGoExprInBraces();
+				attrs.push({ kind: "expr", name: attrName, tokens: exprTokens });
+			} else if (this.peek() === '"') {
+				// Static string attribute
+				const value = this._readQuotedString();
+				attrs.push({ kind: "static", name: attrName, value });
+			} else {
+				// Unquoted attribute value
+				let val = "";
+				while (!this.eof() && !/[\s>]/.test(this.peek())) val += this.advance();
+				attrs.push({ kind: "static", name: attrName, value: val });
+			}
+		}
+
+		const kind = selfClose ? TT.HTML_SELF : TT.HTML_OPEN;
+		this.tokens.push(new Token(kind, { tag: tagName, attrs }, line, col));
+	}
+
+	_lexHtmlCloseTag() {
+		const line = this.line;
+		this.advance(); // <
+		this.advance(); // /
+		this.skipWhitespace();
+		const tagName = this._readHtmlName();
+		this.skipWhitespace();
+		if (this.peek() === ">") this.advance();
+		this.tokens.push(new Token(TT.HTML_CLOSE, tagName, line, this.col));
+	}
+
+	_skipHtmlComment() {
+		// <!-- ... -->
+		this.pos += 4; // <!--
+		while (!this.eof()) {
+			if (this.src.startsWith("-->", this.pos)) {
+				this.pos += 3;
+				return;
+			}
+			if (this.peek() === "\n") {
+				this.line++;
+				this.col = 1;
+			} else {
+				this.col++;
+			}
+			this.pos++;
+		}
+	}
+
+	// ── Expression / control flow inside HTML ─────────────────────
+
+	_lexBraceExprOrControl() {
+		// { children... }  →  TEMPL_CHILDREN
+		// { goExpr }       →  TEMPL_EXPR
+		const saveLine = this.line;
+		const saveCol = this.col;
+
+		this.advance(); // {
+		this.skipWhitespace();
+
+		// Check for children...
+		if (this.src.startsWith("children...", this.pos)) {
+			this.pos += "children...".length;
+			this.skipWhitespace();
+			if (this.peek() === "}") this.advance();
+			this.tokens.push(new Token(TT.TEMPL_CHILDREN, null, saveLine, saveCol));
+			return;
+		}
+
+		// Otherwise lex the Go expression inside the braces
+		const exprTokens = this._readGoExprTokensUntilClose();
+		this.tokens.push(new Token(TT.TEMPL_EXPR, exprTokens, saveLine, saveCol));
+	}
+
+	_lexTemplComponent() {
+		const line = this.line;
+		const col = this.col;
+		this.advance(); // @
+
+		// Read the call expression (Go tokens until end of balanced parens)
+		const nameStart = this.pos;
+		while (!this.eof() && /[\w.]/.test(this.peek())) this.advance();
+		const name = this.src.slice(nameStart, this.pos);
+		if (!name) this.err("expected component name after @");
+
+		this.skipWhitespace();
+		if (this.peek() !== "(") this.err("expected '(' after @component name");
+
+		const argsStart = this.pos;
+		this._skipBalanced("(", ")");
+		const argsSrc = name + this.src.slice(argsStart, this.pos);
+
+		const callTokens = this._goTokens(argsSrc);
+		this.tokens.push(new Token(TT.TEMPL_COMP, callTokens, line, col));
+	}
+
+	_lexTemplIf() {
+		const line = this.line;
+		const condTokens = this._lexGoExprBeforeBrace();
+		this.tokens.push(new Token(TT.TEMPL_IF, condTokens, line, this.col));
+		this._lexHtmlBody(1);
+
+		// Check for else
+		this.skipWhitespace();
+		if (this._peekKeyword("else")) {
+			this._readWord(); // consume `else`
+			this.skipWhitespace();
+			this.tokens.push(new Token(TT.TEMPL_ELSE, null, this.line, this.col));
+			if (this._peekKeyword("if")) {
+				// else if
+				this._lexTemplIf();
+			} else {
+				// else block
+				if (this.peek() !== "{") this.err("expected '{' after else");
+				this.advance(); // {
+				this._lexHtmlBody(1);
+			}
+		}
+	}
+
+	_lexTemplFor() {
+		const line = this.line;
+		const stmtTokens = this._lexGoExprBeforeBrace();
+		this.tokens.push(new Token(TT.TEMPL_FOR, stmtTokens, line, this.col));
+		this._lexHtmlBody(1);
+	}
+
+	_lexTemplSwitch() {
+		const line = this.line;
+		this._readWord(); // consume "switch"
+		this.skipWhitespace();
+		const exprTokens = this._lexGoExprBeforeBrace();
+		this.tokens.push(new Token(TT.TEMPL_SWITCH, exprTokens, line, this.col));
+
+		// Lex case/default entries until the closing }
+		while (!this.eof()) {
+			this.skipWhitespace();
+			if (this.eof()) break;
+
+			if (this.peek() === "}") {
+				this.advance(); // consume }
+				this.tokens.push(new Token(TT.TEMPL_END, "}", this.line, this.col));
+				return;
+			}
+
+			if (this._peekKeyword("case")) {
+				const caseLine = this.line;
+				this._readWord(); // consume "case"
+				this.skipWhitespace();
+				// Read case expression up to ":"
+				const caseExprStart = this.pos;
+				// Skip to the colon, handling string literals
+				while (!this.eof() && this.peek() !== ":") {
+					const c = this.peek();
+					if (c === '"' || c === "'") {
+						this._skipGoString(c);
+					} else if (c === "`") {
+						this._skipRawString();
+					} else {
+						this.advance();
+					}
+				}
+				const caseExprSrc = this.src.slice(caseExprStart, this.pos).trimEnd();
+				if (this.peek() === ":") this.advance(); // :
+				const caseTokens = this._goTokens(caseExprSrc);
+				this.tokens.push(
+					new Token(TT.TEMPL_CASE, caseTokens, caseLine, this.col),
+				);
+				this._lexCaseBody();
+			} else if (this._peekKeyword("default")) {
+				const defaultLine = this.line;
+				this._readWord(); // consume "default"
+				this.skipWhitespace();
+				if (this.peek() === ":") this.advance(); // :
+				this.tokens.push(
+					new Token(TT.TEMPL_DEFAULT, null, defaultLine, this.col),
+				);
+				this._lexCaseBody();
+			} else {
+				// Unexpected token inside switch body — skip to avoid infinite loop
+				this.advance();
+			}
+		}
+	}
+
+	// ── HTML text content ─────────────────────────────────────────
+
+	_lexHtmlText() {
+		let text = "";
+		const line = this.line;
+		while (!this.eof()) {
+			const c = this.peek();
+			if (c === "<" || c === "{" || c === "@" || c === "}") break;
+			// Detect control-flow keywords at line start
+			if (
+				(c === "i" && this._peekKeyword("if")) ||
+				(c === "f" && this._peekKeyword("for")) ||
+				(c === "s" && this._peekKeyword("switch")) ||
+				(c === "c" && this._peekKeyword("case")) ||
+				(c === "d" && this._peekKeyword("default"))
+			)
+				break;
+			text += this.advance();
+		}
+		const trimmed = text.trim();
+		if (trimmed)
+			this.tokens.push(new Token(TT.HTML_TEXT, trimmed, line, this.col));
+	}
+
+	// ── Helpers ───────────────────────────────────────────────────
+
+	_goTokens(src) {
+		return new Lexer(src, this.filename)
+			.tokenize()
+			.filter((t) => t.type !== T.EOF);
+	}
+
+	// Consume text up to the opening `{`, advance past it, and return Go tokens.
+	_lexGoExprBeforeBrace() {
+		const start = this.pos;
+		this._skipToOpenBrace();
+		const src = this.src.slice(start, this.pos).trimEnd();
+		this.advance(); // consume {
+		return src ? this._goTokens(src) : [];
+	}
+
+	// Read Go expression tokens until the closing }
+	_readGoExprInBraces() {
+		this.advance(); // {
+		return this._readGoExprTokensUntilClose();
+	}
+
+	_readGoExprTokensUntilClose() {
+		const start = this.pos;
+		let depth = 1;
+		while (!this.eof() && depth > 0) {
+			const c = this.peek();
+			if (c === "{") depth++;
+			else if (c === "}") {
+				depth--;
+				if (depth === 0) break;
+			} else if (c === '"' || c === "'") {
+				this._skipGoString(c);
+				continue;
+			} else if (c === "`") {
+				this._skipRawString();
+				continue;
+			}
+			this.advance();
+		}
+		const exprSrc = this.src.slice(start, this.pos).trim();
+		if (this.peek() === "}") this.advance(); // consume }
+		if (!exprSrc) return [];
+		return this._goTokens(exprSrc);
+	}
+
+	_readHtmlName() {
+		let name = "";
+		while (!this.eof() && /[a-zA-Z0-9_-]/.test(this.peek()))
+			name += this.advance();
+		return name;
+	}
+
+	_readHtmlAttrName() {
+		let name = "";
+		while (!this.eof() && /[a-zA-Z0-9_:.-]/.test(this.peek()))
+			name += this.advance();
+		return name;
+	}
+
+	_readQuotedString() {
+		this.advance(); // opening "
+		let s = "";
+		while (!this.eof() && this.peek() !== '"') {
+			if (this.peek() === "\\") {
+				this.advance();
+				s += this.advance();
+			} else s += this.advance();
+		}
+		if (this.peek() === '"') this.advance();
+		return s;
+	}
+
+	_readWord() {
+		let w = "";
+		while (!this.eof() && /\w/.test(this.peek())) w += this.advance();
+		return w;
+	}
+
+	_skipToOpenBrace() {
+		// Skip text until we hit an unbalanced {, tracking parens/brackets
+		let paren = 0,
+			bracket = 0;
+		while (!this.eof()) {
+			const c = this.peek();
+			if (c === "(") paren++;
+			else if (c === ")") paren--;
+			else if (c === "[") bracket++;
+			else if (c === "]") bracket--;
+			else if (c === "{" && paren === 0 && bracket === 0) return;
+			else if (c === '"' || c === "'") {
+				this._skipGoString(c);
+				continue;
+			} else if (c === "`") {
+				this._skipRawString();
+				continue;
+			}
+			this.advance();
+		}
+	}
+
+	_skipBalanced(open, close) {
+		let depth = 0;
+		while (!this.eof()) {
+			const c = this.peek();
+			if (c === '"' || c === "'") {
+				this._skipGoString(c);
+				continue;
+			}
+			if (c === "`") {
+				this._skipRawString();
+				continue;
+			}
+			if (c === open) depth++;
+			else if (c === close) {
+				depth--;
+				if (depth === 0) {
+					this.advance();
+					return;
+				}
+			}
+			this.advance();
+		}
+	}
+
+	_skipGoString(quote) {
+		this.advance(); // opening quote
+		while (!this.eof()) {
+			const c = this.peek();
+			if (c === "\\") {
+				this.advance();
+				this.advance();
+				continue;
+			}
+			if (c === quote) {
+				this.advance();
+				return;
+			}
+			this.advance();
+		}
+	}
+
+	_skipRawString() {
+		this.advance(); // `
+		while (!this.eof() && this.peek() !== "`") this.advance();
+		if (!this.eof()) this.advance();
+	}
+
+	// Returns true if src[pos..] starts with `word` followed by a non-word char
+	_peekKeyword(word) {
+		if (!this.src.startsWith(word, this.pos)) return false;
+		const after = this.src[this.pos + word.length];
+		return !after || /\W/.test(after);
+	}
+
+	// Consume a keyword word
+	_matchKeyword(word) {
+		if (!this._peekKeyword(word)) return false;
+		for (let i = 0; i < word.length; i++) this.advance();
+		return true;
+	}
+}