more robust expression parsing

packages/svelte/src/compiler/phases/1-parse/read/expression.js

@@ -3,6 +3,7 @@
 import { parse_expression_at } from '../acorn.js';
 import { regex_whitespace } from '../../patterns.js';
 import * as e from '../../../errors.js';
+import { find_matching_bracket } from '../utils/bracket.js';
 
 /**
  * @param {Parser} parser
@@ -41,33 +42,17 @@ export default function read_expression(parser) {
 	} catch (err) {
 		if (parser.loose) {
 			// Find the next } and treat it as the end of the expression
-			let index = parser.index;
-			let num_braces = 0;
-			while (index < parser.template.length) {
-				const char = parser.template[index];
-				if (char === '{') num_braces += 1;
-				if (char === '}') {
-					if (num_braces === 0) {
-						// We assume that there's some kind of whitespace or the start of the closing tag after the closing brace,
-						// else this hints at a wrong counting of braces (e.g. in the case of foo={'hi}'})
-						if (!/[\s>/]/.test(parser.template[index + 1])) {
-							num_braces += 1;
-							continue;
-						}
-
-						const start = parser.index;
-						parser.index = index;
-						// We don't know what the expression is and signal this by returning an empty identifier
-						return {
-							type: 'Identifier',
-							start,
-							end: index,
-							name: ''
-						};
-					}
-					num_braces -= 1;
-				}
-				index += 1;
+			const end = find_matching_bracket(parser.template, parser.index, '{');
+			if (end) {
+				const start = parser.index;
+				parser.index = end;
+				// We don't know what the expression is and signal this by returning an empty identifier
+				return {
+					type: 'Identifier',
+					start,
+					end,
+					name: ''
+				};
 			}
 		}
 

packages/svelte/src/compiler/phases/1-parse/utils/bracket.js

@@ -1,3 +1,5 @@
+import full_char_code_at from './full_char_code_at.js';
+
 const SQUARE_BRACKET_OPEN = '['.charCodeAt(0);
 const SQUARE_BRACKET_CLOSE = ']'.charCodeAt(0);
 const CURLY_BRACKET_OPEN = '{'.charCodeAt(0);
@@ -33,3 +35,137 @@ export function get_bracket_close(open) {
 		return CURLY_BRACKET_CLOSE;
 	}
 }
+
+/**
+ * @param {number} num
+ * @returns {number} Infinity if {@link num} is negative, else {@link num}.
+ */
+function infinity_if_negative(num) {
+	if (num < 0) {
+		return Infinity;
+	}
+	return num;
+}
+
+/**
+ * @param {string} string The string to search.
+ * @param {number} search_start_index The index to start searching at.
+ * @param {"'" | '"' | '`'} string_start_char The character that started this string.
+ * @returns {number} The index of the end of this string expression, or `Infinity` if not found.
+ */
+function find_string_end(string, search_start_index, string_start_char) {
+	let string_to_search;
+	if (string_start_char === '`') {
+		string_to_search = string;
+	} else {
+		// we could slice at the search start index, but this way the index remains valid
+		string_to_search = string.slice(
+			0,
+			infinity_if_negative(string.indexOf('\n', search_start_index))
+		);
+	}
+
+	return find_unescaped_char(string_to_search, search_start_index, string_start_char);
+}
+
+/**
+ * @param {string} string The string to search.
+ * @param {number} search_start_index The index to start searching at.
+ * @returns {number} The index of the end of this regex expression, or `Infinity` if not found.
+ */
+function find_regex_end(string, search_start_index) {
+	return find_unescaped_char(string, search_start_index, '/');
+}
+
+/**
+ *
+ * @param {string} string The string to search.
+ * @param {number} search_start_index The index to begin the search at.
+ * @param {string} char The character to search for.
+ * @returns {number} The index of the first unescaped instance of {@link char}, or `Infinity` if not found.
+ */
+function find_unescaped_char(string, search_start_index, char) {
+	let i = search_start_index;
+	while (true) {
+		const found_index = string.indexOf(char, i);
+		if (found_index === -1) {
+			return Infinity;
+		}
+		if (count_leading_backslashes(string, found_index - 1) % 2 === 0) {
+			return found_index;
+		}
+		i = found_index + 1;
+	}
+}
+
+/**
+ * Count consecutive leading backslashes before {@link search_start_index}.
+ *
+ * @example
+ * ```js
+ * count_leading_backslashes('\\\\\\foo', 2); // 3 (the backslashes have to be escaped in the string literal, there are three in reality)
+ * ```
+ *
+ * @param {string} string The string to search.
+ * @param {number} search_start_index The index to begin the search at.
+ */
+function count_leading_backslashes(string, search_start_index) {
+	let i = search_start_index;
+	let count = 0;
+	while (string[i] === '\\') {
+		count++;
+		i--;
+	}
+	return count;
+}
+
+/**
+ * Finds the corresponding closing bracket, ignoring brackets found inside comments, strings, or regex expressions.
+ * @param {string} template The string to search.
+ * @param {number} index The index to begin the search at.
+ * @param {string} open The opening bracket (ex: `'{'` will search for `'}'`).
+ * @returns {number | undefined} The index of the closing bracket, or undefined if not found.
+ */
+export function find_matching_bracket(template, index, open) {
+	const open_code = full_char_code_at(open, 0);
+	const close_code = get_bracket_close(open_code);
+	let brackets = 1;
+	let i = index;
+	while (brackets > 0 && i < template.length) {
+		const char = template[i];
+		switch (char) {
+			case "'":
+			case '"':
+			case '`':
+				i = find_string_end(template, i + 1, char) + 1;
+				continue;
+			case '/': {
+				const next_char = template[i + 1];
+				if (!next_char) continue;
+				if (next_char === '/') {
+					i = infinity_if_negative(template.indexOf('\n', i + 1)) + '\n'.length;
+					continue;
+				}
+				if (next_char === '*') {
+					i = infinity_if_negative(template.indexOf('*/', i + 1)) + '*/'.length;
+					continue;
+				}
+				i = find_regex_end(template, i + 1) + '/'.length;
+				continue;
+			}
+			default: {
+				const code = full_char_code_at(template, i);
+				if (code === open_code) {
+					brackets++;
+				} else if (code === close_code) {
+					brackets--;
+				}
+				if (brackets === 0) {
+					return i;
+				}
+				i++;
+			}
+		}
+	}
+	return undefined;
+}

packages/svelte/tests/parser-legacy/samples/loose-invalid-expression/input.svelte

@@ -6,3 +6,6 @@
 <Component onclick={() => x.} />
 
 <input bind:value={a.} />
+
+asd{a.}asd
+{foo[bar.]}

packages/svelte/tests/parser-legacy/samples/loose-invalid-expression/output.json

@@ -2,7 +2,7 @@
 	"html": {
 		"type": "Fragment",
 		"start": 0,
-		"end": 140,
+		"end": 164,
 		"children": [
 			{
 				"type": "Element",
@@ -200,6 +200,42 @@
 					}
 				],
 				"children": []
+			},
+			{
+				"type": "Text",
+				"start": 140,
+				"end": 145,
+				"raw": "\n\nasd",
+				"data": "\n\nasd"
+			},
+			{
+				"type": "MustacheTag",
+				"start": 145,
+				"end": 149,
+				"expression": {
+					"type": "Identifier",
+					"start": 146,
+					"end": 148,
+					"name": ""
+				}
+			},
+			{
+				"type": "Text",
+				"start": 149,
+				"end": 153,
+				"raw": "asd\n",
+				"data": "asd\n"
+			},
+			{
+				"type": "MustacheTag",
+				"start": 153,
+				"end": 164,
+				"expression": {
+					"type": "Identifier",
+					"start": 154,
+					"end": 163,
+					"name": ""
+				}
 			}
 		]
 	}

packages/svelte/tests/parser-modern/samples/loose-invalid-expression/input.svelte

@@ -6,3 +6,6 @@
 <Component onclick={() => x.} />
 
 <input bind:value={a.} />
+
+asd{a.}asd
+{foo[bar.]}

packages/svelte/tests/parser-modern/samples/loose-invalid-expression/output.json

@@ -2,7 +2,7 @@
 	"css": null,
 	"js": [],
 	"start": 0,
-	"end": 140,
+	"end": 164,
 	"type": "Root",
 	"fragment": {
 		"type": "Fragment",
@@ -211,6 +211,42 @@
 					"type": "Fragment",
 					"nodes": []
 				}
+			},
+			{
+				"type": "Text",
+				"start": 140,
+				"end": 145,
+				"raw": "\n\nasd",
+				"data": "\n\nasd"
+			},
+			{
+				"type": "ExpressionTag",
+				"start": 145,
+				"end": 149,
+				"expression": {
+					"type": "Identifier",
+					"start": 146,
+					"end": 148,
+					"name": ""
+				}
+			},
+			{
+				"type": "Text",
+				"start": 149,
+				"end": 153,
+				"raw": "asd\n",
+				"data": "asd\n"
+			},
+			{
+				"type": "ExpressionTag",
+				"start": 153,
+				"end": 164,
+				"expression": {
+					"type": "Identifier",
+					"start": 154,
+					"end": 163,
+					"name": ""
+				}
 			}
 		]
 	},