svelte-on-rails 0.0.1

data/lib/mount_svelte/node_modules/svelte/src/compiler/phases/1-parse/state/element.js

@@ -0,0 +1,823 @@
+/** @import { Expression } from 'estree' */
+/** @import { AST } from '#compiler' */
+/** @import { Parser } from '../index.js' */
+import { is_void } from '../../../../utils.js';
+import read_expression from '../read/expression.js';
+import { read_script } from '../read/script.js';
+import read_style from '../read/style.js';
+import { decode_character_references } from '../utils/html.js';
+import * as e from '../../../errors.js';
+import * as w from '../../../warnings.js';
+import { create_fragment } from '../utils/create.js';
+import { create_attribute, create_expression_metadata, is_element_node } from '../../nodes.js';
+import { get_attribute_expression, is_expression_attribute } from '../../../utils/ast.js';
+import { closing_tag_omitted } from '../../../../html-tree-validation.js';
+import { list } from '../../../utils/string.js';
+import { regex_whitespace } from '../../patterns.js';
+
+const regex_invalid_unquoted_attribute_value = /^(\/>|[\s"'=<>`])/;
+const regex_closing_textarea_tag = /^<\/textarea(\s[^>]*)?>/i;
+const regex_closing_comment = /-->/;
+const regex_whitespace_or_slash_or_closing_tag = /(\s|\/|>)/;
+const regex_token_ending_character = /[\s=/>"']/;
+const regex_starts_with_quote_characters = /^["']/;
+const regex_attribute_value = /^(?:"([^"]*)"|'([^'])*'|([^>\s]+))/;
+const regex_valid_element_name =
+	/^(?:![a-zA-Z]+|[a-zA-Z](?:[a-zA-Z0-9-]*[a-zA-Z0-9])?|[a-zA-Z][a-zA-Z0-9]*:[a-zA-Z][a-zA-Z0-9-]*[a-zA-Z0-9])$/;
+export const regex_valid_component_name =
+	// https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Lexical_grammar#identifiers adjusted for our needs
+	// (must start with uppercase letter if no dots, can contain dots)
+	/^(?:\p{Lu}[$\u200c\u200d\p{ID_Continue}.]*|\p{ID_Start}[$\u200c\u200d\p{ID_Continue}]*(?:\.[$\u200c\u200d\p{ID_Continue}]+)+)$/u;
+
+/** @type {Map<string, AST.ElementLike['type']>} */
+const root_only_meta_tags = new Map([
+	['svelte:head', 'SvelteHead'],
+	['svelte:options', 'SvelteOptions'],
+	['svelte:window', 'SvelteWindow'],
+	['svelte:document', 'SvelteDocument'],
+	['svelte:body', 'SvelteBody']
+]);
+
+/** @type {Map<string, AST.ElementLike['type']>} */
+const meta_tags = new Map([
+	...root_only_meta_tags,
+	['svelte:element', 'SvelteElement'],
+	['svelte:component', 'SvelteComponent'],
+	['svelte:self', 'SvelteSelf'],
+	['svelte:fragment', 'SvelteFragment'],
+	['svelte:boundary', 'SvelteBoundary']
+]);
+
+/** @param {Parser} parser */
+export default function element(parser) {
+	const start = parser.index++;
+
+	let parent = parser.current();
+
+	if (parser.eat('!--')) {
+		const data = parser.read_until(regex_closing_comment);
+		parser.eat('-->', true);
+
+		parser.append({
+			type: 'Comment',
+			start,
+			end: parser.index,
+			data
+		});
+
+		return;
+	}
+
+	const is_closing_tag = parser.eat('/');
+	const name = parser.read_until(regex_whitespace_or_slash_or_closing_tag);
+
+	if (is_closing_tag) {
+		parser.allow_whitespace();
+		parser.eat('>', true);
+
+		if (is_void(name)) {
+			e.void_element_invalid_content(start);
+		}
+
+		// close any elements that don't have their own closing tags, e.g. <div><p></div>
+		while (/** @type {AST.RegularElement} */ (parent).name !== name) {
+			if (parser.loose) {
+				// If the previous element did interpret the next opening tag as an attribute, backtrack
+				if (is_element_node(parent)) {
+					const last = parent.attributes.at(-1);
+					if (last?.type === 'Attribute' && last.name === `<${name}`) {
+						parser.index = last.start;
+						parent.attributes.pop();
+						break;
+					}
+				}
+			}
+
+			if (parent.type !== 'RegularElement' && !parser.loose) {
+				if (parser.last_auto_closed_tag && parser.last_auto_closed_tag.tag === name) {
+					e.element_invalid_closing_tag_autoclosed(start, name, parser.last_auto_closed_tag.reason);
+				} else {
+					e.element_invalid_closing_tag(start, name);
+				}
+			}
+
+			parent.end = start;
+			parser.pop();
+
+			parent = parser.current();
+		}
+
+		parent.end = parser.index;
+		parser.pop();
+
+		if (parser.last_auto_closed_tag && parser.stack.length < parser.last_auto_closed_tag.depth) {
+			parser.last_auto_closed_tag = undefined;
+		}
+
+		return;
+	}
+
+	if (name.startsWith('svelte:') && !meta_tags.has(name)) {
+		const bounds = { start: start + 1, end: start + 1 + name.length };
+		e.svelte_meta_invalid_tag(bounds, list(Array.from(meta_tags.keys())));
+	}
+
+	if (!regex_valid_element_name.test(name) && !regex_valid_component_name.test(name)) {
+		// <div. -> in the middle of typing -> allow in loose mode
+		if (!parser.loose || !name.endsWith('.')) {
+			const bounds = { start: start + 1, end: start + 1 + name.length };
+			e.tag_invalid_name(bounds);
+		}
+	}
+
+	if (root_only_meta_tags.has(name)) {
+		if (name in parser.meta_tags) {
+			e.svelte_meta_duplicate(start, name);
+		}
+
+		if (parent.type !== 'Root') {
+			e.svelte_meta_invalid_placement(start, name);
+		}
+
+		parser.meta_tags[name] = true;
+	}
+
+	const type = meta_tags.has(name)
+		? meta_tags.get(name)
+		: regex_valid_component_name.test(name) || (parser.loose && name.endsWith('.'))
+			? 'Component'
+			: name === 'title' && parent_is_head(parser.stack)
+				? 'TitleElement'
+				: // TODO Svelte 6/7: once slots are removed in favor of snippets, always keep slot as a regular element
+					name === 'slot' && !parent_is_shadowroot_template(parser.stack)
+					? 'SlotElement'
+					: 'RegularElement';
+
+	/** @type {AST.ElementLike} */
+	const element =
+		type === 'RegularElement'
+			? {
+					type,
+					start,
+					end: -1,
+					name,
+					attributes: [],
+					fragment: create_fragment(true),
+					metadata: {
+						svg: false,
+						mathml: false,
+						scoped: false,
+						has_spread: false,
+						path: []
+					}
+				}
+			: /** @type {AST.ElementLike} */ ({
+					type,
+					start,
+					end: -1,
+					name,
+					attributes: [],
+					fragment: create_fragment(true),
+					metadata: {
+						// unpopulated at first, differs between types
+					}
+				});
+
+	parser.allow_whitespace();
+
+	if (parent.type === 'RegularElement' && closing_tag_omitted(parent.name, name)) {
+		parent.end = start;
+		parser.pop();
+		parser.last_auto_closed_tag = {
+			tag: parent.name,
+			reason: name,
+			depth: parser.stack.length
+		};
+	}
+
+	/** @type {string[]} */
+	const unique_names = [];
+
+	const current = parser.current();
+	const is_top_level_script_or_style =
+		(name === 'script' || name === 'style') && current.type === 'Root';
+
+	const read = is_top_level_script_or_style ? read_static_attribute : read_attribute;
+
+	let attribute;
+	while ((attribute = read(parser))) {
+		// animate and transition can only be specified once per element so no need
+		// to check here, use can be used multiple times, same for the on directive
+		// finally let already has error handling in case of duplicate variable names
+		if (
+			attribute.type === 'Attribute' ||
+			attribute.type === 'BindDirective' ||
+			attribute.type === 'StyleDirective' ||
+			attribute.type === 'ClassDirective'
+		) {
+			// `bind:attribute` and `attribute` are just the same but `class:attribute`,
+			// `style:attribute` and `attribute` are different and should be allowed together
+			// so we concatenate the type while normalizing the type for BindDirective
+			const type = attribute.type === 'BindDirective' ? 'Attribute' : attribute.type;
+			if (unique_names.includes(type + attribute.name)) {
+				e.attribute_duplicate(attribute);
+				// <svelte:element bind:this this=..> is allowed
+			} else if (attribute.name !== 'this') {
+				unique_names.push(type + attribute.name);
+			}
+		}
+
+		element.attributes.push(attribute);
+		parser.allow_whitespace();
+	}
+
+	if (element.type === 'SvelteComponent') {
+		const index = element.attributes.findIndex(
+			/** @param {any} attr */
+			(attr) => attr.type === 'Attribute' && attr.name === 'this'
+		);
+		if (index === -1) {
+			e.svelte_component_missing_this(start);
+		}
+
+		const definition = /** @type {AST.Attribute} */ (element.attributes.splice(index, 1)[0]);
+		if (!is_expression_attribute(definition)) {
+			e.svelte_component_invalid_this(definition.start);
+		}
+
+		element.expression = get_attribute_expression(definition);
+	}
+
+	if (element.type === 'SvelteElement') {
+		const index = element.attributes.findIndex(
+			/** @param {any} attr */
+			(attr) => attr.type === 'Attribute' && attr.name === 'this'
+		);
+		if (index === -1) {
+			e.svelte_element_missing_this(start);
+		}
+
+		const definition = /** @type {AST.Attribute} */ (element.attributes.splice(index, 1)[0]);
+
+		if (definition.value === true) {
+			e.svelte_element_missing_this(definition);
+		}
+
+		if (!is_expression_attribute(definition)) {
+			w.svelte_element_invalid_this(definition);
+
+			// note that this is wrong, in the case of e.g. `this="h{n}"` — it will result in `<h>`.
+			// it would be much better to just error here, but we are preserving the existing buggy
+			// Svelte 4 behaviour out of an overabundance of caution regarding breaking changes.
+			// TODO in 6.0, error
+			const chunk = /** @type {Array<AST.ExpressionTag | AST.Text>} */ (definition.value)[0];
+			element.tag =
+				chunk.type === 'Text'
+					? {
+							type: 'Literal',
+							value: chunk.data,
+							raw: `'${chunk.raw}'`,
+							start: chunk.start,
+							end: chunk.end
+						}
+					: chunk.expression;
+		} else {
+			element.tag = get_attribute_expression(definition);
+		}
+	}
+
+	if (is_top_level_script_or_style) {
+		parser.eat('>', true);
+
+		/** @type {AST.Comment | null} */
+		let prev_comment = null;
+		for (let i = current.fragment.nodes.length - 1; i >= 0; i--) {
+			const node = current.fragment.nodes[i];
+
+			if (i === current.fragment.nodes.length - 1 && node.end !== start) {
+				break;
+			}
+
+			if (node.type === 'Comment') {
+				prev_comment = node;
+				break;
+			} else if (node.type !== 'Text' || node.data.trim()) {
+				break;
+			}
+		}
+
+		if (name === 'script') {
+			const content = read_script(parser, start, element.attributes);
+			if (prev_comment) {
+				// We take advantage of the fact that the root will never have leadingComments set,
+				// and set the previous comment to it so that the warning mechanism can later
+				// inspect the root and see if there was a html comment before it silencing specific warnings.
+				content.content.leadingComments = [{ type: 'Line', value: prev_comment.data }];
+			}
+
+			if (content.context === 'module') {
+				if (current.module) e.script_duplicate(start);
+				current.module = content;
+			} else {
+				if (current.instance) e.script_duplicate(start);
+				current.instance = content;
+			}
+		} else {
+			const content = read_style(parser, start, element.attributes);
+			content.content.comment = prev_comment;
+
+			if (current.css) e.style_duplicate(start);
+			current.css = content;
+		}
+		return;
+	}
+
+	parser.append(element);
+
+	const self_closing = parser.eat('/') || is_void(name);
+	const closed = parser.eat('>', true, false);
+
+	// Loose parsing mode
+	if (!closed) {
+		// We may have eaten an opening `<` of the next element and treated it as an attribute...
+		const last = element.attributes.at(-1);
+		if (last?.type === 'Attribute' && last.name === '<') {
+			parser.index = last.start;
+			element.attributes.pop();
+		} else {
+			// ... or we may have eaten part of a following block ...
+			const prev_1 = parser.template[parser.index - 1];
+			const prev_2 = parser.template[parser.index - 2];
+			const current = parser.template[parser.index];
+			if (prev_2 === '{' && prev_1 === '/') {
+				parser.index -= 2;
+			} else if (prev_1 === '{' && (current === '#' || current === '@' || current === ':')) {
+				parser.index -= 1;
+			} else {
+				// ... or we're followed by whitespace, for example near the end of the template,
+				// which we want to take in so that language tools has more room to work with
+				parser.allow_whitespace();
+				if (parser.index === parser.template.length) {
+					while (
+						parser.index < parser.template_untrimmed.length &&
+						regex_whitespace.test(parser.template_untrimmed[parser.index])
+					) {
+						parser.index++;
+					}
+				}
+			}
+		}
+	}
+
+	if (self_closing || !closed) {
+		// don't push self-closing elements onto the stack
+		element.end = parser.index;
+	} else if (name === 'textarea') {
+		// special case
+		element.fragment.nodes = read_sequence(
+			parser,
+			() => regex_closing_textarea_tag.test(parser.template.slice(parser.index)),
+			'inside <textarea>'
+		);
+		parser.read(regex_closing_textarea_tag);
+		element.end = parser.index;
+	} else if (name === 'script' || name === 'style') {
+		// special case
+		const start = parser.index;
+		const data = parser.read_until(new RegExp(`</${name}>`));
+		const end = parser.index;
+
+		/** @type {AST.Text} */
+		const node = {
+			start,
+			end,
+			type: 'Text',
+			data,
+			raw: data
+		};
+
+		element.fragment.nodes.push(node);
+		parser.eat(`</${name}>`, true);
+		element.end = parser.index;
+	} else {
+		parser.stack.push(element);
+		parser.fragments.push(element.fragment);
+	}
+}
+
+/** @param {AST.TemplateNode[]} stack */
+function parent_is_head(stack) {
+	let i = stack.length;
+	while (i--) {
+		const { type } = stack[i];
+		if (type === 'SvelteHead') return true;
+		if (type === 'RegularElement' || type === 'Component') return false;
+	}
+	return false;
+}
+
+/** @param {AST.TemplateNode[]} stack */
+function parent_is_shadowroot_template(stack) {
+	// https://developer.chrome.com/docs/css-ui/declarative-shadow-dom#building_a_declarative_shadow_root
+	let i = stack.length;
+	while (i--) {
+		if (
+			stack[i].type === 'RegularElement' &&
+			/** @type {AST.RegularElement} */ (stack[i]).attributes.some(
+				(a) => a.type === 'Attribute' && a.name === 'shadowrootmode'
+			)
+		) {
+			return true;
+		}
+	}
+	return false;
+}
+
+/**
+ * @param {Parser} parser
+ * @returns {AST.Attribute | null}
+ */
+function read_static_attribute(parser) {
+	const start = parser.index;
+
+	const name = parser.read_until(regex_token_ending_character);
+	if (!name) return null;
+
+	/** @type {true | Array<AST.Text | AST.ExpressionTag>} */
+	let value = true;
+
+	if (parser.eat('=')) {
+		parser.allow_whitespace();
+		let raw = parser.match_regex(regex_attribute_value);
+		if (!raw) {
+			e.expected_attribute_value(parser.index);
+		}
+
+		parser.index += raw.length;
+
+		const quoted = raw[0] === '"' || raw[0] === "'";
+		if (quoted) {
+			raw = raw.slice(1, -1);
+		}
+
+		value = [
+			{
+				start: parser.index - raw.length - (quoted ? 1 : 0),
+				end: quoted ? parser.index - 1 : parser.index,
+				type: 'Text',
+				raw: raw,
+				data: decode_character_references(raw, true)
+			}
+		];
+	}
+
+	if (parser.match_regex(regex_starts_with_quote_characters)) {
+		e.expected_token(parser.index, '=');
+	}
+
+	return create_attribute(name, start, parser.index, value);
+}
+
+/**
+ * @param {Parser} parser
+ * @returns {AST.Attribute | AST.SpreadAttribute | AST.Directive | null}
+ */
+function read_attribute(parser) {
+	const start = parser.index;
+
+	if (parser.eat('{')) {
+		parser.allow_whitespace();
+
+		if (parser.eat('...')) {
+			const expression = read_expression(parser);
+
+			parser.allow_whitespace();
+			parser.eat('}', true);
+
+			/** @type {AST.SpreadAttribute} */
+			const spread = {
+				type: 'SpreadAttribute',
+				start,
+				end: parser.index,
+				expression,
+				metadata: {
+					expression: create_expression_metadata()
+				}
+			};
+
+			return spread;
+		} else {
+			const value_start = parser.index;
+			let name = parser.read_identifier();
+
+			if (name === null) {
+				if (
+					parser.loose &&
+					(parser.match('#') || parser.match('/') || parser.match('@') || parser.match(':'))
+				) {
+					// We're likely in an unclosed opening tag and did read part of a block.
+					// Return null to not crash the parser so it can continue with closing the tag.
+					return null;
+				} else if (parser.loose && parser.match('}')) {
+					// Likely in the middle of typing, just created the shorthand
+					name = '';
+				} else {
+					e.attribute_empty_shorthand(start);
+				}
+			}
+
+			parser.allow_whitespace();
+			parser.eat('}', true);
+
+			/** @type {AST.ExpressionTag} */
+			const expression = {
+				type: 'ExpressionTag',
+				start: value_start,
+				end: value_start + name.length,
+				expression: {
+					start: value_start,
+					end: value_start + name.length,
+					type: 'Identifier',
+					name
+				},
+				metadata: {
+					expression: create_expression_metadata()
+				}
+			};
+
+			return create_attribute(name, start, parser.index, expression);
+		}
+	}
+
+	const name = parser.read_until(regex_token_ending_character);
+	if (!name) return null;
+
+	let end = parser.index;
+
+	parser.allow_whitespace();
+
+	const colon_index = name.indexOf(':');
+	const type = colon_index !== -1 && get_directive_type(name.slice(0, colon_index));
+
+	/** @type {true | AST.ExpressionTag | Array<AST.Text | AST.ExpressionTag>} */
+	let value = true;
+	if (parser.eat('=')) {
+		parser.allow_whitespace();
+
+		if (parser.template[parser.index] === '/' && parser.template[parser.index + 1] === '>') {
+			const char_start = parser.index;
+			parser.index++; // consume '/'
+			value = [
+				{
+					start: char_start,
+					end: char_start + 1,
+					type: 'Text',
+					raw: '/',
+					data: '/'
+				}
+			];
+			end = parser.index;
+		} else {
+			value = read_attribute_value(parser);
+			end = parser.index;
+		}
+	} else if (parser.match_regex(regex_starts_with_quote_characters)) {
+		e.expected_token(parser.index, '=');
+	}
+
+	if (type) {
+		const [directive_name, ...modifiers] = name.slice(colon_index + 1).split('|');
+
+		if (directive_name === '') {
+			e.directive_missing_name({ start, end: start + colon_index + 1 }, name);
+		}
+
+		if (type === 'StyleDirective') {
+			return {
+				start,
+				end,
+				type,
+				name: directive_name,
+				modifiers: /** @type {Array<'important'>} */ (modifiers),
+				value,
+				metadata: {
+					expression: create_expression_metadata()
+				}
+			};
+		}
+
+		const first_value = value === true ? undefined : Array.isArray(value) ? value[0] : value;
+
+		/** @type {Expression | null} */
+		let expression = null;
+
+		if (first_value) {
+			const attribute_contains_text =
+				/** @type {any[]} */ (value).length > 1 || first_value.type === 'Text';
+			if (attribute_contains_text) {
+				e.directive_invalid_value(/** @type {number} */ (first_value.start));
+			} else {
+				// TODO throw a parser error in a future version here if this `[ExpressionTag]` instead of `ExpressionTag`,
+				// which means stringified value, which isn't allowed for some directives?
+				expression = first_value.expression;
+			}
+		}
+
+		/** @type {AST.Directive} */
+		const directive = {
+			start,
+			end,
+			type,
+			name: directive_name,
+			expression,
+			metadata: {
+				expression: create_expression_metadata()
+			}
+		};
+
+		// @ts-expect-error we do this separately from the declaration to avoid upsetting typescript
+		directive.modifiers = modifiers;
+
+		if (directive.type === 'TransitionDirective') {
+			const direction = name.slice(0, colon_index);
+			directive.intro = direction === 'in' || direction === 'transition';
+			directive.outro = direction === 'out' || direction === 'transition';
+		}
+
+		// Directive name is expression, e.g. <p class:isRed />
+		if (
+			(directive.type === 'BindDirective' || directive.type === 'ClassDirective') &&
+			!directive.expression
+		) {
+			directive.expression = /** @type {any} */ ({
+				start: start + colon_index + 1,
+				end,
+				type: 'Identifier',
+				name: directive.name
+			});
+		}
+
+		return directive;
+	}
+
+	return create_attribute(name, start, end, value);
+}
+
+/**
+ * @param {string} name
+ * @returns {any}
+ */
+function get_directive_type(name) {
+	if (name === 'use') return 'UseDirective';
+	if (name === 'animate') return 'AnimateDirective';
+	if (name === 'bind') return 'BindDirective';
+	if (name === 'class') return 'ClassDirective';
+	if (name === 'style') return 'StyleDirective';
+	if (name === 'on') return 'OnDirective';
+	if (name === 'let') return 'LetDirective';
+	if (name === 'in' || name === 'out' || name === 'transition') return 'TransitionDirective';
+	return false;
+}
+
+/**
+ * @param {Parser} parser
+ * @return {AST.ExpressionTag | Array<AST.ExpressionTag | AST.Text>}
+ */
+function read_attribute_value(parser) {
+	const quote_mark = parser.eat("'") ? "'" : parser.eat('"') ? '"' : null;
+	if (quote_mark && parser.eat(quote_mark)) {
+		return [
+			{
+				start: parser.index - 1,
+				end: parser.index - 1,
+				type: 'Text',
+				raw: '',
+				data: ''
+			}
+		];
+	}
+
+	/** @type {Array<AST.ExpressionTag | AST.Text>} */
+	let value;
+	try {
+		value = read_sequence(
+			parser,
+			() => {
+				// handle common case of quote marks existing outside of regex for performance reasons
+				if (quote_mark) return parser.match(quote_mark);
+				return !!parser.match_regex(regex_invalid_unquoted_attribute_value);
+			},
+			'in attribute value'
+		);
+	} catch (/** @type {any} */ error) {
+		if (error.code === 'js_parse_error') {
+			// if the attribute value didn't close + self-closing tag
+			// eg: `<Component test={{a:1} />`
+			// acorn may throw a `Unterminated regular expression` because of `/>`
+			const pos = error.position?.[0];
+			if (pos !== undefined && parser.template.slice(pos - 1, pos + 1) === '/>') {
+				parser.index = pos;
+				e.expected_token(pos, quote_mark || '}');
+			}
+		}
+		throw error;
+	}
+
+	if (value.length === 0 && !quote_mark) {
+		e.expected_attribute_value(parser.index);
+	}
+
+	if (quote_mark) parser.index += 1;
+
+	if (quote_mark || value.length > 1 || value[0].type === 'Text') {
+		return value;
+	} else {
+		return value[0];
+	}
+}
+
+/**
+ * @param {Parser} parser
+ * @param {() => boolean} done
+ * @param {string} location
+ * @returns {any[]}
+ */
+function read_sequence(parser, done, location) {
+	/** @type {AST.Text} */
+	let current_chunk = {
+		start: parser.index,
+		end: -1,
+		type: 'Text',
+		raw: '',
+		data: ''
+	};
+
+	/** @type {Array<AST.Text | AST.ExpressionTag>} */
+	const chunks = [];
+
+	/** @param {number} end */
+	function flush(end) {
+		if (current_chunk.raw) {
+			current_chunk.data = decode_character_references(current_chunk.raw, true);
+			current_chunk.end = end;
+			chunks.push(current_chunk);
+		}
+	}
+
+	while (parser.index < parser.template.length) {
+		const index = parser.index;
+
+		if (done()) {
+			flush(parser.index);
+			return chunks;
+		} else if (parser.eat('{')) {
+			if (parser.match('#')) {
+				const index = parser.index - 1;
+				parser.eat('#');
+				const name = parser.read_until(/[^a-z]/);
+				e.block_invalid_placement(index, name, location);
+			} else if (parser.match('@')) {
+				const index = parser.index - 1;
+				parser.eat('@');
+				const name = parser.read_until(/[^a-z]/);
+				e.tag_invalid_placement(index, name, location);
+			}
+
+			flush(parser.index - 1);
+
+			parser.allow_whitespace();
+			const expression = read_expression(parser);
+			parser.allow_whitespace();
+			parser.eat('}', true);
+
+			/** @type {AST.ExpressionTag} */
+			const chunk = {
+				type: 'ExpressionTag',
+				start: index,
+				end: parser.index,
+				expression,
+				metadata: {
+					expression: create_expression_metadata()
+				}
+			};
+
+			chunks.push(chunk);
+
+			current_chunk = {
+				start: parser.index,
+				end: -1,
+				type: 'Text',
+				raw: '',
+				data: ''
+			};
+		} else {
+			current_chunk.raw += parser.template[parser.index++];
+		}
+	}
+
+	if (parser.loose) {
+		return chunks;
+	} else {
+		e.unexpected_eof(parser.template.length);
+	}
+}