X-Git-Url: https://git.ayoreis.com/relax.git/blobdiff_plain/d7baaea4f90491d77b416fbf53935e090bbf7f7f..18760afc494c80d43ef455b9ef21859069936963:/relax/url-pattern-pattern-string-parser.ts?ds=inline

diff --git a/relax/url-pattern-pattern-string-parser.ts b/relax/url-pattern-pattern-string-parser.ts
index e69de29..662625c 100644
--- a/relax/url-pattern-pattern-string-parser.ts
+++ b/relax/url-pattern-pattern-string-parser.ts
@@ -0,0 +1,615 @@
+// https://tc39.es/ecma262/
+
+/** https://tc39.es/ecma262/#prod-IdentifierStart */
+const IDENTIFIER_START =
+	/^(?:\p{ID_Start}|\$|_|\\u(?:\p{ASCII_Hex_Digit}{4}|\{(?:0?\p{ASCII_Hex_Digit}{1,5}|10\p{ASCII_Hex_Digit}{4})\}))$/u;
+/** https://tc39.es/ecma262/#prod-IdentifierPart */
+const IDENTIFIER_PART =
+	/^(?:\p{ID_Continue}|\$|\\u(?:\p{ASCII_Hex_Digit}{4}|\{(?:0?\p{ASCII_Hex_Digit}{1,5}|10\p{ASCII_Hex_Digit}{4})\}))$/u;
+
+// https://infra.spec.whatwg.org
+
+/** https://infra.spec.whatwg.org/#ascii-code-point */
+const ASCII_CODE_POINT = /^\p{ASCII}$/u;
+
+// https://urlpattern.spec.whatwg.org
+
+/** https://urlpattern.spec.whatwg.org/#token */
+class Token {
+	/** https://urlpattern.spec.whatwg.org/#token-type */
+	type: TokenType = "invalid-char";
+	/** https://urlpattern.spec.whatwg.org/#token-index */
+	index = 0;
+	/** https://urlpattern.spec.whatwg.org/#token-value */
+	value = "";
+}
+
+/** https://urlpattern.spec.whatwg.org/#token-type */
+type TokenType =
+	| "open"
+	| "close"
+	| "regexp"
+	| "name"
+	| "char"
+	| "escaped-char"
+	| "other-modifier"
+	| "asterisk"
+	| "end"
+	| "invalid-char";
+
+/** https://urlpattern.spec.whatwg.org/#tokenize-policy */
+type TokenizePolicy = "strict" | "lenient";
+
+/** https://urlpattern.spec.whatwg.org/#tokenizer */
+class Tokenizer {
+	/** https://urlpattern.spec.whatwg.org/#tokenizer-input */
+	input: string[] = [];
+	/** https://urlpattern.spec.whatwg.org/#tokenizer-policy */
+	policy: TokenizePolicy = "strict";
+	/** https://urlpattern.spec.whatwg.org/#tokenizer-token-list */
+	readonly token_list: Token[] = [];
+	/** https://urlpattern.spec.whatwg.org/#tokenizer-index */
+	index = 0;
+	/** https://urlpattern.spec.whatwg.org/#tokenizer-next-index */
+	next_index = 0;
+	/** https://urlpattern.spec.whatwg.org/#tokenizer-code-point */
+	code_point: string | null = null;
+
+	/** https://urlpattern.spec.whatwg.org/#get-the-next-code-point */
+	get_the_next_code_point() {
+		this.code_point = this.input[this.next_index]!;
+		this.next_index++;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#seek-and-get-the-next-code-point */
+	seek_and_get_the_next_code_point(index: number) {
+		this.next_index = index;
+		this.get_the_next_code_point();
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#add-a-token */
+	add_a_token(
+		type: TokenType,
+		next_position: number,
+		value_position: number,
+		value_length: number,
+	) {
+		const token = new Token();
+		token.type = type;
+		token.index = this.index;
+		token.value = this.input.slice(
+			value_position,
+			value_position + value_length,
+		).join("");
+		this.token_list.push(token);
+		this.index = next_position;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#add-a-token-with-default-length */
+	add_a_token_with_default_length(
+		type: TokenType,
+		next_position: number,
+		value_position: number,
+	) {
+		const computed_length = next_position - value_position;
+		this.add_a_token(type, next_position, value_position, computed_length);
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#add-a-token-with-default-position-and-length */
+	add_a_token_with_default_position_and_length(type: TokenType) {
+		this.add_a_token_with_default_length(type, this.next_index, this.index);
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#process-a-tokenizing-error */
+	process_a_tokenizing_error(next_position: number, value_position: number) {
+		if (this.policy === "strict") throw new TypeError();
+		this.add_a_token_with_default_length(
+			"invalid-char",
+			next_position,
+			value_position,
+		);
+	}
+}
+
+/** https://urlpattern.spec.whatwg.org/#tokenize */
+function tokenize(input: string, policy: TokenizePolicy) {
+	const tokenizer = new Tokenizer();
+	tokenizer.input = Array.from(input);
+	tokenizer.policy = policy;
+
+	while (tokenizer.index < tokenizer.input.length) {
+		tokenizer.seek_and_get_the_next_code_point(tokenizer.index);
+
+		if (tokenizer.code_point === "*") {
+			tokenizer.add_a_token_with_default_position_and_length("asterisk");
+			continue;
+		}
+
+		if (tokenizer.code_point === "+" || tokenizer.code_point === "?") {
+			tokenizer.add_a_token_with_default_position_and_length("other-modifier");
+			continue;
+		}
+
+		if (tokenizer.code_point === "\\") {
+			if (tokenizer.index === tokenizer.input.length - 1) {
+				tokenizer.process_a_tokenizing_error(
+					tokenizer.next_index,
+					tokenizer.index,
+				);
+				continue;
+			}
+
+			const escaped_index = tokenizer.next_index;
+			tokenizer.get_the_next_code_point();
+			tokenizer.add_a_token_with_default_length(
+				"escaped-char",
+				tokenizer.next_index,
+				escaped_index,
+			);
+			continue;
+		}
+
+		if (tokenizer.code_point === "{") {
+			tokenizer.add_a_token_with_default_position_and_length("open");
+			continue;
+		}
+
+		if (tokenizer.code_point === "}") {
+			tokenizer.add_a_token_with_default_position_and_length("close");
+			continue;
+		}
+
+		if (tokenizer.code_point === ":") {
+			let name_position = tokenizer.next_index;
+			const name_start = name_position;
+
+			while (name_position < tokenizer.input.length) {
+				tokenizer.seek_and_get_the_next_code_point(name_position);
+				const first_code_point = name_position === name_start;
+				const valid_code_point = is_a_valid_name_code_point(
+					tokenizer.code_point,
+					first_code_point,
+				);
+				if (!valid_code_point) break;
+				name_position = tokenizer.next_index;
+			}
+
+			if (name_position <= name_start) {
+				tokenizer.process_a_tokenizing_error(name_start, tokenizer.index);
+				continue;
+			}
+
+			tokenizer.add_a_token_with_default_length(
+				"name",
+				name_position,
+				name_start,
+			);
+			continue;
+		}
+
+		if (tokenizer.code_point === "(") {
+			let depth = 1;
+			let regexp_position = tokenizer.next_index;
+			const regexp_start = regexp_position;
+			let error = false;
+
+			while (regexp_position < tokenizer.input.length) {
+				tokenizer.seek_and_get_the_next_code_point(regexp_position);
+
+				if (!ASCII_CODE_POINT.test(tokenizer.code_point)) {
+					tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+					error = true;
+					break;
+				}
+
+				if (
+					regexp_position === regexp_start &&
+					tokenizer.code_point as string === "?"
+				) {
+					tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+					error = true;
+					break;
+				}
+
+				if (tokenizer.code_point as string === "\\") {
+					if (regexp_position === tokenizer.input.length - 1) {
+						tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+						error = true;
+						break;
+					}
+
+					tokenizer.get_the_next_code_point();
+
+					if (!ASCII_CODE_POINT.test(tokenizer.code_point)) {
+						tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+						error = true;
+						break;
+					}
+
+					regexp_position = tokenizer.next_index;
+					continue;
+				}
+
+				if (tokenizer.code_point as string === ")") {
+					depth--;
+
+					if (depth === 0) {
+						regexp_position = tokenizer.next_index;
+						break;
+					}
+				} else if (tokenizer.code_point === "(") {
+					depth++;
+
+					if (regexp_position === tokenizer.input.length - 1) {
+						tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+						error = true;
+						break;
+					}
+
+					const temporary_position = tokenizer.next_index;
+					tokenizer.get_the_next_code_point();
+
+					if (tokenizer.code_point as string !== "?") {
+						tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+						error = true;
+						break;
+					}
+
+					tokenizer.next_index = temporary_position;
+				}
+
+				regexp_position = tokenizer.next_index;
+			}
+
+			if (error) continue;
+
+			if (depth !== 0) {
+				tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+				continue;
+			}
+
+			const regexp_length = regexp_position - regexp_start - 1;
+
+			if (regexp_length === 0) {
+				tokenizer.process_a_tokenizing_error(regexp_start, tokenizer.index);
+				continue;
+			}
+
+			tokenizer.add_a_token(
+				"regexp",
+				regexp_position,
+				regexp_start,
+				regexp_length,
+			);
+			continue;
+		}
+
+		tokenizer.add_a_token_with_default_position_and_length("char");
+	}
+
+	tokenizer.add_a_token_with_default_length(
+		"end",
+		tokenizer.index,
+		tokenizer.index,
+	);
+
+	return tokenizer.token_list;
+}
+
+/** https://urlpattern.spec.whatwg.org/#is-a-valid-name-code-point */
+function is_a_valid_name_code_point(code_point: string, first: boolean) {
+	if (first) return IDENTIFIER_START.test(code_point);
+	return IDENTIFIER_PART.test(code_point);
+}
+
+/** https://urlpattern.spec.whatwg.org/#part */
+export class Part {
+	/** https://urlpattern.spec.whatwg.org/#part-type */
+	readonly type;
+	/** https://urlpattern.spec.whatwg.org/#part-value */
+	readonly value;
+	/** https://urlpattern.spec.whatwg.org/#part-modifier */
+	readonly modifier;
+	/** https://urlpattern.spec.whatwg.org/#part-name */
+	readonly name;
+	/** https://urlpattern.spec.whatwg.org/#part-prefix */
+	readonly prefix;
+	/** https://urlpattern.spec.whatwg.org/#part-suffix */
+	readonly suffix;
+
+	constructor(
+		type: PartType,
+		value: string,
+		modifier: PartModifier,
+		name = "",
+		prefix = "",
+		suffix = "",
+	) {
+		this.type = type;
+		this.value = value;
+		this.modifier = modifier;
+		this.name = name;
+		this.prefix = prefix;
+		this.suffix = suffix;
+	}
+}
+
+/** https://urlpattern.spec.whatwg.org/#part-type */
+type PartType = "fixed-text" | "regexp" | "segment-wildcard" | "full-wildcard";
+/** https://urlpattern.spec.whatwg.org/#part-modifier */
+type PartModifier = "none" | "optional" | "zero-or-more" | "one-or-more";
+
+/** https://urlpattern.spec.whatwg.org/#options */
+class Options {
+	/** https://urlpattern.spec.whatwg.org/#options-delimiter-code-point */
+	readonly delimiter_code_point;
+	/** https://urlpattern.spec.whatwg.org/#options-prefix-code-point */
+	readonly prefix_code_point;
+	/** https://urlpattern.spec.whatwg.org/#options-ignore-case */
+	readonly ignore_case;
+
+	constructor(
+		delimiter_code_point: string,
+		prefix_code_point: string,
+		ignore_case = false,
+	) {
+		this.delimiter_code_point = delimiter_code_point;
+		this.prefix_code_point = prefix_code_point;
+		this.ignore_case = ignore_case;
+	}
+}
+
+/** https://urlpattern.spec.whatwg.org/#default-options */
+export const DEFAULT_OPTIONS = new Options("", "");
+/** https://urlpattern.spec.whatwg.org/#hostname-options */
+export const HOSTNAME_OPTIONS = new Options(".", "");
+/** https://urlpattern.spec.whatwg.org/#pathname-options */
+export const PATHNAME_OPTIONS = new Options("/", "/");
+
+/** https://urlpattern.spec.whatwg.org/#encoding-callback */
+type EncodingCallback = (input: string) => string;
+
+/** https://urlpattern.spec.whatwg.org/#pattern-parser */
+class PatternParser {
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-token-list */
+	token_list: Token[] = [];
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-encoding-callback */
+	readonly encoding_callback;
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-segment-wildcard-regexp */
+	readonly segment_wildcard_regexp;
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-part-list */
+	readonly part_list: Part[] = [];
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-pending-fixed-value */
+	pending_fixed_value = "";
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-index */
+	index = 0;
+	/** https://urlpattern.spec.whatwg.org/#pattern-parser-next-numeric-name */
+	next_numeric_name = 0;
+
+	constructor(
+		encoding_callback: EncodingCallback,
+		segment_wildcard_regexp: string,
+	) {
+		this.encoding_callback = encoding_callback;
+		this.segment_wildcard_regexp = segment_wildcard_regexp;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#try-to-consume-a-token */
+	try_to_consume_a_token(type: TokenType) {
+		const next_token = this.token_list[this.index]!;
+		if (next_token.type !== type) return null;
+		this.index++;
+		return next_token;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#try-to-consume-a-modifier-token */
+	try_to_consume_a_modifier_token() {
+		let token = this.try_to_consume_a_token("other-modifier");
+		if (token !== null) return token;
+		token = this.try_to_consume_a_token("asterisk");
+		return token;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#try-to-consume-a-regexp-or-wildcard-token */
+	try_to_consume_a_regexp_or_wildcard_token(name_token: Token | null) {
+		let token = this.try_to_consume_a_token("regexp");
+
+		if (name_token === null && token === null) {
+			token = this.try_to_consume_a_token("asterisk");
+		}
+
+		return token;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#consume-a-required-token */
+	consume_a_required_token(type: TokenType) {
+		const result = this.try_to_consume_a_token(type);
+		if (result === null) throw new TypeError();
+		return result;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#consume-text */
+	consume_text() {
+		let result = "";
+
+		while (true) {
+			let token = this.try_to_consume_a_token("char");
+			token ??= this.try_to_consume_a_token("escaped-char");
+			if (token === null) break;
+			result += token.value;
+		}
+
+		return result;
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#maybe-add-a-part-from-the-pending-fixed-value */
+	maybe_add_a_part_from_the_pending_fixed_value() {
+		if (this.pending_fixed_value === "") return;
+		const encoded_value = this.encoding_callback(this.pending_fixed_value);
+		this.pending_fixed_value = "";
+		const part = new Part("fixed-text", encoded_value, "none");
+		this.part_list.push(part);
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#add-a-part */
+	add_a_part(
+		prefix: string,
+		name_token: Token | null,
+		regexp_or_wildcard_token: Token | null,
+		suffix: string,
+		modifier_token: Token | null,
+	) {
+		let modifier: PartModifier = "none";
+		if (modifier_token?.value === "?") modifier = "optional";
+		else if (modifier_token?.value === "*") modifier = "zero-or-more";
+		else if (modifier_token?.value === "+") modifier = "one-or-more";
+
+		if (
+			name_token === null && regexp_or_wildcard_token === null &&
+			modifier === "none"
+		) {
+			this.pending_fixed_value += prefix;
+			return;
+		}
+
+		this.maybe_add_a_part_from_the_pending_fixed_value();
+
+		if (name_token === null && regexp_or_wildcard_token === null) {
+			if (prefix === "") return;
+			const encoded_value = this.encoding_callback(prefix);
+			const part = new Part("fixed-text", encoded_value, modifier);
+			this.part_list.push(part);
+			return;
+		}
+
+		let regexp_value = "";
+
+		if (regexp_or_wildcard_token === null) {
+			regexp_value = this.segment_wildcard_regexp;
+		} else if (regexp_or_wildcard_token.type === "asterisk") {
+			regexp_value = FULL_WILDCARD_REGEXP_VALUE;
+		} else {
+			regexp_value = regexp_or_wildcard_token.value;
+		}
+
+		let type: PartType = "regexp";
+
+		if (regexp_value === this.segment_wildcard_regexp) {
+			type = "segment-wildcard";
+			regexp_value = "";
+		} else if (regexp_value === FULL_WILDCARD_REGEXP_VALUE) {
+			type = "full-wildcard";
+			regexp_value = "";
+		}
+
+		let name = "";
+
+		if (name_token !== null) {
+			name = name_token.value;
+		} else if (regexp_or_wildcard_token !== null) {
+			name = String(this.next_numeric_name);
+			this.next_numeric_name++;
+		}
+
+		if (this.is_a_duplicate_name(name)) throw new TypeError();
+
+		const encoded_prefix = this.encoding_callback(prefix);
+		const encoded_suffix = this.encoding_callback(suffix);
+		const part = new Part(
+			type,
+			regexp_value,
+			modifier,
+			name,
+			encoded_prefix,
+			encoded_suffix,
+		);
+		this.part_list.push(part);
+	}
+
+	/** https://urlpattern.spec.whatwg.org/#is-a-duplicate-name */
+	is_a_duplicate_name(name: string) {
+		return this.part_list.some((part) => part.name === name);
+	}
+}
+
+/** https://urlpattern.spec.whatwg.org/#parse-a-pattern-string */
+export function parse_a_pattern_string(
+	input: string,
+	options: Options,
+	encoding_callback: EncodingCallback,
+) {
+	const parser = new PatternParser(
+		encoding_callback,
+		generate_a_segment_wildcard_regexp(options),
+	);
+	parser.token_list = tokenize(input, "strict");
+
+	while (parser.index < parser.token_list.length) {
+		const char_token = parser.try_to_consume_a_token("char");
+		let name_token = parser.try_to_consume_a_token("name");
+		let regexp_or_wildcard_token = parser
+			.try_to_consume_a_regexp_or_wildcard_token(name_token);
+
+		if (name_token !== null || regexp_or_wildcard_token !== null) {
+			let prefix = "";
+			if (char_token !== null) prefix = char_token.value;
+
+			if (prefix !== "" && prefix !== options.prefix_code_point) {
+				parser.pending_fixed_value += prefix;
+				prefix = "";
+			}
+
+			parser.maybe_add_a_part_from_the_pending_fixed_value();
+			const modifier_token = parser.try_to_consume_a_modifier_token();
+			parser.add_a_part(
+				prefix,
+				name_token,
+				regexp_or_wildcard_token,
+				"",
+				modifier_token,
+			);
+			continue;
+		}
+
+		let fixed_token = char_token;
+		fixed_token ??= parser.try_to_consume_a_token("escaped-char");
+
+		if (fixed_token !== null) {
+			parser.pending_fixed_value += fixed_token.value;
+			continue;
+		}
+
+		const open_token = parser.try_to_consume_a_token("open");
+
+		if (open_token !== null) {
+			const prefix = parser.consume_text();
+			name_token = parser.try_to_consume_a_token("name");
+			regexp_or_wildcard_token = parser
+				.try_to_consume_a_regexp_or_wildcard_token(name_token);
+			const suffix = parser.consume_text();
+			parser.consume_a_required_token("close");
+			const modifier_token = parser.try_to_consume_a_modifier_token();
+			parser.add_a_part(
+				prefix,
+				name_token,
+				regexp_or_wildcard_token,
+				suffix,
+				modifier_token,
+			);
+			continue;
+		}
+
+		parser.maybe_add_a_part_from_the_pending_fixed_value();
+		parser.consume_a_required_token("end");
+	}
+
+	return parser.part_list;
+}
+
+/** https://urlpattern.spec.whatwg.org/#full-wildcard-regexp-value */
+const FULL_WILDCARD_REGEXP_VALUE = ".*";
+
+/** https://urlpattern.spec.whatwg.org/#generate-a-segment-wildcard-regexp */
+function generate_a_segment_wildcard_regexp(options: Options) {
+	return `[^${RegExp.escape(options.delimiter_code_point)}]+?`;
+}