std/path/glob_to_regexp.ts

95 lines
3.8 KiB
TypeScript

// Copyright 2018-2024 the Deno authors. All rights reserved. MIT license.
// This module is browser compatible.
import type { GlobOptions } from "./_common/glob_to_reg_exp.ts";
import { isWindows } from "./_os.ts";
import { globToRegExp as posixGlobToRegExp } from "./posix/glob_to_regexp.ts";
import {
globToRegExp as windowsGlobToRegExp,
} from "./windows/glob_to_regexp.ts";
export type { GlobOptions };
/**
* Converts a glob string to a regular expression.
*
* Tries to match bash glob expansion as closely as possible.
*
* Basic glob syntax:
* - `*` - Matches everything without leaving the path segment.
* - `?` - Matches any single character.
* - `{foo,bar}` - Matches `foo` or `bar`.
* - `[abcd]` - Matches `a`, `b`, `c` or `d`.
* - `[a-d]` - Matches `a`, `b`, `c` or `d`.
* - `[!abcd]` - Matches any single character besides `a`, `b`, `c` or `d`.
* - `[[:<class>:]]` - Matches any character belonging to `<class>`.
* - `[[:alnum:]]` - Matches any digit or letter.
* - `[[:digit:]abc]` - Matches any digit, `a`, `b` or `c`.
* - See https://facelessuser.github.io/wcmatch/glob/#posix-character-classes
* for a complete list of supported character classes.
* - `\` - Escapes the next character for an `os` other than `"windows"`.
* - \` - Escapes the next character for `os` set to `"windows"`.
* - `/` - Path separator.
* - `\` - Additional path separator only for `os` set to `"windows"`.
*
* Extended syntax:
* - Requires `{ extended: true }`.
* - `?(foo|bar)` - Matches 0 or 1 instance of `{foo,bar}`.
* - `@(foo|bar)` - Matches 1 instance of `{foo,bar}`. They behave the same.
* - `*(foo|bar)` - Matches _n_ instances of `{foo,bar}`.
* - `+(foo|bar)` - Matches _n > 0_ instances of `{foo,bar}`.
* - `!(foo|bar)` - Matches anything other than `{foo,bar}`.
* - See https://www.linuxjournal.com/content/bash-extended-globbing.
*
* Globstar syntax:
* - Requires `{ globstar: true }`.
* - `**` - Matches any number of any path segments.
* - Must comprise its entire path segment in the provided glob.
* - See https://www.linuxjournal.com/content/globstar-new-bash-globbing-option.
*
* Note the following properties:
* - The generated `RegExp` is anchored at both start and end.
* - Repeating and trailing separators are tolerated. Trailing separators in the
* provided glob have no meaning and are discarded.
* - Absolute globs will only match absolute paths, etc.
* - Empty globs will match nothing.
* - Any special glob syntax must be contained to one path segment. For example,
* `?(foo|bar/baz)` is invalid. The separator will take precedence and the
* first segment ends with an unclosed group.
* - If a path segment ends with unclosed groups or a dangling escape prefix, a
* parse error has occurred. Every character for that segment is taken
* literally in this event.
*
* Limitations:
* - A negative group like `!(foo|bar)` will wrongly be converted to a negative
* look-ahead followed by a wildcard. This means that `!(foo).js` will wrongly
* fail to match `foobar.js`, even though `foobar` is not `foo`. Effectively,
* `!(foo|bar)` is treated like `!(@(foo|bar)*)`. This will work correctly if
* the group occurs not nested at the end of the segment.
*
* @example Usage
* ```ts
* import { globToRegExp } from "@std/path/glob-to-regexp";
* import { assertEquals } from "@std/assert";
*
* if (Deno.build.os === "windows") {
* assertEquals(globToRegExp("*.js"), /^[^\\/]*\.js(?:\\|\/)*$/);
* } else {
* assertEquals(globToRegExp("*.js"), /^[^/]*\.js\/*$/);
* }
* ```
*
* @param glob Glob string to convert.
* @param options Conversion options.
* @returns The regular expression equivalent to the glob.
*/
export function globToRegExp(
glob: string,
options: GlobOptions = {},
): RegExp {
return isWindows
? windowsGlobToRegExp(glob, options)
: posixGlobToRegExp(glob, options);
}