vscode/src/vs/base/common/codicon.ts

/*---------------------------------------------------------------------------------------------
 *  Copyright (c) Microsoft Corporation. All rights reserved.
 *  Licensed under the MIT License. See License.txt in the project root for license information.
 *--------------------------------------------------------------------------------------------*/

import { matchesFuzzy, IMatch } from 'vs/base/common/filters';
import { ltrim } from 'vs/base/common/strings';

export const codiconStartMarker = '$(';

export interface IParsedCodicons {
	readonly text: string;
	readonly codiconOffsets?: readonly number[];
}

export function parseCodicons(text: string): IParsedCodicons {
	const firstCodiconIndex = text.indexOf(codiconStartMarker);
	if (firstCodiconIndex === -1) {
		return { text }; // return early if the word does not include an codicon
	}

	return doParseCodicons(text, firstCodiconIndex);
}

function doParseCodicons(text: string, firstCodiconIndex: number): IParsedCodicons {
	const codiconOffsets: number[] = [];
	let textWithoutCodicons: string = '';

	function appendChars(chars: string) {
		if (chars) {
			textWithoutCodicons += chars;

			for (const _ of chars) {
				codiconOffsets.push(codiconsOffset); // make sure to fill in codicon offsets
			}
		}
	}

	let currentCodiconStart = -1;
	let currentCodiconValue: string = '';
	let codiconsOffset = 0;

	let char: string;
	let nextChar: string;

	let offset = firstCodiconIndex;
	const length = text.length;

	// Append all characters until the first codicon
	appendChars(text.substr(0, firstCodiconIndex));

	// example: $(file-symlink-file) my cool $(other-codicon) entry
	while (offset < length) {
		char = text[offset];
		nextChar = text[offset + 1];

		// beginning of codicon: some value $( <--
		if (char === codiconStartMarker[0] && nextChar === codiconStartMarker[1]) {
			currentCodiconStart = offset;

			// if we had a previous potential codicon value without
			// the closing ')', it was actually not an codicon and
			// so we have to add it to the actual value
			appendChars(currentCodiconValue);

			currentCodiconValue = codiconStartMarker;

			offset++; // jump over '('
		}

		// end of codicon: some value $(some-codicon) <--
		else if (char === ')' && currentCodiconStart !== -1) {
			const currentCodiconLength = offset - currentCodiconStart + 1; // +1 to include the closing ')'
			codiconsOffset += currentCodiconLength;
			currentCodiconStart = -1;
			currentCodiconValue = '';
		}

		// within codicon
		else if (currentCodiconStart !== -1) {
			// Make sure this is a real codicon name
			if (/^[a-z0-9\-]$/i.test(char)) {
				currentCodiconValue += char;
			} else {
				// This is not a real codicon, treat it as text
				appendChars(currentCodiconValue);

				currentCodiconStart = -1;
				currentCodiconValue = '';
			}
		}

		// any value outside of codicons
		else {
			appendChars(char);
		}

		offset++;
	}

	// if we had a previous potential codicon value without
	// the closing ')', it was actually not an codicon and
	// so we have to add it to the actual value
	appendChars(currentCodiconValue);

	return { text: textWithoutCodicons, codiconOffsets };
}

export function matchesFuzzyCodiconAware(query: string, target: IParsedCodicons, enableSeparateSubstringMatching = false): IMatch[] | null {
	const { text, codiconOffsets } = target;

	// Return early if there are no codicon markers in the word to match against
	if (!codiconOffsets || codiconOffsets.length === 0) {
		return matchesFuzzy(query, text, enableSeparateSubstringMatching);
	}

	// Trim the word to match against because it could have leading
	// whitespace now if the word started with an codicon
	const wordToMatchAgainstWithoutCodiconsTrimmed = ltrim(text, ' ');
	const leadingWhitespaceOffset = text.length - wordToMatchAgainstWithoutCodiconsTrimmed.length;

	// match on value without codicons
	const matches = matchesFuzzy(query, wordToMatchAgainstWithoutCodiconsTrimmed, enableSeparateSubstringMatching);

	// Map matches back to offsets with codicons and trimming
	if (matches) {
		for (const match of matches) {
			const codiconOffset = codiconOffsets[match.start + leadingWhitespaceOffset] /* codicon offsets at index */ + leadingWhitespaceOffset /* overall leading whitespace offset */;
			match.start += codiconOffset;
			match.end += codiconOffset;
		}
	}

	return matches;
}