| 1 | /** |
| 2 | * Copyright (c) Meta Platforms, Inc. and affiliates. |
| 3 | * |
| 4 | * This source code is licensed under the MIT license found in the |
| 5 | * LICENSE file in the root directory of this source tree. |
| 6 | */ |
| 7 | |
| 8 | import type {IGrammar} from 'vscode-textmate'; |
| 9 | import type {ThemeColor} from '../../theme'; |
| 10 | import type { |
| 11 | SyntaxWorkerRequest, |
| 12 | SyntaxWorkerResponse, |
| 13 | TokenizedDiffHunks, |
| 14 | TokenizedHunk, |
| 15 | } from './syntaxHighlightingTypes'; |
| 16 | |
| 17 | import {CancellationToken} from 'shared/CancellationToken'; |
| 18 | import FilepathClassifier from 'shared/textmate-lib/FilepathClassifier'; |
| 19 | import {tokenizeLines} from 'shared/textmate-lib/tokenize'; |
| 20 | import {loadWASM} from 'vscode-oniguruma'; |
| 21 | import {grammars, languages} from '../../generated/textmate/TextMateGrammarManifest'; |
| 22 | import {getGrammar, getGrammarStore} from './grammar'; |
| 23 | |
| 24 | const URL_TO_ONIG_WASM = './generated/textmate/onig.wasm'; |
| 25 | |
| 26 | /* This file is intended to be executed in a WebWorker, without access to the DOM. */ |
| 27 | |
| 28 | /** |
| 29 | * Fetch requests inside the webworker must be made relative to the base URI. |
| 30 | * By executing the web worker via a blob: URL, we can't depend on the base being inherited. |
| 31 | */ |
| 32 | let globalBaseUri: string; |
| 33 | |
| 34 | async function loadGrammar( |
| 35 | theme: ThemeColor, |
| 36 | path: string, |
| 37 | postMessage: (msg: SyntaxWorkerResponse) => void, |
| 38 | ): Promise<IGrammar | undefined> { |
| 39 | await ensureOnigurumaIsLoaded(globalBaseUri); |
| 40 | |
| 41 | const scopeName = getFilepathClassifier().findScopeNameForPath(path); |
| 42 | if (!scopeName) { |
| 43 | return undefined; |
| 44 | } |
| 45 | |
| 46 | const store = getGrammarStore(theme, globalBaseUri, colorMap => { |
| 47 | // tell client the newest colorMap |
| 48 | postMessage({type: 'cssColorMap', colorMap} as SyntaxWorkerResponse); |
| 49 | }); |
| 50 | |
| 51 | const grammar = await getGrammar(store, scopeName); |
| 52 | return grammar ?? undefined; |
| 53 | } |
| 54 | |
| 55 | const cancellationTokenForId = new Map<number, CancellationToken>(); |
| 56 | |
| 57 | class WorkQueue { |
| 58 | private queue: Array<() => Promise<void>> = []; |
| 59 | private isProcessing = false; |
| 60 | |
| 61 | public push(work: () => Promise<void>) { |
| 62 | this.queue.push(work); |
| 63 | |
| 64 | if (!this.isProcessing) { |
| 65 | this.processNext(); |
| 66 | } |
| 67 | } |
| 68 | |
| 69 | private async processNext() { |
| 70 | if (this.queue.length > 0) { |
| 71 | const work = this.queue.shift(); |
| 72 | this.isProcessing = true; |
| 73 | // Allow the task queue to be emptied before continuing, |
| 74 | // so we can process cancel messages |
| 75 | await new Promise(res => setTimeout(res, 0)); |
| 76 | await work?.().catch(err => { |
| 77 | // eslint-disable-next-line no-console |
| 78 | console.error(err); |
| 79 | return null; |
| 80 | }); |
| 81 | this.isProcessing = false; |
| 82 | this.processNext(); |
| 83 | } |
| 84 | } |
| 85 | } |
| 86 | |
| 87 | const workQueue = new WorkQueue(); |
| 88 | |
| 89 | export function handleMessage( |
| 90 | postMessage: (msg: SyntaxWorkerResponse & {id?: number}) => unknown, |
| 91 | event: MessageEvent, |
| 92 | ) { |
| 93 | const data = event.data as SyntaxWorkerRequest & {id: number}; |
| 94 | |
| 95 | const token = new CancellationToken(); |
| 96 | if (data.id != null) { |
| 97 | cancellationTokenForId.set(data.id, token); |
| 98 | } |
| 99 | switch (data.type) { |
| 100 | case 'setBaseUri': { |
| 101 | globalBaseUri = data.base; |
| 102 | break; |
| 103 | } |
| 104 | case 'tokenizeContents': { |
| 105 | workQueue.push(async () => { |
| 106 | const grammar = await loadGrammar(data.theme, data.path, postMessage); |
| 107 | const result = tokenizeContent(grammar, data.content, token); |
| 108 | postMessage({type: data.type, id: data.id, result}); |
| 109 | cancellationTokenForId.delete(data.id); |
| 110 | }); |
| 111 | break; |
| 112 | } |
| 113 | case 'tokenizeHunks': { |
| 114 | workQueue.push(async () => { |
| 115 | const grammar = await loadGrammar(data.theme, data.path, postMessage); |
| 116 | const result = tokenizeHunks(grammar, data.hunks, token); |
| 117 | postMessage({type: data.type, id: data.id, result}); |
| 118 | cancellationTokenForId.delete(data.id); |
| 119 | }); |
| 120 | break; |
| 121 | } |
| 122 | case 'cancel': { |
| 123 | const token = cancellationTokenForId.get(data.idToCancel); |
| 124 | token?.cancel(); |
| 125 | } |
| 126 | } |
| 127 | } |
| 128 | |
| 129 | if (typeof self.document === 'undefined') { |
| 130 | // inside WebWorker, use global onmessage and postMessage |
| 131 | onmessage = handleMessage.bind(undefined, postMessage); |
| 132 | // outside of a WebWorker, the exported `handleMessage` function should be used instead. |
| 133 | } |
| 134 | |
| 135 | function tokenizeHunks( |
| 136 | grammar: IGrammar | undefined, |
| 137 | hunks: Array<{lines: Array<string>}>, |
| 138 | cancellationToken: CancellationToken, |
| 139 | ): TokenizedDiffHunks | undefined { |
| 140 | if (grammar == null) { |
| 141 | return undefined; |
| 142 | } |
| 143 | |
| 144 | if (cancellationToken.isCancelled) { |
| 145 | // check for cancellation before doing expensive highlighting |
| 146 | return undefined; |
| 147 | } |
| 148 | |
| 149 | const tokenizedPatches: TokenizedDiffHunks = hunks |
| 150 | .map(hunk => recoverFileContentsFromPatchLines(hunk.lines)) |
| 151 | .map(([before, after]) => [tokenizeLines(before, grammar), tokenizeLines(after, grammar)]); |
| 152 | |
| 153 | return tokenizedPatches; |
| 154 | } |
| 155 | |
| 156 | function tokenizeContent( |
| 157 | grammar: IGrammar | undefined, |
| 158 | content: Array<string>, |
| 159 | cancellationToken: CancellationToken, |
| 160 | ): TokenizedHunk | undefined { |
| 161 | if (grammar == null) { |
| 162 | return undefined; |
| 163 | } |
| 164 | |
| 165 | if (cancellationToken.isCancelled) { |
| 166 | // check for cancellation before doing expensive highlighting |
| 167 | return undefined; |
| 168 | } |
| 169 | |
| 170 | return tokenizeLines(content, grammar); |
| 171 | } |
| 172 | |
| 173 | /** |
| 174 | * Patch lines start with ' ', '+', or '-'. From this we can reconstruct before & after file contents as strings, |
| 175 | * which we can actually use in the syntax highlighting. |
| 176 | */ |
| 177 | function recoverFileContentsFromPatchLines( |
| 178 | lines: Array<string>, |
| 179 | ): [before: Array<string>, after: Array<string>] { |
| 180 | const linesBefore = []; |
| 181 | const linesAfter = []; |
| 182 | for (const line of lines) { |
| 183 | if (line[0] === ' ') { |
| 184 | linesBefore.push(line.slice(1)); |
| 185 | linesAfter.push(line.slice(1)); |
| 186 | } else if (line[0] === '+') { |
| 187 | linesAfter.push(line.slice(1)); |
| 188 | } else if (line[0] === '-') { |
| 189 | linesBefore.push(line.slice(1)); |
| 190 | } |
| 191 | } |
| 192 | |
| 193 | return [linesBefore, linesAfter]; |
| 194 | } |
| 195 | |
| 196 | let onigurumaLoadingJob: Promise<void> | null = null; |
| 197 | function ensureOnigurumaIsLoaded(base: string): Promise<void> { |
| 198 | if (onigurumaLoadingJob === null) { |
| 199 | onigurumaLoadingJob = loadOniguruma(base); |
| 200 | } |
| 201 | return onigurumaLoadingJob; |
| 202 | } |
| 203 | |
| 204 | async function loadOniguruma(base: string): Promise<void> { |
| 205 | const url = new URL(URL_TO_ONIG_WASM, base); |
| 206 | const onigurumaWASMRequest = fetch(url); |
| 207 | const response = await onigurumaWASMRequest; |
| 208 | |
| 209 | const contentType = response.headers.get('content-type'); |
| 210 | const useStreamingParser = contentType === 'application/wasm'; |
| 211 | |
| 212 | if (useStreamingParser) { |
| 213 | await loadWASM(response); |
| 214 | } else { |
| 215 | const dataOrOptions = { |
| 216 | data: await response.arrayBuffer(), |
| 217 | }; |
| 218 | await loadWASM(dataOrOptions); |
| 219 | } |
| 220 | } |
| 221 | |
| 222 | let _classifier: FilepathClassifier | null = null; |
| 223 | |
| 224 | function getFilepathClassifier(): FilepathClassifier { |
| 225 | if (_classifier == null) { |
| 226 | _classifier = new FilepathClassifier(grammars, languages); |
| 227 | } |
| 228 | return _classifier; |
| 229 | } |
| 230 | |