feat(fmt): cache check and list results (#225)

Author: chenjiahan

packages/rstack/src/fmt/runner.ts

@@ -1,4 +1,8 @@
+import { cacheNamespace, createCacheKeyResolver, createOptionsHasher } from './cacheIdentity.ts';
+import { loadFmtCacheStore } from './cacheStore.ts';
+import type { FmtCacheEntry, FmtCacheStore } from './cacheStore.ts';
 import type {
+  FmtFileCache,
   FmtExitCode,
   FmtFileRequest,
   FmtFileResult,
@@ -11,6 +15,18 @@ import type { FmtWorkerPool } from './workerPool.ts';
 type FormatFile = FmtWorkerPool['formatFile'];
 type FmtFileOutcome = FmtFileResult | 'unchanged' | 'unsupported';
 
+interface FmtFileRun {
+  outcome: FmtFileOutcome;
+  key?: string;
+  entry?: FmtCacheEntry;
+}
+
+interface RunCache {
+  store: FmtCacheStore;
+  resolveKey: ReturnType<typeof createCacheKeyResolver>;
+  hashOptions: ReturnType<typeof createOptionsHasher>;
+}
+
 interface FmtWorkerPoolResult {
   files: FmtFileResult[];
   processedFileCount: number;
@@ -32,22 +48,47 @@ const runFmtFile = async (
   file: FmtFileRequest,
   shouldWrite: boolean,
   formatFile: FormatFile,
-): Promise<FmtFileOutcome> => {
-  try {
-    const result = await formatFile(file, shouldWrite);
-    if (result === 'unchanged' || result === 'unsupported') {
-      return result;
+  cache?: RunCache,
+): Promise<FmtFileRun> => {
+  let key: string | undefined;
+  let fileCache: FmtFileCache | undefined;
+
+  if (cache) {
+    key = cache.resolveKey(file.path);
+    if (key !== undefined) {
+      const optionsHash = cache.hashOptions(file.options);
+      if (optionsHash === undefined) {
+        key = undefined;
+      } else {
+        fileCache = {
+          entry: cache.store.get(key),
+          optionsHash,
+        };
+      }
     }
+  }
 
-    return {
-      path: file.path,
-      status: shouldWrite ? 'written' : 'different',
-    };
+  try {
+    const result = await formatFile(file, shouldWrite, fileCache);
+    const outcome: FmtFileOutcome =
+      result.status === 'changed'
+        ? {
+            path: file.path,
+            status: shouldWrite ? 'written' : 'different',
+          }
+        : result.status;
+
+    if (key !== undefined && result.cacheEntry) {
+      return { outcome, key, entry: result.cacheEntry };
+    }
+    return { outcome };
   } catch (error) {
     return {
-      path: file.path,
-      status: 'error',
-      error,
+      outcome: {
+        path: file.path,
+        status: 'error',
+        error,
+      },
     };
   }
 };
@@ -57,7 +98,8 @@ const runPriorityFmtFiles = async (
   files: FmtFileRequest[],
   shouldWrite: boolean,
   formatFile: FormatFile,
-): Promise<FmtFileOutcome[]> => {
+  cache?: RunCache,
+): Promise<FmtFileRun[]> => {
   const priority: number[] = [];
   const rest: number[] = [];
 
@@ -67,9 +109,9 @@ const runPriorityFmtFiles = async (
 
   const order = priority.concat(rest);
   const outcomes = await Promise.all(
-    order.map((index) => runFmtFile(files[index], shouldWrite, formatFile)),
+    order.map((index) => runFmtFile(files[index], shouldWrite, formatFile, cache)),
   );
-  const results = new Array<FmtFileOutcome>(files.length);
+  const results = new Array<FmtFileRun>(files.length);
   for (let index = 0; index < order.length; index++) {
     results[order[index]] = outcomes[index];
   }
@@ -81,28 +123,32 @@ const runFmtFilesInWorkerPool = async (
   files: FmtFileRequest[],
   shouldWrite: boolean,
   maxWorkers?: number,
+  cache?: RunCache,
 ): Promise<FmtWorkerPoolResult> => {
   const { createFmtWorkerPool } = await import('./workerPool.ts');
   const workerPool = await createFmtWorkerPool(files.length, maxWorkers);
 
   try {
     const results =
       workerPool.workerCount >= minPriorityWorkers
-        ? await runPriorityFmtFiles(files, shouldWrite, workerPool.formatFile)
+        ? await runPriorityFmtFiles(files, shouldWrite, workerPool.formatFile, cache)
         : await Promise.all(
-            files.map((file) => runFmtFile(file, shouldWrite, workerPool.formatFile)),
+            files.map((file) => runFmtFile(file, shouldWrite, workerPool.formatFile, cache)),
           );
     const processedFiles: FmtFileResult[] = [];
     let processedFileCount = 0;
 
-    for (const result of results) {
-      if (result === 'unsupported') {
+    for (const { outcome, key, entry } of results) {
+      if (key !== undefined && entry) {
+        cache?.store.set(key, entry);
+      }
+      if (outcome === 'unsupported') {
         continue;
       }
 
       processedFileCount++;
-      if (result !== 'unchanged') {
-        processedFiles.push(result);
+      if (outcome !== 'unchanged') {
+        processedFiles.push(outcome);
       }
     }
 
@@ -133,12 +179,23 @@ const runFmtFiles = async ({
   files,
   mode,
   maxWorkers,
+  cache,
 }: RunFmtFilesOptions): Promise<FmtRunResult> => {
   const shouldWrite = mode === 'write';
+  let runCache: RunCache | undefined;
+  if (files.length > 0 && cache && !shouldWrite) {
+    runCache = {
+      store: await loadFmtCacheStore(cache.filePath, cacheNamespace),
+      resolveKey: createCacheKeyResolver(cache.rootPath),
+      hashOptions: createOptionsHasher(),
+    };
+  }
+
   const result =
     files.length === 0
       ? { files: [], processedFileCount: 0 }
-      : await runFmtFilesInWorkerPool(files, shouldWrite, maxWorkers);
+      : await runFmtFilesInWorkerPool(files, shouldWrite, maxWorkers, runCache);
+  await runCache?.store.save().catch(() => false);
 
   return {
     ...result,

packages/rstack/src/fmt/types.ts

@@ -1,4 +1,5 @@
 import type { Config as PrettierConfig, Options as PrettierOptions } from 'prettier';
+import type { FmtCacheEntry } from './cacheStore.ts';
 
 /** Plugin objects cannot cross worker boundaries and are not planned for support. */
 type FmtPluginSpecifier = string | URL;
@@ -71,6 +72,23 @@ interface FmtFileRequest {
   options: ResolvedFmtOptions;
 }
 
+interface FmtCacheContext {
+  /** Persistent cache file to load and update. */
+  filePath: string;
+  /** Root used to create portable per-file cache keys. */
+  rootPath: string;
+}
+
+interface FmtFileCache {
+  entry: FmtCacheEntry | undefined;
+  optionsHash: string;
+}
+
+interface FmtWorkerResult {
+  status: 'changed' | 'unchanged' | 'unsupported';
+  cacheEntry?: FmtCacheEntry;
+}
+
 type FmtMode = 'write' | 'check' | 'list-different';
 type FmtExitCode = 0 | 1 | 2;
 
@@ -81,6 +99,8 @@ interface RunFmtFilesOptions {
   mode: FmtMode;
   /** Maximum number of formatting workers. */
   maxWorkers?: number;
+  /** Internal persistent cache context. Currently used only by check and list modes. */
+  cache?: FmtCacheContext;
 }
 
 interface SuccessfulFmtFileResult {
@@ -106,14 +126,17 @@ interface FmtRunResult {
 
 export type {
   DiscoverFmtFilesOptions,
+  FmtCacheContext,
   FmtConfig,
   FmtConfigDefinition,
   FmtExitCode,
   FmtFileResult,
   FmtFileRequest,
+  FmtFileCache,
   FmtMode,
   FmtPluginSpecifier,
   FmtRunResult,
+  FmtWorkerResult,
   ResolvedFmtConfig,
   ResolvedFmtOptions,
   RunFmtFilesOptions,

packages/rstack/src/fmt/worker.ts

@@ -1,39 +1,64 @@
 // Derived from @prettier/cli, see THIRD_PARTY_NOTICES.md
 
+import { createHash } from 'node:crypto';
 import { readFileSync, writeFileSync } from 'node:fs';
-import { formatFmtSource } from './format.ts';
-import type { FmtFileRequest } from './types.ts';
-
-type FormatFileResult = 'changed' | 'unchanged' | 'unsupported';
+import type { FmtCacheEntry } from './cacheStore.ts';
+import type { FmtFileCache, FmtFileRequest, FmtWorkerResult } from './types.ts';
 
 interface FormatFileTask {
   file: FmtFileRequest;
   shouldWrite: boolean;
+  cache?: FmtFileCache;
 }
 
+const hashContent = (content: Uint8Array): string =>
+  createHash('sha256').update(content).digest('hex');
+
 /**
  * Use synchronous direct I/O inside the dedicated worker to avoid libuv
  * scheduling overhead. This prioritizes throughput over crash-safe replacement.
  */
-const formatFile = async ({ file, shouldWrite }: FormatFileTask): Promise<FormatFileResult> => {
-  const result = await formatFmtSource(file, () => readFileSync(file.path, 'utf8'));
+const formatFile = async ({
+  file,
+  shouldWrite,
+  cache,
+}: FormatFileTask): Promise<FmtWorkerResult> => {
+  let source: string | undefined;
+  let contentHash: string | undefined;
+
+  if (cache && !shouldWrite) {
+    const content = readFileSync(file.path);
+    contentHash = hashContent(content);
+    source = content.toString('utf8');
+
+    const { entry, optionsHash } = cache;
+    if (entry?.[0] === contentHash && entry[1] === optionsHash) {
+      return { status: entry[2] === 'clean' ? 'unchanged' : 'changed' };
+    }
+  }
+
+  const { formatFmtSource } = await import('./format.ts');
+  const result = await formatFmtSource(file, () => (source ??= readFileSync(file.path, 'utf8')));
   if (result.status === 'unsupported') {
-    return 'unsupported';
+    return { status: 'unsupported' };
   }
 
-  const { source, formatted } = result;
-  if (source === formatted) {
-    return 'unchanged';
+  const unchanged = result.source === result.formatted;
+
+  if (!unchanged && shouldWrite) {
+    writeFileSync(file.path, result.formatted, 'utf8');
   }
 
-  if (shouldWrite) {
-    writeFileSync(file.path, formatted, 'utf8');
+  const status = unchanged ? 'unchanged' : 'changed';
+  if (!cache || contentHash === undefined) {
+    return { status };
   }
 
-  return 'changed';
+  const cacheEntry: FmtCacheEntry = [contentHash, cache.optionsHash, unchanged ? 'clean' : 'dirty'];
+  return { status, cacheEntry };
 };
 
-/** Confirms that the worker module and its runtime dependencies are ready. */
+/** Confirms that the worker module is ready. Formatter dependencies load only on a cache miss. */
 const initializeFmtWorker = (): true => true;
 
 export { formatFile, initializeFmtWorker };

packages/rstack/src/fmt/workerPool.ts

@@ -2,7 +2,7 @@
 
 import { availableParallelism } from 'node:os';
 import Tinypool from 'tinypool';
-import type { FmtFileRequest } from './types.ts';
+import type { FmtFileCache, FmtFileRequest } from './types.ts';
 
 type FmtWorkerMethods = typeof import('./worker.ts');
 
@@ -11,6 +11,7 @@ interface FmtWorkerPool {
   formatFile: (
     file: FmtFileRequest,
     shouldWrite: boolean,
+    cache?: FmtFileCache,
   ) => ReturnType<FmtWorkerMethods['formatFile']>;
   terminate: () => Promise<void>;
 }
@@ -57,7 +58,8 @@ const createFmtWorkerPool = async (
 
   return {
     workerCount,
-    formatFile: (file, shouldWrite) => pool.run({ file, shouldWrite }, { name: 'formatFile' }),
+    formatFile: (file, shouldWrite, cache) =>
+      pool.run({ file, shouldWrite, cache }, { name: 'formatFile' }),
     terminate: () => pool.destroy(),
   };
 };