refactor: add base class for CLI-based runners

Adds a base class with common logic for command-line-based runners.

Author: crisbeto

runner/codegen/base-cli-agent-runner.ts

@@ -0,0 +1,207 @@
+import {ChildProcess, spawn} from 'child_process';
+import {join, relative} from 'path';
+import {existsSync} from 'fs';
+import assert from 'assert';
+import {
+  LlmConstrainedOutputGenerateResponse,
+  LlmGenerateFilesRequestOptions,
+  LlmGenerateFilesResponse,
+  LlmGenerateTextResponse,
+} from './llm-runner.js';
+import {DirectorySnapshot} from './directory-snapshot.js';
+import {LlmResponseFile} from '../shared-interfaces.js';
+import {UserFacingError} from '../utils/errors.js';
+
+/** Base class for a command-line-based runner. */
+export abstract class BaseCliAgentRunner {
+  abstract readonly displayName: string;
+  protected abstract readonly binaryName: string;
+  protected abstract readonly ignoredFilePatterns: string[];
+  protected abstract getCommandLineFlags(options: LlmGenerateFilesRequestOptions): string[];
+  protected abstract writeAgentFiles(options: LlmGenerateFilesRequestOptions): Promise<void>;
+
+  private pendingTimeouts = new Set<ReturnType<typeof setTimeout>>();
+  private pendingProcesses = new Set<ChildProcess>();
+  private binaryPath: string | null = null;
+  private commonIgnoredPatterns = ['**/node_modules/**', '**/dist/**', '**/.angular/**'];
+
+  async generateFiles(options: LlmGenerateFilesRequestOptions): Promise<LlmGenerateFilesResponse> {
+    const {context} = options;
+
+    // TODO: Consider removing these assertions when we have better types.
+    assert(
+      context.buildCommand,
+      'Expected a `buildCommand` to be set in the LLM generate request context',
+    );
+    assert(
+      context.packageManager,
+      'Expected a `packageManager` to be set in the LLM generate request context',
+    );
+
+    const ignoredPatterns = [...this.commonIgnoredPatterns, ...this.ignoredFilePatterns];
+    const initialSnapshot = await DirectorySnapshot.forDirectory(
+      context.directory,
+      ignoredPatterns,
+    );
+
+    await this.writeAgentFiles(options);
+
+    const reasoning = await this.runAgentProcess(options, 2, 10);
+    const finalSnapshot = await DirectorySnapshot.forDirectory(context.directory, ignoredPatterns);
+
+    const diff = finalSnapshot.getChangedOrAddedFiles(initialSnapshot);
+    const files: LlmResponseFile[] = [];
+
+    for (const [absolutePath, code] of diff) {
+      files.push({
+        filePath: relative(context.directory, absolutePath),
+        code,
+      });
+    }
+
+    return {files, reasoning, toolLogs: []};
+  }
+
+  generateText(): Promise<LlmGenerateTextResponse> {
+    // Technically we can make this work, but we don't need it at the time of writing.
+    throw new UserFacingError(`Generating text with ${this.displayName} is not supported.`);
+  }
+
+  generateConstrained(): Promise<LlmConstrainedOutputGenerateResponse<any>> {
+    // We can't support this, because there's no straightforward
+    // way to tell the agent to follow a schema.
+    throw new UserFacingError(`Constrained output with ${this.displayName} is not supported.`);
+  }
+
+  async dispose(): Promise<void> {
+    for (const timeout of this.pendingTimeouts) {
+      clearTimeout(timeout);
+    }
+
+    for (const childProcess of this.pendingProcesses) {
+      childProcess.kill('SIGKILL');
+    }
+
+    this.pendingTimeouts.clear();
+    this.pendingProcesses.clear();
+  }
+
+  private resolveBinaryPath(binaryName: string): string {
+    let dir = import.meta.dirname;
+    let closestRoot: string | null = null;
+
+    // Attempt to resolve the agent CLI binary by starting at the current file and going up until
+    // we find the closest `node_modules`. Note that we can't rely on `import.meta.resolve` here,
+    // because that'll point us to the agent bundle, but not its binary. In some package
+    // managers (pnpm specifically) the `node_modules` in which the file is installed is different
+    // from the one in which the binary is placed.
+    while (dir.length > 1) {
+      if (existsSync(join(dir, 'node_modules'))) {
+        closestRoot = dir;
+        break;
+      }
+
+      const parent = join(dir, '..');
+
+      if (parent === dir) {
+        // We've reached the root, stop traversing.
+        break;
+      } else {
+        dir = parent;
+      }
+    }
+
+    const binaryPath = closestRoot ? join(closestRoot, `node_modules/.bin/${binaryName}`) : null;
+
+    if (!binaryPath || !existsSync(binaryPath)) {
+      throw new UserFacingError(`${this.displayName} is not installed inside the current project`);
+    }
+
+    return binaryPath;
+  }
+
+  private runAgentProcess(
+    options: LlmGenerateFilesRequestOptions,
+    inactivityTimeoutMins: number,
+    totalRequestTimeoutMins: number,
+  ): Promise<string> {
+    return new Promise<string>(resolve => {
+      let stdoutBuffer = '';
+      let stdErrBuffer = '';
+      let isDone = false;
+      const msPerMin = 1000 * 60;
+      const finalize = (finalMessage: string) => {
+        if (isDone) {
+          return;
+        }
+
+        isDone = true;
+
+        if (inactivityTimeout) {
+          clearTimeout(inactivityTimeout);
+          this.pendingTimeouts.delete(inactivityTimeout);
+        }
+
+        clearTimeout(globalTimeout);
+        childProcess.kill('SIGKILL');
+        this.pendingTimeouts.delete(globalTimeout);
+        this.pendingProcesses.delete(childProcess);
+
+        const separator = '\n--------------------------------------------------\n';
+
+        if (stdErrBuffer.length > 0) {
+          stdoutBuffer += separator + 'Stderr output:\n' + stdErrBuffer;
+        }
+
+        stdoutBuffer += separator + finalMessage;
+        resolve(stdoutBuffer);
+      };
+
+      const noOutputCallback = () => {
+        finalize(
+          `There was no output from ${this.displayName} for ${inactivityTimeoutMins} minute(s). ` +
+            `Stopping the process...`,
+        );
+      };
+
+      // The agent can get into a state where it stops outputting code, but it also doesn't exit
+      // the process. Stop if there hasn't been any output for a certain amount of time.
+      let inactivityTimeout = setTimeout(noOutputCallback, inactivityTimeoutMins * msPerMin);
+      this.pendingTimeouts.add(inactivityTimeout);
+
+      // Also add a timeout for the entire codegen process.
+      const globalTimeout = setTimeout(() => {
+        finalize(
+          `${this.displayName} didn't finish within ${totalRequestTimeoutMins} minute(s). ` +
+            `Stopping the process...`,
+        );
+      }, totalRequestTimeoutMins * msPerMin);
+
+      this.binaryPath ??= this.resolveBinaryPath(this.binaryName);
+
+      const childProcess = spawn(this.binaryPath, this.getCommandLineFlags(options), {
+        cwd: options.context.directory,
+        env: {...process.env},
+      });
+
+      childProcess.on('close', code =>
+        finalize(
+          `${this.displayName} process has exited` + (code == null ? '.' : ` with ${code} code.`),
+        ),
+      );
+      childProcess.stdout.on('data', data => {
+        if (inactivityTimeout) {
+          this.pendingTimeouts.delete(inactivityTimeout);
+          clearTimeout(inactivityTimeout);
+        }
+
+        stdoutBuffer += data.toString();
+        inactivityTimeout = setTimeout(noOutputCallback, inactivityTimeoutMins * msPerMin);
+        this.pendingTimeouts.add(inactivityTimeout);
+      });
+      childProcess.stderr.on('data', data => {
+        stdErrBuffer += data.toString();
+      });
+    });
+  }
+}

runner/codegen/gemini-cli/directory-snapshot.ts renamed to runner/codegen/directory-snapshot.ts

@@ -15,7 +15,7 @@ import {
   getGeminiInstructionsFile,
   getGeminiSettingsFile,
 } from './gemini-files.js';
-import {DirectorySnapshot} from './directory-snapshot.js';
+import {DirectorySnapshot} from '../directory-snapshot.js';
 import {LlmResponseFile} from '../../shared-interfaces.js';
 import {UserFacingError} from '../../utils/errors.js';
 import assert from 'assert';