scai/dist/fileRules/fileClassifier.js

> AI-powered CLI tool for commit messages **and** pull request reviews — using local models.

// utils/fileClassifier.ts
import path from 'path';
/**
 * Determines whether a file is likely to be a *generated* or *bundled* file,
 * rather than handwritten source code.
 *
 * This helps filter out files that shouldn't be analyzed for user-authored logic,
 * like minified JS bundles, Webpack chunks, TypeScript output, etc.
 */
export function isGeneratedOrBundledFile(filePath) {
    const base = path.basename(filePath);
    /**
     * Minified file detection:
     * Matches file names like `something.min.js` or `app.min.ts`.
     * These are typically compiled output intended for production and are not original source code.
     */
    const isMinified = /\.min\.(js|ts)$/.test(base);
    /**
     * Hash-named or chunk file detection:
     * Matches file names like `bundle.839abc.js`, `chunk.123abc.ts`, or `main-worker.js`.
     * These are often created by bundlers like Webpack, Vite, or Rollup.
     */
    const isHashNamed = /[-_.](worker|bundle|chunk|[a-f0-9]{6,})\.(js|ts)$/.test(base);
    /**
     * Output folder detection:
     * These folders are commonly used to store compiled or bundled output.
     * Examples: `dist/`, `build/`, `assets/`, `node_modules/`, `plugins/`
     *
     * If a file is inside any of these folders, we consider it generated/bundled.
     */
    const outputDirs = ['dist', 'build', 'assets', 'node_modules', 'plugins'];
    const isInKnownOutputFolder = outputDirs.some(dir => new RegExp(`[\\\\/]${dir}[\\\\/]`, 'i').test(filePath));
    /**
     * Special case: `lib/` folder
     * The `lib` folder may contain either handwritten code or compiled output, depending on the project.
     * To avoid over-filtering, we only treat files in `lib/` as generated if they also look minified or hashed.
     */
    const isInLib = /[\\/]lib[\\/]/i.test(filePath);
    const isLikelyBundledLib = isInLib && (isMinified || isHashNamed);
    /**
     * Return true if *any* of the following conditions are met:
     *  - The file looks minified (e.g., `.min.js`)
     *  - The file has a hash or is a known bundle/chunk/worker
     *  - The file is located in a known output directory
     *  - The file is in `lib/` and has signs of being generated (minified or hashed)
     */
    return isMinified || isHashNamed || isInKnownOutputFolder || isLikelyBundledLib;
}