Merge bcf354a4e2 into 51f2ad6e8a

apps/x/apps/main/bundle.mjs

@@ -7,6 +7,13 @@
  * - Bundling inlines all dependencies into a single file, eliminating node_modules
  * 
  * This script is called by the generateAssets hook in forge.config.js before packaging.
+ * 
+ * Why pdf-parse, xlsx, papaparse, mammoth are marked external:
+ * - builtin-tools.ts loads these via _importDynamic (new Function pattern) to prevent
+ *   esbuild from statically bundling pdfjs-dist's DOM polyfills into the main process.
+ * - Because esbuild cannot see through the dynamic import, these packages must be
+ *   available as real node_modules at runtime instead.
+ * - forge.config.cjs copies them into .package/node_modules after bundling.
  */
 
 import * as esbuild from 'esbuild';
@@ -16,13 +23,17 @@ import * as esbuild from 'esbuild';
 // and we use define to replace all import.meta.url references with it.
 const cjsBanner = `var __import_meta_url = require('url').pathToFileURL(__filename).href;`;
 
+// These packages are loaded at runtime via _importDynamic and cannot be bundled.
+// They must be present in node_modules alongside the app bundle.
+const RUNTIME_EXTERNAL = ['pdf-parse', 'xlsx', 'papaparse', 'mammoth'];
+
 await esbuild.build({
   entryPoints: ['./dist/main.js'],
   bundle: true,
   platform: 'node',
   target: 'node20',
   outfile: './.package/dist/main.cjs',
-  external: ['electron'],  // Provided by Electron runtime
+  external: ['electron', ...RUNTIME_EXTERNAL],
   // Use CommonJS format - many dependencies use require() which doesn't work
   // well with esbuild's ESM shim. CJS handles dynamic requires natively.
   format: 'cjs',
@@ -34,4 +45,4 @@ await esbuild.build({
   },
 });
 
-console.log('✅ Main process bundled to .package/dist-bundle/main.js');
+console.log('✅ Main process bundled to .package/dist/main.cjs');

apps/x/apps/main/forge.config.cjs

@@ -170,6 +170,53 @@ module.exports = {
             fs.mkdirSync(rendererDest, { recursive: true });
             fs.cpSync(rendererSrc, rendererDest, { recursive: true });
 
+            // Copy runtime-external packages and their transitive deps into .package/node_modules.
+            // These are loaded via _importDynamic at runtime and cannot be bundled
+            // by esbuild (doing so would pull pdfjs-dist DOM polyfills into the
+            // Electron main process). They must exist as real node_modules.
+            const runtimeRoots = ['pdf-parse', 'xlsx', 'papaparse', 'mammoth'];
+            const pnpmModules = path.join(__dirname, '../../node_modules/.pnpm');
+            const destModules = path.join(packageDir, 'node_modules');
+            fs.mkdirSync(destModules, { recursive: true });
+
+            // pnpm stores @scope/pkg as @scope+pkg@version in the .pnpm directory.
+            // Inside that versioned dir, the package lives at node_modules/@scope/pkg.
+            function pnpmDirName(pkg) { return pkg.replace('/', '+'); }
+
+            // Recursively collect all transitive + optional deps from the pnpm store.
+            // optionalDeps are included because @napi-rs/canvas (needed by pdfjs-dist)
+            // ships its native binaries as optional platform-specific packages.
+            function collectDeps(pkgName, visited = new Set()) {
+                if (visited.has(pkgName)) return visited;
+                visited.add(pkgName);
+                const dirName = pnpmDirName(pkgName);
+                const entries = fs.readdirSync(pnpmModules).filter(e => e.startsWith(`${dirName}@`));
+                if (!entries.length) return visited;
+                const pkgJson = path.join(pnpmModules, entries[0], 'node_modules', pkgName, 'package.json');
+                if (!fs.existsSync(pkgJson)) return visited;
+                const d = JSON.parse(fs.readFileSync(pkgJson, 'utf8'));
+                const allDeps = { ...d.dependencies, ...d.optionalDependencies };
+                for (const dep of Object.keys(allDeps)) collectDeps(dep, visited);
+                return visited;
+            }
+
+            const allPkgs = new Set();
+            for (const root of runtimeRoots) collectDeps(root, allPkgs);
+
+            for (const pkg of allPkgs) {
+                const dirName = pnpmDirName(pkg);
+                const entries = fs.readdirSync(pnpmModules).filter(e => e.startsWith(`${dirName}@`));
+                if (!entries.length) continue;
+                const pkgSrc = path.join(pnpmModules, entries[0], 'node_modules', pkg);
+                if (!fs.existsSync(pkgSrc)) continue;
+                // Scoped packages (@scope/pkg) need their parent dir created first
+                const pkgDestDir = path.join(destModules, path.dirname(pkg));
+                if (!fs.existsSync(pkgDestDir)) fs.mkdirSync(pkgDestDir, { recursive: true });
+                const pkgDest = path.join(destModules, pkg);
+                if (!fs.existsSync(pkgDest)) fs.cpSync(pkgSrc, pkgDest, { recursive: true });
+            }
+            console.log(`📦 Copied ${allPkgs.size} packages (parseFile runtime deps) → .package/node_modules/`);
+
             console.log('✅ All assets staged in .package/');
         },
     }