feat: extract module paths from code.js via fileFinishedImporting, rename extract output to source paths

src/update/shared.ts

@@ -7,6 +7,7 @@ export const workFolder = "tmp";
 export const apksFolder = join(workFolder, "apks");
 export const codePath = join(workFolder, "code.js");
 export const modulesPath = join(workFolder, "modules");
+export const modulePathsDest = join("..", "data", "module-paths.json");
 export const apkSplits = ["base", "config.hdpi", "config.xxhdpi"] as const;
 
 export const oprevFiles = ["source.jsonl"] as const;

src/update/tasks/code.ts

@@ -2,9 +2,20 @@ import { mkdir, rm, readdir, stat } from "node:fs/promises";
 import { join as pathJoin } from "node:path";
 import { commit } from "../git";
 import type { Progress } from "../progress";
-import { commitAnyway, cuteVersion, modulesPath } from "../shared";
+import { commitAnyway, cuteVersion, modulePathsDest, modulesPath } from "../shared";
 import { join, sortEntries } from "../utils";
 
+function extractModuleId(filename: string): number | undefined {
+	const m1 = filename.match(/^module_(\d+)\.js$/);
+	if (m1) return Number.parseInt(m1[1], 10);
+	const m2 = filename.match(/^(\d+)_/);
+	if (m2) return Number.parseInt(m2[1], 10);
+}
+
+function sanitizePath(p: string): string {
+	return p.replace(/\\/g, "/").replace(/^\.\.\//, "").replace(/[<>:"|?*]/g, "_");
+}
+
 export default async function code(progress: Progress, _code: string[]) {
 	progress.start("code_getting");
 
@@ -12,6 +23,15 @@ export default async function code(progress: Progress, _code: string[]) {
 		throw new Error(`Modules directory not found at ${modulesPath}`);
 	}
 
+	const modulePaths = new Map<number, string>();
+	try {
+		const arr: { id: number; path: string }[] = await Bun.file(modulePathsDest).json();
+		for (const { id, path } of arr) modulePaths.set(id, path);
+		console.log(`Loaded ${modulePaths.size} module path mappings`);
+	} catch {
+		console.log("No module-paths.json found, using extract names");
+	}
+
 	async function scanDir(dir: string, basePath: string): Promise<Map<string, number>> {
 		const files = new Map<string, number>();
 		const entries = await readdir(dir, { withFileTypes: true });
@@ -31,9 +51,24 @@ export default async function code(progress: Progress, _code: string[]) {
 
 	const rawFiles = await scanDir(modulesPath, modulesPath);
 
+	function resolveDest(relativePath: string): string {
+		const id = extractModuleId(relativePath);
+		if (id !== undefined) {
+			const sourcePath = modulePaths.get(id);
+			if (sourcePath) return sanitizePath(sourcePath);
+		}
+		return relativePath;
+	}
+
+	const files = new Map<string, number>();
+	for (const [relativePath, fileSize] of rawFiles) {
+		const dest = resolveDest(relativePath);
+		if (!files.has(dest)) files.set(dest, fileSize);
+	}
+
 	await Bun.write(
 		"../data/source.jsonl",
-		sortEntries([...rawFiles.entries()])
+		sortEntries([...files.entries()])
 			.map(([file, size]) => `{ "file": ${JSON.stringify(file)}, "size": ${size} }`)
 			.join("\n"),
 	);
@@ -48,7 +83,7 @@ export default async function code(progress: Progress, _code: string[]) {
 
 		await Promise.all(
 			[...rawFiles.entries()].map(async ([relativePath]) => {
-				const dest = join(filePrefix, relativePath);
+				const dest = join(filePrefix, resolveDest(relativePath));
 				await mkdir(pathJoin(dest, ".."), { recursive: true });
 				await Bun.write(dest, Bun.file(join(modulesPath, relativePath)));
 			}),

src/update/tasks/decompile.ts

@@ -1,7 +1,7 @@
 import { exists } from "node:fs/promises";
 import { commit } from "../git";
 import type { Progress } from "../progress";
-import { codePath, commitAnyway, cuteVersion, modulesPath, workFolder } from "../shared";
+import { codePath, commitAnyway, cuteVersion, modulePathsDest, modulesPath, workFolder } from "../shared";
 import { handleShellErr, join } from "../utils";
 
 const gzipWorkerURL = new URL("decompile-gzip.ts", import.meta.url).href;
@@ -47,6 +47,20 @@ export default async function decompile(progress: Progress, pathToBundle: string
 
 	progress.update("decompile_decompiling", true);
 
+	progress.start("decompile_paths");
+	const codeJs = await Bun.file(codePath).text();
+	const modulePaths: { id: number; path: string }[] = [];
+	const moduleSections = codeJs.split(/\/\/ === Module (\d+): .+ ===\n?/);
+	for (let i = 1; i < moduleSections.length; i += 2) {
+		const modId = Number.parseInt(moduleSections[i], 10);
+		const content = moduleSections[i + 1] || "";
+		const m = content.match(/fileFinishedImporting\("([^"]+)"\)/);
+		if (m) modulePaths.push({ id: modId, path: m[1] });
+	}
+	await Bun.write(modulePathsDest, JSON.stringify(modulePaths));
+	console.log(`Extracted ${modulePaths.length} module paths`);
+	progress.update("decompile_paths", true);
+
 	if (process.env.NODE_ENV !== "test" && !commitAnyway) {
 		const gzFile = "code.js.gz";
 
@@ -54,7 +68,7 @@ export default async function decompile(progress: Progress, pathToBundle: string
 		progress.start("decompile_gzip");
 		gzipper.addEventListener("message", async ({ data }) => {
 			if (data === true) {
-				await commit([gzFile], `chore: update decompiled code for ${cuteVersion}`);
+				await commit([gzFile, "module-paths.json"], `chore: update decompiled code for ${cuteVersion}`);
 				progress.update("decompile_gzip", true);
 			}
 			gzipper.terminate();