Implements support for external package loading in validator.

src/pyodide/internal/python.ts

@@ -1,6 +1,7 @@
 import { enterJaegerSpan } from 'pyodide-internal:jaeger';
 import {
   SITE_PACKAGES,
+  TRANSITIVE_REQUIREMENTS,
   adjustSysPath,
   mountSitePackages,
   mountWorkerFiles,
@@ -50,6 +51,7 @@ import {
   setUnsafeEval,
   setGetRandomValues,
 } from 'pyodide-internal:generated/emscriptenSetup';
+import { loadPackages } from 'pyodide-internal:loadPackage';
 
 /**
  * After running `instantiateEmscriptenModule` but before calling into any C
@@ -62,6 +64,11 @@ async function prepareWasmLinearMemory(Module: Module): Promise<void> {
   mountSitePackages(Module, SITE_PACKAGES.rootInfo);
   entropyMountFiles(Module);
   Module.noInitialRun = !SHOULD_RESTORE_SNAPSHOT;
+
+  // NB. loadPackages adds the packages to the `SITE_PACKAGES` global which then gets used in
+  // preloadDynamicLibs.
+  await loadPackages(Module, TRANSITIVE_REQUIREMENTS);
+
   enterJaegerSpan('preload_dynamic_libs', () => preloadDynamicLibs(Module));
   enterJaegerSpan('remove_run_dependency', () =>
     Module.removeRunDependency('dynlibs')
@@ -100,9 +107,11 @@ export async function loadPyodide(
   }
   setUnsafeEval(UnsafeEval);
   setGetRandomValues(getRandomValues);
+
   await enterJaegerSpan('prepare_wasm_linear_memory', () =>
     prepareWasmLinearMemory(Module)
   );
+
   maybeSetupSnapshotUpload(Module);
   // Mount worker files after doing snapshot upload so we ensure that data from the files is never
   // present in snapshot memory.
@@ -111,6 +120,7 @@ export async function loadPyodide(
   // Finish setting up Pyodide's ffi so we can use the nice Python interface
   await enterJaegerSpan('finalize_bootstrap', Module.API.finalizeBootstrap);
   const pyodide = Module.API.public_api;
+
   finishSnapshotSetup(pyodide);
   return pyodide;
 }

src/pyodide/internal/setupPackages.ts

@@ -8,6 +8,7 @@ import {
   LOAD_WHEELS_FROM_ARTIFACT_BUNDLER,
 } from 'pyodide-internal:metadata';
 import { simpleRunPython } from 'pyodide-internal:util';
+import { default as EmbeddedPackagesTarReader } from 'pyodide-internal:packages_tar_reader';
 
 const canonicalizeNameRegex = /[-_.]+/g;
 
@@ -44,6 +45,7 @@ class SitePackagesDir {
       path: '',
       name: '',
       parts: [],
+      reader: null,
     };
     this.soFiles = [];
     this.loadedRequirements = new Set();
@@ -125,9 +127,11 @@ class SitePackagesDir {
  *
  * This also returns the list of soFiles in the resulting site-packages
  * directory so we can preload them.
+ *
+ * TODO(later): This needs to be removed when external package loading is enabled.
  */
 export function buildSitePackages(requirements: Set<string>): SitePackagesDir {
-  const [bigTarInfo, bigTarSoFiles] = parseTarInfo();
+  const [bigTarInfo, bigTarSoFiles] = parseTarInfo(EmbeddedPackagesTarReader);
 
   let requirementsInBigBundle = new Set([...STDLIB_PACKAGES]);
 
@@ -171,6 +175,7 @@ function disabledLoadPackage(): never {
 function getTransitiveRequirements(): Set<string> {
   const requirements = REQUIREMENTS.map(canonicalizePackageName);
   // resolve transitive dependencies of requirements and if IN_WORKERD install them from the cdn.
+  // TODO(later): use current package's LOCKFILE instead of the global.
   const packageDatas = recursiveDependencies(LOCKFILE, requirements);
   return new Set(packageDatas.map(({ name }) => canonicalizePackageName(name)));
 }

src/pyodide/internal/snapshot.ts

@@ -5,7 +5,7 @@ import {
   SITE_PACKAGES,
   getSitePackagesPath,
 } from 'pyodide-internal:setupPackages';
-import { default as TarReader } from 'pyodide-internal:packages_tar_reader';
+import { default as EmbeddedPackagesTarReader } from 'pyodide-internal:packages_tar_reader';
 import {
   SHOULD_SNAPSHOT_TO_DISK,
   IS_CREATING_BASELINE_SNAPSHOT,
@@ -136,7 +136,10 @@ export function preloadDynamicLibs(Module: Module): void {
         throw Error('contentsOffset not defined for ' + soFile);
       }
       const wasmModuleData = new Uint8Array(size);
-      TarReader.read(contentsOffset, wasmModuleData);
+      (node.reader ?? EmbeddedPackagesTarReader).read(
+        contentsOffset,
+        wasmModuleData
+      );
       const path = sitePackages + '/' + soFile.join('/');
       PRELOADED_SO_FILES.push(path);
       loadDynlib(Module, path, wasmModuleData);

src/pyodide/internal/tar.ts

@@ -1,5 +1,3 @@
-import { default as TarReader } from 'pyodide-internal:packages_tar_reader';
-
 // This is based on the info about the tar file format on wikipedia
 // And some trial and error with real tar files.
 // https://en.wikipedia.org/wiki/Tar_(computing)#File_format
@@ -44,7 +42,7 @@ function decodeHeader(buf: Uint8Array, reader: Reader): TarFSInfo {
   };
 }
 
-export function parseTarInfo(reader = TarReader): [TarFSInfo, string[]] {
+export function parseTarInfo(reader: Reader): [TarFSInfo, string[]] {
   const directories: TarFSInfo[] = [];
   const soFiles = [];
   const root: TarFSInfo = {

src/pyodide/python-entrypoint-helper.ts

@@ -21,7 +21,6 @@ import {
 import { reportError } from 'pyodide-internal:util';
 import { default as Limiter } from 'pyodide-internal:limiter';
 import { entropyBeforeRequest } from 'pyodide-internal:topLevelEntropy/lib';
-import { loadPackages } from 'pyodide-internal:loadPackage';
 
 function pyimportMainModule(pyodide: Pyodide): PyModule {
   if (!MAIN_MODULE_NAME.endsWith('.py')) {
@@ -74,21 +73,8 @@ async function applyPatch(pyodide: Pyodide, patchName: string): Promise<void> {
   pyodide.pyimport(patchName + '_patch');
 }
 
-/**
- * Set up Python packages:
- *  - patch loadPackage to ignore integrity
- *  - get requirements
- *  - Use tar file + requirements to mount site packages directory
- *  - if in workerd use loadPackage to load packages
- *  - install patches to make various requests packages work
- *
- * TODO: move this into setupPackages.js. Can't now because the patch imports
- * fail from there for some reason.
- */
-export async function setupPackages(pyodide: Pyodide): Promise<void> {
-  return await enterJaegerSpan('setup_packages', async () => {
-    patchLoadPackage(pyodide);
-    await loadPackages(pyodide._module, TRANSITIVE_REQUIREMENTS);
+async function setupPatches(pyodide: Pyodide): Promise<void> {
+  return await enterJaegerSpan('setup_patches', async () => {
     // install any extra packages into the site-packages directory, so calculate where that is.
     const pymajor = pyodide._module._py_version_major();
     const pyminor = pyodide._module._py_version_minor();
@@ -119,7 +105,8 @@ function getMainModule(): Promise<PyModule> {
     }
     mainModulePromise = (async function () {
       const pyodide = await getPyodide();
-      await setupPackages(pyodide);
+      patchLoadPackage(pyodide);
+      await setupPatches(pyodide);
       Limiter.beginStartup();
       try {
         return enterJaegerSpan('pyimport_main_module', () =>
@@ -173,17 +160,6 @@ const handlers: {
 try {
   // Do not setup anything to do with Python in the global scope when tracing. The Jaeger tracing
   // needs to be called inside an IO context.
-  if (!IS_TRACING) {
-    if (IS_WORKERD) {
-      // If we're in workerd, we have to do setupPackages in the IoContext, so don't start it yet.
-      // TODO: fix this.
-      await getPyodide();
-    } else {
-      // If we're not in workerd, setupPackages doesn't require IO so we can do it all here.
-      await getMainModule();
-    }
-  }
-
   if (IS_WORKERD || IS_TRACING) {
     handlers.fetch = makeHandler('on_fetch');
     handlers.test = makeHandler('test');

src/pyodide/types/FS.d.ts

@@ -12,7 +12,7 @@ interface TarFSInfo {
   name: string;
   parts: string[];
   contentsOffset?: number;
-  reader?: Reader;
+  reader: Reader | null;
 }
 
 declare type MetadataDirInfo = Map<string, MetadataDirInfo>;

src/workerd/server/workerd-api.c++

@@ -558,6 +558,7 @@ void WorkerdApi::compileModules(jsg::Lock& lockParam,
           makePyodideMetadataReader(conf, impl->pythonConfig), jsg::ModuleRegistry::Type::INTERNAL);
 
       // Inject packages tar file
+      // TODO(later): This shouldn't exist once featureFlags.getPythonExternalPackages() is true.
       modules->addBuiltinModule("pyodide-internal:packages_tar_reader",
           jsg::alloc<ReadOnlyBuffer>(PYODIDE_PACKAGES_TAR.get()),
           workerd::jsg::ModuleRegistry::Type::INTERNAL);