switch to new package index, decompress into .tar in javascript

build/pyodide_bucket.bzl

@@ -1,7 +1,8 @@
 # Do not edit this file by hand. See docs/pyodide.md for info on how to generate it.
 # These variables are factored out here because they are being shared by the WORKSPACE files in
 # both edgeworker and workerd, as well as src/pyodide/BUILD.bazel
-PYODIDE_PACKAGE_BUCKET_URL = "https://pub-45d734c4145d4285b343833ee450ef38.r2.dev/20240320/"
-PYODIDE_GITHUB_RELEASE_URL = "https://github.com/cloudflare/pyodide-build-scripts/releases/download/20240320/"
-PYODIDE_LOCK_SHA256 = "a176311d4c449aac4ef7a333977af8b6e08224c115a9a6d05c04592c841b8a58"
-PYODIDE_PACKAGES_TAR_ZIP_SHA256 = "e191bae60aad75f6a9c33fac5c0ff1ad2b0e564bdd7a07fbdc848df4b62c60a1"
+PYODIDE_PACKAGE_BUCKET_URL = "https://pub-45d734c4145d4285b343833ee450ef38.r2.dev/20240415-experimental/"
+PYODIDE_GITHUB_RELEASE_URL = "https://github.com/cloudflare/pyodide-build-scripts/releases/download/20240415-experimental/"
+PYODIDE_LOCK_SHA256 = "67d1a24edf4f3ab2cf85c736391c04763ff722bf3aebf9ea3469d96e5f51e1da"
+PYODIDE_PACKAGES_TAR_ZIP_SHA256 = "749967941204154e7ae866fe08f1216a3e5ee58ba6a3757231a5be0d9d4430f8"
+PYODIDE_ALL_WHEELS_ZIP_SHA256 = "9e7c330ee93d81d0356cc2d585f217dfee58b623ad4535282baa6e82bd063eee"

docs/pyodide.md

@@ -5,10 +5,11 @@ workerd is linked against a Pyodide lock file, which is located within an R2 buc
 If you know where the R2 bucket is (See build/pyodide_bucket.bzl) then the `pyodide-lock.json` file is located inside the root of the R2 directory for the Pyodide package bundle release.
 
 This lock file contains some information used by workerd to pull in package requirements, including but not limited to:
+
 - The versions of each package included in the package bundle
 - The file names and SHA hashes of each package available for download in the bucket
 - What the dependencies are for each package
 
 ## Generating pyodide_bucket.bzl
-We have scripts and GitHub actions set up for building and uploading Pyodide package bundles onto R2. These are available [here](https://github.com/cloudflare/pyodide-build-scripts). Simply follow the instructions on that repo to build a new version of Pyodide or a new package bundle release.
 
+We have scripts and GitHub actions set up for building and uploading Pyodide package bundles onto R2. These are available [here](https://github.com/cloudflare/pyodide-build-scripts). Simply follow the instructions on that repo to build a new version of Pyodide or a new package bundle release.

src/pyodide/internal/builtin_wrappers.js

@@ -114,25 +114,3 @@ export async function wasmInstantiate(module, imports) {
   const instance = new WebAssembly.Instance(module, imports);
   return { module, instance };
 }
-
-export function patchFetch(origin) {
-  // Patch fetch to first go through disk cache, but only when url points to origin
-  const origFetch = globalThis.fetch;
-  globalThis.fetch = async function (url, options) {
-    if (url.origin !== origin) {
-      return origFetch(url, options);
-    }
-
-    const fileName = url.pathname.substring(url.pathname.lastIndexOf("/") + 1);
-    const cached = DiskCache.get(fileName);
-    if (cached) {
-      return new Response(cached);
-    }
-
-    // we didn't find it in the disk cache, continue with original fetch
-    const response = await origFetch(url, options);
-    const arrayBuffer = await response.arrayBuffer();
-    DiskCache.put(fileName, arrayBuffer);
-    return new Response(arrayBuffer);
-  };
-}

src/pyodide/internal/loadPackage.js

@@ -0,0 +1,86 @@
+/**
+ * This file contains code that roughly replaces pyodide.loadPackage, with workerd-specific
+ * optimizations:
+ * - Wheels are decompressed with a DecompressionStream instead of in Python
+ * - Wheels are overlaid onto the site-packages dir instead of actually being copied
+ * - Wheels are fetched from a disk cache if available.
+ *
+ * Note that loadPackages is only used in local dev for now, internally we use the full big bundle
+ * that contains all the packages ready to go.
+ */
+
+import { default as LOCKFILE } from "pyodide-internal:generated/pyodide-lock.json";
+import { WORKERD_INDEX_URL } from "pyodide-internal:metadata";
+import { SITE_PACKAGES, LOAD_WHEELS_FROM_R2, getSitePackagesPath } from "pyodide-internal:setupPackages";
+import { parseTarInfo } from "pyodide-internal:tar";
+import { default as DiskCache } from "pyodide-internal:disk_cache";
+import { createTarFS } from "pyodide-internal:tarfs";
+
+async function loadBundle(requirement) {
+  // first check if the disk cache has what we want
+  const filename = LOCKFILE["packages"][requirement]["file_name"];
+  const cached = DiskCache.get(filename);
+  if (cached) {
+    return [requirement, cached];
+  }
+
+  // we didn't find it in the disk cache, continue with original fetch
+  const url = new URL(WORKERD_INDEX_URL + filename);
+  const response = await fetch(url);
+
+  const arrayBuffer = await new Response(response.body.pipeThrough(new DecompressionStream("gzip"))).arrayBuffer();
+
+  DiskCache.put(filename, arrayBuffer);
+  return [requirement, arrayBuffer];
+};
+
+/**
+ * ArrayBufferReader wraps around an arrayBuffer in a way that tar.js is able to read from
+ */
+class ArrayBufferReader {
+  constructor(arrayBuffer) {
+    this.arrayBuffer = arrayBuffer;
+  }
+
+  read(offset, buf){
+    // buf is a Uint8Array
+    const size = this.arrayBuffer.byteLength;
+    if (offset >= size || offset < 0) {
+      return 0;
+    }
+    let toCopy = buf.length;
+    if (size - offset < toCopy) {
+      toCopy = size - offset;
+    }
+    buf.set(new Uint8Array(this.arrayBuffer, offset, toCopy));
+    return toCopy;
+  }
+}
+
+export async function loadPackages(Module, requirements) {
+  if (!LOAD_WHEELS_FROM_R2) return;
+
+  let loadPromises = [];
+  let loading = [];
+  for (const req of requirements) {
+    if (SITE_PACKAGES.loadedRequirements.has(req)) continue;
+    loadPromises.push(loadBundle(req));
+    loading.push(req);
+  }
+
+  console.log("Loading " + loading.join(", "));
+
+  const buffers = await Promise.all(loadPromises);
+  for (const [requirement, buffer] of buffers) {
+    const reader = new ArrayBufferReader(buffer);
+    const [tarInfo, soFiles] = parseTarInfo(reader);
+    SITE_PACKAGES.addSmallBundle(tarInfo, soFiles, requirement);
+  }
+
+  console.log("Loaded " + loading.join(", "));
+
+  const tarFS = createTarFS(Module);
+  const path = getSitePackagesPath(Module);
+  const info = SITE_PACKAGES.rootInfo;
+  Module.FS.mount(tarFS, { info }, path);
+}

src/pyodide/internal/python.js

@@ -1,7 +1,8 @@
 Error.stackTraceLimit = Infinity;
 import { enterJaegerSpan } from "pyodide-internal:jaeger";
 import {
-  SITE_PACKAGES_INFO,
+  TRANSITIVE_REQUIREMENTS,
+  SITE_PACKAGES,
   adjustSysPath,
   mountLib,
 } from "pyodide-internal:setupPackages";
@@ -187,7 +188,7 @@ async function instantiateEmscriptenModule(emscriptenSettings) {
  */
 async function prepareWasmLinearMemory(Module) {
   // Note: if we are restoring from a snapshot, runtime is not initialized yet.
-  mountLib(Module, SITE_PACKAGES_INFO);
+  mountLib(Module, SITE_PACKAGES.rootInfo);
   entropyMountFiles(Module);
   if (SHOULD_RESTORE_SNAPSHOT) {
     restoreSnapshot(Module);

src/pyodide/internal/setupPackages.js

@@ -3,7 +3,6 @@ import { createTarFS } from "pyodide-internal:tarfs";
 import { createMetadataFS } from "pyodide-internal:metadatafs";
 import { default as LOCKFILE } from "pyodide-internal:generated/pyodide-lock.json";
 import { REQUIREMENTS, WORKERD_INDEX_URL } from "pyodide-internal:metadata";
-import { patchFetch } from "pyodide-internal:builtin_wrappers";
 import { simpleRunPython } from "pyodide-internal:util";
 
 const canonicalizeNameRegex = /[-_.]+/g;
@@ -23,6 +22,86 @@ const STDLIB_PACKAGES = Object.values(LOCKFILE.packages)
   .filter(({ install_dir }) => install_dir === "stdlib")
   .map(({ name }) => canonicalizePackageName(name));
 
+/**
+ * SitePackagesDir keeps track of the virtualized view of the site-packages
+ * directory generated for each worker.
+ */
+class SitePackagesDir {
+  constructor() {
+    this.rootInfo = {
+      children: new Map(),
+      mode: 0o777,
+      type: 5,
+      modtime: 0,
+      size: 0,
+      path: "",
+      name: "",
+      parts: [],
+    };
+    this.soFiles = [];
+    this.loadedRequirements = new Set();
+  }
+
+  /**
+   * mountOverlay "overlays" a directory onto the site-packages root directory.
+   * All files and subdirectories in the overlay will be accessible at site-packages by the worker.
+   * If a file or directory already exists, an error is thrown.
+   * @param {TarInfo} overlayInfo The directory that is to be "copied" into site-packages
+   */
+  mountOverlay(overlayInfo) {
+    overlayInfo.children.forEach((val, key) => {
+      if (this.rootInfo.children.has(key)) {
+        throw new Error(
+          `File/folder ${key} being written by multiple packages`,
+        );
+      }
+      this.rootInfo.children.set(key, val);
+    });
+  }
+
+  /**
+   * A small bundle contains just a single package. The entire bundle will be overlaid onto site-packages.
+   * A small bundle can basically be thought of as a wheel.
+   * @param {TarInfo} tarInfo The root tarInfo for the small bundle (See tar.js)
+   * @param {List<String>} soFiles A list of .so files contained in the small bundle
+   * @param {String} requirement The canonicalized package name this small bundle corresponds to
+   */
+  addSmallBundle(tarInfo, soFiles, requirement) {
+    for (const soFile of soFiles) {
+      this.soFiles.push(soFile.split("/"));
+    }
+    this.mountOverlay(tarInfo);
+    this.loadedRequirements.add(requirement);
+  }
+
+  /**
+   * A big bundle contains multiple packages, each package contained in a folder whose name is the canonicalized package name.
+   * This function overlays the requested packages onto the site-packages directory.
+   * @param {TarInfo} tarInfo The root tarInfo for the big bundle (See tar.js)
+   * @param {List<String>} soFiles A list of .so files contained in the big bundle
+   * @param {List<String>} requirements canonicalized list of packages to pick from the big bundle
+   */
+  addBigBundle(tarInfo, soFiles, requirements) {
+    // add all the .so files we will need to preload from the big bundle
+    for (const soFile of soFiles) {
+      // If folder is in list of requirements include .so file in list to preload.
+      const [pkg, ...rest] = soFile.split("/");
+      if (requirements.has(pkg)) {
+        this.soFiles.push(rest);
+      }
+    }
+
+    for (const req of requirements) {
+      const child = tarInfo.children.get(req);
+      if (!child) {
+        throw new Error(`Requirement ${req} not found in pyodide packages tar`);
+      }
+      this.mountOverlay(child);
+      this.loadedRequirements.add(req);
+    }
+  }
+};
+
 /**
  * This stitches together the view of the site packages directory. Each
  * requirement corresponds to a folder in the original tar file. For each
@@ -33,52 +112,19 @@ const STDLIB_PACKAGES = Object.values(LOCKFILE.packages)
  * directory so we can preload them.
  */
 export function buildSitePackages(requirements) {
-  const [origTarInfo, origSoFiles] = parseTarInfo();
-  // We'd like to set USE_LOAD_PACKAGE = IS_WORKERD but we also build a funny
-  // workerd with the downstream package set. We can distinguish between them by
-  // looking at the contents. This uses the fact that the downstream set is
-  // larger, but there are a lot of differences...
-  const USE_LOAD_PACKAGE = origTarInfo.children.size < 10;
-  if (USE_LOAD_PACKAGE) {
-    requirements = new Set([...STDLIB_PACKAGES]);
-  } else {
-    requirements = new Set([...STDLIB_PACKAGES, ...requirements]);
-  }
-  const soFiles = [];
-  for (const soFile of origSoFiles) {
-    // If folder is in list of requirements include .so file in list to preload.
-    const [pkg, ...rest] = soFile.split("/");
-    if (requirements.has(pkg)) {
-      soFiles.push(rest);
-    }
-  }
-  const newTarInfo = {
-    children: new Map(),
-    mode: 0o777,
-    type: 5,
-    modtime: 0,
-    size: 0,
-    path: "",
-    name: "",
-    parts: [],
-  };
-
-  for (const req of requirements) {
-    const child = origTarInfo.children.get(req);
-    if (!child) {
-      throw new Error(`Requirement ${req} not found in pyodide packages tar`);
-    }
-    child.children.forEach((val, key) => {
-      if (newTarInfo.children.has(key)) {
-        throw new Error(
-          `File/folder ${key} being written by multiple packages`,
-        );
-      }
-      newTarInfo.children.set(key, val);
-    });
+  const [bigTarInfo, bigTarSoFiles] = parseTarInfo();
+
+  let LOAD_WHEELS_FROM_R2 = true;
+  let requirementsInBigBundle = new Set([...STDLIB_PACKAGES]);
+  if (bigTarInfo.children.size > 10) {
+    LOAD_WHEELS_FROM_R2 = false;
+    requirements.forEach(r => requirementsInBigBundle.add(r));
   }
 
-  return [newTarInfo, soFiles, USE_LOAD_PACKAGE];
+  const res = new SitePackagesDir();
+  res.addBigBundle(bigTarInfo, bigTarSoFiles, requirementsInBigBundle);
+
+  return [res, LOAD_WHEELS_FROM_R2];
 }
 
 /**
@@ -89,23 +135,12 @@ export function buildSitePackages(requirements) {
  * TODO: stop using loadPackage in workerd.
  */
 export function patchLoadPackage(pyodide) {
-  if (!USE_LOAD_PACKAGE) {
-    pyodide.loadPackage = disabledLoadPackage;
-    return;
-  }
-  patchFetch(new URL(WORKERD_INDEX_URL).origin);
-  const origLoadPackage = pyodide.loadPackage;
-  function loadPackage(packages, options) {
-    return origLoadPackage(packages, {
-      checkIntegrity: false,
-      ...options,
-    });
-  }
-  pyodide.loadPackage = loadPackage;
+  pyodide.loadPackage = disabledLoadPackage;
+  return;
 }
 
 function disabledLoadPackage() {
-  throw new Error("We only use loadPackage in workerd");
+  throw new Error("pyodide.loadPackage is disabled because packages are encoded in the binary");
 }
 
 /**
@@ -138,7 +173,12 @@ export function mountLib(Module, info) {
   const site_packages = getSitePackagesPath(Module);
   Module.FS.mkdirTree(site_packages);
   Module.FS.mkdirTree("/session/metadata");
-  Module.FS.mount(tarFS, { info }, site_packages);
+  if (!LOAD_WHEELS_FROM_R2) {
+    // if we are not loading additional wheels from R2, then we're done
+    // with site-packages and we can mount it here. Otherwise, we must mount it in
+    // loadPackages().
+    Module.FS.mount(tarFS, { info }, site_packages);
+  }
   Module.FS.mount(mdFS, {}, "/session/metadata");
 }
 
@@ -191,5 +231,4 @@ function addPackageToLoad(lockfile, name, toLoad) {
 
 export { REQUIREMENTS };
 export const TRANSITIVE_REQUIREMENTS = getTransitiveRequirements();
-export const [SITE_PACKAGES_INFO, SITE_PACKAGES_SO_FILES, USE_LOAD_PACKAGE] =
-  buildSitePackages(TRANSITIVE_REQUIREMENTS);
+export const [SITE_PACKAGES, LOAD_WHEELS_FROM_R2] = buildSitePackages(TRANSITIVE_REQUIREMENTS);