File
Blob: src/pyodide/internal/util.ts
| 1 | // Copyright (c) 2026 Cloudflare, Inc. |
| 2 | // Licensed under the Apache 2.0 license found in the LICENSE file or at: |
| 3 | // https://opensource.org/licenses/Apache-2.0 |
| 4 | |
| 5 | // Callback used to report PythonWorkersInternalError construction to C++ metrics (via the |
| 6 | // WorkerFatalReporter module). Registered by `python.ts` at module init in the main workerd |
| 7 | // context. |
| 8 | // |
| 9 | // We can't `import` `pyodide-internal:fatal-reporter` (a C++ extension module) directly here |
| 10 | // because `util.ts` is also esbuild-bundled into the pool context (`pool/emscriptenSetup.ts` |
| 11 | // runs in a vanilla V8 isolate with no C++ extension module support). A static import would fail |
| 12 | // esbuild's resolver; a dynamic import fails at runtime because workerd's dynamic module |
| 13 | // resolver doesn't surface INTERNAL C++-backed modules. The callback indirection keeps |
| 14 | // `util.ts` free of any reference to the C++ module while still letting the main context wire |
| 15 | // in the real reporter. |
| 16 | // |
| 17 | // TODO: If we ever remove the Python pool, `util.ts` will no longer be pool-bundled and we can |
| 18 | // drop this callback in favor of a direct static import of FatalReporter. |
| 19 | let _reportInternalError: (() => void) | null = null; |
| 20 | |
| 21 | export function setInternalErrorReporter(cb: () => void): void { |
| 22 | _reportInternalError = cb; |
| 23 | } |
| 24 | |
| 25 | /** |
| 26 | * This is an exception we should be throwing whenever there is something unexpected in our runtime |
| 27 | * that is **not** a result of the user doing something wrong, i.e. it's an internal error that is |
| 28 | * a result of a bug in our runtime. |
| 29 | */ |
| 30 | export class PythonWorkersInternalError extends Error { |
| 31 | constructor(message?: string) { |
| 32 | super(message); |
| 33 | try { |
| 34 | _reportInternalError?.(); |
| 35 | } catch (_) {} |
| 36 | } |
| 37 | |
| 38 | override get name(): string { |
| 39 | return this.constructor.name; |
| 40 | } |
| 41 | } |
| 42 | |
| 43 | /** |
| 44 | * This is an exception we throw whenever there is an issue with the user's code, i.e. it's a result |
| 45 | * of the user doing something wrong. |
| 46 | */ |
| 47 | export class PythonUserError extends Error { |
| 48 | override get name(): string { |
| 49 | return this.constructor.name; |
| 50 | } |
| 51 | } |
| 52 | |
| 53 | // Split the stack into lines and print them individually. |
| 54 | // We do this because edgeworker's test runner will put a multiline log all on one line. This is |
| 55 | // very hard to read. |
| 56 | export function reportError(e: Error): never { |
| 57 | e.stack?.split('\n').forEach((s: string) => { |
| 58 | console.warn(s); |
| 59 | }); |
| 60 | throw e; |
| 61 | } |
| 62 | |
| 63 | /** |
| 64 | * Simple as possible runPython function which works with no foreign function |
| 65 | * interface. We need to use this rather than the normal easier to use |
| 66 | * interface because the normal interface doesn't work until after |
| 67 | * `API.finalizeBootstrap`, but `API.finalizeBootstrap` makes changes inside |
| 68 | * and outside the linear memory which have to stay in sync. It's hard to keep |
| 69 | * track of the invariants that `finalizeBootstrap` introduces between JS land |
| 70 | * and the linear memory so we do this. |
| 71 | * |
| 72 | * We wrap API.rawRun which does the following steps: |
| 73 | * 1. use textEncoder.encode to convert `code` into UTF8 bytes |
| 74 | * 2. malloc space for `code` in the wasm linear memory and copy the encoded |
| 75 | * `code` to this pointer |
| 76 | * 3. redirect standard error to a temporary buffer |
| 77 | * 4. call `PyRun_SimpleString`, which either works and returns 0 or formats a |
| 78 | * traceback to stderr and returns -1 |
| 79 | * https://docs.python.org/3/c-api/veryhigh.html?highlight=simplestring#c.PyRun_SimpleString |
| 80 | * 5. frees the `code` pointer |
| 81 | * 6. Returns the return value from `PyRun_SimpleString` and whatever |
| 82 | * information went to stderr. |
| 83 | * |
| 84 | * PyRun_SimpleString executes the code at top level in the `__main__` module, |
| 85 | * so all variables defined get leaked into the global namespace unless we |
| 86 | * clean them up explicitly. |
| 87 | */ |
| 88 | export function simpleRunPython( |
| 89 | emscriptenModule: Module, |
| 90 | code: string |
| 91 | ): string { |
| 92 | const [status, cause] = emscriptenModule.API.rawRun(code); |
| 93 | // status 0: Ok |
| 94 | // status -1: Error |
| 95 | if (status === -1) { |
| 96 | // PyRun_SimpleString will have written a Python traceback to stderr. |
| 97 | console.warn('Command failed:', code); |
| 98 | console.warn(cause); |
| 99 | throw new PythonWorkersInternalError( |
| 100 | 'Failed to run Python code:\n' + code + '\n\nError:\n' + cause |
| 101 | ); |
| 102 | } |
| 103 | return cause; |
| 104 | } |
| 105 | |
| 106 | export function invalidateCaches(Module: Module): void { |
| 107 | simpleRunPython( |
| 108 | Module, |
| 109 | `from importlib import invalidate_caches; invalidate_caches(); del invalidate_caches` |
| 110 | ); |
| 111 | } |
| 112 | |
| 113 | export function unreachable(obj: never, msg?: string): never { |
| 114 | if (msg === undefined) { |
| 115 | msg = obj; |
| 116 | } |
| 117 | throw new PythonWorkersInternalError(`Unreachable: ${msg}`); |
| 118 | } |
| 119 | |
| 120 | /** |
| 121 | * Loads a Python source file (bundled as a Uint8Array) into a fresh anonymous |
| 122 | * module and returns it. This is used for internal Python helpers that need to |
| 123 | * be invoked from JS but should not pollute the global namespace. |
| 124 | * |
| 125 | * `moduleName` is the bare name of the module (e.g. "introspection"); typically |
| 126 | * the source is the default export from `pyodide-internal:<moduleName>.py`. |
| 127 | */ |
| 128 | export function loadPythonMod( |
| 129 | pyodide: Pyodide, |
| 130 | moduleName: string, |
| 131 | source: Uint8Array |
| 132 | ): { __dict__: PyDict } { |
| 133 | const mod = pyodide.runPython( |
| 134 | `from types import ModuleType; ModuleType('${moduleName}')` |
| 135 | ) as { __dict__: PyDict }; |
| 136 | const decoder = new TextDecoder(); |
| 137 | pyodide.runPython(decoder.decode(source), { |
| 138 | globals: mod.__dict__, |
| 139 | filename: `${moduleName}.py`, |
| 140 | }); |
| 141 | return mod; |
| 142 | } |