Don't fail at upload time when module modifies sys.path

We try to execute all the top level imports but we don't execute
changes to the path so we can get crashes.

Author: hoodmane

src/pyodide/internal/snapshot.ts

@@ -13,7 +13,11 @@ import {
   MEMORY_SNAPSHOT_READER,
   REQUIREMENTS,
 } from 'pyodide-internal:metadata';
-import { reportError, simpleRunPython } from 'pyodide-internal:util';
+import {
+  reportError,
+  simpleRunPython,
+  SimpleRunPythonError,
+} from 'pyodide-internal:util';
 import { default as MetadataReader } from 'pyodide-internal:runtime-generated/metadata';
 
 let LOADED_BASELINE_SNAPSHOT: number;
@@ -262,8 +266,22 @@ function memorySnapshotDoImports(Module: Module): string[] {
   const deduplicatedModules = [...new Set(importedModules)];
 
   // Import the modules list so they are included in the snapshot.
-  if (deduplicatedModules.length > 0) {
-    simpleRunPython(Module, 'import ' + deduplicatedModules.join(','));
+  for (const mod of deduplicatedModules) {
+    try {
+      // If the module manipulates sys.path or sys.meta_path, we hopefully won't be able to import
+      // some of the things it imports. If they import a package that attempts to import from js,
+      // then this might leave the package in an unusable state. TODO: finalizeBootstrap before
+      // taking the snapshot.
+      simpleRunPython(Module, `import ${mod}`);
+    } catch (e) {
+      if (!(e instanceof SimpleRunPythonError)) {
+        // This probably means we segfaulted.
+        throw e;
+      }
+      continue;
+    }
+    // Delete the imported module to avoid polluting the namespace
+    simpleRunPython(Module, `del ${mod.split('.', 1)[0]}`);
   }
 
   return deduplicatedModules;

src/pyodide/internal/util.ts

@@ -1,8 +1,34 @@
 export function reportError(e: any): never {
-  e.stack?.split('\n').forEach((s: any) => console.warn(s));
+  if (e instanceof SimpleRunPythonError) {
+    // PyRun_SimpleString will have written a Python traceback to stderr.
+    console.warn('Command failed:', e.pyCode);
+    console.warn('Error was:');
+    for (const line of e.pyTraceback.split('\n')) {
+      console.warn(line);
+    }
+  } else {
+    e.stack?.split('\n').forEach((s: any) => console.warn(s));
+  }
   throw e;
 }
 
+function setErrorName(errClass: any) {
+  Object.defineProperty(errClass.prototype, 'name', {
+    value: errClass.name,
+  });
+}
+
+export class SimpleRunPythonError extends Error {
+  pyCode: string;
+  pyTraceback: string;
+  constructor(pyCode: string, pyTraceback: string) {
+    super('Failed to run Python code: ' + pyTraceback);
+    this.pyCode = pyCode;
+    this.pyTraceback = pyTraceback;
+  }
+}
+setErrorName(SimpleRunPythonError);
+
 /**
  *  Simple as possible runPython function which works with no foreign function
  *  interface. We need to use this rather than the normal easier to use
@@ -36,13 +62,7 @@ export function simpleRunPython(
   // status 0: Ok
   // status -1: Error
   if (status) {
-    // PyRun_SimpleString will have written a Python traceback to stderr.
-    console.warn('Command failed:', code);
-    console.warn('Error was:');
-    for (const line of err.split('\n')) {
-      console.warn(line);
-    }
-    throw new Error('Failed to run Python code: ' + err);
+    throw new SimpleRunPythonError(code, err);
   }
   return err;
 }