Adapt per review

packages/safe-chain/src/registryProxy/interceptors/pip/pipInterceptor.customRegistries.spec.js

@@ -2,7 +2,7 @@ import { describe, it, mock } from "node:test";
 import assert from "node:assert";
 
 describe("pipInterceptor custom registries", async () => {
-  let lastPackage;
+  let scannedPackages;
   let malwareResponse = false;
   let customRegistries = [];
 
@@ -27,7 +27,7 @@ describe("pipInterceptor custom registries", async () => {
   mock.module("../../../scanning/audit/index.js", {
     namedExports: {
       isMalwarePackage: async (packageName, version) => {
-        lastPackage = { packageName, version };
+        scannedPackages.push({ packageName, version });
         return malwareResponse;
       },
     },
@@ -46,6 +46,7 @@ describe("pipInterceptor custom registries", async () => {
   });
 
   it("should parse package from custom registry URL", async () => {
+    scannedPackages = [];
     customRegistries = ["my-custom-registry.example.com"];
     const url =
       "https://my-custom-registry.example.com/packages/xx/yy/foobar-1.2.3.tar.gz";
@@ -55,13 +56,16 @@ describe("pipInterceptor custom registries", async () => {
 
     await interceptor.handleRequest(url);
 
-    assert.deepEqual(lastPackage, {
+    assert.ok(
-      packageName: "foobar",
+      scannedPackages.some(
-      version: "1.2.3",
+        ({ packageName, version }) =>
-    });
+          packageName === "foobar" && version === "1.2.3"
+      )
+    );
   });
 
   it("should parse wheel package from custom registry URL", async () => {
+    scannedPackages = [];
     customRegistries = ["private-pypi.internal.com"];
     const url =
       "https://private-pypi.internal.com/packages/foo_bar-2.0.0-py3-none-any.whl";
@@ -71,10 +75,12 @@ describe("pipInterceptor custom registries", async () => {
 
     await interceptor.handleRequest(url);
 
-    assert.deepEqual(lastPackage, {
+    assert.ok(
-      packageName: "foo-bar",
+      scannedPackages.some(
-      version: "2.0.0",
+        ({ packageName, version }) =>
-    });
+          packageName === "foo-bar" && version === "2.0.0"
+      )
+    );
   });
 
   it("should handle multiple custom registries", async () => {
@@ -96,6 +102,7 @@ describe("pipInterceptor custom registries", async () => {
   });
 
   it("should block malicious package from custom registry", async () => {
+    scannedPackages = [];
     customRegistries = ["my-custom-registry.example.com"];
     malwareResponse = true;
 
@@ -115,6 +122,7 @@ describe("pipInterceptor custom registries", async () => {
   });
 
   it("should still work with known registries when custom registries are set", async () => {
+    scannedPackages = [];
     customRegistries = ["my-custom-registry.example.com"];
 
     const url =
@@ -126,10 +134,12 @@ describe("pipInterceptor custom registries", async () => {
 
     await interceptor.handleRequest(url);
 
-    assert.deepEqual(lastPackage, {
+    assert.ok(
-      packageName: "foobar",
+      scannedPackages.some(
-      version: "1.2.3",
+        ({ packageName, version }) =>
-    });
+          packageName === "foobar" && version === "1.2.3"
+      )
+    );
   });
 
   it("should not create interceptor for unknown registry when custom registries are set", () => {
@@ -152,6 +162,7 @@ describe("pipInterceptor custom registries", async () => {
   });
 
   it("should parse .whl.metadata from custom registry", async () => {
+    scannedPackages = [];
     customRegistries = ["private-pypi.internal.com"];
     const url =
       "https://private-pypi.internal.com/packages/foo_bar-2.0.0-py3-none-any.whl.metadata";
@@ -161,13 +172,16 @@ describe("pipInterceptor custom registries", async () => {
 
     await interceptor.handleRequest(url);
 
-    assert.deepEqual(lastPackage, {
+    assert.ok(
-      packageName: "foo-bar",
+      scannedPackages.some(
-      version: "2.0.0",
+        ({ packageName, version }) =>
-    });
+          packageName === "foo-bar" && version === "2.0.0"
+      )
+    );
   });
 
   it("should parse .tar.gz.metadata from custom registry", async () => {
+    scannedPackages = [];
     customRegistries = ["private-pypi.internal.com"];
     const url =
       "https://private-pypi.internal.com/packages/foo_bar-2.0.0.tar.gz.metadata";
@@ -177,9 +191,11 @@ describe("pipInterceptor custom registries", async () => {
 
     await interceptor.handleRequest(url);
 
-    assert.deepEqual(lastPackage, {
+    assert.ok(
-      packageName: "foo-bar",
+      scannedPackages.some(
-      version: "2.0.0",
+        ({ packageName, version }) =>
-    });
+          packageName === "foo-bar" && version === "2.0.0"
+      )
+    );
   });
 });

packages/safe-chain/src/registryProxy/interceptors/pip/pipInterceptor.js

@@ -1,8 +1,10 @@
 import {
+  ECOSYSTEM_PY,
   getPipCustomRegistries,
   skipMinimumPackageAge,
 } from "../../../config/settings.js";
 import { isMalwarePackage } from "../../../scanning/audit/index.js";
+import { getEquivalentPackageNames } from "../../../scanning/packageNameVariants.js";
 import { openNewPackagesDatabase } from "../../../scanning/newPackagesListCache.js";
 import { interceptRequests } from "../interceptorBuilder.js";
 import { isExcludedFromMinimumPackageAge } from "../minimumPackageAgeExclusions.js";
@@ -50,14 +52,21 @@ function createPipRequestHandler(registry) {
       registry
     );
 
-    // PyPI treats hyphens and underscores as equivalent distribution names.
+    if (!packageName) {
-    const hyphenName = packageName?.includes("_")
+      return;
-      ? packageName.replace(/_/g, "-")
+    }
-      : packageName;
 
-    const isMalicious =
+    const equivalentPackageNames = getEquivalentPackageNames(
-      await isMalwarePackage(packageName, version) ||
+      packageName,
-      await isMalwarePackage(hyphenName, version);
+      ECOSYSTEM_PY
+    );
+    let isMalicious = false;
+    for (const equivalentPackageName of equivalentPackageNames) {
+      if (await isMalwarePackage(equivalentPackageName, version)) {
+        isMalicious = true;
+        break;
+      }
+    }
 
     if (isMalicious) {
       reqContext.blockMalware(packageName, version);
@@ -65,7 +74,6 @@ function createPipRequestHandler(registry) {
     }
 
     if (
-      packageName &&
       version &&
       !skipMinimumPackageAge() &&
       !isExcludedFromMinimumPackageAge(packageName)

packages/safe-chain/src/registryProxy/interceptors/pip/pipInterceptor.packageDownload.spec.js

@@ -2,13 +2,13 @@ import { describe, it, mock } from "node:test";
 import assert from "node:assert";
 
 describe("pipInterceptor", async () => {
-  let lastPackage;
+  let scannedPackages;
   let malwareResponse = false;
 
   mock.module("../../../scanning/audit/index.js", {
     namedExports: {
       isMalwarePackage: async (packageName, version) => {
-        lastPackage = { packageName, version };
+        scannedPackages.push({ packageName, version });
         return malwareResponse;
       },
     },
@@ -111,12 +111,24 @@ describe("pipInterceptor", async () => {
 
   parserCases.forEach(({ url, expected }, index) => {
     it(`should parse URL ${index + 1}: ${url}`, async () => {
+      scannedPackages = [];
       const interceptor = pipInterceptorForUrl(url);
       assert.ok(interceptor, "Interceptor should be created for known pip registry");
 
       await interceptor.handleRequest(url);
 
-      assert.deepEqual(lastPackage, expected);
+      if (expected.packageName === undefined) {
+        assert.deepEqual(scannedPackages, []);
+        return;
+      }
+
+      assert.ok(
+        scannedPackages.some(
+          ({ packageName, version }) =>
+            packageName === expected.packageName &&
+            version === expected.version
+        )
+      );
     });
   });
 
@@ -127,6 +139,7 @@ describe("pipInterceptor", async () => {
   });
 
   it("should block malicious package", async () => {
+    scannedPackages = [];
     const url =
       "https://files.pythonhosted.org/packages/xx/yy/malicious_package-1.0.0.tar.gz";
     malwareResponse = true;

packages/safe-chain/src/scanning/packageNameVariants.js

@@ -10,7 +10,10 @@ export function getEquivalentPackageNames(packageName, ecosystem) {
     return [packageName];
   }
 
-  return [...new Set([packageName, ...["-", "_", "."].map((separator) =>
+  const pythonSeparatorPattern = /[._-]/g;
-    packageName.replaceAll(/[._-]/g, separator)
+  const hyphenName = packageName.replaceAll(pythonSeparatorPattern, "-");
-  )])];
+  const underscoreName = packageName.replaceAll(pythonSeparatorPattern, "_");
+  const dotName = packageName.replaceAll(pythonSeparatorPattern, ".");
+
+  return [...new Set([packageName, hyphenName, underscoreName, dotName])];
 }