Move npm and pip mitm interception to separate files

packages/safe-chain/src/registryProxy/interceptors/createInterceptorForEcoSystem.js

@@ -0,0 +1,25 @@
+import {
+  ECOSYSTEM_JS,
+  ECOSYSTEM_PY,
+  getEcoSystem,
+} from "../../config/settings.js";
+import { npmInterceptorForUrl } from "./npmInterceptor.js";
+import { pipInterceptorForUrl } from "./pipInterceptor.js";
+
+/**
+ * @param {string} url
+ * @returns {import("./interceptorBuilder.js").Interceptor | undefined}
+ */
+export function createInterceptorForUrl(url) {
+  const ecosystem = getEcoSystem();
+
+  if (ecosystem === ECOSYSTEM_JS) {
+    return npmInterceptorForUrl(url);
+  }
+
+  if (ecosystem === ECOSYSTEM_PY) {
+    return pipInterceptorForUrl(url);
+  }
+
+  return undefined;
+}

packages/safe-chain/src/registryProxy/interceptors/npmInterceptor.js

@@ -0,0 +1,82 @@
+import { isMalwarePackage } from "../../scanning/audit/index.js";
+import { createInterceptorBuilder } from "./interceptorBuilder.js";
+
+const knownJsRegistries = ["registry.npmjs.org", "registry.yarnpkg.com"];
+
+/**
+ * @param {string} url
+ * @returns {import("./interceptorBuilder.js").Interceptor | undefined}
+ */
+export function npmInterceptorForUrl(url) {
+  const registry = knownJsRegistries.find((reg) => url.includes(reg));
+
+  if (registry) {
+    return buildNpmInterceptor(registry);
+  }
+
+  return undefined;
+}
+
+/**
+ * @param {string} registry
+ * @returns {import("./interceptorBuilder.js").Interceptor | undefined}
+ */
+function buildNpmInterceptor(registry) {
+  const builder = createInterceptorBuilder();
+
+  builder.onRequest(async (req) => {
+    const { packageName, version } = parseNpmPackageUrl(
+      req.targetUrl,
+      registry
+    );
+    if (await isMalwarePackage(packageName, version)) {
+      req.blockRequest(403, "Forbidden - blocked by safe-chain");
+    }
+  });
+
+  return builder.build();
+}
+
+/**
+ * @param {string} url
+ * @param {string} registry
+ * @returns {{packageName: string | undefined, version: string | undefined}}
+ */
+export function parseNpmPackageUrl(url, registry) {
+  let packageName, version;
+  if (!registry || !url.endsWith(".tgz")) {
+    return { packageName, version };
+  }
+
+  const registryIndex = url.indexOf(registry);
+  const afterRegistry = url.substring(registryIndex + registry.length + 1); // +1 to skip the slash
+
+  const separatorIndex = afterRegistry.indexOf("/-/");
+  if (separatorIndex === -1) {
+    return { packageName, version };
+  }
+
+  packageName = afterRegistry.substring(0, separatorIndex);
+  const filename = afterRegistry.substring(
+    separatorIndex + 3,
+    afterRegistry.length - 4
+  ); // Remove /-/ and .tgz
+
+  // Extract version from filename
+  // For scoped packages like @babel/core, the filename is core-7.21.4.tgz
+  // For regular packages like lodash, the filename is lodash-4.17.21.tgz
+  if (packageName.startsWith("@")) {
+    const scopedPackageName = packageName.substring(
+      packageName.lastIndexOf("/") + 1
+    );
+    if (filename.startsWith(scopedPackageName + "-")) {
+      version = filename.substring(scopedPackageName.length + 1);
+    }
+  } else {
+    if (filename.startsWith(packageName + "-")) {
+      version = filename.substring(packageName.length + 1);
+    }
+  }
+
+  return { packageName, version };
+}

packages/safe-chain/src/registryProxy/interceptors/npmInterceptor.spec.js

@@ -0,0 +1,163 @@
+import { describe, it, mock } from "node:test";
+import assert from "node:assert";
+
+describe("npmInterceptor", async () => {
+  let lastPackage;
+  let malwareResponse = false;
+
+  mock.module("../../scanning/audit/index.js", {
+    namedExports: {
+      isMalwarePackage: async (packageName, version) => {
+        lastPackage = { packageName, version };
+        return malwareResponse;
+      },
+    },
+  });
+
+  const { npmInterceptorForUrl } = await import("./npmInterceptor.js");
+
+  const parserCases = [
+    // Regular packages
+    {
+      url: "https://registry.npmjs.org/lodash/-/lodash-4.17.21.tgz",
+      expected: { packageName: "lodash", version: "4.17.21" },
+    },
+    {
+      url: "https://registry.npmjs.org/express/-/express-4.18.2.tgz",
+      expected: { packageName: "express", version: "4.18.2" },
+    },
+    // Packages with hyphens in name
+    {
+      url: "https://registry.npmjs.org/safe-chain-test/-/safe-chain-test-1.0.0.tgz",
+      expected: { packageName: "safe-chain-test", version: "1.0.0" },
+    },
+    {
+      url: "https://registry.npmjs.org/web-vitals/-/web-vitals-3.5.0.tgz",
+      expected: { packageName: "web-vitals", version: "3.5.0" },
+    },
+    // Preview/prerelease versions
+    {
+      url: "https://registry.npmjs.org/safe-chain-test/-/safe-chain-test-0.0.1-security.tgz",
+      expected: { packageName: "safe-chain-test", version: "0.0.1-security" },
+    },
+    {
+      url: "https://registry.npmjs.org/lodash/-/lodash-5.0.0-beta.1.tgz",
+      expected: { packageName: "lodash", version: "5.0.0-beta.1" },
+    },
+    {
+      url: "https://registry.npmjs.org/react/-/react-18.3.0-canary-abc123.tgz",
+      expected: { packageName: "react", version: "18.3.0-canary-abc123" },
+    },
+    // Scoped packages
+    {
+      url: "https://registry.npmjs.org/@babel/core/-/core-7.21.4.tgz",
+      expected: { packageName: "@babel/core", version: "7.21.4" },
+    },
+    {
+      url: "https://registry.npmjs.org/@types/node/-/node-20.10.5.tgz",
+      expected: { packageName: "@types/node", version: "20.10.5" },
+    },
+    {
+      url: "https://registry.npmjs.org/@angular/common/-/common-17.0.8.tgz",
+      expected: { packageName: "@angular/common", version: "17.0.8" },
+    },
+    // Scoped packages with hyphens
+    {
+      url: "https://registry.npmjs.org/@safe-chain/test-package/-/test-package-2.1.0.tgz",
+      expected: { packageName: "@safe-chain/test-package", version: "2.1.0" },
+    },
+    {
+      url: "https://registry.npmjs.org/@aws-sdk/client-s3/-/client-s3-3.465.0.tgz",
+      expected: { packageName: "@aws-sdk/client-s3", version: "3.465.0" },
+    },
+    // Scoped packages with preview versions
+    {
+      url: "https://registry.npmjs.org/@babel/core/-/core-8.0.0-alpha.1.tgz",
+      expected: { packageName: "@babel/core", version: "8.0.0-alpha.1" },
+    },
+    {
+      url: "https://registry.npmjs.org/@safe-chain/security-test/-/security-test-1.0.0-security.tgz",
+      expected: {
+        packageName: "@safe-chain/security-test",
+        version: "1.0.0-security",
+      },
+    },
+    // Yarn registry
+    {
+      url: "https://registry.yarnpkg.com/lodash/-/lodash-4.17.21.tgz",
+      expected: { packageName: "lodash", version: "4.17.21" },
+    },
+    {
+      url: "https://registry.yarnpkg.com/@babel/core/-/core-7.21.4.tgz",
+      expected: { packageName: "@babel/core", version: "7.21.4" },
+    },
+    // URL to get package info, not tarball
+    {
+      url: "https://registry.npmjs.org/lodash",
+      expected: { packageName: undefined, version: undefined },
+    },
+    // Complex version patterns
+    {
+      url: "https://registry.npmjs.org/package-with-many-hyphens/-/package-with-many-hyphens-1.0.0-rc.1+build.123.tgz",
+      expected: {
+        packageName: "package-with-many-hyphens",
+        version: "1.0.0-rc.1+build.123",
+      },
+    },
+    {
+      url: "https://registry.npmjs.org/@scope/package-name-with-hyphens/-/package-name-with-hyphens-2.0.0-beta.2.tgz",
+      expected: {
+        packageName: "@scope/package-name-with-hyphens",
+        version: "2.0.0-beta.2",
+      },
+    },
+  ];
+
+  parserCases.forEach(({ url, expected }, index) => {
+    it(`should parse URL ${index + 1}: ${url}`, async () => {
+      const interceptor = npmInterceptorForUrl(url);
+      assert.ok(
+        interceptor,
+        "Interceptor should be created for known npm registry"
+      );
+
+      await interceptor.handleRequest(url);
+
+      assert.deepEqual(lastPackage, expected);
+    });
+  });
+
+  it("should not create interceptor for unknown registry", () => {
+    const url = "https://example.com/some-package/-/some-package-1.0.0.tgz";
+
+    const interceptor = npmInterceptorForUrl(url);
+
+    assert.equal(
+      interceptor,
+      undefined,
+      "Interceptor should be undefined for unknown registry"
+    );
+  });
+
+  it("should block malicious package", async () => {
+    const url =
+      "https://registry.npmjs.org/malicious-package/-/malicious-package-1.0.0.tgz";
+    malwareResponse = true;
+
+    const interceptor = npmInterceptorForUrl(url);
+
+    const result = await interceptor.handleRequest(url);
+
+    assert.ok(result.blockResponse, "Should contain a blockResponse");
+    assert.equal(
+      result.blockResponse.statusCode,
+      403,
+      "Block response should have status code 403"
+    );
+    assert.equal(
+      result.blockResponse.message,
+      "Forbidden - blocked by safe-chain",
+      "Block response should have correct status message"
+    );
+  });
+});

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

@@ -0,0 +1,119 @@
+import { isMalwarePackage } from "../../scanning/audit/index.js";
+import { createInterceptorBuilder } from "./interceptorBuilder.js";
+
+const knownPipRegistries = [
+  "files.pythonhosted.org",
+  "pypi.org",
+  "pypi.python.org",
+  "pythonhosted.org",
+];
+
+/**
+ * @param {string} url
+ * @returns {import("./interceptorBuilder.js").Interceptor | undefined}
+ */
+export function pipInterceptorForUrl(url) {
+  const registry = knownPipRegistries.find((reg) => url.includes(reg));
+
+  if (registry) {
+    return buildPipInterceptor(registry);
+  }
+
+  return undefined;
+}
+
+/**
+ * @param {string} registry
+ * @returns {import("./interceptorBuilder.js").Interceptor | undefined}
+ */
+function buildPipInterceptor(registry) {
+  const builder = createInterceptorBuilder();
+
+  builder.onRequest(async (req) => {
+    const { packageName, version } = parsePipPackageFromUrl(
+      req.targetUrl,
+      registry
+    );
+    if (await isMalwarePackage(packageName, version)) {
+      req.blockRequest(403, "Forbidden - blocked by safe-chain");
+    }
+  });
+
+  return builder.build();
+}
+
+/**
+ * @param {string} url
+ * @param {string} registry
+ * @returns {{packageName: string | undefined, version: string | undefined}}
+ */
+function parsePipPackageFromUrl(url, registry) {
+  let packageName, version;
+
+  // Basic validation
+  if (!registry || typeof url !== "string") {
+    return { packageName, version };
+  }
+
+  // Quick sanity check on the URL + parse
+  let urlObj;
+  try {
+    urlObj = new URL(url);
+  } catch {
+    return { packageName, version };
+  }
+
+  // Get the last path segment (filename) and decode it (strip query & fragment automatically)
+  const lastSegment = urlObj.pathname.split("/").filter(Boolean).pop();
+  if (!lastSegment) {
+    return { packageName, version };
+  }
+
+  const filename = decodeURIComponent(lastSegment);
+
+  // Parse Python package downloads from PyPI/pythonhosted.org
+  // Example wheel: https://files.pythonhosted.org/packages/xx/yy/requests-2.28.1-py3-none-any.whl
+  // Example sdist: https://files.pythonhosted.org/packages/xx/yy/requests-2.28.1.tar.gz
+
+  // Wheel (.whl)
+  if (filename.endsWith(".whl")) {
+    const base = filename.slice(0, -4); // remove ".whl"
+    const firstDash = base.indexOf("-");
+    if (firstDash > 0) {
+      const dist = base.slice(0, firstDash); // may contain underscores
+      const rest = base.slice(firstDash + 1); // version + the rest of tags
+      const secondDash = rest.indexOf("-");
+      const rawVersion = secondDash >= 0 ? rest.slice(0, secondDash) : rest;
+      packageName = dist; // preserve underscores
+      version = rawVersion;
+      // Reject "latest" as it's a placeholder, not a real version
+      // When version is "latest", this signals the URL doesn't contain actual version info
+      // Returning undefined allows the request (see registryProxy.js isAllowedUrl)
+      if (version === "latest" || !packageName || !version) {
+        return { packageName: undefined, version: undefined };
+      }
+      return { packageName, version };
+    }
+  }
+
+  // Source dist (sdist)
+  const sdistExtMatch = filename.match(/\.(tar\.gz|zip|tar\.bz2|tar\.xz)$/i);
+  if (sdistExtMatch) {
+    const base = filename.slice(0, -sdistExtMatch[0].length);
+    const lastDash = base.lastIndexOf("-");
+    if (lastDash > 0 && lastDash < base.length - 1) {
+      packageName = base.slice(0, lastDash);
+      version = base.slice(lastDash + 1);
+      // Reject "latest" as it's a placeholder, not a real version
+      // When version is "latest", this signals the URL doesn't contain actual version info
+      // Returning undefined allows the request (see registryProxy.js isAllowedUrl)
+      if (version === "latest" || !packageName || !version) {
+        return { packageName: undefined, version: undefined };
+      }
+      return { packageName, version };
+    }
+  }
+
+  // Unknown file type or invalid
+  return { packageName: undefined, version: undefined };
+}

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

@@ -0,0 +1,135 @@
+import { describe, it, mock } from "node:test";
+import assert from "node:assert";
+
+describe("pipInterceptor", async () => {
+  let lastPackage;
+  let malwareResponse = false;
+
+  mock.module("../../scanning/audit/index.js", {
+    namedExports: {
+      isMalwarePackage: async (packageName, version) => {
+        lastPackage = { packageName, version };
+        return malwareResponse;
+      },
+    },
+  });
+
+  const { pipInterceptorForUrl } = await import("./pipInterceptor.js");
+
+  const parserCases = [
+    // Valid pip URLs
+    {
+      url: "https://files.pythonhosted.org/packages/xx/yy/foobar-1.2.3.tar.gz",
+      expected: { packageName: "foobar", version: "1.2.3" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foobar/foobar-1.2.3.tar.gz",
+      expected: { packageName: "foobar", version: "1.2.3" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo-bar/foo-bar-0.9.0.tar.gz",
+      expected: { packageName: "foo-bar", version: "0.9.0" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0-py3-none-any.whl",
+      expected: { packageName: "foo_bar", version: "2.0.0" },
+    },
+    {
+      url: "https://files.pythonhosted.org/packages/xx/yy/foo_bar-2.0.0-py3-none-any.whl",
+      expected: { packageName: "foo_bar", version: "2.0.0" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo.bar/foo.bar-1.0.0.tar.gz",
+      expected: { packageName: "foo.bar", version: "1.0.0" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0b1.tar.gz",
+      expected: { packageName: "foo_bar", version: "2.0.0b1" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0rc1.tar.gz",
+      expected: { packageName: "foo_bar", version: "2.0.0rc1" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0.post1.tar.gz",
+      expected: { packageName: "foo_bar", version: "2.0.0.post1" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0.dev1.tar.gz",
+      expected: { packageName: "foo_bar", version: "2.0.0.dev1" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0a1.tar.gz",
+      expected: { packageName: "foo_bar", version: "2.0.0a1" },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-2.0.0-cp38-cp38-manylinux1_x86_64.whl",
+      expected: { packageName: "foo_bar", version: "2.0.0" },
+    },
+    // Invalid pip URLs
+    {
+      url: "https://pypi.org/simple/",
+      expected: { packageName: undefined, version: undefined },
+    },
+    {
+      url: "https://pypi.org/project/foobar/",
+      expected: { packageName: undefined, version: undefined },
+    },
+    {
+      url: "https://files.pythonhosted.org/packages/xx/yy/foobar-latest.tar.gz",
+      expected: { packageName: undefined, version: undefined },
+    },
+    {
+      url: "https://pypi.org/packages/source/f/foo_bar/foo_bar-latest.tar.gz",
+      expected: { packageName: undefined, version: undefined },
+    },
+  ];
+
+  parserCases.forEach(({ url, expected }, index) => {
+    it(`should parse URL ${index + 1}: ${url}`, async () => {
+      const interceptor = pipInterceptorForUrl(url);
+      assert.ok(
+        interceptor,
+        "Interceptor should be created for known npm registry"
+      );
+
+      await interceptor.handleRequest(url);
+
+      assert.deepEqual(lastPackage, expected);
+    });
+  });
+
+  it("should not create interceptor for unknown registry", () => {
+    const url = "https://example.com/packages/xx/yy/foobar-1.2.3.tar.gz";
+
+    const interceptor = pipInterceptorForUrl(url);
+
+    assert.equal(
+      interceptor,
+      undefined,
+      "Interceptor should be undefined for unknown registry"
+    );
+  });
+
+  it("should block malicious package", async () => {
+    const url =
+      "https://files.pythonhosted.org/packages/xx/yy/malicious_package-1.0.0.tar.gz";
+    malwareResponse = true;
+
+    const interceptor = pipInterceptorForUrl(url);
+
+    const result = await interceptor.handleRequest(url);
+
+    assert.ok(result.blockResponse, "Should contain a blockResponse");
+    assert.equal(
+      result.blockResponse.statusCode,
+      403,
+      "Block response should have status code 403"
+    );
+    assert.equal(
+      result.blockResponse.message,
+      "Forbidden - blocked by safe-chain",
+      "Block response should have correct status message"
+    );
+  });
+});