Merge pull request #142 from AikidoSec/feature/pypi-ci

[PYPI] Add CI Shims

package-lock.json

@@ -2096,6 +2096,8 @@
         "aikido-npx": "bin/aikido-npx.js",
         "aikido-pip": "bin/aikido-pip.js",
         "aikido-pip3": "bin/aikido-pip3.js",
+        "aikido-python": "bin/aikido-python.js",
+        "aikido-python3": "bin/aikido-python3.js",
         "aikido-pnpm": "bin/aikido-pnpm.js",
         "aikido-pnpx": "bin/aikido-pnpx.js",
         "aikido-yarn": "bin/aikido-yarn.js",

packages/safe-chain/bin/aikido-pip.js

@@ -3,17 +3,16 @@
 import { main } from "../src/main.js";
 import { initializePackageManager } from "../src/packagemanager/currentPackageManager.js";
 import { setEcoSystem, ECOSYSTEM_PY } from "../src/config/settings.js";
-
-// Defaults
-let packageManagerName = "pip";
-// Pass through user args as-is
-const argv = process.argv.slice(2);
+import { setCurrentPipInvocation, PIP_INVOCATIONS, PIP_PACKAGE_MANAGER } from "../src/packagemanager/pip/pipSettings.js";
 
 // Set eco system
-// This can be used in other parts of the code to determine which eco system we are working with
 setEcoSystem(ECOSYSTEM_PY);
 
-initializePackageManager(packageManagerName);
-var exitCode = await main(argv);
+// Set current invocation
+setCurrentPipInvocation(PIP_INVOCATIONS.PIP);
 
+initializePackageManager(PIP_PACKAGE_MANAGER);
+
+// Pass through only user-supplied pip args
+var exitCode = await main(process.argv.slice(2));
 process.exit(exitCode);

packages/safe-chain/bin/aikido-pip3.js

@@ -3,17 +3,17 @@
 import { main } from "../src/main.js";
 import { initializePackageManager } from "../src/packagemanager/currentPackageManager.js";
 import { setEcoSystem, ECOSYSTEM_PY } from "../src/config/settings.js";
+import { setCurrentPipInvocation, PIP_INVOCATIONS, PIP_PACKAGE_MANAGER } from "../src/packagemanager/pip/pipSettings.js";
 
-// Explicit pip3 entrypoint
-const packageManagerName = "pip3";
-
-// Copy argv as-is
-const argv = process.argv.slice(2);
-
-// Set ecosystem to Python
+// Set eco system
 setEcoSystem(ECOSYSTEM_PY);
 
-initializePackageManager(packageManagerName);
-var exitCode = await main(argv);
+// Set current invocation
+setCurrentPipInvocation(PIP_INVOCATIONS.PIP3);
 
+// Create package manager
+initializePackageManager(PIP_PACKAGE_MANAGER);
+
+// Pass through only user-supplied pip args
+var exitCode = await main(process.argv.slice(2));
 process.exit(exitCode);

packages/safe-chain/bin/aikido-python.js

@@ -0,0 +1,28 @@
+#!/usr/bin/env node
+
+import { initializePackageManager } from "../src/packagemanager/currentPackageManager.js";
+import { setCurrentPipInvocation, PIP_INVOCATIONS, PIP_PACKAGE_MANAGER } from "../src/packagemanager/pip/pipSettings.js";
+import { setEcoSystem, ECOSYSTEM_PY } from "../src/config/settings.js";
+import { main } from "../src/main.js";
+
+// Set eco system
+setEcoSystem(ECOSYSTEM_PY);
+
+// Strip nodejs and wrapper script from args
+let argv = process.argv.slice(2);
+
+if (argv[0] === '-m' && (argv[1] === 'pip' || argv[1] === 'pip3')) {
+	setEcoSystem(ECOSYSTEM_PY);
+	setCurrentPipInvocation(argv[1] === 'pip3' ? PIP_INVOCATIONS.PY_PIP3 : PIP_INVOCATIONS.PY_PIP);
+	initializePackageManager(PIP_PACKAGE_MANAGER);
+
+  // Strip off the '-m pip' or '-m pip3' from the args
+  argv = argv.slice(2);
+
+  var exitCode = await main(argv);
+	process.exit(exitCode);
+} else {
+	// Forward to real python binary for non-pip flows
+	const { spawn } = await import('child_process');
+	spawn('python', argv, { stdio: 'inherit' });
+}

packages/safe-chain/bin/aikido-python3.js

@@ -0,0 +1,28 @@
+#!/usr/bin/env node
+
+import { initializePackageManager } from "../src/packagemanager/currentPackageManager.js";
+import { setCurrentPipInvocation, PIP_INVOCATIONS, PIP_PACKAGE_MANAGER } from "../src/packagemanager/pip/pipSettings.js";
+import { setEcoSystem, ECOSYSTEM_PY } from "../src/config/settings.js";
+import { main } from "../src/main.js";
+
+// Set eco system
+setEcoSystem(ECOSYSTEM_PY);
+
+// Strip nodejs and wrapper script from args
+let argv = process.argv.slice(2);
+
+if (argv[0] === '-m' && (argv[1] === 'pip' || argv[1] === 'pip3')) {
+	setEcoSystem(ECOSYSTEM_PY);
+	setCurrentPipInvocation(argv[1] === 'pip3' ? PIP_INVOCATIONS.PY3_PIP3 : PIP_INVOCATIONS.PY3_PIP);
+	initializePackageManager(PIP_PACKAGE_MANAGER);
+
+  // Strip off the '-m pip' or '-m pip3' from the args
+	argv = argv.slice(2);
+
+  var exitCode = await main(argv);
+	process.exit(exitCode);
+} else {
+	// Forward to real python3 binary for non-pip flows
+	const { spawn } = await import('child_process');
+	spawn('python3', argv, { stdio: 'inherit' });
+}

packages/safe-chain/package.json

@@ -17,6 +17,8 @@
     "aikido-bunx": "bin/aikido-bunx.js",
     "aikido-pip": "bin/aikido-pip.js",
     "aikido-pip3": "bin/aikido-pip3.js",
+    "aikido-python": "bin/aikido-python.js",
+    "aikido-python3": "bin/aikido-python3.js",
     "safe-chain": "bin/safe-chain.js"
   },
   "type": "module",

packages/safe-chain/src/packagemanager/currentPackageManager.js

@@ -52,8 +52,8 @@ export function initializePackageManager(packageManagerName) {
     state.packageManagerName = createBunPackageManager();
   } else if (packageManagerName === "bunx") {
     state.packageManagerName = createBunxPackageManager();
-  } else if (packageManagerName === "pip" || packageManagerName === "pip3") {
-    state.packageManagerName = createPipPackageManager(packageManagerName);
+  } else if (packageManagerName === "pip") {
+    state.packageManagerName = createPipPackageManager();
   } else {
     throw new Error("Unsupported package manager: " + packageManagerName);
   }

packages/safe-chain/src/packagemanager/pip/createPackageManager.js

@@ -1,12 +1,18 @@
 import { runPip } from "./runPipCommand.js";
-
+import { getCurrentPipInvocation } from "./pipSettings.js";
 /**
- * @param {string} [command]
  * @returns {import("../currentPackageManager.js").PackageManager}
  */
-export function createPipPackageManager(command = "pip") {
+export function createPipPackageManager() {
   return {
-    runCommand: /** @param {string[]} args */ (args) => runPip(command, args),
+    /**
+     * @param {string[]} args
+     */
+    runCommand: (args) => {
+      const invocation = getCurrentPipInvocation();
+      const fullArgs = [...invocation.args, ...args];
+      return runPip(invocation.command, fullArgs);
+    },
     // For pip, rely solely on MITM proxy to detect/deny downloads from known registries.
     isSupportedCommand: () => false,
     getDependencyUpdatesForCommand: () => [],

packages/safe-chain/src/packagemanager/pip/pipSettings.js

@@ -0,0 +1,30 @@
+export const PIP_PACKAGE_MANAGER = "pip";
+
+// All supported python/pip invocations for Safe Chain interception
+export const PIP_INVOCATIONS = {
+  PIP: { command: "pip", args: [] },
+  PIP3: { command: "pip3", args: [] },
+  PY_PIP: { command: "python", args: ["-m", "pip"] },
+  PY3_PIP: { command: "python3", args: ["-m", "pip"] },
+  PY_PIP3: { command: "python", args: ["-m", "pip3"] },
+  PY3_PIP3: { command: "python3", args: ["-m", "pip3"] }
+};
+
+/**
+ * @type {{ command: string, args: string[] }}
+ */
+let currentInvocation = PIP_INVOCATIONS.PY3_PIP; // Default to python3 -m pip
+
+/**
+ * @param {{ command: string, args: string[] }} invocation
+ */
+export function setCurrentPipInvocation(invocation) {
+  currentInvocation = invocation;
+}
+
+/**
+ * @returns {{ command: string, args: string[] }}
+ */
+export function getCurrentPipInvocation() {
+  return currentInvocation;
+}

packages/safe-chain/src/packagemanager/pip/runPipCommand.js

@@ -29,7 +29,8 @@ export async function runPip(command, args) {
     if (error.status) {
       return { status: error.status };
     } else {
-      ui.writeError("Error executing command:", error.message);
+      ui.writeError(`Error executing command: ${error.message}`);
+      ui.writeError(`Is '${command}' installed and available on your system?`);
       return { status: 1 };
     }
   }

packages/safe-chain/src/scanning/audit/index.js

@@ -57,8 +57,6 @@ export async function auditChanges(changes) {
   );
 
   for (const change of changes) {
-    //Uncomment next line during manual testing
-    //console.log(" Safe-chain: auditing package:", change);
     const malwarePackage = malwarePackages.find(
       (pkg) => pkg.name === change.name && pkg.version === change.version
     );

packages/safe-chain/src/shell-integration/helpers.js

@@ -22,6 +22,8 @@ export const knownAikidoTools = [
   { tool: "bunx", aikidoCommand: "aikido-bunx" },
   { tool: "pip", aikidoCommand: "aikido-pip" },
   { tool: "pip3", aikidoCommand: "aikido-pip3" },
+  { tool: "python", aikidoCommand: "aikido-python" },
+  { tool: "python3", aikidoCommand: "aikido-python3" },
   // When adding a new tool here, also update the documentation for the new tool in the README.md
 ];
 

packages/safe-chain/src/shell-integration/path-wrappers/templates/unix-wrapper.template.sh

@@ -19,4 +19,4 @@ else
     echo "Error: Could not find original {{PACKAGE_MANAGER}}" >&2
     exit 1
   fi
-fi
+fi

packages/safe-chain/src/shell-integration/setup-ci.js

@@ -51,13 +51,9 @@ function createUnixShims(shimsDir) {
 
   const template = fs.readFileSync(templatePath, "utf-8");
 
-  // Create a shim for each tool except pip (CI support not yet implemented)
+  // Create a shim for each tool
   let created = 0;
   for (const toolInfo of knownAikidoTools) {
-    if (toolInfo.tool === "pip") {
-      continue; // Skip pip shims in CI for now
-    }
-
     const shimContent = template
       .replaceAll("{{PACKAGE_MANAGER}}", toolInfo.tool)
       .replaceAll("{{AIKIDO_COMMAND}}", toolInfo.aikidoCommand);
@@ -98,18 +94,14 @@ function createWindowsShims(shimsDir) {
 
   const template = fs.readFileSync(templatePath, "utf-8");
 
-  // Create a shim for each tool except pip (CI support not yet implemented)
+  // Create a shim for each tool
   let created = 0;
   for (const toolInfo of knownAikidoTools) {
-    if (toolInfo.tool === "pip") {
-      continue; // Skip pip shims in CI for now
-    }
-
     const shimContent = template
       .replaceAll("{{PACKAGE_MANAGER}}", toolInfo.tool)
       .replaceAll("{{AIKIDO_COMMAND}}", toolInfo.aikidoCommand);
 
-    const shimPath = path.join(shimsDir, `${toolInfo.tool}.cmd`);
+  const shimPath = `${shimsDir}/${toolInfo.tool}.cmd`;
     fs.writeFileSync(shimPath, shimContent, "utf-8");
     created++;
   }

packages/safe-chain/src/shell-integration/setup-ci.spec.js

@@ -147,4 +147,4 @@ describe("Setup CI shell integration", () => {
       assert.ok(!fs.existsSync(unixNpmShim), "Unix npm shim should not exist on Windows");
     });
   });
-});
+});

packages/safe-chain/src/shell-integration/startup-scripts/init-fish.fish

@@ -79,26 +79,10 @@ end
 
 # `python -m pip`, `python -m pip3`.
 function python
-    if test (count $argv) -ge 2; and test $argv[1] = "-m"; and string match -qr '^pip(3)?$' -- $argv[2]
-        set mod $argv[2]
-        set args $argv[3..-1]
-        if test $mod = "pip3"
-            wrapSafeChainCommand "pip3" "aikido-pip3" $args
-        else
-            wrapSafeChainCommand "pip" "aikido-pip" $args
-        end
-    else
-        command python $argv
-    end
+    wrapSafeChainCommand "python" "aikido-python" $argv
 end
 
 # `python3 -m pip`, `python3 -m pip3'.
 function python3
-    if test (count $argv) -ge 2; and test $argv[1] = "-m"; and string match -qr '^pip(3)?$' -- $argv[2]
-        set args $argv[3..-1]
-        # python3 always uses pip3, regardless of whether user types `pip` or `pip3`
-        wrapSafeChainCommand "pip3" "aikido-pip3" $args
-    else
-        command python3 $argv
-    end
+    wrapSafeChainCommand "python3" "aikido-python3" $argv
 end

packages/safe-chain/src/shell-integration/startup-scripts/init-posix.sh

@@ -71,26 +71,10 @@ function pip3() {
 
 # `python -m pip`, `python -m pip3`.
 function python() {
-  if [[ "$1" == "-m" && "$2" == pip* ]]; then
-    local mod="$2"
-    shift 2
-    if [[ "$mod" == "pip3" ]]; then
-      wrapSafeChainCommand "pip3" "aikido-pip3" "$@"
-    else
-      wrapSafeChainCommand "pip" "aikido-pip" "$@"
-    fi
-  else
-    command python "$@"
-  fi
+  wrapSafeChainCommand "python" "aikido-python" "$@"
 }
 
 # `python3 -m pip`, `python3 -m pip3'.
 function python3() {
-  if [[ "$1" == "-m" && "$2" == pip* ]]; then
-    shift 2
-    # python3 always uses pip3, regardless of whether user types `pip` or `pip3`
-    wrapSafeChainCommand "pip3" "aikido-pip3" "$@"
-  else
-    command python3 "$@"
-  fi
+  wrapSafeChainCommand "python3" "aikido-python3" "$@"
 }

packages/safe-chain/src/shell-integration/startup-scripts/init-pwsh.ps1

@@ -97,27 +97,11 @@ function pip3 {
 
 # `python -m pip`, `python -m pip3`.
 function python {
-    param([Parameter(ValueFromRemainingArguments=$true)]$Args)
-    if ($Args.Length -ge 2 -and $Args[0] -eq '-m' -and $Args[1] -match '^pip(3)?$') {
-        $pipArgs = if ($Args.Length -gt 2) { $Args | Select-Object -Skip 2 } else { @() }
-        if ($Args[1] -eq 'pip3') { Invoke-WrappedCommand 'pip3' 'aikido-pip3' $pipArgs }
-        else { Invoke-WrappedCommand 'pip' 'aikido-pip' $pipArgs }
-    }
-    else {
-        Invoke-RealCommand 'python' $Args
-    }
+    Invoke-WrappedCommand 'python' 'aikido-python' $args
 }
 
 # `python3 -m pip`, `python3 -m pip3'.
 function python3 {
-    param([Parameter(ValueFromRemainingArguments=$true)]$Args)
-    if ($Args.Length -ge 2 -and $Args[0] -eq '-m' -and $Args[1] -match '^pip(3)?$') {
-        # python3 always uses pip3, regardless of whether user types `pip` or `pip3`
-        $pipArgs = if ($Args.Length -gt 2) { $Args | Select-Object -Skip 2 } else { @() }
-        Invoke-WrappedCommand 'pip3' 'aikido-pip3' $pipArgs
-    }
-    else {
-        Invoke-RealCommand 'python3' $Args
-    }
+    Invoke-WrappedCommand 'python3' 'aikido-python3' $args
 }
 

test/e2e/Dockerfile

@@ -53,7 +53,19 @@ RUN curl -fsSL https://bun.sh/install | bash
 # Install Python and pip (pip3)
 RUN apt-get update && apt-get install -y python${PYTHON_VERSION} python3-pip && \
     ln -sf /usr/bin/python${PYTHON_VERSION} /usr/local/bin/python3 && \
-    ln -sf /usr/bin/pip3 /usr/local/bin/pip3
+    ln -sf /usr/bin/python${PYTHON_VERSION} /usr/local/bin/python && \
+    ln -sf /usr/bin/pip3 /usr/local/bin/pip3 && \
+    cat <<'EOF' > /usr/lib/python3/dist-packages/pip3.py
+"""
+Shim module so 'python[3] -m pip3 …' resolves to pip's CLI entry point.
+"""
+try:
+    import pip._internal
+    pip._internal.main()
+except Exception as exc:
+    print("pip3 module shim failed:", exc)
+    raise
+EOF
 
 # Copy and install Safe chain
 COPY --from=builder /app/*.tgz /pkgs/

test/e2e/pip-ci.e2e.spec.js

@@ -0,0 +1,171 @@
+import { describe, it, before, beforeEach, afterEach } from "node:test";
+import { DockerTestContainer } from "./DockerTestContainer.js";
+import assert from "node:assert";
+
+describe("E2E: safe-chain setup-ci command for pip/pip3", () => {
+  let container;
+
+  before(async () => {
+    DockerTestContainer.buildImage();
+  });
+
+  beforeEach(async () => {
+    container = new DockerTestContainer();
+    await container.start();
+  });
+
+  afterEach(async () => {
+    if (container) {
+      await container.stop();
+      container = null;
+    }
+  });
+
+  describe("E2E: pip CI support", () => {
+    it("does not intercept python3 --version", async () => {
+      const shell = await container.openShell("zsh");
+      const result = await shell.runCommand("python3 --version");
+      assert.ok(result.output.match(/Python \d+\.\d+\.\d+/), `Output was: ${result.output}`);
+      assert.ok(!result.output.includes("Safe-chain"), "Safe Chain should not intercept generic python3 command");
+    });
+
+    it("does not intercept python3 -c 'print(\"hello\")'", async () => {
+      const shell = await container.openShell("zsh");
+      const result = await shell.runCommand("python3 -c 'print(\"hello\")'");
+      assert.ok(result.output.includes("hello"), `Output was: ${result.output}`);
+      assert.ok(!result.output.includes("Safe-chain"), "Safe Chain should not intercept generic python3 -c command");
+    });
+
+    it("does not intercept python3 test.py", async () => {
+      const shell = await container.openShell("zsh");
+      await shell.runCommand("echo 'print(\"Hello from test.py!\")' > test.py");
+      const result = await shell.runCommand("python3 test.py");
+      assert.ok(result.output.includes("Hello from test.py!"), `Output was: ${result.output}`);
+      assert.ok(!result.output.includes("Safe-chain"), "Safe Chain should not intercept generic python3 script execution");
+    });
+
+    it("does not intercept python test.py", async () => {
+      const shell = await container.openShell("zsh");
+      await shell.runCommand("echo 'print(\"Hello from test.py!\")' > test.py");
+      const result = await shell.runCommand("python test.py");
+      assert.ok(result.output.includes("Hello from test.py!"), `Output was: ${result.output}`);
+      assert.ok(!result.output.includes("Safe-chain"), "Safe Chain should not intercept generic python script execution");
+    });
+  });
+
+  for (let shell of ["bash", "zsh"]) {
+    it(`safe-chain setup-ci wraps pip3 command with PATH shim after installation for ${shell}`, async () => {
+      // Setup safe-chain CI shims
+      const installationShell = await container.openShell(shell);
+      await installationShell.runCommand("safe-chain setup-ci");
+
+      // Add $HOME/.safe-chain/shims to PATH for subsequent shells
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.zshrc"
+      );
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.bashrc"
+      );
+
+      const projectShell = await container.openShell(shell);
+      // Use --break-system-packages to avoid Debian/Ubuntu external management restrictions
+      const result = await projectShell.runCommand(
+        "pip3 install --break-system-packages certifi"
+      );
+
+      const hasExpectedOutput = result.output.includes(
+        "no malware found."
+      );
+      assert.ok(
+        hasExpectedOutput,
+        hasExpectedOutput
+          ? "Expected pip3 command to be wrapped by safe-chain"
+          : `Output did not contain \"no malware found.\": \n${result.output}`
+      );
+    });
+
+    it(`setup-ci routes python -m pip through safe-chain for ${shell}`, async () => {
+      const installationShell = await container.openShell(shell);
+      await installationShell.runCommand("safe-chain setup-ci");
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.zshrc"
+      );
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.bashrc"
+      );
+
+      const projectShell = await container.openShell(shell);
+      const result = await projectShell.runCommand(
+        "python -m pip install --break-system-packages certifi"
+      );
+
+      assert.ok(
+        result.output.includes("no malware found."),
+        `Output did not contain scan message. Output was:\n${result.output}`
+      );
+    });
+
+    it(`setup-ci routes python3 -m pip through safe-chain for ${shell}`, async () => {
+      const installationShell = await container.openShell(shell);
+      await installationShell.runCommand("safe-chain setup-ci");
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.zshrc"
+      );
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.bashrc"
+      );
+
+      const projectShell = await container.openShell(shell);
+      const result = await projectShell.runCommand(
+        "python3 -m pip install --break-system-packages certifi"
+      );
+
+      assert.ok(
+        result.output.includes("no malware found."),
+        `Output did not contain scan message. Output was:\n${result.output}`
+      );
+    });
+
+    it(`setup-ci routes pip through safe-chain for ${shell}`, async () => {
+      const installationShell = await container.openShell(shell);
+      await installationShell.runCommand("safe-chain setup-ci");
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.zshrc"
+      );
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.bashrc"
+      );
+
+      const projectShell = await container.openShell(shell);
+      const result = await projectShell.runCommand(
+        "pip install --break-system-packages certifi"
+      );
+
+      assert.ok(
+        result.output.includes("no malware found."),
+        `Output did not contain scan message. Output was:\n${result.output}`
+      );
+    });
+
+    it(`setup-ci routes pip3 through safe-chain for ${shell}`, async () => {
+      const installationShell = await container.openShell(shell);
+      await installationShell.runCommand("safe-chain setup-ci");
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.zshrc"
+      );
+      await installationShell.runCommand(
+        "echo 'export PATH=\"$HOME/.safe-chain/shims:$PATH\"' >> ~/.bashrc"
+      );
+
+      const projectShell = await container.openShell(shell);
+      const result = await projectShell.runCommand(
+        "pip3 install --break-system-packages certifi"
+      );
+
+      assert.ok(
+        result.output.includes("no malware found."),
+        `Output did not contain scan message. Output was:\n${result.output}`
+      );
+    });
+  }
+});

test/e2e/pip.e2e.spec.js

@@ -323,4 +323,12 @@ describe("E2E: pip coverage", () => {
       `Should not have SSL/certificate errors for tunneled hosts. Output was:\n${result.output}`
     );
   });
+
+  it(`pip3 install requests with --safe-chain-logging=verbose`, async () => {
+    const shell = await container.openShell("zsh");
+    const result = await shell.runCommand(
+      "pip3 install --break-system-packages requests --safe-chain-logging=verbose"
+    );
+    assert.ok(result.output.includes("no malware found."), `Output did not include expected text. Output was:\n${result.output}`);
+  });
 });