fix: change signature and pull node and python data

lib/index.ts

@@ -12,7 +12,7 @@ import {
   UpdateDockerfileBaseImageNameErrorCode,
 } from "./dockerfile/types";
 import * as facts from "./facts";
-import { extractContent, scan } from "./scan";
+import { collectApplicationFiles, scan } from "./scan";
 import {
   AutoDetectedUserInstructions,
   ContainerTarget,
@@ -28,7 +28,7 @@ export {
   scan,
   display,
   dockerFile,
-  extractContent,
+  collectApplicationFiles,
   facts,
   ScanResult,
   PluginResponse,

lib/scan.ts

@@ -1,17 +1,22 @@
 import * as fs from "fs";
 import * as path from "path";
 
+import { getApplicationFiles } from "./analyzer/applications/runtime-common";
 import { getImageArchive } from "./analyzer/image-inspector";
 import { readDockerfileAndAnalyse } from "./dockerfile";
 import { DockerFileAnalysis } from "./dockerfile/types";
 import { extractImageContent } from "./extractor";
 import { ImageName } from "./extractor/image";
-import { ExtractAction, ExtractionResult } from "./extractor/types";
+import { ImageIdFact } from "./facts";
 import { fullImageSavePath } from "./image-save-path";
 import { getArchivePath, getImageType } from "./image-type";
+import { getFileContent } from "./inputs";
+import { getNodeJsTsAppFileContentAction } from "./inputs/node/static";
+import { getPythonAppFileContentAction } from "./inputs/python/static";
 import { isNumber, isTrue } from "./option-utils";
 import * as staticModule from "./static";
-import { ImageType, PluginOptions, PluginResponse } from "./types";
+
+import { ImageType, PluginOptions, PluginResponse, ScanResult } from "./types";
 
 // Registry credentials may also be provided by env vars. When both are set, flags take precedence.
 export function mergeEnvVarsIntoCredentials(
@@ -213,10 +218,9 @@ export function appendLatestTagIfMissing(targetImage: string): string {
   return targetImage;
 }
 
-export async function extractContent(
-  extractActions: ExtractAction[],
+export async function collectApplicationFiles(
   options?: Partial<PluginOptions>,
-): Promise<ExtractionResult> {
+): Promise<PluginResponse> {
   const {
     targetImage,
     imageType,
@@ -245,10 +249,56 @@ export async function extractContent(
       throw new Error("Unhandled image type for image " + targetImage);
   }
 
-  return extractImageContent(
+  const extractActions = [
+    getNodeJsTsAppFileContentAction,
+    getPythonAppFileContentAction,
+  ];
+
+  const content = await extractImageContent(
     imageType,
     imagePath,
     extractActions,
     updatedOptions,
   );
+
+  const nodeApplicationFilesScanResults = getApplicationFiles(
+    getFileContent(
+      content.extractedLayers,
+      getNodeJsTsAppFileContentAction.actionName,
+    ),
+    "node",
+    "npm",
+  );
+
+  const pythonApplicationFilesScanResults = getApplicationFiles(
+    getFileContent(
+      content.extractedLayers,
+      getPythonAppFileContentAction.actionName,
+    ),
+    "python",
+    "python",
+  );
+
+  const imageIdFact: ImageIdFact = {
+    type: "imageId",
+    data: content.imageId,
+  };
+
+  const applicationDependenciesScanResults: ScanResult[] = [
+    ...nodeApplicationFilesScanResults,
+    ...pythonApplicationFilesScanResults,
+  ].map((appDepsScanResult) => {
+    appDepsScanResult.facts.push(imageIdFact);
+
+    return {
+      ...appDepsScanResult,
+      target: {
+        image: targetImage,
+      },
+    };
+  });
+
+  return {
+    scanResults: applicationDependenciesScanResults,
+  };
 }

test/system/application-scans/gomodules.spec.ts

@@ -1,6 +1,6 @@
 import * as elf from "elfy";
 
-import { extractContent, scan } from "../../../lib";
+import { collectApplicationFiles, scan } from "../../../lib";
 import { getGoModulesContentAction } from "../../../lib/go-parser";
 import { getFixture } from "../../util";
 
@@ -24,19 +24,6 @@ describe("gomodules binaries scanning", () => {
     expect(pluginResult).toMatchSnapshot();
   });
 
-  it("should extract image content successfully", async () => {
-    const fixturePath = getFixture(
-      "docker-archives/docker-save/testgo-1.17.tar",
-    );
-    const imageNameAndTag = `docker-archive:${fixturePath}`;
-    const result = await extractContent([getGoModulesContentAction], {
-      path: imageNameAndTag,
-    });
-    const testgoBinary = result.extractedLayers["/testgo"];
-    expect(testgoBinary).toBeTruthy();
-    expect("gomodules" in testgoBinary).toBeTruthy();
-  });
-
   it("return plugin result when Go binary cannot be parsed do not break layer iterator", async () => {
     const elfParseMock = jest.spyOn(elf, "parse").mockImplementation(() => {
       throw new Error("Cannot read property 'type' of undefined");

test/system/application-scans/node.spec.ts

@@ -4,15 +4,15 @@ import { legacy } from "@snyk/dep-graph";
 import * as lockFileParser from "snyk-nodejs-lockfile-parser";
 import { NodeLockfileVersion } from "snyk-nodejs-lockfile-parser";
 import * as resolveDeps from "snyk-resolve-deps";
-import { extractContent, scan } from "../../../lib";
+import { collectApplicationFiles, scan } from "../../../lib";
 import {
   getLockFileVersion,
   shouldBuildDepTree,
 } from "../../../lib/analyzer/applications/node";
 import * as nodeUtils from "../../../lib/analyzer/applications/node-modules-utils";
 import { getAppFilesRootDir } from "../../../lib/analyzer/applications/runtime-common";
 import { FilePathToContent } from "../../../lib/analyzer/applications/types";
-import { getNodeAppFileContentAction } from "../../../lib/inputs/node/static";
+import { ApplicationFilesFact } from "../../../lib/facts";
 import { getFixture, getObjFromFixture } from "../../util";
 
 describe("node application scans", () => {
@@ -216,21 +216,25 @@ describe("node application scans", () => {
     expect(depGraphNpmFromGlobalNodeModules.rootPkg.name).toEqual("lib");
   });
 
-  it("should extract image content successfully", async () => {
+  it("should collect application files successfully", async () => {
     const fixturePath = getFixture("npm/multi-project-image.tar");
     const imageNameAndTag = `docker-archive:${fixturePath}`;
-    const result = await extractContent([getNodeAppFileContentAction], {
+    const result = await collectApplicationFiles({
       path: imageNameAndTag,
     });
-    expect(Object.keys(result.extractedLayers).length).toEqual(608);
-    Object.keys(result.extractedLayers).forEach((fileName) => {
-      expect(
-        fileName.endsWith("/package.json") ||
-          fileName.endsWith("/package-lock.json") ||
-          fileName.endsWith("/yarn.lock"),
-      ).toBeTruthy();
-      expect("node-app-files" in result.extractedLayers[fileName]).toBeTruthy();
-    });
+    const applicationFilesFact: ApplicationFilesFact =
+      result.scanResults[0].facts.find(
+        (fact) => fact.type === "applicationFiles",
+      )!.data;
+
+    const fileList = applicationFilesFact[0].fileHierarchy.map((fh) => fh.path);
+    expect(fileList.length).toEqual(4);
+    expect(fileList).toEqual([
+      "bin/yarn.js",
+      "lib/cli.js",
+      "lib/v8-compile-cache.js",
+      "package.json",
+    ]);
   });
 
   it("should generate a scanResult that contains a npm7 depGraph generated from node modules manifest files", async () => {

test/system/application-scans/python/pip.spec.ts

@@ -1,6 +1,6 @@
-import { extractContent, scan } from "../../../../lib";
+import { collectApplicationFiles, scan } from "../../../../lib";
 import { getAppFilesRootDir } from "../../../../lib/analyzer/applications/runtime-common";
-import { getPythonAppFileContentAction } from "../../../../lib/inputs/python/static";
+import { ApplicationFilesFact } from "../../../../lib/facts";
 import { getFixture } from "../../../util";
 
 describe("pip application scan", () => {
@@ -61,15 +61,19 @@ describe("pip application scan", () => {
     expect(pluginResultExcludeAppVulnsTrueBoolean.scanResults).toHaveLength(1);
   });
 
-  it("should extract image content successfully", async () => {
+  it("should collect application files content successfully", async () => {
     const fixturePath = getFixture("docker-archives/docker-save/pip-flask.tar");
     const imageNameAndTag = `docker-archive:${fixturePath}`;
-    const result = await extractContent([getPythonAppFileContentAction], {
+    const result = await collectApplicationFiles({
       path: imageNameAndTag,
     });
-    const serverPyFile = result.extractedLayers["/app/server.py"];
-    expect(serverPyFile).toBeTruthy();
-    expect("python-app-files" in serverPyFile).toBeTruthy();
+    const applicationFilesFact: ApplicationFilesFact =
+      result.scanResults[0].facts.find(
+        (fact) => fact.type === "applicationFiles",
+      )!.data;
+    const fileList = applicationFilesFact[0].fileHierarchy.map((fh) => fh.path);
+    expect(fileList.length).toBe(1);
+    expect(fileList.includes("server.py")).toBeTruthy();
   });
 
   it("should handle --collect-application-files", async () => {