fix: layer extractor should match python metadata files on windows [c… (#770)

fix: layer extractor should match python metadata files on windows [cn-874]

Author: jcambier

lib/inputs/python/static.ts

@@ -1,12 +1,13 @@
-import { basename } from "path";
+import { basename, sep } from "path";
 
 import { ExtractAction } from "../../extractor/types";
 import { streamToString } from "../../stream-utils";
 
 const poetryManifestFiles = ["pyproject.toml", "poetry.lock"];
 const pipManifestFiles = ["requirements.txt"];
+// Match both forward slashes (POSIX/macOS/Linux) and backslashes (Windows)
 const pythonMetadataFilesRegex =
-  /\/lib\/python.*?\/(?:dist|site)-packages\/.*?\.dist-info\/METADATA/;
+  /[\/\\]lib[\/\\]python.*?[\/\\](?:dist|site)-packages[\/\\].*?\.dist-info[\/\\]METADATA/;
 const deletedPoetryAppFiles = poetryManifestFiles.map((file) => ".wh." + file);
 const deletedPipAppFiles = pipManifestFiles.map((file) => ".wh." + file);
 const pythonApplicationFileSuffixes = [".py", "requirements.txt", "Pipfile"];
@@ -42,10 +43,9 @@ export const getPipAppFileContentAction: ExtractAction = {
 
 function pythonApplicationFilePathMatches(filePath: string): boolean {
   return (
-    !filePath.includes("/site-packages/") &&
-    !filePath.includes("/dist-packages/") &&
-    // "/usr/" should not include 1st party code
-    !filePath.startsWith("/usr/") &&
+    !filePath.includes(`${sep}site-packages${sep}`) &&
+    !filePath.includes(`${sep}dist-packages${sep}`) &&
+    !filePath.startsWith(`${sep}usr${sep}`) &&
     pythonApplicationFileSuffixes.some((suffix) => filePath.endsWith(suffix))
   );
 }

test/lib/inputs/python/static.spec.ts

@@ -0,0 +1,79 @@
+import { sep } from "path";
+import {
+  getPipAppFileContentAction,
+  getPythonAppFileContentAction,
+} from "../../../../lib/inputs/python/static";
+
+describe("Python pip file path matching", () => {
+  const { filePathMatches } = getPipAppFileContentAction;
+
+  describe("requirements.txt matching", () => {
+    it("should match requirements.txt", () => {
+      expect(filePathMatches("/app/requirements.txt")).toBe(true);
+    });
+  });
+
+  describe("METADATA file matching", () => {
+    it("should match METADATA files with forward slashes", () => {
+      expect(
+        filePathMatches(
+          "/usr/lib/python3.9/dist-packages/Django-4.1.2.dist-info/METADATA",
+        ),
+      ).toBe(true);
+      expect(
+        filePathMatches(
+          "/usr/lib/python3.9/site-packages/requests-2.28.0.dist-info/METADATA",
+        ),
+      ).toBe(true);
+    });
+
+    it("should match METADATA files with backslashes (Windows)", () => {
+      expect(
+        filePathMatches(
+          "\\usr\\lib\\python3.9\\dist-packages\\Django-4.1.2.dist-info\\METADATA",
+        ),
+      ).toBe(true);
+      expect(
+        filePathMatches(
+          "\\usr\\lib\\python3.9\\site-packages\\requests-2.28.0.dist-info\\METADATA",
+        ),
+      ).toBe(true);
+    });
+
+    it("should not match non-METADATA files in dist-info", () => {
+      expect(
+        filePathMatches(
+          "/usr/lib/python3.9/dist-packages/Django-4.1.2.dist-info/RECORD",
+        ),
+      ).toBe(false);
+    });
+
+    it("should not match METADATA files outside of site/dist-packages", () => {
+      expect(filePathMatches("/app/METADATA")).toBe(false);
+      expect(filePathMatches("/usr/lib/METADATA")).toBe(false);
+    });
+  });
+});
+
+describe("Python application file path matching", () => {
+  const { filePathMatches } = getPythonAppFileContentAction;
+
+  it("should match .py, Pipfile and requirements.txt files", () => {
+    expect(filePathMatches(`${sep}app${sep}main.py`)).toBe(true);
+    expect(filePathMatches(`${sep}app${sep}Pipfile`)).toBe(true);
+    expect(filePathMatches(`${sep}app${sep}requirements.txt`)).toBe(true);
+  });
+
+  it("should exclude files in site-packages and dist-packages", () => {
+    expect(filePathMatches(`${sep}lib${sep}site-packages${sep}foo.py`)).toBe(
+      false,
+    );
+    expect(filePathMatches(`${sep}lib${sep}dist-packages${sep}bar.py`)).toBe(
+      false,
+    );
+  });
+
+  it("should exclude files under /usr/", () => {
+    expect(filePathMatches(`${sep}usr${sep}lib${sep}script.py`)).toBe(false);
+  });
+});