Improve regex used in tag_known_software #238

    * Update tests with more paths to test regex patterns

Signed-off-by: Jono Yang <[email protected]>

scanpipe/pipes/rootfs.py

@@ -362,7 +362,8 @@ def tag_ignorable_codebase_resources(project):
 
 def tag_data_files_with_no_clues(project):
     """
-    Tag CodebaseResources that have a file type of `data` and no detected clues to be uninteresting.
+    Tag CodebaseResources that have a file type of `data` and no detected clues
+    to be uninteresting.
     """
     lookup = Q(
         file_type="data",

scanpipe/pipes/windows.py

@@ -138,30 +138,31 @@ def tag_known_software(project):
     then a version number of "nv" will be set.
     """
     qs = project.codebaseresources.no_status()
-    python_root_directory_name_pattern = r"(^.*Python(\d*))/.*$"
+    python_root_directory_name_pattern = r"(^/(Files/)?Python(\d+)?)/.*$"
     python_root_directory_name_pattern_compiled = re.compile(
         python_root_directory_name_pattern
     )
-    python_paths_by_versions = {}
-    lookup = Q(rootfs_path__regex=python_root_directory_name_pattern)
-    for python_codebase_resource in qs.filter(lookup):
-        _, python_root_dir, version, _ = re.split(
+    python_versions_by_path = {}
+    for python_codebase_resource in qs.filter(
+        rootfs_path__regex=python_root_directory_name_pattern
+    ):
+        _, python_root_dir, _, version, _ = re.split(
             python_root_directory_name_pattern_compiled,
             python_codebase_resource.rootfs_path,
         )
+        if python_root_dir in python_versions_by_path:
+            continue
         if not version:
             version = "nv"
-        if version in python_paths_by_versions:
-            continue
         if version != "nv":
             version = ".".join(digit for digit in version)
-        python_paths_by_versions[version] = python_root_dir
+        python_versions_by_path[python_root_dir] = version
 
     # We do not want to tag the files in the `site-packages` directory as being
     # from Python proper. The packages found here are oftentime third-party
     # packages from outside the Python foundation
     q_objects = [~Q(rootfs_path__icontains="site-packages")]
-    for python_version, python_path in python_paths_by_versions.items():
+    for python_path, python_version in python_versions_by_path.items():
         python_package = win_reg.InstalledWindowsProgram(
             name="Python",
             version=python_version,
@@ -177,27 +178,27 @@ def tag_known_software(project):
         )
 
     qs = project.codebaseresources.no_status()
-    openjdk_root_directory_name_pattern = r"(^.*/(open)?jdk(-((\d*)(\.\d+)*))*)/.*$"
+    openjdk_root_directory_name_pattern = (
+        r"^(/(Files/)?(open)?jdk(-((\d*)(\.\d+)*))*)/.*$"
+    )
     openjdk_root_directory_name_pattern_compiled = re.compile(
         openjdk_root_directory_name_pattern
     )
-    openjdk_paths_by_versions = {}
+    openjdk_versions_by_path = {}
     for openjdk_codebase_resource in qs.filter(
         rootfs_path__regex=openjdk_root_directory_name_pattern
     ):
-        _, openjdk_root_path, open_prefix, _, openjdk_version, _, _, _ = re.split(
+        _, openjdk_root_path, _, _, _, openjdk_version, _, _, _ = re.split(
             openjdk_root_directory_name_pattern_compiled,
             openjdk_codebase_resource.rootfs_path,
         )
+        if openjdk_root_path in openjdk_versions_by_path:
+            continue
         if not openjdk_version:
             openjdk_version = "nv"
-        if not open_prefix:
-            open_prefix = ""
-        if openjdk_version in openjdk_paths_by_versions:
-            continue
-        openjdk_paths_by_versions[openjdk_version] = openjdk_root_path
+        openjdk_versions_by_path[openjdk_root_path] = openjdk_version
 
-    for openjdk_version, openjdk_path in openjdk_paths_by_versions.items():
+    for openjdk_path, openjdk_version in openjdk_versions_by_path.items():
         openjdk_package = win_reg.InstalledWindowsProgram(
             name="OpenJDK",
             version=openjdk_version,
@@ -232,7 +233,7 @@ def tag_program_files(project):
         program_files_one_directory_below_pattern
     )
     program_files_dirname_by_path = {}
-    for program_file in qs.filter(rootfs_path__regex="^.*/Program Files( \(x86\))?"):
+    for program_file in qs.filter(rootfs_path__regex=r"^.*/Program Files( \(x86\))?"):
         _, program_files_subdir, _, dirname, _ = re.split(
             program_files_one_directory_below_pattern_compiled, program_file.rootfs_path
         )

scanpipe/tests/test_pipes.py

@@ -820,18 +820,124 @@ def test_scanpipe_pipes_windows_tag_known_software(self):
             path="root/Files/Python39/Lib/site-packages/pip-21.1.3.dist-info/WHEEL",
             rootfs_path="/Files/Python39/Lib/site-packages/pip-21.1.3.dist-info/WHEEL",
         )
+        resource6 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Files/jdk-11.0.1/readme.txt",
+            rootfs_path="/Files/jdk-11.0.1/readme.txt",
+        )
+        resource7 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Files/openjdk-11.0.1/readme.txt",
+            rootfs_path="/Files/openjdk-11.0.1/readme.txt",
+        )
+        resource8 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Files/jdk/readme.txt",
+            rootfs_path="/Files/jdk/readme.txt",
+        )
+        resource9 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Files/openjdk/readme.txt",
+            rootfs_path="/Files/openjdk/readme.txt",
+        )
+        resource10 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Files/Program Files/something-else/jdk/readme.txt",
+            rootfs_path="/Files/Program Files/something-else/jdk/readme.txt",
+        )
+        resource11 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Python/py.exe",
+            rootfs_path="/Python/py.exe",
+        )
+        resource12 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Python27/python2.exe",
+            rootfs_path="/Python27/python2.exe",
+        )
+        resource13 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Python3/python3.exe",
+            rootfs_path="/Python3/python3.exe",
+        )
+        resource14 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Python39/python3.9",
+            rootfs_path="/Python39/python3.9.exe",
+        )
+        resource15 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Python39/Lib/site-packages/pip-21.1.3.dist-info/WHEEL",
+            rootfs_path="/Python39/Lib/site-packages/pip-21.1.3.dist-info/WHEEL",
+        )
+        resource16 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/jdk-11.0.1/readme.txt",
+            rootfs_path="/jdk-11.0.1/readme.txt",
+        )
+        resource17 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/openjdk-11.0.1/readme.txt",
+            rootfs_path="/openjdk-11.0.1/readme.txt",
+        )
+        resource18 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/jdk/readme.txt",
+            rootfs_path="/jdk/readme.txt",
+        )
+        resource19 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/openjdk/readme.txt",
+            rootfs_path="/openjdk/readme.txt",
+        )
+        resource20 = CodebaseResource.objects.create(
+            project=p1,
+            path="root/Program Files/something-else/jdk/readme.txt",
+            rootfs_path="/Program Files/something-else/jdk/readme.txt",
+        )
 
         windows.tag_known_software(p1)
+        resource11.refresh_from_db()
+        resource12.refresh_from_db()
+        resource13.refresh_from_db()
+        resource14.refresh_from_db()
+        resource15.refresh_from_db()
+        resource16.refresh_from_db()
+        resource17.refresh_from_db()
+        resource18.refresh_from_db()
+        resource19.refresh_from_db()
+        resource20.refresh_from_db()
         resource1.refresh_from_db()
         resource2.refresh_from_db()
         resource3.refresh_from_db()
         resource4.refresh_from_db()
         resource5.refresh_from_db()
+        resource6.refresh_from_db()
+        resource7.refresh_from_db()
+        resource8.refresh_from_db()
+        resource9.refresh_from_db()
+        resource10.refresh_from_db()
+
         self.assertEqual("installed-package", resource1.status)
         self.assertEqual("installed-package", resource2.status)
         self.assertEqual("installed-package", resource3.status)
         self.assertEqual("installed-package", resource4.status)
         self.assertEqual("", resource5.status)
+        self.assertEqual("installed-package", resource6.status)
+        self.assertEqual("installed-package", resource7.status)
+        self.assertEqual("installed-package", resource8.status)
+        self.assertEqual("installed-package", resource9.status)
+        self.assertEqual("", resource10.status)
+        self.assertEqual("installed-package", resource11.status)
+        self.assertEqual("installed-package", resource12.status)
+        self.assertEqual("installed-package", resource13.status)
+        self.assertEqual("installed-package", resource14.status)
+        self.assertEqual("", resource15.status)
+        self.assertEqual("installed-package", resource16.status)
+        self.assertEqual("installed-package", resource17.status)
+        self.assertEqual("installed-package", resource18.status)
+        self.assertEqual("installed-package", resource19.status)
+        self.assertEqual("", resource20.status)
 
     def test_scanpipe_pipes_windows_tag_program_files(self):
         p1 = Project.objects.create(name="Analysis")