Fix: wildcard excludes in unarchive with tar archives (#40935)

* fix: exclude using wildcards for tar archives

Fixes #37842, #22947

* fix: Remove quote() as it munges the exclude format

* test: Refactor to use single archive structure

A common structure archived by different methods should simplify some of
the feature tests.

* test: Use common archive layout to validate exclude feature

* test: Use the same exclude checks for zip/tar archives

lib/ansible/modules/files/unarchive.py

@@ -647,7 +647,7 @@ class TgzArchive(object):
         if self.opts:
             cmd.extend(['--show-transformed-names'] + self.opts)
         if self.excludes:
-            cmd.extend(['--exclude=' + quote(f) for f in self.excludes])
+            cmd.extend(['--exclude=' + f for f in self.excludes])
         cmd.extend(['-f', self.src])
         rc, out, err = self.module.run_command(cmd, cwd=self.dest, environ_update=dict(LANG='C', LC_ALL='C', LC_MESSAGES='C'))
         if rc != 0:
@@ -658,14 +658,18 @@ class TgzArchive(object):
             # filename = filename.decode('string_escape')
             filename = to_native(codecs.escape_decode(filename)[0])
 
-            if filename and filename not in self.excludes:
-                # We don't allow absolute filenames.  If the user wants to unarchive rooted in "/"
-                # they need to use "dest: '/'".  This follows the defaults for gtar, pax, etc.
-                # Allowing absolute filenames here also causes bugs: https://github.com/ansible/ansible/issues/21397
-                if filename.startswith('/'):
-                    filename = filename[1:]
+            # We don't allow absolute filenames.  If the user wants to unarchive rooted in "/"
+            # they need to use "dest: '/'".  This follows the defaults for gtar, pax, etc.
+            # Allowing absolute filenames here also causes bugs: https://github.com/ansible/ansible/issues/21397
+            if filename.startswith('/'):
+                filename = filename[1:]
 
-                self._files_in_archive.append(filename)
+            if self.excludes:
+                for exclude in self.excludes:
+                    if not fnmatch.fnmatch(filename, exclude):
+                        self._files_in_archive.append(to_native(filename))
+            else:
+                self._files_in_archive.append(to_native(filename))
 
         return self._files_in_archive
 
@@ -682,7 +686,7 @@ class TgzArchive(object):
         if self.module.params['keep_newer']:
             cmd.append('--keep-newer-files')
         if self.excludes:
-            cmd.extend(['--exclude=' + quote(f) for f in self.excludes])
+            cmd.extend(['--exclude=' + f for f in self.excludes])
         cmd.extend(['-f', self.src])
         rc, out, err = self.module.run_command(cmd, cwd=self.dest, environ_update=dict(LANG='C', LC_ALL='C', LC_MESSAGES='C'))
 
@@ -729,7 +733,7 @@ class TgzArchive(object):
         if self.module.params['keep_newer']:
             cmd.append('--keep-newer-files')
         if self.excludes:
-            cmd.extend(['--exclude=' + quote(f) for f in self.excludes])
+            cmd.extend(['--exclude=' + f for f in self.excludes])
         cmd.extend(['-f', self.src])
         rc, out, err = self.module.run_command(cmd, cwd=self.dest, environ_update=dict(LANG='C', LC_ALL='C', LC_MESSAGES='C'))
         return dict(cmd=cmd, rc=rc, out=out, err=err)

test/integration/targets/unarchive/tasks/main.yml

@@ -60,6 +60,39 @@
 - name: prep a zip file
   shell: zip test-unarchive.zip foo-unarchive.txt foo-unarchive-777.txt chdir={{output_dir}}
 
+- name: Prepare - Create test dirs
+  file:
+    path: "{{output_dir}}/{{item}}"
+    state: directory
+  with_items:
+    - created/include
+    - created/exclude
+    - created/other
+
+- name: Prepare - Create test files
+  file:
+    path: "{{output_dir}}/created/{{item}}"
+    state: touch
+  with_items:
+    - include/include-1.txt
+    - include/include-2.txt
+    - include/include-3.txt
+    - exclude/exclude-1.txt
+    - exclude/exclude-2.txt
+    - exclude/exclude-3.txt
+    - other/include-1.ext
+    - other/include-2.ext
+    - other/exclude-1.ext
+    - other/exclude-2.ext
+    - other/other-1.ext
+    - other/other-2.ext
+
+- name: Prepare - zip file
+  shell: zip -r {{output_dir}}/unarchive-00.zip * chdir={{output_dir}}/created/
+
+- name: Prepare - tar file
+  shell: tar czvf {{output_dir}}/unarchive-00.tar * chdir={{output_dir}}/created/
+
 - name: add a file with Windows permissions to zip file
   shell: zip -k test-unarchive.zip FOO-UNAR.TXT chdir={{output_dir}}
 
@@ -187,22 +220,33 @@
 - name: "Create {{ output_dir }}/exclude directory"
   file:
     state: directory
-    path: "{{ output_dir }}/exclude"
+    path: "{{ output_dir }}/exclude-{{item}}"
+  with_items:
+    - zip
+    - tar
 
-- name: Unpack zip file excluding one file.
+- name: Unpack archive file excluding glob files.
   unarchive:
-    src: "{{ output_dir }}/test-unarchive.zip"
-    dest: "{{ output_dir }}/exclude"
-    exclude: "foo-unarchive-*.txt"
+    src: "{{ output_dir }}/unarchive-00.{{item}}"
+    dest: "{{ output_dir }}/exclude-{{item}}"
+    exclude: "exclude/exclude-*.txt"
+  with_items:
+    - zip
+    - tar
 
 - name: verify that the file was unarchived
-  shell: find {{ output_dir }}/exclude chdir={{ output_dir }}
-  register: unarchive_dir01
+  shell: find {{ output_dir }}/exclude-{{item}} chdir={{ output_dir }}
+  register: unarchive00
+  with_items:
+    - zip
+    - tar
 
-- name: verify that zip extraction excluded file
+- name: verify that archive extraction excluded the files
   assert:
     that:
-      - "'foo-unarchive-777.txt' not in unarchive_dir01.stdout"
+      - "'exclude/exclude-1.txt' not in item.stdout"
+  with_items:
+    - "{{ unarchive00.results }}"
 
 - name: remove our zip unarchive destination
   file: path={{output_dir}}/test-unarchive-zip state=absent