archive - Adding exclusion_patterns option (#2616)

* Adding exclusion_patterns option

* Adding changelog fragment and Python 2.6 compatability

* Minor refactoring for readability

* Removing unneccessary conditional

* Applying initial review suggestions

* Adding missed review suggestion

changelogs/fragments/2616-archive-exclusion_patterns-option.yml

@@ -0,0 +1,2 @@
+minor_changes:
+  - archive - added ``exclusion_patterns`` option to exclude files or subdirectories from archives (https://github.com/ansible-collections/community.general/pull/2616).

plugins/modules/files/archive.py

@@ -41,8 +41,16 @@ options:
   exclude_path:
     description:
       - Remote absolute path, glob, or list of paths or globs for the file or files to exclude from I(path) list and glob expansion.
+      - Use I(exclusion_patterns) to instead exclude files or subdirectories below any of the paths from the I(path) list.
     type: list
     elements: path
+  exclusion_patterns:
+    description:
+      - Glob style patterns to exclude files or directories from the resulting archive.
+      - This differs from I(exclude_path) which applies only to the source paths from I(path).
+    type: list
+    elements: path
+    version_added: 3.2.0
   force_archive:
     description:
       - Allows you to force the module to treat this as an archive even if only a single file is specified.
@@ -163,6 +171,8 @@ import re
 import shutil
 import tarfile
 import zipfile
+from fnmatch import fnmatch
+from sys import version_info
 from traceback import format_exc
 
 from ansible.module_utils.basic import AnsibleModule, missing_required_lib
@@ -186,6 +196,8 @@ else:
         LZMA_IMP_ERR = format_exc()
         HAS_LZMA = False
 
+PY27 = version_info[0:2] >= (2, 7)
+
 
 def to_b(s):
     return to_bytes(s, errors='surrogate_or_strict')
@@ -214,6 +226,59 @@ def expand_paths(paths):
     return expanded_path, is_globby
 
 
+def matches_exclusion_patterns(path, exclusion_patterns):
+    return any(fnmatch(path, p) for p in exclusion_patterns)
+
+
+def get_filter(exclusion_patterns, format):
+    def zip_filter(path):
+        return matches_exclusion_patterns(path, exclusion_patterns)
+
+    def tar_filter(tarinfo):
+        return None if matches_exclusion_patterns(tarinfo.name, exclusion_patterns) else tarinfo
+
+    return zip_filter if format == 'zip' or not PY27 else tar_filter
+
+
+def get_archive_contains(format):
+    def archive_contains(archive, name):
+        try:
+            if format == 'zip':
+                archive.getinfo(name)
+            else:
+                archive.getmember(name)
+        except KeyError:
+            return False
+
+        return True
+
+    return archive_contains
+
+
+def get_add_to_archive(format, filter):
+    def add_to_zip_archive(archive_file, path, archive_name):
+        try:
+            if not filter(path):
+                archive_file.write(path, archive_name)
+        except Exception as e:
+            return e
+
+        return None
+
+    def add_to_tar_archive(archive_file, path, archive_name):
+        try:
+            if PY27:
+                archive_file.add(path, archive_name, recursive=False, filter=filter)
+            else:
+                archive_file.add(path, archive_name, recursive=False, exclude=filter)
+        except Exception as e:
+            return e
+
+        return None
+
+    return add_to_zip_archive if format == 'zip' else add_to_tar_archive
+
+
 def main():
     module = AnsibleModule(
         argument_spec=dict(
@@ -221,6 +286,7 @@ def main():
             format=dict(type='str', default='gz', choices=['bz2', 'gz', 'tar', 'xz', 'zip']),
             dest=dict(type='path'),
             exclude_path=dict(type='list', elements='path'),
+            exclusion_patterns=dict(type='list', elements='path'),
             force_archive=dict(type='bool', default=False),
             remove=dict(type='bool', default=False),
         ),
@@ -242,6 +308,8 @@ def main():
     changed = False
     state = 'absent'
 
+    exclusion_patterns = params['exclusion_patterns'] or []
+
     # Simple or archive file compression (inapplicable with 'zip' since it's always an archive)
     b_successes = []
 
@@ -262,6 +330,10 @@ def main():
     # Only attempt to expand the exclude paths if it exists
     b_expanded_exclude_paths = expand_paths(exclude_paths)[0] if exclude_paths else []
 
+    filter = get_filter(exclusion_patterns, fmt)
+    archive_contains = get_archive_contains(fmt)
+    add_to_archive = get_add_to_archive(fmt, filter)
+
     # Only try to determine if we are working with an archive or not if we haven't set archive to true
     if not force_archive:
         # If we actually matched multiple files or TRIED to, then
@@ -384,38 +456,31 @@ def main():
                                     n_fullpath = to_na(b_fullpath)
                                     n_arcname = to_native(b_match_root.sub(b'', b_fullpath), errors='surrogate_or_strict')
 
-                                    try:
+                                    err = add_to_archive(arcfile, n_fullpath, n_arcname)
-                                        if fmt == 'zip':
+                                    if err:
-                                            arcfile.write(n_fullpath, n_arcname)
+                                        errors.append('%s: %s' % (n_fullpath, to_native(err)))
-                                        else:
-                                            arcfile.add(n_fullpath, n_arcname, recursive=False)
-
-                                    except Exception as e:
-                                        errors.append('%s: %s' % (n_fullpath, to_native(e)))
 
                                 for b_filename in b_filenames:
                                     b_fullpath = b_dirpath + b_filename
                                     n_fullpath = to_na(b_fullpath)
                                     n_arcname = to_n(b_match_root.sub(b'', b_fullpath))
 
-                                    try:
+                                    err = add_to_archive(arcfile, n_fullpath, n_arcname)
-                                        if fmt == 'zip':
+                                    if err:
-                                            arcfile.write(n_fullpath, n_arcname)
+                                        errors.append('Adding %s: %s' % (to_native(b_path), to_native(err)))
-                                        else:
-                                            arcfile.add(n_fullpath, n_arcname, recursive=False)
 
+                                    if archive_contains(arcfile, n_arcname):
                                         b_successes.append(b_fullpath)
-                                    except Exception as e:
-                                        errors.append('Adding %s: %s' % (to_native(b_path), to_native(e)))
                         else:
                             path = to_na(b_path)
                             arcname = to_n(b_match_root.sub(b'', b_path))
-                            if fmt == 'zip':
-                                arcfile.write(path, arcname)
-                            else:
-                                arcfile.add(path, arcname, recursive=False)
 
-                            b_successes.append(b_path)
+                            err = add_to_archive(arcfile, path, arcname)
+                            if err:
+                                errors.append('Adding %s: %s' % (to_native(b_path), to_native(err)))
+
+                            if archive_contains(arcfile, arcname):
+                                b_successes.append(b_path)
 
                 except Exception as e:
                     expanded_fmt = 'zip' if fmt == 'zip' else ('tar.' + fmt)

tests/integration/targets/archive/tasks/main.yml

@@ -363,6 +363,19 @@
 - name: remove nonascii test
   file: path="{{ output_dir }}/test-archive-nonascii-くらとみ.zip" state=absent
 
+- name: Test exclusion_patterns option
+  archive:
+    path: "{{ output_dir }}/*.txt"
+    dest: "{{ output_dir }}/test-archive-exclustion-patterns.tgz"
+    exclusion_patterns: b?r.*
+  register: exclusion_patterns_result
+
+- name: Assert that exclusion_patterns only archives included files
+  assert:
+    that:
+      - exclusion_patterns_result is changed
+      - "'bar.txt' not in exclusion_patterns_result.archived"
+
 - name: Remove backports.lzma if previously installed (pip)
   pip: name=backports.lzma state=absent
   when: backports_lzma_pip is changed