Fix 'ansible-galaxy collection verify' to display new files/dirs (#76690)

* Fix 'ansible-galaxy collection verify' to report files/directories not listed in the FILES.json

* changelog

* Fix bug in 2.13+

changelogs/fragments/76690-fix-ansible-galaxy-collection-verify-modified-content.yaml

@@ -0,0 +1,2 @@
+bugfixes:
+  - ansible-galaxy collection verify - display files/directories not included in the FILES.json as modified content.

lib/ansible/galaxy/collection/__init__.py

@@ -298,11 +298,46 @@ def verify_local_collection(
     _verify_file_hash(b_collection_path, file_manifest_filename, expected_hash, modified_content)
     file_manifest = get_json_from_validation_source(file_manifest_filename)
 
+    collection_dirs = set()
+    collection_files = {
+        os.path.join(b_collection_path, b'MANIFEST.json'),
+        os.path.join(b_collection_path, b'FILES.json'),
+    }
+
     # Use the file manifest to verify individual file checksums
     for manifest_data in file_manifest['files']:
+        name = manifest_data['name']
+
         if manifest_data['ftype'] == 'file':
+            collection_files.add(
+                os.path.join(b_collection_path, to_bytes(name, errors='surrogate_or_strict'))
+            )
             expected_hash = manifest_data['chksum_%s' % manifest_data['chksum_type']]
-            _verify_file_hash(b_collection_path, manifest_data['name'], expected_hash, modified_content)
+            _verify_file_hash(b_collection_path, name, expected_hash, modified_content)
+
+        if manifest_data['ftype'] == 'dir':
+            collection_dirs.add(
+                os.path.join(b_collection_path, to_bytes(name, errors='surrogate_or_strict'))
+            )
+
+    # Find any paths not in the FILES.json
+    for root, dirs, files in os.walk(b_collection_path):
+        for name in files:
+            full_path = os.path.join(root, name)
+            path = to_text(full_path[len(b_collection_path) + 1::], errors='surrogate_or_strict')
+
+            if full_path not in collection_files:
+                modified_content.append(
+                    ModifiedContent(filename=path, expected='the file does not exist', installed='the file exists')
+                )
+        for name in dirs:
+            full_path = os.path.join(root, name)
+            path = to_text(full_path[len(b_collection_path) + 1::], errors='surrogate_or_strict')
+
+            if full_path not in collection_dirs:
+                modified_content.append(
+                    ModifiedContent(filename=path, expected='the directory does not exist', installed='the directory exists')
+                )
 
     if modified_content:
         result.success = False

lib/ansible/galaxy/dependency_resolution/dataclasses.py

@@ -176,13 +176,7 @@ class _ComputedReqKindsMixin:
         if not self.may_have_offline_galaxy_info:
             self._source_info = None
         else:
-            # Store Galaxy metadata adjacent to the namespace of the collection
-            # Chop off the last two parts of the path (/ns/coll) to get the dir containing the ns
-            b_src = to_bytes(self.src, errors='surrogate_or_strict')
-            b_path_parts = b_src.split(to_bytes(os.path.sep))[0:-2]
-            b_path = to_bytes(os.path.sep).join(b_path_parts)
-
-            info_path = self.construct_galaxy_info_path(b_path)
+            info_path = self.construct_galaxy_info_path(to_bytes(self.src, errors='surrogate_or_strict'))
 
             self._source_info = get_validated_source_info(
                 info_path,
@@ -447,10 +441,16 @@ class _ComputedReqKindsMixin:
             return False
         return True
 
-    def construct_galaxy_info_path(self, b_metadata_dir):
+    def construct_galaxy_info_path(self, b_collection_path):
         if not self.may_have_offline_galaxy_info and not self.type == 'galaxy':
             raise TypeError('Only installed collections from a Galaxy server have offline Galaxy info')
 
+        # Store Galaxy metadata adjacent to the namespace of the collection
+        # Chop off the last two parts of the path (/ns/coll) to get the dir containing the ns
+        b_src = to_bytes(b_collection_path, errors='surrogate_or_strict')
+        b_path_parts = b_src.split(to_bytes(os.path.sep))[0:-2]
+        b_metadata_dir = to_bytes(os.path.sep).join(b_path_parts)
+
         # ns.coll-1.0.0.info
         b_dir_name = to_bytes(f"{self.namespace}.{self.name}-{self.ver}.info", errors="surrogate_or_strict")
 

test/integration/targets/ansible-galaxy-collection/tasks/verify.yml

@@ -251,6 +251,16 @@
 - name: append a newline to a module to modify the checksum
   shell: "echo '' >> {{ module_path }}"
 
+- name: create a new module file
+  file:
+    path: '{{ galaxy_dir }}/ansible_collections/ansible_test/verify/plugins/modules/test_new_file.py'
+    state: touch
+
+- name: create a new directory
+  file:
+    path: '{{ galaxy_dir }}/ansible_collections/ansible_test/verify/plugins/modules/test_new_dir'
+    state: directory
+
 - name: verify modified collection locally-only (should fail)
   command: ansible-galaxy collection verify --offline ansible_test.verify
   register: verify
@@ -261,6 +271,8 @@
       - verify.rc != 0
       - "'Collection ansible_test.verify contains modified content in the following files:' in verify.stdout"
       - "'plugins/modules/test_module.py' in verify.stdout"
+      - "'plugins/modules/test_new_file.py' in verify.stdout"
+      - "'plugins/modules/test_new_dir' in verify.stdout"
 
 # TODO: add a test for offline Galaxy signature metadata