fix issues around duplicate taskfile loading (#195)

Signed-off-by: hirokuni-kitahara <[email protected]>

ansible_risk_insight/dependency_finder.py

@@ -34,6 +34,7 @@
 requirements_yml = "requirements.yml"
 galaxy_yml = "galaxy.yml"
 GALAXY_yml = "GALAXY.yml"
+github_workflows_dir = ".github/workflows"
 
 
 def find_dependency(type, target, dependency_dir, use_ansible_path=False):
@@ -154,6 +155,7 @@ def find_role_dependency(target):
         ],
         recursive=True,
     )
+    role_meta_files = [fpath for fpath in role_meta_files if github_workflows_dir not in fpath]
     main_yaml = ""
     if len(role_meta_files) > 0:
         for rf in role_meta_files:
@@ -205,6 +207,7 @@ def find_collection_dependency(target):
     requirements = {}
     # collection dir installed by ansible-galaxy command
     manifest_json_files = safe_glob(os.path.join(target, "**", collection_manifest_json), recursive=True)
+    manifest_json_files = [fpath for fpath in manifest_json_files if github_workflows_dir not in fpath]
     logger.debug("found meta files {}".format(manifest_json_files))
     manifest_json = ""
     if len(manifest_json_files) > 0:
@@ -301,6 +304,7 @@ def load_dependency_from_galaxy(path):
     yaml_path = ""
     galaxy_yml_files = safe_glob(os.path.join(path, "**", galaxy_yml), recursive=True)
     galaxy_yml_files.extend(safe_glob(os.path.join(path, "**", GALAXY_yml), recursive=True))
+    galaxy_yml_files = [fpath for fpath in galaxy_yml_files if github_workflows_dir not in fpath]
     logger.debug("found meta files {}".format(galaxy_yml_files))
     if len(galaxy_yml_files) > 0:
         for g in galaxy_yml_files:
@@ -324,6 +328,7 @@ def load_existing_dependency_dir(dependency_dir):
     #     recursive=True,
     # )
     collection_meta_files = safe_glob(os.path.join(dependency_dir, "**", collection_manifest_json), recursive=True)
+    collection_meta_files = [fpath for fpath in collection_meta_files if github_workflows_dir not in fpath]
     requirements = {
         "roles": [],
         "collections": [],

ansible_risk_insight/finder.py

@@ -43,6 +43,8 @@
 
 playbook_taskfile_dir_patterns = ["tasks", "playbooks"]
 
+github_workflows_dir = ".github/workflows"
+
 
 class Singleton(type):
     _instances = {}
@@ -321,6 +323,7 @@ def find_best_repo_root_path(path):
 def find_collection_name_of_repo(path):
     pattern = os.path.join(path, "**/galaxy.yml")
     found_galaxy_ymls = safe_glob(pattern, recursive=True)
+    found_galaxy_ymls = [fpath for fpath in found_galaxy_ymls if github_workflows_dir not in fpath]
 
     # skip galaxy ymls found in collections/roles in the repository
     _galaxy_ymls = []

ansible_risk_insight/parser.py

@@ -261,8 +261,16 @@ def run(self, load_data=None, load_json_path="", collection_name_of_project=""):
             mappings["roles"].append([role_path, r.key])
 
         taskfiles = [tf for r in roles for tf in r.taskfiles if r.fqcn != ld.target_name]
+        loaded_absolute_path_list = []
+        for r in roles:
+            for tf in r.taskfiles:
+                if r.fqcn != ld.target_name:
+                    loaded_absolute_path_list.append(os.path.join(basedir, r.defined_in, tf.defined_in))
         for taskfile_path in ld.taskfiles:
             try:
+                abs_path = os.path.join(basedir, taskfile_path)
+                if abs_path in loaded_absolute_path_list:
+                    continue
                 tf = load_taskfile(
                     path=taskfile_path,
                     yaml_str=ld.taskfile_yaml,