ARCHBOM-1551: update generate_code_owner_mappings.py (#25309)

update generate_code_owner_mappings.py:

* Uses updated 'owner.theme' and 'owner.squad' columns.
* Outputs new CODE_OWNER_THEMES to detail themes.
* Output lowercase names only

ARCHBOM-1551

Co-authored-by: Tim McCormack <tmccormack@edx.org>

lms/djangoapps/monitoring/scripts/generate_code_owner_mappings.py

@@ -3,7 +3,7 @@ This script generates code owner mappings for monitoring LMS.
 
 Sample usage::
 
-    python lms/djangoapps/monitoring/scripts/generate_code_owner_mappings.py --repo-csv "Individual Repo Ownership.csv" --app-csv "edx-platform Apps Ownership.csv" --dep-csv "edx-platform 3rd-party Ownership.csv"
+    python lms/djangoapps/monitoring/scripts/generate_code_owner_mappings.py --repo-csv "Own Repos.csv" --app-csv "Own edx-platform Apps.csv" --dep-csv "Own edx-platform Libs.csv"
 
 Or for more details::
 
@@ -92,12 +92,15 @@ def main(repo_csv, app_csv, dep_csv):
     Final output only includes paths which might contain views.
 
     """
+    # Maps theme name to a list of code owners in the theme, and squad to full code owner name.
+    # Code owner is a string combining theme and squad information.
+    owner_map = {'theme_to_owners_map': {}, 'squad_to_theme_map': {}}
     # Maps owner names to a list of dotted module paths.
     # For example: { 'team-red': [ 'openedx.core.djangoapps.api_admin', 'openedx.core.djangoapps.auth_exchange' ] }
     owner_to_paths_map = {}
-    _map_repo_apps('edx-repo', repo_csv, EDX_REPO_APPS, owner_to_paths_map)
-    _map_repo_apps('3rd-party', dep_csv, THIRD_PARTY_APPS, owner_to_paths_map)
-    _map_edx_platform_apps(app_csv, owner_to_paths_map)
+    _map_repo_apps('edx-repo', repo_csv, EDX_REPO_APPS, owner_map, owner_to_paths_map)
+    _map_repo_apps('3rd-party', dep_csv, THIRD_PARTY_APPS, owner_map, owner_to_paths_map)
+    _map_edx_platform_apps(app_csv, owner_map, owner_to_paths_map)
 
     print('# Do not hand edit CODE_OWNER_MAPPINGS. Generated by {}'.format(os.path.basename(__file__)))
     print('CODE_OWNER_MAPPINGS:')
@@ -107,16 +110,30 @@ def main(repo_csv, app_csv, dep_csv):
         for path in path_list:
             print("  - {}".format(path))
 
+    owner_with_mappings_set = set(owner_to_paths_map.keys())
+    print('# Do not hand edit CODE_OWNER_THEMES. Generated by {}'.format(os.path.basename(__file__)))
+    print('CODE_OWNER_THEMES:')
+    for theme, owner_list in sorted(owner_map['theme_to_owners_map'].items()):
+        theme_owner_set = set(owner_list)
+        # only include the theme's list of owners that have mappings
+        theme_owner_with_mappings_list = list(theme_owner_set & owner_with_mappings_set)
+        if theme_owner_with_mappings_list:
+            print("  {}:".format(theme))
+            theme_owner_with_mappings_list.sort()
+            for owner in theme_owner_with_mappings_list:
+                print("  - {}".format(owner))
 
-def _map_repo_apps(csv_type, repo_csv, app_to_repo_map, owner_to_paths_map):
+
+def _map_repo_apps(csv_type, repo_csv, app_to_repo_map, owner_map, owner_to_paths_map):
     """
-    Reads CSV of repo ownership and uses app_to_repo_map to updates owner_to_paths_map
+    Reads CSV of repo ownership and uses app_to_repo_map to update owner_map and owner_to_paths_map
 
     Arguments:
         csv_type (string): Either 'edx-repo' or '3rd-party' for warning message
         repo_csv (string): File name for the edx-repo or 3rd-party repo csv
         app_to_repo_map (dict): Dict mapping Django apps to repo urls
-        owner_to_paths_map (dict): Holds results mapping owner to paths.
+        owner_map (dict): Dict of owner details
+        owner_to_paths_map (dict): Holds results mapping owner to paths
 
     """
     with open(repo_csv, 'r') as file:
@@ -125,7 +142,8 @@ def _map_repo_apps(csv_type, repo_csv, app_to_repo_map, owner_to_paths_map):
 
     csv_repo_to_owner_map = {}
     for row in reader:
-        csv_repo_to_owner_map[row.get('repo url')] = row.get('owner.squad')
+        owner = _get_and_map_code_owner(row, owner_map)
+        csv_repo_to_owner_map[row.get('repo url')] = owner
 
     for app, repo_url in app_to_repo_map.items():
         owner = csv_repo_to_owner_map.get(repo_url, None)
@@ -137,7 +155,7 @@ def _map_repo_apps(csv_type, repo_csv, app_to_repo_map, owner_to_paths_map):
             print('WARNING: Repo {} was not found in {} csv. Needed for app {}.'.format(repo_url, csv_type, app))
 
 
-def _map_edx_platform_apps(app_csv, owner_to_paths_map):
+def _map_edx_platform_apps(app_csv, owner_map, owner_to_paths_map):
     """
     Reads CSV of edx-platform app ownership and updates mappings
     """
@@ -146,7 +164,7 @@ def _map_edx_platform_apps(app_csv, owner_to_paths_map):
     reader = csv.DictReader(csv_data.splitlines())
     for row in reader:
         path = row.get('Path')
-        owner = row.get('owner.squad')
+        owner = _get_and_map_code_owner(row, owner_map)
 
         # add paths that may have views
         may_have_views = re.match(r'.*djangoapps', path) or re.match(r'[./]*openedx\/features', path)
@@ -166,5 +184,48 @@ def _map_edx_platform_apps(app_csv, owner_to_paths_map):
             owner_to_paths_map[owner].append(path)
 
 
+def _get_and_map_code_owner(row, owner_map):
+    """
+    From a csv row, takes the theme and squad, update ownership maps, and return the code_owner.
+
+    Will also warn if the squad appears in multiple themes.
+
+    Arguments:
+        row: A csv row that should have 'owner.theme' and 'owner.squad'.
+        owner_map: A dict with 'theme_to_owners_map' and 'squad_to_theme_map' keys.
+
+    Returns:
+        The code_owner for the row.  This is made from the theme+squad (or squad if there is no theme).
+
+    """
+    theme = row.get('owner.theme')
+    squad = row.get('owner.squad')
+    assert squad, 'Csv row is missing required owner.squad: %s' % row
+
+    # use lower case names only
+    squad = squad.lower()
+    if theme:
+        theme = theme.lower()
+
+    owner = '{}-{}'.format(theme, squad) if theme else squad
+    theme = theme or squad
+
+    if squad not in owner_map['squad_to_theme_map']:
+        # store the theme for each squad for a later data integrity check
+        owner_map['squad_to_theme_map'][squad] = theme
+
+        # add to the list of owners for each theme
+        if theme not in owner_map['theme_to_owners_map']:
+            owner_map['theme_to_owners_map'][theme] = []
+        owner_map['theme_to_owners_map'][theme].append(owner)
+
+    # assert that squads have a unique theme. otherwise we have a data integrity issues in the csv.
+    assert owner_map['squad_to_theme_map'][squad] == theme, \
+        'Squad %s is associated with theme %s in row %s, but theme %s elsewhere in the csv.' % \
+        (squad, theme, row, owner_map['squad_to_theme_map'][squad])
+
+    return owner
+
+
 if __name__ == "__main__":
     main()  # pylint: disable=no-value-for-parameter