fix: Use auto-approved users allowlist for queue eligibility (#7156)

Author: ReenigneArcher

.github/workflows/auto-update-db.yml

@@ -95,7 +95,7 @@ jobs:
         if: github.event.label.name == 'request-theme'
         env:
           ISSUE_AUTHOR_USER_ID: ${{ github.event.issue.user.id }}
-        run: uv run --no-sync python -m src.queue_eligibility --submission-file submission.json --database-root database
+        run: uv run --no-sync python -m src.queue_eligibility
 
       - name: Update
         id: update

auto_approved_users.json

@@ -0,0 +1,14 @@
+[
+  {
+    "user_id": 30657709,
+    "username": "LuckyNoS7evin"
+  },
+  {
+    "user_id": 42013603,
+    "username": "ReenigneArcher"
+  },
+  {
+    "user_id": 88998541,
+    "username": "jjnether"
+  }
+]

src/queue_eligibility.py

@@ -6,21 +6,10 @@
 import os
 from dataclasses import dataclass
 from pathlib import Path
-from urllib.parse import urlparse
+from typing import Iterable
 
 
-CONTRIBUTION_THRESHOLD = 15
-IGDB_HOST = 'igdb.com'
-TMDB_HOST = 'themoviedb.org'
-
-CATEGORY_BY_HOST_AND_PATH = {
-    (IGDB_HOST, 'games'): 'games',
-    (IGDB_HOST, 'collections'): 'game_collections',
-    (IGDB_HOST, 'franchises'): 'game_franchises',
-    (TMDB_HOST, 'movie'): 'movies',
-    (TMDB_HOST, 'collection'): 'movie_collections',
-    (TMDB_HOST, 'tv'): 'tv_shows',
-}
+AUTO_APPROVED_USERS_FILE = 'auto_approved_users.json'
 
 
 @dataclass(frozen=True)
@@ -32,169 +21,102 @@ class QueueEligibilityResult:
     ----------
     queue_eligible : bool
         Whether the submission should receive the queue label.
-    category : str
-        Database category derived from the submission URL.
-    contribution_count : int
-        Existing contribution count for the submitter in the derived category.
+    user_id : str
+        GitHub user id of the issue author.
     reason : str
         Machine-readable reason for the eligibility decision.
     """
 
     queue_eligible: bool
-    category: str
-    contribution_count: int
+    user_id: str
     reason: str
 
 
-def get_submission_category(database_url: str) -> str:
+def normalize_user_id(user_id: object) -> str:
     """
-    Return the database category for a submission URL.
+    Normalize a GitHub user id for matching.
 
     Parameters
     ----------
-    database_url : str
-        Database URL from the issue submission.
+    user_id : object
+        GitHub user id to normalize.
 
     Returns
     -------
     str
-        Top-level database category directory.
-
-    Raises
-    ------
-    ValueError
-        If the URL cannot be mapped to a known database category.
-    """
-    parsed_url = urlparse(database_url.strip())
-    hostname = parsed_url.hostname or ''
-    hostname = hostname.removeprefix('www.')
-    path_parts = [part for part in parsed_url.path.split('/') if part]
-
-    if not path_parts:
-        raise ValueError(f'Unsupported database URL: {database_url}')
-
-    category = CATEGORY_BY_HOST_AND_PATH.get((hostname, path_parts[0]))
-    if not category:
-        raise ValueError(f'Unsupported database URL: {database_url}')
-
-    return category
-
-
-def get_contribution_count(contributor_data: dict, user_id: str) -> int:
-    """
-    Return a contributor's total contributions from contributor metadata.
-
-    Parameters
-    ----------
-    contributor_data : dict
-        Parsed contents of a category ``contributors.json`` file.
-    user_id : str
-        GitHub user ID to look up.
-
-    Returns
-    -------
-    int
-        Sum of the contributor's added and edited item counts, or zero when the
-        metadata is missing or malformed.
+        Trimmed user id string.
     """
-    if not isinstance(contributor_data, dict):
-        return 0
+    if user_id is None:
+        return ''
 
-    contributor = contributor_data.get(str(user_id), {})
-    if not isinstance(contributor, dict):
-        return 0
+    return str(user_id).strip()
 
-    try:
-        return int(contributor.get('items_added', 0)) + int(contributor.get('items_edited', 0))
-    except (TypeError, ValueError):
-        return 0
 
-
-def load_contribution_count(database_root: Path, category: str, user_id: str) -> int:
+def load_auto_approved_user_ids(auto_approved_users_file: Path) -> frozenset[str]:
     """
-    Load a contributor's category contribution count from disk.
+    Load auto-approved GitHub user ids from a JSON file.
 
     Parameters
     ----------
-    database_root : pathlib.Path
-        Root directory containing category database folders.
-    category : str
-        Category directory to inspect.
-    user_id : str
-        GitHub user ID to look up.
+    auto_approved_users_file : pathlib.Path
+        JSON file containing auto-approved user objects.
 
     Returns
     -------
-    int
-        Contributor count for the category, or zero when the contributors file
-        is absent or unreadable.
+    frozenset[str]
+        Normalized GitHub user ids.
     """
-    contributors_file = database_root / category / 'contributors.json'
-    if not contributors_file.exists():
-        return 0
-
     try:
-        with contributors_file.open() as contributor_f:
-            contributor_data = json.load(contributor_f)
+        with auto_approved_users_file.open(encoding='utf-8') as auto_approved_users_f:
+            auto_approved_users = json.load(auto_approved_users_f)
     except (OSError, json.JSONDecodeError):
-        return 0
+        return frozenset()
+
+    if not isinstance(auto_approved_users, list):
+        return frozenset()
 
-    return get_contribution_count(contributor_data=contributor_data, user_id=user_id)
+    return frozenset(
+        normalized_user_id
+        for user in auto_approved_users
+        if isinstance(user, dict)
+        if (normalized_user_id := normalize_user_id(user.get('user_id', '')))
+    )
 
 
-def evaluate_queue_eligibility(submission: dict,
-                               database_root: Path,
-                               user_id: str,
-                               threshold: int = CONTRIBUTION_THRESHOLD) -> QueueEligibilityResult:
+def evaluate_queue_eligibility(user_id: object,
+                               auto_approved_user_ids: Iterable[object]) -> QueueEligibilityResult:
     """
     Evaluate whether a submission is eligible to be queued automatically.
 
     Parameters
     ----------
-    submission : dict
-        Parsed issue submission data.
-    database_root : pathlib.Path
-        Root directory containing category database folders.
-    user_id : str
-        GitHub user ID of the issue author.
-    threshold : int, optional
-        Minimum prior category contribution count that must be exceeded.
+    user_id : object
+        GitHub user id of the issue author.
+    auto_approved_user_ids : Iterable[object]
+        GitHub user ids eligible for automatic queueing.
 
     Returns
     -------
     QueueEligibilityResult
         Eligibility decision and supporting metadata.
     """
-    if not user_id:
+    normalized_user_id = normalize_user_id(user_id=user_id)
+    if not normalized_user_id:
         return QueueEligibilityResult(
             queue_eligible=False,
-            category='',
-            contribution_count=0,
+            user_id='',
             reason='missing-user-id',
         )
 
-    try:
-        category = get_submission_category(database_url=submission['database_url'])
-    except (KeyError, ValueError):
-        return QueueEligibilityResult(
-            queue_eligible=False,
-            category='',
-            contribution_count=0,
-            reason='unsupported-category',
-        )
-
-    contribution_count = load_contribution_count(
-        database_root=database_root,
-        category=category,
-        user_id=user_id,
-    )
-    queue_eligible = contribution_count > threshold
-    reason = 'eligible' if queue_eligible else 'below-threshold'
+    queue_eligible = normalized_user_id in {
+        normalize_user_id(user_id=auto_approved_user_id)
+        for auto_approved_user_id in auto_approved_user_ids
+    }
+    reason = 'auto-approved-user' if queue_eligible else 'not-auto-approved-user'
 
     return QueueEligibilityResult(
         queue_eligible=queue_eligible,
-        category=category,
-        contribution_count=contribution_count,
+        user_id=normalized_user_id,
         reason=reason,
     )
 
@@ -214,8 +136,7 @@ def write_github_outputs(result: QueueEligibilityResult) -> None:
     """
     outputs = {
         'queue_eligible': str(result.queue_eligible).lower(),
-        'category': result.category,
-        'contribution_count': str(result.contribution_count),
+        'user_id': result.user_id,
         'reason': result.reason,
     }
     output_lines = [f'{key}={value}' for key, value in outputs.items()]
@@ -239,9 +160,7 @@ def parse_args() -> argparse.Namespace:
         Parsed command-line arguments.
     """
     parser = argparse.ArgumentParser(description='Check whether a submission is eligible for queueing.')
-    parser.add_argument('--submission-file', default='submission.json')
-    parser.add_argument('--database-root', default='database')
-    parser.add_argument('--threshold', type=int, default=CONTRIBUTION_THRESHOLD)
+    parser.add_argument('--auto-approved-users-file', default=AUTO_APPROVED_USERS_FILE)
     parser.add_argument('--user-id', default=os.environ.get('ISSUE_AUTHOR_USER_ID', ''))
     return parser.parse_args()
 
@@ -255,24 +174,13 @@ def main() -> None:
     None
     """
     args = parse_args()
-
-    try:
-        with open(args.submission_file) as submission_f:
-            submission = json.load(submission_f)
-    except (OSError, json.JSONDecodeError):
-        result = QueueEligibilityResult(
-            queue_eligible=False,
-            category='',
-            contribution_count=0,
-            reason='submission-read-error',
-        )
-    else:
-        result = evaluate_queue_eligibility(
-            submission=submission,
-            database_root=Path(args.database_root),
-            user_id=args.user_id,
-            threshold=args.threshold,
-        )
+    auto_approved_user_ids = load_auto_approved_user_ids(
+        auto_approved_users_file=Path(args.auto_approved_users_file),
+    )
+    result = evaluate_queue_eligibility(
+        user_id=args.user_id,
+        auto_approved_user_ids=auto_approved_user_ids,
+    )
     write_github_outputs(result=result)