fix: scope issues count to tracked repos

gittensor/utils/github_api_tools.py

@@ -36,7 +36,6 @@
     query($userId: ID!, $limit: Int!, $cursor: String) {
       node(id: $userId) {
         ... on User {
-          issues(states: [OPEN]) { totalCount }
           pullRequests(first: $limit, states: [MERGED, OPEN, CLOSED], orderBy: {field: CREATED_AT, direction: DESC}, after: $cursor) {
             pageInfo {
               hasNextPage
@@ -942,6 +941,69 @@ def should_skip_merged_pr(
     return (False, None)
 
 
+_USER_OPEN_ISSUES_QUERY = """
+query($userId: ID!, $cursor: String) {
+  node(id: $userId) {
+    ... on User {
+      issues(states: [OPEN], first: 100, after: $cursor) {
+        pageInfo { hasNextPage endCursor }
+        nodes { repository { nameWithOwner } }
+      }
+    }
+  }
+}
+"""
+
+
+def count_tracked_open_issues(
+    token: str,
+    github_user_node_id: str,
+    tracked_repo_names: set,
+    max_pages: int = 10,
+) -> Optional[int]:
+    """Count open issues filed against tracked repos only.
+
+    Returns None on API failure so callers can fall back to 0 — a broken
+    fetch must not silently demote a miner via the spam multiplier.
+    """
+    if not token or not tracked_repo_names:
+        return 0
+
+    # GitHub returns nameWithOwner in original case; normalize both sides.
+    tracked_lower = {name.lower() for name in tracked_repo_names}
+
+    count = 0
+    cursor: Optional[str] = None
+    for _ in range(max_pages):
+        result = execute_graphql_query(
+            query=_USER_OPEN_ISSUES_QUERY,
+            variables={'userId': github_user_node_id, 'cursor': cursor},
+            token=token,
+            max_attempts=3,
+        )
+        if not result or 'errors' in result:
+            bt.logging.warning(f'count_tracked_open_issues: GraphQL failure for user {github_user_node_id}')
+            return None
+
+        # data.node can be explicitly null (deleted account, wrong id) — guard it.
+        node = ((result.get('data') or {}).get('node')) or {}
+        issues_block = node.get('issues') or {}
+        for issue_node in issues_block.get('nodes') or []:
+            repo_name = ((issue_node or {}).get('repository') or {}).get('nameWithOwner')
+            if repo_name and repo_name.lower() in tracked_lower:
+                count += 1
+
+        page_info = issues_block.get('pageInfo') or {}
+        if not page_info.get('hasNextPage'):
+            return count
+        cursor = page_info.get('endCursor')
+
+    bt.logging.warning(
+        f'count_tracked_open_issues: hit {max_pages}-page cap for user {github_user_node_id}, returning partial count'
+    )
+    return count
+
+
 def load_miners_prs(
     miner_eval: MinerEvaluation, master_repositories: Dict[str, RepositoryConfig], max_prs: int = 1000
 ) -> None:
@@ -998,10 +1060,6 @@ def load_miners_prs(
                 bt.logging.warning('User not found or no pull requests')
                 break
 
-            # Extract open issue count from first page (User-level field, not paginated)
-            if cursor is None:
-                miner_eval.total_open_issues = user_data.get('issues', {}).get('totalCount', 0)
-
             pr_data: Dict = user_data.get('pullRequests', {})
             prs: List = pr_data.get('nodes', [])
             page_info: Dict = pr_data.get('pageInfo', {})
@@ -1058,9 +1116,18 @@ def load_miners_prs(
     except Exception as e:
         bt.logging.error(f'Unexpected error fetching PRs via GraphQL: {e}')
 
+    # Count open issues scoped to tracked repos (anti-spam gate).
+    # Must be scoped — a global count penalises miners for unrelated personal projects.
+    scoped_count = count_tracked_open_issues(
+        miner_eval.github_pat,
+        global_user_id,
+        set(master_repositories.keys()),
+    )
+    miner_eval.total_open_issues = scoped_count if scoped_count is not None else 0
+
     bt.logging.info(
         f'Fetched {len(miner_eval.merged_pull_requests)} merged PRs, {len(miner_eval.open_pull_requests)} open PRs, '
-        f'{len(miner_eval.closed_pull_requests)} closed'
+        f'{len(miner_eval.closed_pull_requests)} closed, {miner_eval.total_open_issues} tracked-repo open issues'
     )
 
 

tests/utils/test_count_tracked_open_issues.py

@@ -0,0 +1,111 @@
+"""Scoped open issue count — filters User.issues to tracked repos only.
+
+Fixes the bug where the spam multiplier gated on the global User.issues.totalCount
+(including personal projects and non-tracked upstreams), silently zeroing
+miner scores.
+"""
+
+from unittest.mock import patch
+
+from gittensor.utils.github_api_tools import count_tracked_open_issues
+
+TRACKED = {'owner/repo-a', 'owner/repo-b'}
+
+
+def _page(nodes, has_next=False, cursor=None):
+    return {
+        'data': {
+            'node': {
+                'issues': {
+                    'pageInfo': {'hasNextPage': has_next, 'endCursor': cursor},
+                    'nodes': nodes,
+                }
+            }
+        }
+    }
+
+
+def _issue(repo):
+    return {'repository': {'nameWithOwner': repo}}
+
+
+def test_counts_only_tracked_repos():
+    response = _page(
+        [
+            _issue('owner/repo-a'),
+            _issue('owner/repo-a'),
+            _issue('someone/personal-sandbox'),  # not tracked → must be excluded
+            _issue('owner/repo-b'),
+            _issue('other/upstream'),  # not tracked → must be excluded
+        ]
+    )
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query', return_value=response):
+        assert count_tracked_open_issues('tok', 'node-id', TRACKED) == 3
+
+
+def test_empty_tracked_set_returns_zero():
+    # Short-circuit without hitting the API.
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query') as mock:
+        assert count_tracked_open_issues('tok', 'node-id', set()) == 0
+        mock.assert_not_called()
+
+
+def test_missing_token_returns_zero():
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query') as mock:
+        assert count_tracked_open_issues('', 'node-id', TRACKED) == 0
+        mock.assert_not_called()
+
+
+def test_graphql_failure_returns_none():
+    """None = signal to caller; caller defaults to 0 (no penalty) on failure.
+    Safer direction — a broken fetch must not silently demote a miner."""
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query', return_value=None):
+        assert count_tracked_open_issues('tok', 'node-id', TRACKED) is None
+
+
+def test_graphql_error_block_returns_none():
+    with patch(
+        'gittensor.utils.github_api_tools.execute_graphql_query',
+        return_value={'errors': [{'message': 'boom'}]},
+    ):
+        assert count_tracked_open_issues('tok', 'node-id', TRACKED) is None
+
+
+def test_pagination_accumulates_across_pages():
+    pages = [
+        _page([_issue('owner/repo-a'), _issue('other/upstream')], has_next=True, cursor='c1'),
+        _page([_issue('owner/repo-b'), _issue('owner/repo-a')], has_next=False),
+    ]
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query', side_effect=pages):
+        assert count_tracked_open_issues('tok', 'node-id', TRACKED) == 3
+
+
+def test_hits_max_pages_cap_returns_partial():
+    """Pathological account with huge open-issue history — return partial rather than crash."""
+    cap = 2
+    pages = [_page([_issue('owner/repo-a')], has_next=True, cursor=f'c{i}') for i in range(cap)]
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query', side_effect=pages):
+        result = count_tracked_open_issues('tok', 'node-id', TRACKED, max_pages=cap)
+        assert result == cap  # one hit per page
+
+
+def test_case_insensitive_repo_match():
+    """master_repositories is lowercased by load_master_repo_weights, but
+    GitHub returns nameWithOwner in original case. Both sides must normalize."""
+    response = _page(
+        [
+            _issue('Owner/Repo-A'),  # mixed case from GitHub
+            _issue('OWNER/REPO-B'),  # all caps
+            _issue('Other/Upstream'),  # not tracked
+        ]
+    )
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query', return_value=response):
+        assert count_tracked_open_issues('tok', 'node-id', TRACKED) == 2
+
+
+def test_null_node_does_not_crash():
+    """data.node can be explicitly null (deleted account, wrong id) — must not
+    raise AttributeError on None.get('issues')."""
+    response = {'data': {'node': None}}
+    with patch('gittensor.utils.github_api_tools.execute_graphql_query', return_value=response):
+        assert count_tracked_open_issues('tok', 'node-id', TRACKED) == 0