Added few changes to repository_contributor file

backend/apps/github/graphql/queries/repository_contributor.py

@@ -1,8 +1,6 @@
 """OWASP repository contributor GraphQL queries."""
 
 import graphene
-from django.db.models import F, Window
-from django.db.models.functions import Rank
 
 from apps.common.graphql.queries import BaseQuery
 from apps.github.graphql.nodes.repository_contributor import RepositoryContributorNode
@@ -16,66 +14,34 @@ class RepositoryContributorQuery(BaseQuery):
         RepositoryContributorNode,
         limit=graphene.Int(default_value=15),
         organization=graphene.String(required=False),
+        project=graphene.String(required=False),
     )
 
-    def resolve_top_contributors(root, info, limit, organization=None):
-        """Resolve top contributors only for repositories with projects.
+    def resolve_top_contributors(root, info, limit, organization=None, project=None):
+        """Resolve top contributors.
 
         Args:
             root (Any): The root query object.
             info (ResolveInfo): The GraphQL execution context.
             limit (int): Maximum number of contributors to return.
             organization (str, optional): Organization login to filter by.
+            project (str, optional): Project key to filter by.
 
         Returns:
             list: List of top contributors with their details.
 
         """
-        queryset = (
-            RepositoryContributor.objects.by_humans()
-            .to_community_repositories()
-            .filter(repository__project__isnull=False)
-            .select_related("repository__project", "user")
-            .annotate(
-                rank=Window(
-                    expression=Rank(),
-                    order_by=F("contributions_count").desc(),
-                    partition_by=F("user__login"),
-                )
-            )
-        )
-
-        if organization:
-            queryset = queryset.select_related(
-                "repository__organization",
-            ).filter(
-                repository__organization__login=organization,
-            )
-
-        top_contributors = (
-            queryset.filter(rank=1)
-            .annotate(
-                project_name=F("repository__project__name"),
-                project_key=F("repository__project__key"),
-            )
-            .values(
-                "contributions_count",
-                "user__avatar_url",
-                "user__login",
-                "user__name",
-                "project_key",
-                "project_name",
-            )
-            .order_by("-contributions_count")[:limit]
+        top_contributors = RepositoryContributor.get_top_contributors(
+            organization=organization, project=project, limit=limit
         )
 
         return [
             RepositoryContributorNode(
-                avatar_url=trc["user__avatar_url"],
+                avatar_url=trc["avatar_url"],
                 contributions_count=trc["contributions_count"],
-                login=trc["user__login"],
-                name=trc["user__name"],
-                project_key=trc["project_key"].replace("www-project-", ""),
+                login=trc["login"],
+                name=trc["name"],
+                project_key=trc["project_key"],
                 project_name=trc["project_name"],
             )
             for trc in top_contributors

backend/apps/github/models/mixins/repository.py

@@ -1,10 +1,8 @@
 """GitHub repository mixins."""
 
 from apps.github.models.repository_contributor import (
-    TOP_CONTRIBUTORS_LIMIT,
     RepositoryContributor,
 )
-from apps.github.models.user import User
 
 
 class RepositoryIndexMixin:
@@ -103,23 +101,7 @@ def idx_subscribers_count(self):
     @property
     def idx_top_contributors(self):
         """Return top contributors for indexing."""
-        return [
-            {
-                "avatar_url": tc["user__avatar_url"],
-                "contributions_count": tc["contributions_count"],
-                "login": tc["user__login"],
-                "name": tc["user__name"],
-            }
-            for tc in RepositoryContributor.objects.filter(repository=self)
-            .exclude(user__login__in=User.get_non_indexable_logins())
-            .values(
-                "contributions_count",
-                "user__avatar_url",
-                "user__login",
-                "user__name",
-            )
-            .order_by("-contributions_count")[:TOP_CONTRIBUTORS_LIMIT]
-        ]
+        return RepositoryContributor.get_top_contributors(repositories=[self])
 
     @property
     def idx_topics(self):

backend/apps/github/models/repository_contributor.py

@@ -1,6 +1,8 @@
 """Github app label model."""
 
 from django.db import models
+from django.db.models import F, Sum, Window
+from django.db.models.functions import Rank
 from django.template.defaultfilters import pluralize
 
 from apps.common.models import BulkSaveModel, TimestampedModel
@@ -102,3 +104,88 @@ def update_data(gh_contributor, repository, user, save=True):
             repository_contributor.save()
 
         return repository_contributor
+
+    @classmethod
+    def get_top_contributors(
+        cls, repositories=None, organization=None, project=None, limit=TOP_CONTRIBUTORS_LIMIT
+    ):
+        """Get top contributors across repositories, organization, or project.
+
+        Args:
+            repositories (QuerySet or list, optional): Repositories to filter by.
+            organization (str, optional): Organization login to filter by.
+            project (str, optional): Project key to filter by.
+            limit (int, optional): Maximum number of contributors to return.
+
+        Returns:
+            list: List of dictionaries containing contributor information.
+
+        """
+        queryset = (
+            cls.objects.by_humans()
+            .to_community_repositories()
+            .select_related("repository__project", "user")
+        )
+
+        if repositories:
+            queryset = queryset.filter(repository__in=repositories)
+
+        if organization:
+            queryset = queryset.select_related(
+                "repository__organization",
+            ).filter(
+                repository__organization__login=organization,
+            )
+
+        if project:
+            queryset = queryset.filter(
+                repository__project__key__iexact=f"www-project-{project}" if project else None
+            )
+
+        if not repositories:
+            # Only apply project filter when not filtering by repositories
+            queryset = queryset.filter(repository__project__isnull=False)
+
+            # Rank by contributions when looking across repositories
+            queryset = queryset.annotate(
+                rank=Window(
+                    expression=Rank(),
+                    order_by=F("contributions_count").desc(),
+                    partition_by=F("user__login"),
+                )
+            )
+
+            # Get top contributor per user
+            queryset = queryset.filter(rank=1)
+
+        # Aggregate total contributions for users
+        result = queryset.values(
+            "user__avatar_url",
+            "user__login",
+            "user__name",
+        ).annotate(total_contributions=Sum("contributions_count"))
+
+        # Add project information if we're not querying by project or organization
+        if not project and not organization and not repositories:
+            result = result.annotate(
+                project_name=F("repository__project__name"),
+                project_key=F("repository__project__key"),
+            )
+
+        # Order and limit
+        result = result.order_by("-total_contributions")[:limit]
+
+        # Format the result
+        return [
+            {
+                "avatar_url": item["user__avatar_url"],
+                "contributions_count": item["total_contributions"],
+                "login": item["user__login"],
+                "name": item["user__name"],
+                "project_key": item.get("project_key", "").replace("www-project-", "")
+                if item.get("project_key")
+                else None,
+                "project_name": item.get("project_name"),
+            }
+            for item in result
+        ]

backend/apps/owasp/models/common.py

@@ -7,15 +7,13 @@
 
 import yaml
 from django.db import models
-from django.db.models import Sum
 
 from apps.common.open_ai import OpenAi
 from apps.github.constants import (
     GITHUB_REPOSITORY_RE,
     GITHUB_USER_RE,
 )
 from apps.github.models.repository_contributor import (
-    TOP_CONTRIBUTORS_LIMIT,
     RepositoryContributor,
 )
 from apps.github.utils import get_repository_file_content
@@ -197,23 +195,7 @@ def get_related_url(self, url, exclude_domains=(), include_domains=()):
 
     def get_top_contributors(self, repositories=()):
         """Get top contributors."""
-        return [
-            {
-                "avatar_url": tc["user__avatar_url"],
-                "contributions_count": tc["total_contributions"],
-                "login": tc["user__login"],
-                "name": tc["user__name"],
-            }
-            for tc in RepositoryContributor.objects.by_humans()
-            .filter(repository__in=repositories)
-            .values(
-                "user__avatar_url",
-                "user__login",
-                "user__name",
-            )
-            .annotate(total_contributions=Sum("contributions_count"))
-            .order_by("-total_contributions")[:TOP_CONTRIBUTORS_LIMIT]
-        ]
+        return RepositoryContributor.get_top_contributors(repositories=repositories)
 
     def parse_tags(self, tags):
         """Parse entity tags."""