feat: add github-repo to collaborator transformer (#8)

* feat: add github-repo to collaborator transformer

* chore: bump version

Author: jbristow

nodestream_github/client/githubclient.py

@@ -23,6 +23,7 @@
 
 import nodestream_github.types as types
 from nodestream_github.logging import get_plugin_logger
+from nodestream_github.types.enums import CollaboratorAffiliation
 
 DEFAULT_REQUEST_RATE_LIMIT_PER_MINUTE = int(13000 / 60)
 DEFAULT_MAX_RETRIES = 20
@@ -81,6 +82,7 @@ def __init__(
         max_retries: int | None = None,
         rate_limit_per_minute: int | None = None,
         max_retry_wait_seconds: int | None = None,
+        **_kwargs: any,
     ):
         if per_page is None:
             per_page = DEFAULT_PAGE_SIZE
@@ -97,8 +99,10 @@ def __init__(
         self._auth_token = auth_token
         if github_hostname == "api.github.com" or github_hostname is None:
             self._base_url = "https://api.github.com"
+            self._is_default_hostname = True
         else:
             self._base_url = f"https://{github_hostname}/api/v3"
+            self._is_default_hostname = False
 
         self._per_page = per_page
         self._limit_storage = MemoryStorage()
@@ -181,6 +185,10 @@ def base_url(self) -> str:
     def per_page(self) -> int:
         return self._per_page
 
+    @property
+    def is_default_hostname(self) -> bool:
+        return self._is_default_hostname
+
     async def _get(
         self,
         url: str,
@@ -390,7 +398,11 @@ async def fetch_webhooks_for_repo(
             _fetch_problem(f"webhooks for repo {owner_login}/{repo_name}", e)
 
     async def fetch_collaborators_for_repo(
-        self, owner_login: str, repo_name: str, affiliation: str
+        self,
+        *,
+        owner_login: str,
+        repo_name: str,
+        affiliation: CollaboratorAffiliation,
     ) -> AsyncGenerator[types.GithubUser]:
         """Try to get collaborator data for this repo.
 

nodestream_github/repos.py

@@ -14,6 +14,7 @@
 from .interpretations.relationship.user import simplify_user
 from .logging import get_plugin_logger
 from .types import GithubRepo, RepositoryRecord
+from .types.enums import CollaboratorAffiliation
 
 logger = get_plugin_logger(__name__)
 
@@ -106,15 +107,15 @@ async def _extract_repo(self, repo: GithubRepo) -> RepositoryRecord:
         repo["collaborators"] = []
 
         async for user in self.client.fetch_collaborators_for_repo(
-            owner["login"],
-            repo["name"],
-            "direct",
+            owner_login=owner["login"],
+            repo_name=repo["name"],
+            affiliation=CollaboratorAffiliation.DIRECT,
         ):
             repo["collaborators"].append(simplify_user(user, affiliation="direct"))
         async for user in self.client.fetch_collaborators_for_repo(
-            owner["login"],
-            repo["name"],
-            "outside",
+            owner_login=owner["login"],
+            repo_name=repo["name"],
+            affiliation=CollaboratorAffiliation.OUTSIDE,
         ):
             repo["collaborators"].append(simplify_user(user, affiliation="outside"))
 

nodestream_github/transformer/__init__.py

@@ -0,0 +1,53 @@
+import logging
+from collections.abc import AsyncGenerator
+
+from nodestream.pipeline import Transformer
+
+from nodestream_github import types
+from nodestream_github.client import GithubRestApiClient
+from nodestream_github.interpretations.relationship.repository import simplify_repo
+from nodestream_github.logging import get_plugin_logger
+from nodestream_github.types.enums import CollaboratorAffiliation
+
+logger = get_plugin_logger(__name__)
+
+
+class RepoToCollaboratorsTransformer(Transformer):
+    def __init__(
+        self,
+        *,
+        full_name_key: str = "full_name",
+        **kwargs: any,
+    ):
+
+        self.client = GithubRestApiClient(**kwargs)
+        self.full_name_key = full_name_key
+
+    async def transform_record(
+        self,
+        record: types.GithubRepo,
+    ) -> AsyncGenerator[types.GithubUser]:
+        (repo_owner, repo_name) = record[self.full_name_key].split("/")
+        logging.debug("Transforming record %s/%s", repo_owner, repo_name)
+
+        simplified_repo = simplify_repo(record)
+
+        async for collaborator in self.client.fetch_collaborators_for_repo(
+            owner_login=repo_owner,
+            repo_name=repo_name,
+            affiliation=CollaboratorAffiliation.DIRECT,
+        ):
+            yield collaborator | {
+                "repository": simplified_repo,
+                "affiliation": CollaboratorAffiliation.DIRECT,
+            }
+
+        async for collaborator in self.client.fetch_collaborators_for_repo(
+            owner_login=repo_owner,
+            repo_name=repo_name,
+            affiliation=CollaboratorAffiliation.OUTSIDE,
+        ):
+            yield collaborator | {
+                "repository": simplified_repo,
+                "affiliation": CollaboratorAffiliation.OUTSIDE,
+            }

nodestream_github/transformer/repo.py

@@ -0,0 +1,7 @@
+from enum import StrEnum
+
+
+class CollaboratorAffiliation(StrEnum):
+    ALL = "all"
+    OUTSIDE = "outside"
+    DIRECT = "direct"