Support non GitHub repo hosts, ignore metrics (#325)

* Support non GitHub repo hosts, ignore metrics

* Fix parsing

* Update changelog

* lint

Author: banesullivan

CHANGELOG.md

@@ -6,6 +6,8 @@ See [GitHub releases](https://github.com/pyOpenSci/pyosMeta/releases) page for a
 
 ## [Unreleased]
 
+* Feature: allow non-GitHub repository hosts and establish a pattern for collecting repository metrics (@banesullivan, #325)
+
 ## [v1.7.6] - 2025-10-10
 * Bug fix: handle markdown styling in issue template for package submission
 * Feature: allow empty reviewers for in issue template for package submission

src/pyosmeta/cli/process_reviews.py

@@ -53,7 +53,7 @@ def main():
     # Contrib count is only available via rest api
     logger.info("Getting GitHub metrics for all packages...")
     repo_paths = process_review.get_repo_paths(accepted_reviews)
-    all_reviews = github_api.get_gh_metrics(repo_paths, accepted_reviews)
+    all_reviews = github_api.get_metrics(repo_paths, accepted_reviews)
 
     with open("all_reviews.pickle", "wb") as f:
         pickle.dump(all_reviews, f)

src/pyosmeta/github_api.py

@@ -18,6 +18,7 @@
 from dotenv import load_dotenv
 
 from pyosmeta.models import ReviewModel
+from pyosmeta.models.base import RepositoryHost
 
 from .logging import logger
 
@@ -181,7 +182,7 @@ def _get_response_rest(self, url: str) -> list[dict[str, Any]]:
 
         return results
 
-    def get_gh_metrics(
+    def get_metrics(
         self,
         endpoints: dict[dict[str, str]],
         reviews: dict[str, ReviewModel],
@@ -204,7 +205,15 @@ def get_gh_metrics(
         """
 
         for pkg_name, owner_repo in endpoints.items():
-            reviews[pkg_name].gh_meta = self.get_repo_meta(owner_repo)
+            review = reviews[pkg_name]
+            if review.repository_host == RepositoryHost.github:
+                reviews[pkg_name].gh_meta = self.get_repo_meta_github(
+                    owner_repo
+                )
+            else:
+                logger.warning(
+                    f"Unsupported repository host for {pkg_name}: {review.repository_host}"
+                )
 
         return reviews
 
@@ -363,7 +372,7 @@ def _get_metrics_graphql(
             )
             return None
 
-    def get_repo_meta(
+    def get_repo_meta_github(
         self, repo_info: dict[str, str]
     ) -> dict[str, Any] | None:
         """Get GitHub metrics from the GitHub GraphQL API for a repository.
@@ -392,6 +401,11 @@ def get_repo_meta(
 
         return metrics
 
+    def get_repo_meta_gitlab(
+        self, repo_info: dict[str, str]
+    ) -> dict[str, Any] | None:
+        raise NotImplementedError
+
     def get_user_info(
         self, gh_handle: str, name: Optional[str] = None
     ) -> dict[str, Union[str, Any]]:

src/pyosmeta/models/base.py

@@ -15,6 +15,7 @@
     Field,
     field_serializer,
     field_validator,
+    model_validator,
 )
 
 from pyosmeta.logging import logger
@@ -32,6 +33,56 @@ class Partnerships(str, Enum):
     pangeo = "pangeo"
 
 
+class RepositoryHost(str, Enum):
+    other = "other"
+    github = "github"
+    gitlab = "gitlab"
+    codeberg = "codeberg"
+
+    @classmethod
+    def from_url(cls, url: str) -> "RepositoryHost":
+        """Determine the repository host from a given URL.
+
+        Parameters
+        ----------
+        url : str
+            The URL of the repository.
+
+        Returns
+        -------
+        RepositoryHost
+            The corresponding RepositoryHost enum value.
+        """
+        if "github.com" in url:
+            return cls.github
+        elif "gitlab.com" in url:
+            return cls.gitlab
+        elif "codeberg.org" in url:
+            return cls.codeberg
+        else:
+            return cls.other
+
+    def parse_url(self, url: str) -> tuple[str, str]:
+        """Parse the URL to extract the repository host name.
+
+        Parameters
+        ----------
+        url : str
+            The URL of the repository.
+
+        Returns
+        -------
+        tuple[str, str]
+            The owner and repository name.
+        """
+        match = re.match(r"https?://[^/]+/([^/]+)/([^/]+)", url)
+        if match:
+            owner, repo = match.group(1), match.group(2)
+            return owner, repo
+        else:
+            raise ValueError(f"Could not parse owner/repo from URL: {url}")
+
+
 class UrlValidatorMixin:
     """A mixin to validate classes that are of the same type across
     several models.
@@ -252,6 +303,7 @@ class ReviewModel(BaseModel):
     all_current_maintainers: list[ReviewUser] = Field(default_factory=list)
     # Support presubmissions with an alias
     repository_link: str = Field(..., alias="repository_link_(if_existing)")
+    repository_host: RepositoryHost = Field(default=None)
     version_submitted: Optional[str] = None
     categories: Optional[list[str]] = None
     editor: ReviewUser | list[ReviewUser] | None = None
@@ -273,6 +325,15 @@ class ReviewModel(BaseModel):
     labels: list[str] = Field(default_factory=list)
     active: bool = True  # To indicate if package is maintained or archived
 
+    @model_validator(mode="after")
+    def set_repository_host_from_link(self):
+        """Set repository_host based on repository_link if not already set."""
+        if self.repository_host is None and self.repository_link:
+            self.repository_host = RepositoryHost.from_url(
+                self.repository_link
+            )
+        return self
+
     @field_validator(
         "date_accepted",
         mode="before",
@@ -388,6 +449,7 @@ def listify(cls, item: Any):
             return item
 
     @field_validator("labels", mode="before")
+    @classmethod
     def extract_label(cls, labels: list[str | Labels]) -> list[str]:
         """
         Get just the ``name`` from the Labels model, if given

src/pyosmeta/parse_issues.py

@@ -8,6 +8,7 @@
 from tqdm.contrib.logging import logging_redirect_tqdm
 
 from pyosmeta.models import ReviewModel, ReviewUser
+from pyosmeta.models.base import RepositoryHost
 from pyosmeta.models.github import Issue, Labels, LabelType
 
 from .github_api import GitHubAPI
@@ -388,16 +389,8 @@ def get_repo_paths(
         all_repos = {}
         for a_package in review_issues.keys():
             repo_url = review_issues[a_package].repository_link
-            # for now skip if it's a gitlab repo
-            if "gitlab" in repo_url:
-                continue
-            owner, repo = (
-                repo_url.replace("https://github.com/", "")
-                .replace("https://www.github.com/", "")
-                .rstrip("/")
-                .split("/", 1)
-            )
-
+            host = RepositoryHost(review_issues[a_package].repository_host)
+            owner, repo = host.parse_url(repo_url)
             all_repos[a_package] = {"owner": owner, "repo_name": repo}
         return all_repos
 

tests/data/reviews/github_submission.txt

@@ -0,0 +1,27 @@
+Submitting Author: Fakename (@fakeauthor)
+All current maintainers: (@fakeauthor1, @fakeauthor2)
+Package Name: fake_package
+One-Line Description of Package: A fake python package
+Repository Link: https://github.com/fakeauthor1/fake_package
+Version submitted: v1.0.0
+EiC: @fakeeic1
+Editor: @fakeeditor1
+Reviewers: @fakereviewer1 , @fakereviewer2, @fakereviewer3
+Reviews Expected By: fake date
+Archive: [![DOI](https://example.com/fakearchive)](https://zenodo.org/records/8415866)
+JOSS DOI: [![DOI](https://example.com/fakearchive)](https://doi.org/10.21105/joss.01450)
+Version accepted: 2.0.0 ([repo](https://example.com/fakeauthor1/fake_package/releases/tag/v2.0.0), [pypi](https://pypi.org/project/fake_project/2.0.0), [archive](https://example.com/fakearchive))
+Date accepted (month/day/year): 06/29/2024
+
+---
+
+## Scope
+
+- [x] I agree to abide by [pyOpenSci's Code of Conduct][PyOpenSciCodeOfConduct] during the review process and in maintaining my package after should it be accepted.
+- [x] I have read and will commit to package maintenance after the review as per the [pyOpenSci Policies Guidelines][Commitment].
+(etc)
+
+## Community Partnerships
+
+- [ ] etc
+- [ ] aaaaaa

tests/data/reviews/gitlab_submission.txt

@@ -0,0 +1,27 @@
+Submitting Author: Fakename (@fakeauthor)
+All current maintainers: (@fakeauthor1, @fakeauthor2)
+Package Name: fake_package
+One-Line Description of Package: A fake python package
+Repository Link: https://gitlab.com/fakeauthor1/fake_package
+Version submitted: v1.0.0
+EiC: @fakeeic1
+Editor: @fakeeditor1
+Reviewers: @fakereviewer1 , @fakereviewer2, @fakereviewer3
+Reviews Expected By: fake date
+Archive: [![DOI](https://example.com/fakearchive)](https://zenodo.org/records/8415866)
+JOSS DOI: [![DOI](https://example.com/fakearchive)](https://doi.org/10.21105/joss.01450)
+Version accepted: 2.0.0 ([repo](https://example.com/fakeauthor1/fake_package/releases/tag/v2.0.0), [pypi](https://pypi.org/project/fake_project/2.0.0), [archive](https://example.com/fakearchive))
+Date accepted (month/day/year): 06/29/2024
+
+---
+
+## Scope
+
+- [x] I agree to abide by [pyOpenSci's Code of Conduct][PyOpenSciCodeOfConduct] during the review process and in maintaining my package after should it be accepted.
+- [x] I have read and will commit to package maintenance after the review as per the [pyOpenSci Policies Guidelines][Commitment].
+(etc)
+
+## Community Partnerships
+
+- [ ] etc
+- [ ] aaaaaa

tests/integration/test_parse_issues.py

@@ -143,3 +143,21 @@ def test_multiple_editors_and_eic(process_issues, data_file):
     review = data_file("reviews/multiple_editors.txt", True)
     review = process_issues.parse_issue(review)
     assert review.package_name == "fake_package"
+
+
+def test_repository_host_github(process_issues, data_file):
+    """
+    Test handling of submissions with GitHub repository hosts.
+    """
+    review = data_file("reviews/github_submission.txt", True)
+    review = process_issues.parse_issue(review)
+    assert review.repository_host == "github"
+
+
+def test_repository_host_gitlab(process_issues, data_file):
+    """
+    Test handling of submissions with GitLab repository hosts.
+    """
+    review = data_file("reviews/gitlab_submission.txt", True)
+    review = process_issues.parse_issue(review)
+    assert review.repository_host == "gitlab"