')
@@ -699,6 +706,98 @@ def render_replay_detail(investigation: dict[str, Any]) -> str:
return "\n".join(html_parts)
+def render_code_changes(code_changes: list[dict[str, Any]]) -> str:
+ """Render related code changes as HTML with clickable links.
+
+ Args:
+ code_changes: List of code change dicts with commit/PR metadata.
+
+ Returns:
+ HTML string with styled code change links.
+ """
+ if not code_changes:
+ return ""
+
+ html_parts = [
+ '
',
+ "",
+ "
Related Code Changes
",
+ "
",
+ ]
+
+ for change in code_changes[:10]: # Limit to 10 changes
+ url = change.get("url", "#")
+ pr_number = change.get("pr_number")
+ pr_title = change.get("pr_title")
+ commit_hash = change.get("commit_hash", "")
+ message = change.get("message", "")
+ author = change.get("author", "")
+
+ html_parts.append("- ")
+
+ if pr_number:
+ # PR with badge
+ html_parts.append(f'PR #{pr_number}')
+ if pr_title:
+ display_text = html.escape(pr_title)
+ else:
+ display_text = f"Pull Request #{pr_number}"
+ html_parts.append(f'{display_text}')
+ else:
+ # Commit link
+ short_hash = commit_hash[:7] if commit_hash else "unknown"
+ if message:
+ truncated_msg = message[:50] + "..." if len(message) > 50 else message
+ display_text = f"{short_hash}: {html.escape(truncated_msg)}"
+ else:
+ display_text = short_hash
+ html_parts.append(
+ f''
+ f'{display_text}'
+ f""
+ )
+
+ if author:
+ html_parts.append(f' by {html.escape(author)}')
+
+ html_parts.append("
")
+
+ if len(code_changes) > 10:
+ html_parts.append(f"- ...and {len(code_changes) - 10} more changes
")
+
+ html_parts.append("
")
+ html_parts.append("
")
+
+ return "\n".join(html_parts)
+
+
def render_comparison_table(inv1: dict[str, Any], inv2: dict[str, Any]) -> str:
"""Render side-by-side comparison of two investigations as HTML.
diff --git a/python-packages/dataing/migrations/034_code_changes_pr_metadata.sql b/python-packages/dataing/migrations/034_code_changes_pr_metadata.sql
new file mode 100644
index 00000000..f9028bed
--- /dev/null
+++ b/python-packages/dataing/migrations/034_code_changes_pr_metadata.sql
@@ -0,0 +1,22 @@
+-- PR metadata for code changes
+-- Epic fn-38: PR/Commit Link in Investigation Output
+
+ALTER TABLE code_changes
+ ADD COLUMN IF NOT EXISTS pr_number INTEGER,
+ ADD COLUMN IF NOT EXISTS pr_url TEXT,
+ ADD COLUMN IF NOT EXISTS pr_title TEXT,
+ ADD COLUMN IF NOT EXISTS pr_author TEXT,
+ ADD COLUMN IF NOT EXISTS pr_merged_at TIMESTAMPTZ,
+ ADD COLUMN IF NOT EXISTS provider TEXT;
+
+-- Index for PR lookups by commit
+CREATE INDEX IF NOT EXISTS idx_code_changes_pr_number
+ ON code_changes(repo_id, pr_number)
+ WHERE pr_number IS NOT NULL;
+
+COMMENT ON COLUMN code_changes.pr_number IS 'PR/MR number in the provider';
+COMMENT ON COLUMN code_changes.pr_url IS 'Full URL to the PR/MR in the provider UI';
+COMMENT ON COLUMN code_changes.pr_title IS 'Title of the PR/MR';
+COMMENT ON COLUMN code_changes.pr_author IS 'Author login/username of the PR/MR';
+COMMENT ON COLUMN code_changes.pr_merged_at IS 'When the PR/MR was merged';
+COMMENT ON COLUMN code_changes.provider IS 'Git provider: github, gitlab, bitbucket';
diff --git a/python-packages/dataing/src/dataing/adapters/db/code_changes.py b/python-packages/dataing/src/dataing/adapters/db/code_changes.py
index 58358468..762cc8e9 100644
--- a/python-packages/dataing/src/dataing/adapters/db/code_changes.py
+++ b/python-packages/dataing/src/dataing/adapters/db/code_changes.py
@@ -65,6 +65,12 @@ async def get_relevant_code_changes(
cc.message,
cc.committed_at,
cc.affected_assets,
+ cc.pr_number,
+ cc.pr_url,
+ cc.pr_title,
+ cc.pr_author,
+ cc.pr_merged_at,
+ cc.provider,
1.0::float AS relevance_score,
'directly_affects_asset' AS relevance_reason
FROM code_changes cc
@@ -80,6 +86,12 @@ async def get_relevant_code_changes(
cc.message,
cc.committed_at,
cc.affected_assets,
+ cc.pr_number,
+ cc.pr_url,
+ cc.pr_title,
+ cc.pr_author,
+ cc.pr_merged_at,
+ cc.provider,
0.7::float AS relevance_score,
'affects_upstream_dependency' AS relevance_reason
FROM code_changes cc
@@ -101,6 +113,12 @@ async def get_relevant_code_changes(
cc.message,
cc.committed_at,
cc.affected_assets,
+ cc.pr_number,
+ cc.pr_url,
+ cc.pr_title,
+ cc.pr_author,
+ cc.pr_merged_at,
+ cc.provider,
0.4::float AS relevance_score,
'matches_file_path_pattern' AS relevance_reason
FROM code_changes cc
@@ -140,6 +158,12 @@ async def get_relevant_code_changes(
message,
committed_at,
affected_assets,
+ pr_number,
+ pr_url,
+ pr_title,
+ pr_author,
+ pr_merged_at,
+ provider,
relevance_score,
relevance_reason
FROM all_changes
@@ -189,4 +213,11 @@ def _row_to_domain(self, row: dict[str, Any]) -> RelevantCodeChange:
affected_assets=asset_names,
relevance_score=float(row["relevance_score"]),
relevance_reason=row["relevance_reason"],
+ # PR metadata (populated by bond-agent tools during sync)
+ pr_number=row.get("pr_number"),
+ pr_url=row.get("pr_url"),
+ pr_title=row.get("pr_title"),
+ pr_author=row.get("pr_author"),
+ pr_merged_at=row.get("pr_merged_at"),
+ provider=row.get("provider"),
)
diff --git a/python-packages/dataing/src/dataing/adapters/git/__init__.py b/python-packages/dataing/src/dataing/adapters/git/__init__.py
index 2f51f32b..a2ae8484 100644
--- a/python-packages/dataing/src/dataing/adapters/git/__init__.py
+++ b/python-packages/dataing/src/dataing/adapters/git/__init__.py
@@ -1,12 +1,27 @@
"""Git provider adapters for repository sync and commit parsing."""
from dataing.adapters.git.github import GitHubProvider
+from dataing.adapters.git.pr_enrichment import PREnrichmentService, PRMetadata
from dataing.adapters.git.provider import GitCommit, GitProvider
from dataing.adapters.git.sync import GitSyncService
+from dataing.adapters.git.url_templates import (
+ GitProvider as UrlGitProvider,
+)
+from dataing.adapters.git.url_templates import (
+ build_best_link,
+ build_commit_url,
+ build_pr_url,
+)
__all__ = [
"GitCommit",
"GitHubProvider",
"GitProvider",
"GitSyncService",
+ "PREnrichmentService",
+ "PRMetadata",
+ "UrlGitProvider",
+ "build_best_link",
+ "build_commit_url",
+ "build_pr_url",
]
diff --git a/python-packages/dataing/src/dataing/adapters/git/pr_enrichment.py b/python-packages/dataing/src/dataing/adapters/git/pr_enrichment.py
new file mode 100644
index 00000000..7614e997
--- /dev/null
+++ b/python-packages/dataing/src/dataing/adapters/git/pr_enrichment.py
@@ -0,0 +1,184 @@
+"""PR metadata enrichment using bond-agent tools.
+
+This module provides a service for enriching commits with PR metadata
+using bond-agent's githunter and github toolsets rather than building
+custom API clients.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+import structlog
+
+if TYPE_CHECKING:
+ from bond.tools.github import GitHubAdapter
+ from bond.tools.githunter import GitHunterAdapter
+
+logger = structlog.get_logger()
+
+
+@dataclass
+class PRMetadata:
+ """PR metadata extracted from bond-agent tools."""
+
+ pr_number: int
+ pr_url: str
+ pr_title: str
+ pr_author: str
+ pr_merged_at: datetime | None
+
+
+class PREnrichmentService:
+ """Enriches commits with PR metadata using bond-agent.
+
+ This service wraps bond-agent's githunter and github tools to provide
+ a simple interface for looking up PR metadata for commits. It uses:
+
+ - githunter.find_pr_discussion: Local git-based PR lookup (no API calls)
+ - github.get_pr: Full PR details from GitHub API (when token available)
+
+ The service gracefully falls back when GitHub API is not configured,
+ returning partial metadata from the local git lookup.
+ """
+
+ def __init__(
+ self,
+ repo_path: Path | None = None,
+ github_token: str | None = None,
+ ) -> None:
+ """Initialize the enrichment service.
+
+ Args:
+ repo_path: Path to the local git repository (for githunter).
+ github_token: Optional GitHub API token for full PR details.
+ """
+ self.repo_path = repo_path
+ self._githunter: GitHunterAdapter | None = None
+ self._github: GitHubAdapter | None = None
+ self._github_token = github_token
+
+ @property
+ def githunter(self) -> GitHunterAdapter:
+ """Lazy-load the GitHunter adapter."""
+ if self._githunter is None:
+ from bond.tools.githunter import GitHunterAdapter
+
+ self._githunter = GitHunterAdapter()
+ return self._githunter
+
+ @property
+ def github(self) -> GitHubAdapter | None:
+ """Lazy-load the GitHub adapter (if token available)."""
+ if self._github is None and self._github_token:
+ from bond.tools.github import GitHubAdapter
+
+ self._github = GitHubAdapter(token=self._github_token)
+ return self._github
+
+ async def get_pr_for_commit(
+ self,
+ commit_hash: str,
+ owner: str | None = None,
+ repo: str | None = None,
+ ) -> PRMetadata | None:
+ """Find PR metadata for a commit.
+
+ Uses githunter.find_pr_discussion for local git lookup,
+ optionally falls back to github.get_pr for full metadata.
+
+ Args:
+ commit_hash: The git commit SHA to look up.
+ owner: Repository owner (for GitHub API fallback).
+ repo: Repository name (for GitHub API fallback).
+
+ Returns:
+ PRMetadata if a PR is found, None otherwise.
+ """
+ if not self.repo_path:
+ logger.debug("pr_enrichment_no_repo_path", commit_hash=commit_hash)
+ return None
+
+ try:
+ # Try local git first (no API calls)
+ pr_discussion = self.githunter.find_pr_discussion(
+ repo_path=self.repo_path,
+ commit_hash=commit_hash,
+ )
+
+ if pr_discussion is None:
+ return None
+
+ # If we have GitHub API access, get full PR details
+ if self.github and owner and repo:
+ try:
+ pr = self.github.get_pr(
+ owner=owner,
+ repo=repo,
+ number=pr_discussion.number,
+ )
+ return PRMetadata(
+ pr_number=pr.number,
+ pr_url=pr.url,
+ pr_title=pr.title,
+ pr_author=pr.author.login if pr.author else "",
+ pr_merged_at=pr.merged_at,
+ )
+ except Exception as e:
+ logger.warning(
+ "pr_enrichment_github_api_failed",
+ commit_hash=commit_hash,
+ pr_number=pr_discussion.number,
+ error=str(e),
+ )
+ # Fall through to use local data
+
+ # Fallback: use what we got from git
+ return PRMetadata(
+ pr_number=pr_discussion.number,
+ pr_url=pr_discussion.url or "",
+ pr_title=pr_discussion.title,
+ pr_author="", # Not available from git alone
+ pr_merged_at=None,
+ )
+
+ except Exception as e:
+ logger.warning(
+ "pr_enrichment_failed",
+ commit_hash=commit_hash,
+ error=str(e),
+ )
+ return None
+
+ async def enrich_commits(
+ self,
+ commit_hashes: list[str],
+ owner: str | None = None,
+ repo: str | None = None,
+ ) -> dict[str, PRMetadata]:
+ """Batch enrich multiple commits with PR metadata.
+
+ Args:
+ commit_hashes: List of commit SHAs to enrich.
+ owner: Repository owner (for GitHub API fallback).
+ repo: Repository name (for GitHub API fallback).
+
+ Returns:
+ Dictionary mapping commit_hash to PRMetadata for commits that have PRs.
+ """
+ results: dict[str, PRMetadata] = {}
+ for commit_hash in commit_hashes:
+ pr = await self.get_pr_for_commit(commit_hash, owner, repo)
+ if pr:
+ results[commit_hash] = pr
+ return results
+
+ async def close(self) -> None:
+ """Close any open connections."""
+ if self._githunter:
+ await self._githunter.close()
+ if self._github:
+ await self._github.close()
diff --git a/python-packages/dataing/src/dataing/adapters/git/url_templates.py b/python-packages/dataing/src/dataing/adapters/git/url_templates.py
new file mode 100644
index 00000000..b33433f3
--- /dev/null
+++ b/python-packages/dataing/src/dataing/adapters/git/url_templates.py
@@ -0,0 +1,128 @@
+"""URL templates for git provider deep links.
+
+This module provides URL template rendering for commit and PR links across
+GitHub, GitLab, and Bitbucket. Each provider has different URL patterns,
+and self-hosted instances use custom base URLs.
+"""
+
+from enum import Enum
+
+
+class GitProvider(str, Enum):
+ """Supported git providers."""
+
+ GITHUB = "github"
+ GITLAB = "gitlab"
+ BITBUCKET = "bitbucket"
+
+
+# Default base URLs (overridden for self-hosted instances)
+DEFAULT_BASE_URLS: dict[GitProvider, str] = {
+ GitProvider.GITHUB: "https://github.com",
+ GitProvider.GITLAB: "https://gitlab.com",
+ GitProvider.BITBUCKET: "https://bitbucket.org",
+}
+
+# URL patterns per provider
+COMMIT_URL_TEMPLATES: dict[GitProvider, str] = {
+ GitProvider.GITHUB: "{base_url}/{owner}/{repo}/commit/{commit_hash}",
+ GitProvider.GITLAB: "{base_url}/{owner}/{repo}/-/commit/{commit_hash}",
+ GitProvider.BITBUCKET: "{base_url}/{owner}/{repo}/commits/{commit_hash}",
+}
+
+PR_URL_TEMPLATES: dict[GitProvider, str] = {
+ GitProvider.GITHUB: "{base_url}/{owner}/{repo}/pull/{pr_number}",
+ GitProvider.GITLAB: "{base_url}/{owner}/{repo}/-/merge_requests/{pr_number}",
+ GitProvider.BITBUCKET: "{base_url}/{owner}/{repo}/pull-requests/{pr_number}",
+}
+
+
+def build_commit_url(
+ provider: GitProvider | str,
+ owner: str,
+ repo: str,
+ commit_hash: str,
+ base_url: str | None = None,
+) -> str:
+ """Build a commit URL for the given provider.
+
+ Args:
+ provider: The git provider (github, gitlab, bitbucket).
+ owner: Repository owner or organization.
+ repo: Repository name.
+ commit_hash: The git commit SHA.
+ base_url: Optional custom base URL for self-hosted instances.
+
+ Returns:
+ The full commit URL.
+ """
+ if isinstance(provider, str):
+ provider = GitProvider(provider)
+ base = (base_url or DEFAULT_BASE_URLS[provider]).rstrip("/")
+ template = COMMIT_URL_TEMPLATES[provider]
+ return template.format(
+ base_url=base,
+ owner=owner,
+ repo=repo,
+ commit_hash=commit_hash,
+ )
+
+
+def build_pr_url(
+ provider: GitProvider | str,
+ owner: str,
+ repo: str,
+ pr_number: int,
+ base_url: str | None = None,
+) -> str:
+ """Build a PR/MR URL for the given provider.
+
+ Args:
+ provider: The git provider (github, gitlab, bitbucket).
+ owner: Repository owner or organization.
+ repo: Repository name.
+ pr_number: The PR/MR number.
+ base_url: Optional custom base URL for self-hosted instances.
+
+ Returns:
+ The full PR/MR URL.
+ """
+ if isinstance(provider, str):
+ provider = GitProvider(provider)
+ base = (base_url or DEFAULT_BASE_URLS[provider]).rstrip("/")
+ template = PR_URL_TEMPLATES[provider]
+ return template.format(
+ base_url=base,
+ owner=owner,
+ repo=repo,
+ pr_number=pr_number,
+ )
+
+
+def build_best_link(
+ provider: GitProvider | str,
+ owner: str,
+ repo: str,
+ commit_hash: str,
+ pr_number: int | None = None,
+ base_url: str | None = None,
+) -> str:
+ """Return PR URL if available, otherwise commit URL.
+
+ This is a convenience function that prefers linking to the PR when
+ available, as PRs typically provide more context than individual commits.
+
+ Args:
+ provider: The git provider (github, gitlab, bitbucket).
+ owner: Repository owner or organization.
+ repo: Repository name.
+ commit_hash: The git commit SHA.
+ pr_number: Optional PR/MR number (if available).
+ base_url: Optional custom base URL for self-hosted instances.
+
+ Returns:
+ The PR URL if pr_number is provided, otherwise the commit URL.
+ """
+ if pr_number is not None:
+ return build_pr_url(provider, owner, repo, pr_number, base_url)
+ return build_commit_url(provider, owner, repo, commit_hash, base_url)
diff --git a/python-packages/dataing/src/dataing/core/domain_types.py b/python-packages/dataing/src/dataing/core/domain_types.py
index b3b845a3..5650f262 100644
--- a/python-packages/dataing/src/dataing/core/domain_types.py
+++ b/python-packages/dataing/src/dataing/core/domain_types.py
@@ -305,6 +305,12 @@ class RelevantCodeChange(BaseModel):
affected_assets: List of asset identifiers affected by this commit.
relevance_score: How relevant this change is (1.0=exact, 0.7=upstream, 0.4=path).
relevance_reason: Human-readable explanation of why this change is relevant.
+ pr_number: PR/MR number in the provider (if available).
+ pr_url: Full URL to the PR/MR in the provider UI.
+ pr_title: Title of the PR/MR.
+ pr_author: Author login/username of the PR/MR.
+ pr_merged_at: When the PR/MR was merged.
+ provider: Git provider (github, gitlab, bitbucket).
"""
model_config = ConfigDict(frozen=True)
@@ -316,3 +322,37 @@ class RelevantCodeChange(BaseModel):
affected_assets: list[str] = []
relevance_score: float
relevance_reason: str
+ # PR metadata (populated by bond-agent tools during sync)
+ pr_number: int | None = None
+ pr_url: str | None = None
+ pr_title: str | None = None
+ pr_author: str | None = None
+ pr_merged_at: datetime | None = None
+ provider: str | None = None
+
+
+class CodeChangeLink(BaseModel):
+ """A link to a code change (commit or PR) for investigation output.
+
+ Used to render clickable links in CLI, notebook, and markdown outputs
+ when an investigation identifies a code change as relevant.
+
+ Attributes:
+ commit_hash: The git commit SHA.
+ message: The commit message (if available).
+ url: Best link (PR URL if available, else commit URL).
+ pr_number: PR/MR number (if available).
+ pr_title: Title of the PR/MR (if available).
+ pr_url: Direct URL to the PR/MR (if available).
+ author: Author login/username.
+ """
+
+ model_config = ConfigDict(frozen=True)
+
+ commit_hash: str
+ message: str | None = None
+ url: str # Best link (PR if available, else commit)
+ pr_number: int | None = None
+ pr_title: str | None = None
+ pr_url: str | None = None
+ author: str | None = None
diff --git a/python-packages/dataing/src/dataing/models/code_change.py b/python-packages/dataing/src/dataing/models/code_change.py
index bc0d593f..c0742b03 100644
--- a/python-packages/dataing/src/dataing/models/code_change.py
+++ b/python-packages/dataing/src/dataing/models/code_change.py
@@ -4,7 +4,7 @@
from typing import TYPE_CHECKING, Any
from uuid import UUID
-from sqlalchemy import ARRAY, ForeignKey, Text
+from sqlalchemy import ARRAY, ForeignKey, Integer, String, Text
from sqlalchemy.dialects.postgresql import JSONB
from sqlalchemy.orm import Mapped, mapped_column, relationship
@@ -29,5 +29,13 @@ class CodeChange(BaseModel):
raw_diff: Mapped[str | None] = mapped_column(Text, nullable=True)
files_changed: Mapped[list[str] | None] = mapped_column(ARRAY(Text), nullable=True)
+ # PR metadata (populated by bond-agent tools during sync)
+ pr_number: Mapped[int | None] = mapped_column(Integer, nullable=True)
+ pr_url: Mapped[str | None] = mapped_column(Text, nullable=True)
+ pr_title: Mapped[str | None] = mapped_column(Text, nullable=True)
+ pr_author: Mapped[str | None] = mapped_column(Text, nullable=True)
+ pr_merged_at: Mapped[datetime | None] = mapped_column(nullable=True)
+ provider: Mapped[str | None] = mapped_column(String(20), nullable=True)
+
# Relationships
repository: Mapped["GitRepository"] = relationship(back_populates="code_changes")
diff --git a/python-packages/dataing/tests/unit/adapters/git/test_pr_enrichment.py b/python-packages/dataing/tests/unit/adapters/git/test_pr_enrichment.py
new file mode 100644
index 00000000..2e341495
--- /dev/null
+++ b/python-packages/dataing/tests/unit/adapters/git/test_pr_enrichment.py
@@ -0,0 +1,216 @@
+"""Unit tests for PR enrichment service."""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from dataing.adapters.git.pr_enrichment import PREnrichmentService, PRMetadata
+
+
+class TestPREnrichmentService:
+ """Tests for PREnrichmentService."""
+
+ @pytest.fixture
+ def repo_path(self, tmp_path: Path) -> Path:
+ """Create a mock repo path."""
+ return tmp_path / "repo"
+
+ @pytest.fixture
+ def service(self, repo_path: Path) -> PREnrichmentService:
+ """Create a service instance with mocked adapters."""
+ return PREnrichmentService(repo_path=repo_path, github_token="test-token")
+
+ async def test_get_pr_for_commit_no_repo_path(self) -> None:
+ """Should return None when no repo_path is configured."""
+ service = PREnrichmentService(repo_path=None)
+ result = await service.get_pr_for_commit("abc123")
+ assert result is None
+
+ async def test_get_pr_for_commit_not_found(self, service: PREnrichmentService) -> None:
+ """Should return None when no PR is found for commit."""
+ mock_githunter = MagicMock()
+ mock_githunter.find_pr_discussion.return_value = None
+
+ with patch.object(service, "_githunter", mock_githunter):
+ result = await service.get_pr_for_commit("abc123")
+
+ assert result is None
+ mock_githunter.find_pr_discussion.assert_called_once()
+
+ async def test_get_pr_for_commit_found_local_only(self, service: PREnrichmentService) -> None:
+ """Should return PR metadata from local git when GitHub API not used."""
+ mock_pr_discussion = MagicMock()
+ mock_pr_discussion.number = 42
+ mock_pr_discussion.url = "https://github.com/owner/repo/pull/42"
+ mock_pr_discussion.title = "Fix null handling"
+
+ mock_githunter = MagicMock()
+ mock_githunter.find_pr_discussion.return_value = mock_pr_discussion
+
+ # No owner/repo provided, so GitHub API won't be called
+ service_no_token = PREnrichmentService(repo_path=service.repo_path, github_token=None)
+
+ with patch.object(service_no_token, "_githunter", mock_githunter):
+ result = await service_no_token.get_pr_for_commit("abc123")
+
+ assert result is not None
+ assert result.pr_number == 42
+ assert result.pr_url == "https://github.com/owner/repo/pull/42"
+ assert result.pr_title == "Fix null handling"
+ assert result.pr_author == "" # Not available from git alone
+ assert result.pr_merged_at is None
+
+ async def test_get_pr_for_commit_with_github_api(self, service: PREnrichmentService) -> None:
+ """Should return full PR metadata when GitHub API is available."""
+ mock_pr_discussion = MagicMock()
+ mock_pr_discussion.number = 42
+
+ mock_githunter = MagicMock()
+ mock_githunter.find_pr_discussion.return_value = mock_pr_discussion
+
+ merged_at = datetime(2024, 1, 15, 10, 30, 0, tzinfo=UTC)
+ mock_pr = MagicMock()
+ mock_pr.number = 42
+ mock_pr.url = "https://github.com/owner/repo/pull/42"
+ mock_pr.title = "Fix null handling in orders"
+ mock_pr.author = MagicMock()
+ mock_pr.author.login = "developer123"
+ mock_pr.merged_at = merged_at
+
+ mock_github = MagicMock()
+ mock_github.get_pr.return_value = mock_pr
+
+ with (
+ patch.object(service, "_githunter", mock_githunter),
+ patch.object(service, "_github", mock_github),
+ ):
+ result = await service.get_pr_for_commit("abc123", owner="owner", repo="repo")
+
+ assert result is not None
+ assert result.pr_number == 42
+ assert result.pr_url == "https://github.com/owner/repo/pull/42"
+ assert result.pr_title == "Fix null handling in orders"
+ assert result.pr_author == "developer123"
+ assert result.pr_merged_at == merged_at
+
+ mock_github.get_pr.assert_called_once_with(owner="owner", repo="repo", number=42)
+
+ async def test_get_pr_for_commit_github_api_fails_gracefully(
+ self, service: PREnrichmentService
+ ) -> None:
+ """Should fall back to local data when GitHub API fails."""
+ mock_pr_discussion = MagicMock()
+ mock_pr_discussion.number = 42
+ mock_pr_discussion.url = "https://github.com/owner/repo/pull/42"
+ mock_pr_discussion.title = "Fix null handling"
+
+ mock_githunter = MagicMock()
+ mock_githunter.find_pr_discussion.return_value = mock_pr_discussion
+
+ mock_github = MagicMock()
+ mock_github.get_pr.side_effect = Exception("API rate limited")
+
+ with (
+ patch.object(service, "_githunter", mock_githunter),
+ patch.object(service, "_github", mock_github),
+ ):
+ result = await service.get_pr_for_commit("abc123", owner="owner", repo="repo")
+
+ # Should fall back to local data
+ assert result is not None
+ assert result.pr_number == 42
+ assert result.pr_title == "Fix null handling"
+ assert result.pr_author == "" # GitHub API failed
+
+ async def test_enrich_commits_batch(self, service: PREnrichmentService) -> None:
+ """Should enrich multiple commits in batch."""
+ mock_pr_discussion_1 = MagicMock()
+ mock_pr_discussion_1.number = 42
+ mock_pr_discussion_1.url = "https://github.com/owner/repo/pull/42"
+ mock_pr_discussion_1.title = "PR 42"
+
+ mock_pr_discussion_2 = MagicMock()
+ mock_pr_discussion_2.number = 43
+ mock_pr_discussion_2.url = "https://github.com/owner/repo/pull/43"
+ mock_pr_discussion_2.title = "PR 43"
+
+ mock_githunter = MagicMock()
+ # First commit has PR, second doesn't, third has PR
+ mock_githunter.find_pr_discussion.side_effect = [
+ mock_pr_discussion_1,
+ None,
+ mock_pr_discussion_2,
+ ]
+
+ service_no_token = PREnrichmentService(repo_path=service.repo_path, github_token=None)
+
+ with patch.object(service_no_token, "_githunter", mock_githunter):
+ results = await service_no_token.enrich_commits(["commit1", "commit2", "commit3"])
+
+ assert len(results) == 2
+ assert "commit1" in results
+ assert "commit2" not in results
+ assert "commit3" in results
+ assert results["commit1"].pr_number == 42
+ assert results["commit3"].pr_number == 43
+
+ async def test_enrich_commits_handles_errors(self, service: PREnrichmentService) -> None:
+ """Should continue processing when individual commits fail."""
+ mock_pr_discussion = MagicMock()
+ mock_pr_discussion.number = 42
+ mock_pr_discussion.url = "https://github.com/owner/repo/pull/42"
+ mock_pr_discussion.title = "PR 42"
+
+ mock_githunter = MagicMock()
+ # First commit raises, second succeeds
+ mock_githunter.find_pr_discussion.side_effect = [
+ Exception("Git error"),
+ mock_pr_discussion,
+ ]
+
+ service_no_token = PREnrichmentService(repo_path=service.repo_path, github_token=None)
+
+ with patch.object(service_no_token, "_githunter", mock_githunter):
+ results = await service_no_token.enrich_commits(["commit1", "commit2"])
+
+ # First commit failed, but second should still succeed
+ assert len(results) == 1
+ assert "commit2" in results
+ assert results["commit2"].pr_number == 42
+
+
+class TestPRMetadata:
+ """Tests for PRMetadata dataclass."""
+
+ def test_pr_metadata_creation(self) -> None:
+ """Should create PRMetadata with all fields."""
+ merged_at = datetime(2024, 1, 15, 10, 30, 0, tzinfo=UTC)
+ metadata = PRMetadata(
+ pr_number=42,
+ pr_url="https://github.com/owner/repo/pull/42",
+ pr_title="Fix null handling",
+ pr_author="developer123",
+ pr_merged_at=merged_at,
+ )
+
+ assert metadata.pr_number == 42
+ assert metadata.pr_url == "https://github.com/owner/repo/pull/42"
+ assert metadata.pr_title == "Fix null handling"
+ assert metadata.pr_author == "developer123"
+ assert metadata.pr_merged_at == merged_at
+
+ def test_pr_metadata_optional_merged_at(self) -> None:
+ """Should allow None for pr_merged_at."""
+ metadata = PRMetadata(
+ pr_number=42,
+ pr_url="https://github.com/owner/repo/pull/42",
+ pr_title="Fix null handling",
+ pr_author="developer123",
+ pr_merged_at=None,
+ )
+
+ assert metadata.pr_merged_at is None
diff --git a/python-packages/dataing/tests/unit/adapters/git/test_url_templates.py b/python-packages/dataing/tests/unit/adapters/git/test_url_templates.py
new file mode 100644
index 00000000..e99b25db
--- /dev/null
+++ b/python-packages/dataing/tests/unit/adapters/git/test_url_templates.py
@@ -0,0 +1,218 @@
+"""Unit tests for URL template rendering."""
+
+from __future__ import annotations
+
+import pytest
+
+from dataing.adapters.git.url_templates import (
+ GitProvider,
+ build_best_link,
+ build_commit_url,
+ build_pr_url,
+)
+
+
+class TestBuildCommitUrl:
+ """Tests for build_commit_url function."""
+
+ def test_github_commit_url(self) -> None:
+ """Should generate correct GitHub commit URL."""
+ url = build_commit_url(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ )
+ assert url == "https://github.com/acme/data-pipeline/commit/abc123def456"
+
+ def test_gitlab_commit_url(self) -> None:
+ """Should generate correct GitLab commit URL."""
+ url = build_commit_url(
+ provider=GitProvider.GITLAB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ )
+ assert url == "https://gitlab.com/acme/data-pipeline/-/commit/abc123def456"
+
+ def test_bitbucket_commit_url(self) -> None:
+ """Should generate correct Bitbucket commit URL."""
+ url = build_commit_url(
+ provider=GitProvider.BITBUCKET,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ )
+ assert url == "https://bitbucket.org/acme/data-pipeline/commits/abc123def456"
+
+ def test_github_commit_url_string_provider(self) -> None:
+ """Should accept string provider name."""
+ url = build_commit_url(
+ provider="github",
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ )
+ assert url == "https://github.com/acme/data-pipeline/commit/abc123def456"
+
+ def test_custom_base_url(self) -> None:
+ """Should use custom base URL for self-hosted instances."""
+ url = build_commit_url(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ base_url="https://github.internal.company.com",
+ )
+ assert url == "https://github.internal.company.com/acme/data-pipeline/commit/abc123def456"
+
+ def test_custom_base_url_trailing_slash(self) -> None:
+ """Should handle trailing slash in custom base URL."""
+ url = build_commit_url(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ base_url="https://github.internal.company.com/",
+ )
+ assert url == "https://github.internal.company.com/acme/data-pipeline/commit/abc123def456"
+
+ def test_invalid_provider_raises(self) -> None:
+ """Should raise ValueError for invalid provider string."""
+ with pytest.raises(ValueError, match="'invalid' is not a valid GitProvider"):
+ build_commit_url(
+ provider="invalid",
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ )
+
+
+class TestBuildPrUrl:
+ """Tests for build_pr_url function."""
+
+ def test_github_pr_url(self) -> None:
+ """Should generate correct GitHub PR URL."""
+ url = build_pr_url(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ pr_number=42,
+ )
+ assert url == "https://github.com/acme/data-pipeline/pull/42"
+
+ def test_gitlab_mr_url(self) -> None:
+ """Should generate correct GitLab merge request URL."""
+ url = build_pr_url(
+ provider=GitProvider.GITLAB,
+ owner="acme",
+ repo="data-pipeline",
+ pr_number=42,
+ )
+ assert url == "https://gitlab.com/acme/data-pipeline/-/merge_requests/42"
+
+ def test_bitbucket_pr_url(self) -> None:
+ """Should generate correct Bitbucket pull request URL."""
+ url = build_pr_url(
+ provider=GitProvider.BITBUCKET,
+ owner="acme",
+ repo="data-pipeline",
+ pr_number=42,
+ )
+ assert url == "https://bitbucket.org/acme/data-pipeline/pull-requests/42"
+
+ def test_gitlab_mr_url_string_provider(self) -> None:
+ """Should accept string provider name."""
+ url = build_pr_url(
+ provider="gitlab",
+ owner="acme",
+ repo="data-pipeline",
+ pr_number=42,
+ )
+ assert url == "https://gitlab.com/acme/data-pipeline/-/merge_requests/42"
+
+ def test_custom_base_url(self) -> None:
+ """Should use custom base URL for self-hosted GitLab."""
+ url = build_pr_url(
+ provider=GitProvider.GITLAB,
+ owner="acme",
+ repo="data-pipeline",
+ pr_number=42,
+ base_url="https://gitlab.internal.company.com",
+ )
+ assert url == "https://gitlab.internal.company.com/acme/data-pipeline/-/merge_requests/42"
+
+
+class TestBuildBestLink:
+ """Tests for build_best_link function."""
+
+ def test_returns_pr_url_when_pr_number_provided(self) -> None:
+ """Should return PR URL when pr_number is provided."""
+ url = build_best_link(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ pr_number=42,
+ )
+ assert url == "https://github.com/acme/data-pipeline/pull/42"
+
+ def test_returns_commit_url_when_no_pr_number(self) -> None:
+ """Should return commit URL when pr_number is None."""
+ url = build_best_link(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ pr_number=None,
+ )
+ assert url == "https://github.com/acme/data-pipeline/commit/abc123def456"
+
+ def test_returns_commit_url_by_default(self) -> None:
+ """Should return commit URL when pr_number not provided."""
+ url = build_best_link(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ )
+ assert url == "https://github.com/acme/data-pipeline/commit/abc123def456"
+
+ def test_custom_base_url_with_pr(self) -> None:
+ """Should use custom base URL for PR link."""
+ url = build_best_link(
+ provider=GitProvider.GITHUB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ pr_number=42,
+ base_url="https://github.enterprise.com",
+ )
+ assert url == "https://github.enterprise.com/acme/data-pipeline/pull/42"
+
+ def test_custom_base_url_with_commit(self) -> None:
+ """Should use custom base URL for commit link."""
+ url = build_best_link(
+ provider=GitProvider.GITLAB,
+ owner="acme",
+ repo="data-pipeline",
+ commit_hash="abc123def456",
+ base_url="https://gitlab.enterprise.com",
+ )
+ assert url == "https://gitlab.enterprise.com/acme/data-pipeline/-/commit/abc123def456"
+
+
+class TestGitProvider:
+ """Tests for GitProvider enum."""
+
+ def test_provider_values(self) -> None:
+ """Should have correct string values."""
+ assert GitProvider.GITHUB.value == "github"
+ assert GitProvider.GITLAB.value == "gitlab"
+ assert GitProvider.BITBUCKET.value == "bitbucket"
+
+ def test_provider_from_string(self) -> None:
+ """Should create provider from string."""
+ assert GitProvider("github") == GitProvider.GITHUB
+ assert GitProvider("gitlab") == GitProvider.GITLAB
+ assert GitProvider("bitbucket") == GitProvider.BITBUCKET