Merge pull request #63 from github/copilot/fix-5a44b335-4f8a-4c35-b8fb-9f8941ab7116

Author: zkoppert

.env-example

@@ -6,3 +6,4 @@ GH_ENTERPRISE_URL = ""
 GH_TOKEN = ""
 OUTPUT_FILE = ""
 REPORT_TITLE = ""
+OWNING_TEAM = ""

README.md

@@ -214,6 +214,17 @@ The tool determines team boundaries using this algorithm:
    - Are managers of anyone in the team
 4. **Classify Contributors**: Any contributor not in the team list is considered an InnerSource contributor
 
+**Overriding Team Determination**: You can override this algorithm by setting the `OWNING_TEAM` environment variable with a comma-separated list of GitHub usernames. When set, these users will be considered the owning team, and the algorithm above will be bypassed. This is useful when:
+- The first commit author doesn't accurately represent the current owning team
+- The org chart doesn't align with actual repository ownership
+- You want to explicitly define team boundaries
+
+Example:
+```yaml
+env:
+  OWNING_TEAM: "alice,bob,charlie"
+```
+
 #### Example Team Boundary Calculation
 
 <details>
@@ -517,6 +528,40 @@ jobs:
           CHUNK_SIZE: "100"
 ```
 
+#### Using Custom Team Ownership
+
+To override the automatic team determination and explicitly specify the owning team:
+
+```yaml
+name: InnerSource with Custom Team
+
+on:
+  schedule:
+    - cron: "0 0 * * 0"
+  workflow_dispatch:
+
+jobs:
+  measure-innersource:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Measure InnerSource
+        uses: github/measure-innersource@v1
+        env:
+          REPOSITORY: "org/repository"
+          GH_TOKEN: ${{ secrets.GH_TOKEN }}
+          OWNING_TEAM: "alice,bob,charlie,david"
+          REPORT_TITLE: "InnerSource Report with Custom Team"
+```
+
+This is useful when:
+- The first commit author doesn't represent the current team
+- The org chart doesn't align with actual ownership
+- You want to explicitly define team boundaries
+
+
 ### Configuration
 
 Below are the allowed configuration options:
@@ -550,6 +595,7 @@ This action can be configured to authenticate with GitHub App Installation or Pe
 | `REPORT_TITLE`      | False    | `"InnerSource Report"`  | Title to have on the report issue.                                                                                                    |
 | `REPOSITORY`        | True     | `""`                    | The name of the repository you are trying to measure. Format `owner/repo` ie. `github/measure-innersource`                            |
 | `CHUNK_SIZE`        | False    | `100`                   | Number of items to process at once when fetching data. Increasing can improve performance but uses more memory. Minimum value is 10.  |
+| `OWNING_TEAM`       | False    | `""`                    | Comma-separated list of GitHub usernames that own the repository. Overrides the built-in team determination algorithm. Example: `alice,bob,charlie` |
 
 ## Understanding the Results
 
@@ -709,6 +755,7 @@ The tool automatically splits large files, but you can:
 - [ ] `OUTPUT_FILE` (default: "innersource_report.md")
 - [ ] `CHUNK_SIZE` (default: 100, minimum: 10)
 - [ ] `RATE_LIMIT_BYPASS` (default: false)
+- [ ] `OWNING_TEAM` (comma-separated usernames to override team determination)
 
 #### File Requirements Checklist
 

config.py

@@ -36,6 +36,7 @@ class EnvVars:
         output_file (str): The name of the file to write the report to
         rate_limit_bypass (bool): If set to TRUE, bypass the rate limit for the GitHub API
         chunk_size (int): The number of items to process at once when fetching data (for memory efficiency)
+        owning_team (list[str] | None): Optional list of usernames that comprise the owning team (overrides algorithm)
     """
 
     def __init__(
@@ -52,6 +53,7 @@ def __init__(
         output_file: str,
         rate_limit_bypass: bool = False,
         chunk_size: int = 100,
+        owning_team: list[str] | None = None,
     ):
         self.gh_app_id = gh_app_id
         self.gh_app_installation_id = gh_app_installation_id
@@ -65,6 +67,7 @@ def __init__(
         self.output_file = output_file
         self.rate_limit_bypass = rate_limit_bypass
         self.chunk_size = chunk_size
+        self.owning_team = owning_team
 
     def __repr__(self):
         return (
@@ -80,7 +83,8 @@ def __repr__(self):
             f"{self.repo},"
             f"{self.output_file},"
             f"{self.rate_limit_bypass},"
-            f"{self.chunk_size}"
+            f"{self.chunk_size},"
+            f"{self.owning_team}"
             ")"
         )
 
@@ -183,6 +187,8 @@ def get_env_vars(test: bool = False) -> EnvVars:
         - OUTPUT_FILE: Output filename (default: "innersource_report.md")
         - RATE_LIMIT_BYPASS: Set to "true" to bypass rate limiting
         - CHUNK_SIZE: Number of items to process at once (default: 100, minimum: 10)
+        - OWNING_TEAM: Comma-separated list of GitHub usernames that own the repository
+                      (overrides the built-in team determination algorithm)
 
     Examples:
         >>> os.environ['GH_TOKEN'] = 'ghp_...'
@@ -243,6 +249,20 @@ def get_env_vars(test: bool = False) -> EnvVars:
         # Default to DEFAULT_CHUNK_SIZE if not a valid integer
         chunk_size = DEFAULT_CHUNK_SIZE
 
+    # Get optional owning team override (comma-separated list of usernames)
+    owning_team_str = os.getenv("OWNING_TEAM", "").strip()
+    owning_team = None
+    if owning_team_str:
+        # Parse comma-separated list and strip whitespace from each username
+        owning_team = [
+            username.strip()
+            for username in owning_team_str.split(",")
+            if username.strip()
+        ]
+        # If the list is empty after stripping, set to None
+        if not owning_team:
+            owning_team = None
+
     return EnvVars(
         gh_app_id,
         gh_app_installation_id,
@@ -256,4 +276,5 @@ def get_env_vars(test: bool = False) -> EnvVars:
         output_file,
         rate_limit_bypass,
         chunk_size,
+        owning_team,
     )

markdown_writer.py

@@ -40,6 +40,7 @@ def write_to_markdown(
     innersource_contributors=None,
     innersource_contribution_counts=None,
     team_member_contribution_counts=None,
+    team_ownership_explicitly_specified=False,
 ) -> None:
     """
     Generate a comprehensive InnerSource collaboration report in markdown format.
@@ -84,6 +85,10 @@ def write_to_markdown(
                                                                            mapping team member
                                                                            usernames to their
                                                                            contribution counts.
+        team_ownership_explicitly_specified (bool, optional): Flag indicating whether team
+                                                            ownership is explicitly specified
+                                                            rather than derived from commit history.
+                                                            Defaults to False.
 
     Returns:
         None: This function creates a markdown file as a side effect.
@@ -126,7 +131,22 @@ def write_to_markdown(
         ...     all_contributors=["alice", "bob", "charlie", "dave", "eve"],
         ...     innersource_contributors=["dave", "eve"],
         ...     innersource_contribution_counts={"dave": 15, "eve": 8},
-        ...     team_member_contribution_counts={"alice": 25, "bob": 12, "charlie": 5}
+        ...     team_member_contribution_counts={"alice": 25, "bob": 12, "charlie": 5},
+        ...     team_ownership_explicitly_specified=False
+        ... )
+
+        >>> # Generate a report with explicitly specified team ownership
+        >>> write_to_markdown(
+        ...     report_title="Team Ownership Report",
+        ...     output_file="team_report.md",
+        ...     innersource_ratio=0.38,
+        ...     repo_data=repo_object,
+        ...     team_members_that_own_the_repo=["david", "emily", "frank"],
+        ...     all_contributors=["david", "emily", "frank", "greg", "hannah"],
+        ...     innersource_contributors=["greg", "hannah"],
+        ...     innersource_contribution_counts={"greg": 12, "hannah": 6},
+        ...     team_member_contribution_counts={"david": 18, "emily": 9, "frank": 7},
+        ...     team_ownership_explicitly_specified=True
         ... )
     """
     output_file_name = output_file if output_file else "innersource_report.md"
@@ -138,9 +158,20 @@ def write_to_markdown(
             return
         report_file.write(f"## Repository: {repo_data.full_name}\n\n")
         report_file.write(f"### InnerSource Ratio: {innersource_ratio:.2%}\n\n")
-        report_file.write(
-            f"### Original Commit Author: {original_commit_author} (Manager: {original_commit_author_manager})\n\n"
-        )
+        if team_ownership_explicitly_specified:
+            report_file.write("### Team ownership is explicitly specified\n\n")
+        elif original_commit_author and original_commit_author_manager:
+            report_file.write(
+                f"### Original Commit Author: {original_commit_author} (Manager: {original_commit_author_manager})\n\n"
+            )
+        elif original_commit_author:
+            report_file.write(
+                f"### Original Commit Author: {original_commit_author}\n\n"
+            )
+        else:
+            report_file.write(
+                "### Original commit author information not available\n\n"
+            )
         report_file.write("## Team Members that Own the Repo:\n")
         if team_members_that_own_the_repo:
             for member in team_members_that_own_the_repo: