fix(algorithm): correct bfs to not abort on previously visited node

semantic_release/version/algorithm.py

@@ -71,48 +71,66 @@ def _bfs_for_latest_version_in_history(
     `merge_base`'s parents' history. If no commits in the history correspond
     to a released version, return None
     """
-
+    tag_sha_2_version_lookup = {
+        tag.commit.hexsha: version for tag, version in full_release_tags_and_versions
+    }
     # Step 3. Latest full release version within the history of the current branch
     # Breadth-first search the merge-base and its parent commits for one which matches
     # the tag of the latest full release tag in history
-    def bfs(visited: set[Commit], q: Queue[Commit]) -> Version | None:
-        if q.empty():
-            log.debug("queue is empty, returning none")
-            return None
+    def bfs(start_commit: Commit | TagObject | Blob | Tree) -> Version | None:
+        # Derived from Geeks for Geeks
+        # https://www.geeksforgeeks.org/python-program-for-breadth-first-search-or-bfs-for-a-graph/?ref=lbp
 
-        node = q.get()
-        if node in visited:
-            log.debug("commit %s already visited, returning none", node.hexsha)
-            return None
+        # Create a queue for BFS
+        q: Queue[Commit | TagObject | Blob | Tree] = Queue()
 
-        for tag, version in full_release_tags_and_versions:
-            log.debug(
-                "checking if tag %r (%s) matches commit %s",
-                tag.name,
-                tag.commit.hexsha,
-                node.hexsha,
-            )
-            if tag.commit == node:
+        # Create a set to store visited graph nodes (commit objects in this case)
+        visited: set[Commit | TagObject | Blob | Tree] = set()
+
+        # Add the source node in the queue & mark as visited to start the search
+        q.put(start_commit)
+        visited.add(start_commit)
+
+        # Initialize the result to None
+        result = None
+
+        # Traverse the git history until it finds a version tag if one exists
+        while not q.empty():
+            node = q.get()
+            visited.add(node)
+
+            log.debug("checking if commit %s matches any tags", node.hexsha)
+            version = tag_sha_2_version_lookup.get(node.hexsha, None)
+
+            if version is not None:
                 log.info(
                     "found latest version in branch history: %r (%s)",
                     str(version),
                     node.hexsha[:7],
                 )
-                return version
+                result = version
+                break
+
             log.debug("commit %s doesn't match any tags", node.hexsha)
 
-        visited.add(node)
-        for parent in node.parents:
-            log.debug("queuing parent commit %s", parent.hexsha)
-            q.put(parent)
+            # Add all parent commits to the queue if they haven't been visited
+            for parent in node.parents:
+                if parent in visited:
+                    log.debug("parent commit %s already visited", node.hexsha)
+                    continue
+
+                log.debug("queuing parent commit %s", parent.hexsha)
+                q.put(parent)
 
-        return bfs(visited, q)
+        return result
 
-    q: Queue[Commit] = Queue()
-    q.put(merge_base)
-    latest_version = bfs(set(), q)
+    # Run a Breadth First Search to find the latest version in the history
+    latest_version = bfs(merge_base)
+    if latest_version is not None:
+        log.info("the latest version in this branch's history is %s", latest_version)
+    else:
+        log.info("no version tags found in this branch's history")
 
-    log.info("the latest version in this branch's history is %s", latest_version)
     return latest_version
 
 

tests/unit/semantic_release/version/test_algorithm.py

@@ -1,12 +1,62 @@
 import pytest
-from git import Repo
+from git import Commit, Repo, TagReference
 
 from semantic_release.enums import LevelBump
-from semantic_release.version.algorithm import _increment_version, tags_and_versions
+from semantic_release.version.algorithm import (
+    _bfs_for_latest_version_in_history,
+    _increment_version,
+    tags_and_versions,
+)
 from semantic_release.version.translator import VersionTranslator
 from semantic_release.version.version import Version
 
 
+def test_bfs_for_latest_version_in_history():
+    # Setup fake git graph
+    """
+    * merge commit 6 (start)
+    |\
+    | * commit 5
+    | * commit 4
+    |/
+    * commit 3
+    * commit 2
+    * commit 1
+    * v1.0.0
+    """
+    repo = Repo()
+    expected_version = Version.parse("1.0.0")
+    v1_commit = Commit(repo, binsha=b"0" * 20)
+    class TagReferenceOverride(TagReference):
+        commit = v1_commit # type: ignore - mocking the commit property
+
+    v1_tag = TagReferenceOverride(repo, "refs/tags/v1.0.0", check_path=False)
+
+    trunk = Commit(repo, binsha=b"3" * 20, parents=[
+        Commit(repo, binsha=b"2" * 20, parents=[
+            Commit(repo, binsha=b"1" * 20, parents=[v1_commit]),
+        ]),
+    ])
+    start_commit = Commit(
+        repo,
+        binsha=b"6" * 20,
+        parents=[
+            trunk,
+            Commit(repo, binsha=b"5" * 20, parents=[
+                Commit(repo, binsha=b"4" * 20, parents=[trunk]),
+            ]),
+        ]
+    )
+
+    # Execute
+    actual = _bfs_for_latest_version_in_history(start_commit, [
+        (v1_tag, expected_version),
+    ])
+
+    # Verify
+    assert expected_version == actual
+
+
 @pytest.mark.parametrize(
     "tags, sorted_tags",
     [