Handle Pegleg-generated commits in deployment data

It is possible for Pegleg to generate a commit on top of a repo if
the repo is dirty (aka, has uncommited/untracked files). This
effectively makes the repo appear "clean", and also changes the head
of the repo. This can potentially interfere with the deployment_data
generation that analyzes the cleanliness of the repo as well as the
commit at the head of the repo.

This patch set updates the deployment_data generation logic, to be
able to detect Pegleg-generated commits at the head of a repo, and
instead go off of the Pegleg-generated commit's parent commit when
generating the data. It also ensures the repo in the data is always
marked dirty if a Pegleg-generated commit is seen, because the
Pegleg-generated commit would not exist unless the repo was dirty.

Change-Id: I863b3f2f661f11c36ba939ee3023f78733021b96

pegleg/engine/site.py

@@ -26,6 +26,7 @@ from pegleg import config
 from pegleg.engine import util
 from pegleg.engine.util import files
 from pegleg.engine.util.files import add_representer_ordered_dict
+from pegleg.engine.util.git import TEMP_PEGLEG_COMMIT_MSG
 
 __all__ = ('collect', 'list_', 'show', 'render')
 
@@ -219,6 +220,20 @@ def _get_repo_deployment_data_stanza(repo_path):
     try:
         repo = git.Repo(repo_path)
         commit = repo.commit()
+        contains_pegleg_commit = TEMP_PEGLEG_COMMIT_MSG in commit.message
+
+        # The repo may not appear dirty if Pegleg has made a temporary commit
+        # on top of changed/untracked files, but we know if that temporary
+        # commit happened the repo is indeed dirty
+        dirty = (repo.is_dirty() or contains_pegleg_commit)
+
+        if contains_pegleg_commit:
+            # The commit grabbed above isn't really what we want this data to
+            # reflect, because it was a commit made by Pegleg itself.
+            # Grab the commit before Pegleg made its temporary commit(s)
+            while (TEMP_PEGLEG_COMMIT_MSG in commit.message
+                   and len(commit.parents) > 0):
+                commit = commit.parents[0]
 
         # If we're at a particular tag, reference it
         tag = [tag.name for tag in repo.tags if tag.commit == commit]
@@ -233,6 +248,6 @@ def _get_repo_deployment_data_stanza(repo_path):
                     tag = "Detached HEAD"
                 else:
                     raise e
-        return {"commit": commit.hexsha, "tag": tag, "dirty": repo.is_dirty()}
+        return {"commit": commit.hexsha, "tag": tag, "dirty": dirty}
     except git.InvalidGitRepositoryError:
         return {"commit": "None", "tag": "None", "dirty": "None"}

pegleg/engine/util/git.py

@@ -30,6 +30,8 @@ __all__ = (
     'git_handler', 'is_repository', 'is_equal', 'repo_url', 'repo_name',
     'normalize_repo_path')
 
+TEMP_PEGLEG_COMMIT_MSG = 'Temporary Pegleg commit'
+
 
 def git_handler(
         repo_url, ref=None, proxy_server=None, auth_key=None, clone_path=None):
@@ -107,7 +109,7 @@ def git_handler(
                 'tracked/untracked changes to ref=%s', repo_name(repo_url),
                 ref)
             repo.git.add(all=True)
-            repo.index.commit('Temporary Pegleg commit')
+            repo.index.commit(TEMP_PEGLEG_COMMIT_MSG)
 
         try:
             # Check whether the ref exists locally.