def explode(self, commits, deps_from, deps_on):
"""
Walk the dependency tree breadth-first starting with the
leaves at the bottom.
For each commit, figure out whether it should be exploded
:param commits: dict mapping SHA1 hashes to pygit2.Commit objects
:param deps_from: dict mapping dependents to dependencies
:param deps_on: dict mapping in opposite direction
"""
todo = self.get_leaves(commits, deps_from)
# Each time we explode a commit, we'll remove it from any
# dict which is a value of this dict.
unexploded_deps_from = copy.deepcopy(deps_from)
self.logger.debug("Initial queue of leaves:")
for commit in todo:
self.logger.debug(' ' + GitUtils.commit_summary(commit))
self.current_branch = None
while todo:
commit = todo.pop(0)
sha = commit.hex
self.logger.debug("Exploding %s" % GitUtils.commit_summary(commit))
if unexploded_deps_from[sha]:
abort("BUG: unexploded deps from %s" %
GitUtils.commit_summary(commit))
deps = deps_from[sha]
self.prepare_cherrypick_base(sha, deps, commits)
self.cherry_pick(sha)
self.queue_new_leaves(todo, commit, commits, deps_on,
unexploded_deps_from)
def __init__(self, repo, base, head, debug, context_lines):
self.logger = standard_logger('git-explode', debug)
self.debug = debug
self.repo = repo
self.base = base
self.base_commit = GitUtils.ref_commit(repo, base)
self.logger.debug("base commit %s is %s" %
(base, GitUtils.commit_summary(self.base_commit)))
self.head = head
self.context_lines = context_lines
self.topic_mgr = TopicManager('topic%d', self.logger)
# Map commits to their exploded version
self.exploded = {}
def queue_new_leaves(self, todo, exploded_commit, commits, deps_on,
unexploded_deps_from):
"""When a commit is exploded, there may be other commits in the
dependency tree which only had a single dependency on this
commit. In that case they have effectively become leaves on
the dependency tree of unexploded commits, so they should be
added to the explode queue.
"""
sha1 = exploded_commit.hex
for dependent in deps_on[sha1]:
del unexploded_deps_from[dependent][sha1]
if not unexploded_deps_from[dependent]:
new = commits[dependent]
self.logger.debug("+ pushed to queue: %s" %
GitUtils.commit_summary(new))
todo.insert(0, new)
def register_new_dependent(self, dependent, dependent_sha1):
if dependent_sha1 not in self.dependencies:
self.logger.info(" New dependent: %s" %
GitUtils.commit_summary(dependent))
self.dependencies[dependent_sha1] = {}
self.notify_listeners("new_dependent", dependent)
def blame_hunk(self, dependent, parent, path, hunk):
"""Run git blame on the parts of the hunk which exist in the older
commit in the diff. The commits generated by git blame are
the commits which the newer commit in the diff depends on,
because without the lines from those commits, the hunk would
not apply correctly.
"""
line_range_before = "-%d,%d" % (hunk.old_start, hunk.old_lines)
line_range_after = "+%d,%d" % (hunk.new_start, hunk.new_lines)
self.logger.debug(" Blaming hunk %s @ %s" %
(line_range_before, parent.hex[:8]))
if not self.tree_lookup(path, parent):
# This is probably because dependent added a new directory
# which was not previously in the parent.
return
cmd = [
'git', 'blame', '--porcelain', '-L',
"%d,+%d" % (hunk.old_start, hunk.old_lines), parent.hex, '--', path
]
blame = subprocess.check_output(cmd, universal_newlines=True)
dependent_sha1 = dependent.hex
if dependent_sha1 not in self.dependencies:
self.logger.debug(" New dependent: %s" %
GitUtils.commit_summary(dependent))
self.dependencies[dependent_sha1] = {}
self.notify_listeners("new_dependent", dependent)
line_to_culprit = {}
for line in blame.split('\n'):
self.logger.debug(" !" + line.rstrip())
m = re.match('^([0-9a-f]{40}) (\d+) (\d+)( \d+)?$', line)
if not m:
continue
dependency_sha1, orig_line_num, line_num = m.group(1, 2, 3)
line_num = int(line_num)
dependency = self.get_commit(dependency_sha1)
line_to_culprit[line_num] = dependency.hex
if self.is_excluded(dependency):
self.logger.debug(
" Excluding dependency %s from line %s (%s)" %
(dependency_sha1[:8], line_num,
GitUtils.oneline(dependency)))
continue
if dependency_sha1 not in self.dependencies[dependent_sha1]:
if not self.seen_commit(dependency):
self.notify_listeners("new_commit", dependency)
self.dependencies[dependent_sha1][dependency_sha1] = {}
self.notify_listeners("new_dependency", dependent, dependency,
path, line_num)
self.logger.debug(
" New dependency %s -> %s via line %s (%s)" %
(dependent_sha1[:8], dependency_sha1[:8], line_num,
GitUtils.oneline(dependency)))
if dependency_sha1 in self.todo_d:
self.logger.debug(
" Dependency on %s via line %s already in TODO"
% (
dependency_sha1[:8],
line_num,
))
continue
if dependency_sha1 in self.done_d:
self.logger.debug(
" Dependency on %s via line %s already done" % (
dependency_sha1[:8],
line_num,
))
continue
if dependency_sha1 not in self.dependencies:
if self.options.recurse:
self.todo.append(dependency)
self.todo_d[dependency.hex] = True
self.logger.debug(" + Added %s to TODO" %
dependency.hex[:8])
dep_sources = self.dependencies[dependent_sha1][dependency_sha1]
if path not in dep_sources:
dep_sources[path] = {}
self.notify_listeners('new_path', dependent, dependency, path,
line_num)
if line_num in dep_sources[path]:
abort("line %d already found when blaming %s:%s\n"
"old:\n %s\n"
"new:\n %s" % (line_num, parent.hex[:8], path,
dep_sources[path][line_num], line))
dep_sources[path][line_num] = line
self.logger.debug(" New line for %s -> %s: %s" %
(dependent_sha1[:8], dependency_sha1[:8], line))
self.notify_listeners('new_line', dependent, dependency, path,
line_num)
diff_format = ' |%8.8s %5s %s%s'
hunk_header = '@@ %s %s @@' % (line_range_before, line_range_after)
self.logger.debug(diff_format % ('--------', '-----', '', hunk_header))
line_num = hunk.old_start
for line in hunk.lines:
if "\n\\ No newline at end of file" == line.content.rstrip():
break
if line.origin == '+':
rev = ln = ''
else:
rev = line_to_culprit[line_num]
ln = line_num
line_num += 1
self.logger.debug(diff_format %
(rev, ln, line.origin, line.content.rstrip()))