def analyze_patch(self):
"""
Analyze loaded patch to extract modified lines
and statistics
"""
assert self.patch is not None, "Missing patch"
assert isinstance(self.patch, str), "Invalid patch type"
# List all modified lines from current revision changes
patch_stats = rs_parsepatch.get_lines(self.patch)
assert len(patch_stats) > 0, "Empty patch"
self.lines = {stat["filename"]: stat["added_lines"] for stat in patch_stats}
# Shortcut to files modified
self.files = self.lines.keys()
# Report nb of files and lines analyzed
stats.add_metric("analysis.files", len(self.files))
stats.add_metric(
"analysis.lines", sum(len(line) for line in self.lines.values())
)
def test_phabricator(mock_config, mock_revision):
"""
Test a phabricator revision
"""
assert not hasattr(mock_revision, "mercurial")
assert mock_revision.diff_id == 42
assert mock_revision.diff_phid == "PHID-DIFF-test"
assert mock_revision.url == "https://phabricator.test/D51"
assert repr(mock_revision) == "PHID-DIFF-test"
assert mock_revision.id == 51 # revision
# Patch is automatically loaded from Phabricator
assert mock_revision.patch is not None
assert isinstance(mock_revision.patch, str)
assert len(mock_revision.patch.split("\n")) == 15
patch = rs_parsepatch.get_diffs(mock_revision.patch)
assert patch == [
{
"binary": False,
"copied_from": None,
"deleted": False,
"filename": "test.txt",
"lines": [(1, 1, b"Hello World"), (None, 2, b"Second line")],
"modes": {},
"new": False,
"renamed_from": None,
},
{
"binary": False,
"copied_from": None,
"deleted": False,
"filename": "test.cpp",
"lines": [(None, 1, b"Hello World")],
"modes": {"new": 33188},
"new": True,
"renamed_from": None,
},
]
patch = rs_parsepatch.get_lines(mock_revision.patch)
assert patch == [
{
"added_lines": [2],
"binary": False,
"copied_from": None,
"deleted": False,
"deleted_lines": [],
"filename": "test.txt",
"modes": {},
"new": False,
"renamed_from": None,
},
{
"added_lines": [1],
"binary": False,
"copied_from": None,
"deleted": False,
"deleted_lines": [],
"filename": "test.cpp",
"modes": {"new": 33188},
"new": True,
"renamed_from": None,
},
]
def transform(hg: hglib.client, repo_dir: str, commit: Commit):
hg_modified_files(hg, commit)
if commit.ignored or len(commit.backsout) > 0 or commit.bug_id is None:
return commit
assert code_analysis_server is not None
source_code_sizes = []
other_sizes = []
test_sizes = []
metrics_file_count = 0
patch = hg.export(revs=[commit.node.encode("ascii")], git=True)
try:
patch_data = rs_parsepatch.get_lines(patch)
except Exception:
logger.error(f"Exception while analyzing {commit.node}")
raise
for stats in patch_data:
path = stats["filename"]
if stats["binary"]:
if not is_test(path):
commit.types.add("binary")
continue
size = None
after = None
if not stats["deleted"]:
try:
after = hg.cat(
[os.path.join(repo_dir, path).encode("utf-8")],
rev=commit.node.encode("ascii"),
)
size = after.count(b"\n")
except hglib.error.CommandError as e:
if b"no such file in rev" not in e.err:
raise
type_ = get_type(path)
if is_test(path):
commit.test_files_modified_num += 1
commit.test_added += len(stats["added_lines"])
commit.test_deleted += len(stats["deleted_lines"])
if size is not None:
test_sizes.append(size)
# We don't have a 'test' equivalent of types, as most tests are JS,
# so this wouldn't add useful information.
elif type_ in SOURCE_CODE_TYPES_TO_EXT:
commit.source_code_files_modified_num += 1
commit.source_code_added += len(stats["added_lines"])
commit.source_code_deleted += len(stats["deleted_lines"])
if size is not None:
source_code_sizes.append(size)
if type_ != "IDL/IPDL/WebIDL":
metrics = code_analysis_server.metrics(path,
after,
unit=False)
if metrics.get("spaces"):
metrics_file_count += 1
error = get_metrics(commit, metrics["spaces"])
if error:
logger.debug(
f"rust-code-analysis error on commit {commit.node}, path {path}"
)
touched_functions = get_touched_functions(
metrics["spaces"],
stats["deleted_lines"],
stats["added_lines"],
)
if len(touched_functions) > 0:
commit.functions[path] = list(touched_functions)
# Replace type with "Objective-C/C++" if rust-code-analysis detected this is an Objective-C/C++ file.
if type_ == "C/C++" and metrics.get(
"language") == "obj-c/c++":
type_ = "Objective-C/C++"
commit.types.add(type_)
else:
commit.other_files_modified_num += 1
commit.other_added += len(stats["added_lines"])
commit.other_deleted += len(stats["deleted_lines"])
if size is not None:
other_sizes.append(size)
if type_:
commit.types.add(type_)
commit.total_source_code_file_size = sum(source_code_sizes)
commit.average_source_code_file_size = (
commit.total_source_code_file_size /
len(source_code_sizes) if len(source_code_sizes) > 0 else 0)
commit.maximum_source_code_file_size = max(source_code_sizes, default=0)
commit.minimum_source_code_file_size = min(source_code_sizes, default=0)
commit.total_other_file_size = sum(other_sizes)
commit.average_other_file_size = (commit.total_other_file_size /
len(other_sizes)
if len(other_sizes) > 0 else 0)
commit.maximum_other_file_size = max(other_sizes, default=0)
commit.minimum_other_file_size = min(other_sizes, default=0)
commit.total_test_file_size = sum(test_sizes)
commit.average_test_file_size = (commit.total_test_file_size /
len(test_sizes)
if len(test_sizes) > 0 else 0)
commit.maximum_test_file_size = max(test_sizes, default=0)
commit.minimum_test_file_size = min(test_sizes, default=0)
if metrics_file_count:
commit.average_cyclomatic = commit.total_cyclomatic / metrics_file_count
commit.average_halstead_n2 = commit.total_halstead_n2 / metrics_file_count
commit.average_halstead_N2 = commit.total_halstead_N2 / metrics_file_count
commit.average_halstead_n1 = commit.total_halstead_n1 / metrics_file_count
commit.average_halstead_N1 = commit.total_halstead_N1 / metrics_file_count
commit.average_source_loc = commit.total_source_loc / metrics_file_count
commit.average_logical_loc = commit.total_logical_loc / metrics_file_count
else:
# these values are initialized with sys.maxsize (because we take the min)
# if no files, then reset them to 0 (it'd be stupid to have min > max)
commit.minimum_cyclomatic = 0
commit.minimum_halstead_N2 = 0
commit.minimum_halstead_n2 = 0
commit.minimum_halstead_N1 = 0
commit.minimum_halstead_n1 = 0
commit.minimum_source_loc = 0
commit.minimum_logical_loc = 0
return commit
def read(patch):
counts = pp.get_counts(patch)
diffs = pp.get_diffs(patch)
lines = pp.get_lines(patch)
# compare counts & diffs
assert len(diffs) == len(counts)
for count, ppd in zip(counts, diffs):
diff = copy.deepcopy(ppd)
del diff["lines"]
diff.update({"added_lines": 0, "deleted_lines": 0})
for n, o, _ in ppd["lines"]:
if n is None:
diff["added_lines"] += 1
elif o is None:
diff["deleted_lines"] += 1
assert diff == count
# compare lines & diffs
assert len(diffs) == len(lines)
for line, ppd in zip(lines, diffs):
diff = copy.deepcopy(ppd)
del diff["lines"]
diff.update({"added_lines": [], "deleted_lines": []})
for n, o, _ in ppd["lines"]:
if n is None:
diff["added_lines"].append(o)
elif o is None:
diff["deleted_lines"].append(n)
assert diff == line
if not isinstance(patch, str):
patch = patch.decode("utf-8")
wp_diffs = wp.parse_patch(patch)
wp_diffs = list(wp_diffs)
assert len(wp_diffs) == len(diffs)
# compare wtp and pp outputs
for ppd, wpd in zip(diffs, wp_diffs):
wnew_path = get_filename(wpd.header.new_path)
wold_path = get_filename(wpd.header.old_path)
assert ppd["filename"] == wnew_path
if wnew_path != wold_path:
assert ppd["renamed_from"] == wold_path or ppd[
"copied_from"] == wold_path
else:
assert ppd["renamed_from"] is None
if not wpd.changes:
assert not ppd["lines"]
continue
changes = list(wpd.changes)
assert len(ppd["lines"]) == len(changes)
for pline, wline in zip(ppd["lines"], changes):
wn, wo, wc, whc = wline
pn, po, pc = pline
assert wn == pn
assert wo == po
pc = pc.decode("utf-8")
assert wc == pc
