def transform()

in bugbug/bug_features.py [0:0]

• 45 lines of code
• 13 McCabe index (conditional complexity)

    def transform(self, bugs):
        bugs_iter = iter(bugs())

        reporter_experience_map = defaultdict(int)
        author_ids = get_author_ids() if self.commit_data else None

        def apply_transform(bug):
            data = {}

            for feature_extractor in self.feature_extractors:
                res = feature_extractor(
                    bug,
                    reporter_experience=reporter_experience_map[bug["creator"]],
                    author_ids=author_ids,
                )

                if hasattr(feature_extractor, "name"):
                    feature_extractor_name = feature_extractor.name
                else:
                    feature_extractor_name = feature_extractor.__class__.__name__

                if res is None:
                    continue

                if isinstance(res, (list, set)):
                    for item in res:
                        data[sys.intern(f"{item} in {feature_extractor_name}")] = True
                    continue

                data[feature_extractor_name] = res

            reporter_experience_map[bug["creator"]] += 1

            summary = bug["summary"]
            comments = [c["text"] for c in bug["comments"]]
            for cleanup_function in self.cleanup_functions:
                summary = cleanup_function(summary)
                comments = [cleanup_function(comment) for comment in comments]

            return {
                "data": data,
                "title": summary,
                "first_comment": "" if len(comments) == 0 else comments[0],
                "comments": " ".join(comments),
            }

        def apply_rollback(bugs_iter):
            with Pool() as p:
                yield from p.imap(
                    partial(bug_snapshot.rollback, when=self.rollback_when),
                    bugs_iter,
                    chunksize=1024,
                )

        if self.rollback:
            bugs_iter = apply_rollback(bugs_iter)

        return pd.DataFrame(apply_transform(bug) for bug in bugs_iter)