From adaa1c7fd4a4290cde9a55521f52c9bd79697895 Mon Sep 17 00:00:00 2001 From: Karma Riuk Date: Mon, 17 Mar 2025 15:45:19 +0100 Subject: [PATCH] changed the structure of the entries --- dataset.py | 9 ++------- pull_requests.py | 6 +++--- 2 files changed, 5 insertions(+), 10 deletions(-) diff --git a/dataset.py b/dataset.py index 8497b7b..19d7113 100644 --- a/dataset.py +++ b/dataset.py @@ -18,18 +18,13 @@ class Metadata: reason_for_failure: str = "" last_cmd_error_msg: str = "" -@dataclass -class Diff: - filename: str - patch: str - @dataclass class DatasetEntry: metadata: Metadata files: Dict[str, FileData] # filename -> file data, files before the PR (before the first PR commits) - diffs_before: Dict[str, Diff] # filename -> diff, diffs between the opening of the PR and the comment + diffs_before: Dict[str, str] # filename -> diff, diffs between the opening of the PR and the comment comment: str - diffs_after: Dict[str, Diff] # filename -> diff, changes after the comment + diffs_after: Dict[str, str] # filename -> diff, changes after the comment @dataclass class Dataset: diff --git a/pull_requests.py b/pull_requests.py index 4bca598..1e1d756 100644 --- a/pull_requests.py +++ b/pull_requests.py @@ -82,16 +82,16 @@ def process_pull(repo: Repository, pr: PullRequest, dataset: Dataset, repos_dir: first_commit = commits[0] last_commit = commits[-1] - diffs_before = [Diff(file.filename, file.patch) for file in repo.compare(pr.base.sha, first_commit.sha).files] + diffs_before = {file.filename: file.patch for file in repo.compare(pr.base.sha, first_commit.sha).files} comments = list(pr.get_review_comments()) assert len(comments) == 1 comment_text = comments[0].body if comments else "" - diffs_after = [Diff(file.filename, file.patch) for file in repo.compare(first_commit.sha, last_commit.sha).files] + diffs_after = {file.filename: file.patch for file in repo.compare(first_commit.sha, last_commit.sha).files} entry = DatasetEntry( metadata=Metadata(repo.full_name, pr.number, pr.merge_commit_sha), - files=[FileData(file.filename) for file in pr.get_files()], + files={file.filename: FileData(file.filename) for file in pr.get_files()}, diffs_before=diffs_before, comment=comment_text, diffs_after=diffs_after,