tamada / oinkie / 28728335077

74%

main: 74%

Build Type

Pull #16

github

Committed by tamada

Commit Message

fix: escape CSV fields consistently and parse result files with the csv crate

escape_csv_string was applied only to matrix row/column names, while
Metadata::csv_info, Program::csv_info, Birthmark::csv_info and
CompareResult::to_csv wrote file names and paths verbatim. A path
containing a comma (or quote/newline) corrupted the result CSVs, and
the naive split(',') parsers in CompareResult::parse and
read_result_file then read back wrong paths, which silently poisons
--skip reruns and reaggregation.

- Escape every user-controlled field (names, paths) in the CSV writers
- Parse results.csv lines and per-pair CSV headers with the csv crate,
  which understands the quoting produced by the writers
- Export escape_csv_string through the prelude so the CLI crate shares
  the same escaping
- Add roundtrip tests with commas and quotes in paths

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Pull Request Pull Request #16: fix(security): CSV 出力のエスケープ漏れとパース不整合を修正

Coverage Stats

143 of 178 new or added lines in 7 files covered. (80.34%)

1 existing line in 1 file now uncovered.

1611 of 2186 relevant lines covered (73.7%)

12.22 hits per line