Use GitBlameLineInfo for processing git blame output in aggregateBlameAuthorModifiedAndDateInfo

[gok99]

What feature(s) would you like to see in RepoSense

RepoSense/src/main/java/reposense/authorship/FileInfoAnalyzer.java

Lines 158 to 183 in bf78bf2

	for (int lineCount = 0; lineCount < blameResultLines.length; lineCount += 5) {
	String commitHash = blameResultLines[lineCount].substring(0, FULL_COMMIT_HASH_LENGTH);
	String authorName = blameResultLines[lineCount + 1].substring(AUTHOR_NAME_OFFSET);
	String authorEmail = blameResultLines[lineCount + 2]
	.substring(AUTHOR_EMAIL_OFFSET).replaceAll("<|>", "");
	Long commitDateInMs = Long.parseLong(blameResultLines[lineCount + 3].substring(AUTHOR_TIME_OFFSET)) * 1000;
	LocalDateTime commitDate = LocalDateTime.ofInstant(Instant.ofEpochMilli(commitDateInMs),
	config.getZoneId());
	Author author = config.getAuthor(authorName, authorEmail);
	if (!fileInfo.isFileLineTracked(lineCount / 5) || author.isIgnoringFile(filePath)
	|| CommitHash.isInsideCommitList(commitHash, config.getIgnoreCommitList())
	|| commitDate.compareTo(sinceDate) < 0 || commitDate.compareTo(untilDate) > 0) {
	author = Author.UNKNOWN_AUTHOR;
	}
	if (config.isLastModifiedDateIncluded()) {
	if (config.isShallowCloningPerformed()) {
	logger.warning(String.format(
	MESSAGE_SHALLOW_CLONING_LAST_MODIFIED_DATE_CONFLICT, config.getRepoName()));
	}
	fileInfo.setLineLastModifiedDate(lineCount / 5, commitDate);
	}
	fileInfo.setLineAuthor(lineCount / 5, author);
	}

aggregateBlameAuthorModifiedAndDateInfo in FileInfoAnalyzer.java does some manual string wrangling of git blame output which is hard to interpret, contains magic numbers and breaks the git blame abstraction.

If possible, describe the solution

#2140 Introduces GitBlameLineInfo, returned by blameLine, which encapsulates this string wrangling logic. We can replace this manual string processing by using this method in aggregateBlameAuthorModifiedAndDateInfo.

[logical-1985516]

May I try this issue?

[ckcherry23]

@logical-1985516 Sure, please do let us know if you have any questions!

[logical-1985516]

Hi, should we change commit-time in processGitBlameResultLine to author-time? Originally, author-time is used, and the discrepancy between author-time and commit-time has caused some test cases to fail.
I have also noticed that timestamp for author-time is in seconds. Should we change timestampMilliseconds in GitBlameLineInfo to timestampSeconds?

[ckcherry23]

@logical-1985516 Thanks for investigating what caused the test cases to fail. You can make the changes you described.

Also, since you already have a PR open, you can discuss this problem in PR #2232 itself, as that gives us the most context.