git dm draft

git-dm-draft.R

objects_output <- system("git cat-file --batch-check --batch-all-objects", intern = TRUE)
objects_list <- strsplit(objects_output, "\\s")
objects <- data.frame(ObjectID = sapply(objects_list, "[[", 1),
                      Type = sapply(objects_list, "[[", 2),
                      Size = as.numeric(sapply(objects_list, "[[", 3)))


# commits

commits_output <- system('git log --pretty=format:"%H|/|%an|/|%ae|/|%cn|/|%ce|/|%s|/|%P|/|%T|/|%ct" --all', intern = TRUE)
commits_list <- strsplit(commits_output, "\\|/\\|")
commits <- data.frame(CommitID = sapply(commits_list, "[[", 1),
                      Author = sapply(commits_list, "[[", 2),
                      AuthorEmail = sapply(commits_list, "[[", 3),
                      Committer = sapply(commits_list, "[[", 4),
                      CommitterEmail = sapply(commits_list, "[[", 5),
                      CommitMessage = sapply(commits_list, "[[", 6),
                      ParentCommit = sapply(commits_list, "[[", 7),
                      TreeID = sapply(commits_list, "[[", 8),
                      Timestamp = as.POSIXct(as.numeric(sapply(commits_list, "[[", 9)), origin = "1970-01-01"))

# trees table

trees_output <- system("git ls-tree --full-tree -r HEAD", intern = TRUE)
trees_list <- strsplit(trees_output, "\\t| ")
trees <- data.frame(
  tree_id = sapply(trees_list, "[[", 1),
  type = sapply(trees_list, "[[", 2),
  object_id = sapply(trees_list, "[[", 3),
  path = sapply(trees_list, "[[", 4)
)
                    
# tags

# tags_output <- system("git show-ref --tags --abbrev=40 --hash --dereference", intern = TRUE)
# tags_list <- strsplit(tags_output, "\\s")
# tags <- data.frame(TagID = sapply(tags_list, "[[", 1),
#                    ObjectID = sapply(tags_list, "[[", 2),
#                    TagName = sapply(tags_list, "[[", 3),
#                    Tagger = NA,
#                    TaggerEmail = NA,
#                    TagMessage = NA,
#                    Timestamp = NA)

# branches

branches_output <- system("git show-ref --heads --abbrev=40 --hash", intern = TRUE)
branches_list <- strsplit(branches_output, "\\s")
branches <- data.frame(BranchName = sapply(branches_list, "[[", 2),
                       CommitID = sapply(branches_list, "[[", 1))

# authors 

blobs_output <- system("git cat-file --batch-check --batch-all-objects | grep blob", intern = TRUE)
blobs_list <- strsplit(blobs_output, "\\s")
blobs <- data.frame(BlobID = sapply(blobs_list, "[[", 1),
                    Size = as.numeric(sapply(blobs_list, "[[", 3)),
                    Content = NA)