diffstat

Additional Notes

diffstat reads the output of diff and produces a histogram showing how many lines were inserted, deleted, or modified in each file. It is often used to summarize the size of patch files or version control diffs.

The output shows each filename followed by a horizontal bar chart of added and removed lines. The total number of insertions, deletions, and modified files is shown at the bottom. It is a quick way to assess the scope of a change set.

Syntax

diffstat [options] [file...]

Parameters

• file: A diff or patch file to analyze. If omitted, reads from standard input.

Common Options

• -c: Print each changed line prefixed with + or -.
• -d: Suppress display of files with no insertions or deletions.
• -f format: Set the output format to 0 (normal), 1 (with relative percentages), 2 (histogram only), 3 (histogram plus relative percentages), 4 (machine-readable CSV), or 5 (XML).
• -k: Suppress filenames in the output.
• -l: Show only filenames, one per line.
• -m: Merge insertions and deletions into a single "modification" count.
• -n width: Set the minimum width for filenames (default is the longest filename).
• -p width: Override the histogram width (default is the terminal width).
• -r: Show relative percentages instead of histograms.
• -t: Override the tab size for calculating column alignment.
• -u: Print output sorted by percentage of changes.
• -w: Wrap lines exceeding terminal width.
• -K: Keep the original filenames even when they are temporary.
• -N: Do not show the number of inserted/deleted lines.
• -R: Assume patch was created with reversed old and new order.

Examples

diff file1.c file2.c | diffstat

Show a histogram of changes between two files.

diffstat patch.diff

Summarize the changes in a patch file.

git diff | diffstat

Summarize the unstaged changes in a Git repository.

diffstat -p 40 -n 20 changes.patch

Set the histogram width to 40 characters and minimum filename width to 20.

Practical Notes

• diffstat works with unified, context, and normal diff formats.
• Pipe git diff, svn diff, or diff -u output into diffstat for version control summaries.
• Use the -m option when you do not care about distinguishing inserts from deletes.
• The CSV output (-f 4) is useful for programmatic processing in scripts.