English | 日本語

Comparison options

The text, sorted, and csv subcommands share WinMerge-style comparison options. Some of them change how lines or fields are matched (normalization); others control how much output is produced and how the diff engine re-synchronizes after a difference. Normalization only affects comparison — the output always shows the original lines.

Text options are also exposed through the GUI /api/diff API.

Normalization (how lines are matched)

--ignore-case

Compare lines case-insensitively. Lines that differ only in letter case are treated as equal.

ayame-diff text --ignore-case a.txt b.txt

--ignore-whitespace

Control how whitespace is treated when matching lines.

--ignore-whitespace none | change | all

Value	Behaviour
none	Whitespace is significant (default).
change	Collapse each run of whitespace to a single space and trim the ends.
all	Remove all whitespace before comparing.

# Treat runs of spaces/tabs as one space, ignore leading/trailing whitespace
ayame-diff text --ignore-whitespace change a.txt b.txt

# Ignore whitespace entirely
ayame-diff text --ignore-whitespace all a.txt b.txt

GNU-compatible aliases --ignore-space-change and --ignore-all-space select the corresponding modes.

Note

change and all normalize only for the comparison. The printed lines are the untouched originals.

Line endings

EOLs are significant in text mode by default. --ignore-eol ignores every LF/CRLF difference, while --ignore-trailing-eol ignores only whether the last line has a terminator. CSV parsing is record-based, so these differences are structurally ignored there.

ayame-diff text --ignore-eol windows.txt unix.txt
ayame-diff text --ignore-trailing-eol generated.txt checked-in.txt

--filter-line

Remove every match of a Go regular expression from the comparison view. Repeat the flag to compose filters. A whole-line match therefore ignores that line's contents; a partial match can remove volatile timestamps or request IDs. Output still contains the original text. In CSV mode each field is filtered before key and value comparison.

ayame-diff text --filter-line 'timestamp=\S+' --filter-line 'request_id=\d+' a.log b.log

CSV value controls

--ignore-column NAME / --ignore-column-index N removes a column from value comparison. With the default all-column key it is removed from the key too; with an explicit key, identity remains controlled by that key.

Numeric values can be compared using an absolute tolerance. A global tolerance requires an explicit key; per-column tolerance automatically removes that column from the default key. Tolerance columns cannot themselves be explicit key columns. Non-numeric values continue to compare exactly.

ayame-diff csv --left a.csv --right b.csv --key id \
  --ignore-column updated_at --tolerance 0.0001 --out diff.tsv

ayame-diff csv --left a.csv --right b.csv \
  --column-tolerance price=0.01 --column-tolerance-index 4=0.1 --out diff.tsv

Duplicate-key groups use maximum matching, so a tolerance-compatible pairing is not lost merely because an earlier row also had another possible partner.

Word-level highlighting

--word

In the unified (default) output, --word highlights just the words that changed inside a Replace hunk, instead of marking the whole line. Deletions are wrapped in [-...-] and insertions in {+...+}.

ayame-diff text --word old.txt new.txt

--word applies to the unified format. The GUI shows equivalent word-level highlighting in its side-by-side grid.

Resync and output limits

For huge files, the diff engine looks ahead only a bounded distance to re-synchronize after lines diverge, and it caps how much it prints. These knobs tune that behaviour.

--window

The resync look-ahead window used when lines differ (default 128). A larger window can re-align after bigger blocks of inserted/deleted lines, at the cost of more work per mismatch.

ayame-diff text --window 512 old.txt new.txt

--max-hunks

The maximum number of hunks to print (default 200). Hunks beyond the limit are still counted — the summary and JSON report the total — but are not printed.

ayame-diff text --max-hunks 50 old.txt new.txt

--max-lines

The maximum number of lines shown per hunk side (default 200). Longer hunks are truncated in the output.

ayame-diff text --max-lines 40 old.txt new.txt

--width

Total column width for --side-by-side output (default 160).

ayame-diff text --side-by-side --width 200 old.txt new.txt

Combining options

The options compose freely:

ayame-diff text \
  --ignore-case \
  --ignore-whitespace change \
      --ignore-eol \
      --filter-line 'timestamp=\S+' \
  --word \
  --window 256 \
  --max-hunks 100 \
  old.txt new.txt

All of these also work with sorted, alongside its --numeric / -n and --reverse / -r sort controls:

ayame-diff sorted --numeric --ignore-whitespace all a.txt b.txt