`diff`

Line-by-line and word diff - LCS-based diff of two line arrays, diff of two strings, unified-style format output, check whether a diff has any changes, count additions and removals, and word-level diff of two strings.

Load with: use diff

What this module does

diff computes the difference between two sequences of lines (or words). The core algorithm is LCS (longest common subsequence) via a dynamic-programming table - the same approach as GNU diff. Each diff result is a list of records {t, ln} where t is "+" (added), "-" (removed), or " " (unchanged) and ln is the line content.

This is used by dif3 for three-way merges and is useful for test output comparison, patch generation, and code review tooling.

Quick example

use diff

# Diff two strings line by line
a = "hello\nworld\nfoo"
b = "hello\nearth\nfoo\nbar"

d = diffs(a, b)
# [{t:" ", ln:"hello"}, {t:"-", ln:"world"}, {t:"+", ln:"earth"}, {t:" ", ln:"foo"}, {t:"+", ln:"bar"}]

# Format as unified output
prn(difff(d, nil))

# Did anything change?
prn(diffo(d))   # fls (there are changes)

# Count changes
prn(diffc(d))   # {a: 2, r: 1}

# Word diff
wd = diffw("the quick brown fox", "the slow red fox")

Functions

Core diff

diffl(a, b)

LCS-based diff of two line arrays. Builds an O(m×n) DP table, backtracks to produce the minimal edit sequence, and returns a list of {t, ln} records where t is "+", "-", or " ". This is the primary high-quality diff function used by dif3.

diffs(s1, s2)

Splits two strings on newlines and calls diffl. Convenience wrapper for string inputs.

Output and analysis

difff(diff, ctx)

Formats a diff list as a unified-style string. Each record becomes "{t}{ln}\n" - added lines start with +, removed with -, unchanged with a space. The ctx parameter is reserved for future context-line support.

diffo(diff)

Returns fls if the diff contains any "+" or "-" records (i.e. the sequences differ), tru if they are identical.

diffc(diff)

Counts the number of added and removed lines. Returns {a: additions, r: removals}.

Word diff

diffw(s1, s2)

Splits two strings on spaces and runs an LCS diff on the resulting word arrays. Returns the same {t, ln} record format, where ln is now a word rather than a line.

Notes

The LCS algorithm is O(m×n) space and time - avoid very large inputs in memory-constrained environments.
Used as a dependency by dif3 for three-way merges.
Requires trl and txt.