About this tool
Text diff finds additions, deletions, and unchanged content between versions of copy, configuration, logs, and documents. Results depend on comparison granularity and whitespace handling: lines suit code and paragraphs, while words or characters reveal inline edits.
The tool shows surface changes and does not understand the business meaning of clauses, numbers, or code. Rewrapping, indentation, and line-ending changes can create large diffs, so decide whether whitespace matters and use consistent encoding.
How to use it
Place baseline and revision
Make clear that the first or left text is older so additions and deletions are interpreted correctly.
Choose an appropriate view
Review the overall change, then focus on differing lines and words.
Judge impact manually
Inspect numbers, negations, punctuation, and structure to confirm intent.
Input and output example
Line-level location plus word-level highlighting. Only one number changed on the first line, but the whole line is flagged first.
// left
port 8080
timeout 30s
// right
port 3000
timeout 30s~ line 1: port [8080 → 3000]
line 2: unchangedSupported range and limits
- Comparison granularity
- Differences are located line by line, then highlighted word by word within changed lines; both shown at once
- Algorithm
- Longest common subsequence from an open-source diff library, the same approach git diff uses
- Detects
- Added lines, removed lines, and word-level edits within a line
- Does not detect
- A moved block shows as one deletion plus one insertion rather than as a move
- No understanding of meaning
- Paraphrases and reordered clauses appear as large differences, while a critical change to a number or unit may be one character
- Better for JSON
- When both sides are JSON, use JSON Diff; it compares by field and ignores formatting
When you would use it
Reviewing copy revisions
Confirm that only agreed paragraphs changed and nothing was accidentally removed.
Diagnosing configuration drift
Compare environment configuration or log fragments to locate changes.
Checking a final contract against the previous draft
Word-level highlighting exposes single-character changes to amounts, dates and liability clauses; exactly what a read-through misses.
What to know before you start
- Matching text does not prove identical system behavior because external context can differ.
- LF versus CRLF, trailing spaces, and Unicode normalization create nonvisual differences.
- Sensitive contracts and secrets remain subject to organizational copying and handling rules.
Related concepts
- diff
- A representation of additions and deletions that transform one text sequence into another.
- whitespace
- Spaces, tabs, and line breaks that do not appear as ordinary glyphs.