# Text word diff

`text-word-diff` · version 1.0.0 · Text · free, no key needed

Diff two texts as whitespace versus non-whitespace word tokens.

**Use when you need to: text word diff · word token text diff · diff texts by words.**

## Supported

- text word diff
- word token text diff
- diff texts by words

## Not supported

- locale word break
- Unicode UAX #29 word boundaries
- unicode normalization
- fuzzy matching
- patience/histogram diff
- treat U+00A0 as whitespace
- filesystem access

## Behavior

- left and right are tokenized with splitWords: maximal runs of ASCII space, tab, CR, or LF versus maximal runs of everything else.
- U+00A0 is a word character, not whitespace. Empty text yields no tokens.
- The Myers SES uses string === equality; when D-paths tie, the library prefers delete from the left.
- ops is the per-token SES. keep includes left_index and right_index; delete includes left_index; insert includes right_index.
- keeps, deletions, and insertions count those op kinds. Adjacent keeps are not collapsed.
- Unknown input fields are rejected. Prototype-like tokens such as "__proto__" are ordinary strings.

## Input

- `left` (string, required): max length 262144
- `right` (string, required): max length 262144

## Output

- `ops` (array of object, required): max items 4096
- `keeps` (integer, required): min 0
- `deletions` (integer, required): min 0
- `insertions` (integer, required): min 0

## Limits

- max text bytes: 262144
- max token count: 8192
- max edit ops: 4096
- max output bytes: 1048576

## Example

Request input:

```json
{
  "left": "ab cd",
  "right": "ab xy"
}
```

Response:

```json
{
  "result": {
    "ops": [
      {
        "op": "keep",
        "value": "ab",
        "left_index": 0,
        "right_index": 0
      },
      {
        "op": "keep",
        "value": " ",
        "left_index": 1,
        "right_index": 1
      },
      {
        "op": "delete",
        "value": "cd",
        "left_index": 2
      },
      {
        "op": "insert",
        "value": "xy",
        "right_index": 2
      }
    ],
    "keeps": 2,
    "deletions": 1,
    "insertions": 1
  }
}
```

## How to call it

### MCP

Connect `https://computefirst.net/mcp` ([setup](/docs#connect)), then call `execute` with:

```json
{
  "id": "text-word-diff",
  "version": "1.0.0",
  "input": {
    "left": "ab cd",
    "right": "ab xy"
  }
}
```

### HTTP (no key)

```sh
curl -X POST https://computefirst.net/v1/tools/text-word-diff/versions/1.0.0/execute \
  -H "Content-Type: application/json" \
  -d '{"left":"ab cd","right":"ab xy"}'
```

The machine-readable contract is at [/v1/tools/text-word-diff/versions/1.0.0](/v1/tools/text-word-diff/versions/1.0.0).

### CLI

```sh
node cli.mjs run text-word-diff 1.0.0 --input input.json --base-url https://computefirst.net
```

Get the client at [/clients/cli/](/clients/cli/).

## Related tools

- [Text line diff](/tools/text-line-diff): Diff two texts as line records, preserving each line terminator.
- [Text apply unified patch](/tools/text-apply-unified-patch): Apply a one-file unified patch to text at original line addresses with exact hunk matches.
- [Text diff stats](/tools/text-diff-stats): Summarize a line-record diff as keep, insert, delete, and hunk counts.
- [Text invert unified patch](/tools/text-invert-unified-patch): Invert a one-file unified patch by swapping file names and delete/insert ops.
- [Text lines chunk](/tools/text-lines-chunk): Split parsed lines into consecutive chunks of size lines each.
- [Text parse unified patch](/tools/text-parse-unified-patch): Parse a one-file unified patch into path names and 0-based hunks.