uh-filter

1import type { Register } from 'claude-code'
2
3import { cleanInsertedText, dropTypedFiller, stripFillers } from './filter'
4
5export const register: Register = on => {
6  // Terminal prompt box. A chunk landing at once (dictation, a paste) is cleaned
7  // before it shows; when typing, a filler word goes as soon as a space ends it.
8  on('prompt.edit', async ($, e, next) => {
9    if (!e.key) {
10      const inputText = cleanInsertedText(e.inputText)
11
12      return next(inputText === undefined ? e : { ...e, inputText })
13    }
14
15    const box = await next(e)
16    const isTypedSpace = e.inputText === ' ' && e.start === e.end
17
18    return (isTypedSpace && dropTypedFiller(box)) || box
19  })
20
21  // Last pass on submit, for anything the box didn't catch (and for the desktop app,
22  // whose prompt box mods can't reach).
23  on('prompt.submit', ($, e, next) => {
24    // Slash commands ignore a rewrite, and meta prompts aren't the person's words.
25    if (e.command || e.isMeta) {
26      return next(e)
27    }
28
29    let text = e.text
30    try {
31      text = stripFillers(e.text)
32    } catch {
33      // A filter bug must never cost the person their prompt.
34    }
35
36    return next(text === e.text ? e : { ...e, text })
37  })
38}
39

1// Spoken fillers, matched as whole tokens: "uh", "uhh", "um", "umm", "uhm", "erm", "hmm", "mm".
2// A hyphen on either side leaves it alone, so "uh-huh" and "uh-oh" survive.
3const FILLER = /(?<![\w-])(?:u+h+m*|u+m+|e+r+m+|h+m+|m{2,})(?![\w-])[,.]?/gi
4
5// One to three words said twice in a row: "check check", "where you where you".
6const REPEAT = /\b(\w+(?:\s+\w+){0,2})(?:[\s,]+\1\b)+/gi
7
8// Single words that legitimately double up in English.
9const KEEP_DOUBLED = new Set(['that', 'had', 'is', 'do', 'very', 'no', 'bye'])
10
11// Inline code and fenced blocks are passed through untouched.
12const CODE = /(```[\s\S]*?```|`[^`\n]*`)/
13
14function cleanProse(text: string): string {
15  return text
16    .replace(FILLER, '')
17    .replace(REPEAT, (match, phrase: string) =>
18      !phrase.includes(' ') && KEEP_DOUBLED.has(phrase.toLowerCase()) ? match : phrase,
19    )
20    .replace(/[ \t]{2,}/g, ' ')
21    .replace(/[ \t]+([,.;:!?])/g, '$1')
22    .replace(/([,;:])(?:\s*[,;:])+/g, '$1')
23    .replace(/^[ \t]*[,;:][ \t]*/gm, '')
24    .replace(/[ \t]+(?=\n)/g, '')
25}
26
27export function stripFillers(text: string): string {
28  const cleaned = text
29    .split(CODE)
30    .map((part, i) => (i % 2 === 1 ? part : cleanProse(part)))
31    .join('')
32    .trim()
33
34  if (cleaned === '') {
35    return text
36  }
37
38  const wasCapitalized = /^[A-Z]/.test(text.trimStart())
39
40  return wasCapitalized ? cleaned.charAt(0).toUpperCase() + cleaned.slice(1) : cleaned
41}
42
43// Only plain spoken prose: letters, digits, spaces and sentence punctuation.
44// Anything else (braces, operators, newlines) is likely pasted code and kept as is.
45const PROSE = /^[\p{L}\p{N} ,.'’?!-]+$/u
46
47// Cleans a chunk landing in the prompt box at once (dictation or a paste),
48// keeping its edge spaces so it still joins the draft around it.
49// Returns undefined when the chunk should go in untouched.
50export function cleanInsertedText(inputText: string): string | undefined {
51  if (inputText.length < 2 || !PROSE.test(inputText)) {
52    return undefined
53  }
54
55  const lead = /^ */.exec(inputText)?.[0] ?? ''
56  const trail = / *$/.exec(inputText)?.[0] ?? ''
57  const body = inputText.trim()
58  const cleaned = stripFillers(body)
59
60  if (cleaned === body) {
61    return FILLER_ONLY.test(body) ? '' : undefined
62  }
63
64  return lead + cleaned + trail
65}
66
67// A chunk that is nothing but fillers, which stripFillers keeps rather than empty.
68const FILLER_ONLY = /^(?:(?:u+h+m*|u+m+|e+r+m+|h+m+|m{2,})[,.]?\s*)+$/i
69
70// The word just before the caret, when it is a filler followed by the space just typed.
71const TYPED_FILLER = /(^|\s)((?:u+h+m*|u+m+|e+r+m+|h+m+|m{2,})[,.]?) $/i
72
73// After a space is typed: drops the filler word it ends, with that space,
74// so "hello um " becomes "hello ". Skipped inside an open inline-code span.
75export function dropTypedFiller(box: { text: string; cursor: number }): { text: string; cursor: number } | undefined {
76  const before = box.text.slice(0, box.cursor)
77  const match = TYPED_FILLER.exec(before)
78  const isInCode = (before.match(/`/g)?.length ?? 0) % 2 === 1
79
80  if (!match || isInCode) {
81    return undefined
82  }
83
84  const start = box.cursor - (match[2]?.length ?? 0) - 1
85
86  return { text: box.text.slice(0, start) + box.text.slice(box.cursor), cursor: start }
87}
88