go-whisper-api/garbage/filter.go

package garbage

import (
    "regexp"
    "strings"
)

var spaceCollapse = regexp.MustCompile(`\s+`)

// Word is a timed token for garbage filtering (mirrors whisper.Word JSON shape).
type Word struct {
    Word  string `json:"word"`
    Start int    `json:"start"`
    Stop  int    `json:"stop"`
}

// FilterText removes configured artifact substrings and normalizes whitespace.
func FilterText(text string, patterns []string) string {
    for _, p := range patterns {
        p = strings.TrimSpace(p)
        if p == "" {
            continue
        }
        text = strings.ReplaceAll(text, p, " ")
    }
    return strings.TrimSpace(spaceCollapse.ReplaceAllString(text, " "))
}

// FilterWords drops tokens that match any garbage pattern.
func FilterWords(words []Word, patterns []string) []Word {
    if len(words) == 0 {
        return words
    }
    out := make([]Word, 0, len(words))
    for _, w := range words {
        if matchesGarbage(w.Word, patterns) {
            continue
        }
        out = append(out, w)
    }
    return out
}

func matchesGarbage(word string, patterns []string) bool {
    word = strings.TrimSpace(word)
    for _, p := range patterns {
        p = strings.TrimSpace(p)
        if p == "" {
            continue
        }
        if word == p || strings.Contains(word, p) {
            return true
        }
    }
    return false
}