1package subs
 2
 3import (
 4	"encoding/xml"
 5	"fmt"
 6	"io"
 7	"log"
 8	"strings"
 9)
10
11type TimedText struct {
12	XMLName xml.Name `xml:"timedtext"`
13	Body    Body     `xml:"body"`
14}
15
16type Body struct {
17	Paragraphs []Paragraph `xml:"p"`
18}
19
20type Paragraph struct {
21	Start     int        `xml:"t,attr"`    // Start time in milliseconds
22	Length    int        `xml:"d,attr"`    // Duration in milliseconds
23	Text      string     `xml:",chardata"` // Direct text (for cases without <s> tags)
24	Sentences []Sentence `xml:"s"`         // List of <s> tags (for cases with individual words/phrases)
25}
26
27type Sentence struct {
28	Text string `xml:",chardata"` // Text inside the <s> tag
29	Time int    `xml:"t,attr"`    // Optional start time (not always present)
30}
31
32func writeVTT(output io.Writer, i, startTime, endTime int, sentence string) {
33	output.Write(
34		[]byte(
35			fmt.Sprintf(
36				"%d\n%s --> %s\n%s\n\n",
37				i,
38				millisecondsToTimestamp(startTime),
39				millisecondsToTimestamp(endTime),
40				sentence,
41			),
42		))
43}
44
45// Convert milliseconds to WebVTT timestamp format: HH:MM:SS.mmm
46func millisecondsToTimestamp(ms int) string {
47	seconds := ms / 1000
48	milliseconds := ms % 1000
49	return fmt.Sprintf("%02d:%02d:%02d.%03d", seconds/3600, (seconds%3600)/60, seconds%60, milliseconds)
50}
51
52func Convert(reader io.Reader, output io.Writer) error {
53	content, err := io.ReadAll(reader)
54	if err != nil {
55		return err
56	}
57
58	var timedText TimedText
59	err = xml.Unmarshal(content, &timedText)
60	if err != nil {
61		log.Println("Error unmarshalling XML:", err)
62		return err
63	}
64
65	output.Write([]byte("WEBVTT\n\n"))
66
67	var lastEndTime int
68	for i, p := range timedText.Body.Paragraphs {
69		startTimeMS := p.Start
70		endTimeMS := p.Start + p.Length
71
72		if startTimeMS < lastEndTime {
73			startTimeMS = lastEndTime
74		}
75
76		var sentence string
77		if len(p.Sentences) > 0 {
78			for _, s := range p.Sentences {
79				sentence += s.Text
80			}
81		} else {
82			sentence = p.Text
83		}
84
85		sentence = strings.TrimSpace(sentence)
86		if sentence == "" {
87			continue
88		}
89
90		lastEndTime = endTimeMS
91		writeVTT(output, i+1, startTimeMS, endTimeMS, sentence)
92	}
93
94	return nil
95}