initial

go.mod

@@ -0,0 +1,5 @@
+module git.broken-by-design.fr/fmaury/srt_to_yaml
+
+go 1.22.7
+
+require gopkg.in/yaml.v3 v3.0.1

go.sum

@@ -0,0 +1,4 @@
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
+gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

main.go

@@ -0,0 +1,113 @@
+package main
+
+import (
+	"bufio"
+	"flag"
+	"fmt"
+	"io"
+	"log"
+	"os"
+	"regexp"
+	"strings"
+
+	"gopkg.in/yaml.v3"
+)
+
+var (
+	IdentifierLine = regexp.MustCompile("^\ufeff?\\d+$")
+	TimeStampLine  = regexp.MustCompile(`^((?:\d{2}:){2}\d{2},\d{3}) --> ((?:\d{2}:){2}\d{2},\d{3})$`)
+)
+
+type Entry struct {
+	Start   string `yaml:"start"`
+	End     string `yaml:"end"`
+	Speaker string `yaml:"speaker"`
+	Line1   string `yaml:"line"`
+	Line2   string `yaml:"line2"`
+}
+
+func parseSrt(r io.Reader) (entries []Entry, err error) {
+	srtScan := bufio.NewScanner(r)
+	for {
+		if !srtScan.Scan() {
+			break
+		}
+		firstLine := strings.TrimSpace(srtScan.Text())
+		if !IdentifierLine.MatchString(firstLine) {
+			err = fmt.Errorf("invalid identifier %q", string(firstLine))
+			return
+		}
+
+		if !srtScan.Scan() {
+			err = fmt.Errorf("unexpected end of file after identifier")
+			return
+		}
+		secondLine := strings.TrimSpace(srtScan.Text())
+		extractedValues := TimeStampLine.FindAllStringSubmatch(secondLine, -1)
+		if extractedValues == nil || len(extractedValues) != 1 {
+			err = fmt.Errorf("invalid timestamp line: %q", secondLine)
+			return
+		}
+		startTime := extractedValues[0][1]
+		endTime := extractedValues[0][2]
+
+		if !srtScan.Scan() {
+			err = fmt.Errorf("unexpected end of file after timestamps")
+			return
+		}
+		thirdLine := strings.TrimSpace(srtScan.Text())
+		if thirdLine == "" {
+			err = fmt.Errorf("unexpected empty line after timestamp")
+			return
+		}
+
+		fourthLine := ""
+		if srtScan.Scan() {
+			fourthLine = strings.TrimSpace(srtScan.Text())
+			if fourthLine != "" {
+				_ = srtScan.Scan()
+				if srtScan.Text() != "" {
+					err = fmt.Errorf("unexpected non empty line after the second line of text")
+					return
+				}
+			}
+		}
+		entry := Entry{
+			Start: startTime,
+			End:   endTime,
+			Line1: thirdLine,
+			Line2: fourthLine,
+		}
+		entries = append(entries, entry)
+	}
+	return
+}
+
+func main() {
+	srtFile := flag.String("srt", "", "File path to the srt file to convert")
+	outFile := flag.String("out", "", "File path to the yaml file")
+	flag.Parse()
+
+	srtFd, err := os.Open(*srtFile)
+	if err != nil {
+		log.Fatalf("failed to open SRT file: %q", err.Error())
+	}
+	defer srtFd.Close()
+
+	entries, err := parseSrt(srtFd)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	yamlContent, err := yaml.Marshal(&entries)
+	if err != nil {
+		log.Fatalf("failed to marshal content, %q", err.Error())
+	}
+
+	outFd, err := os.OpenFile(*outFile, os.O_CREATE|os.O_EXCL|os.O_WRONLY, 0o600)
+	if err != nil {
+		log.Fatalf("failed to open YAML file: %q", err.Error())
+	}
+	defer outFd.Close()
+	outFd.Write(yamlContent)
+}

main_test.go

@@ -0,0 +1,139 @@
+package main
+
+import (
+	"bytes"
+	"reflect"
+	"strings"
+	"testing"
+)
+
+func TestParseSingleEntry(t *testing.T) {
+	document := strings.Join([]string{
+		"1",
+		"00:00:00,000 --> 00:00:14,000",
+		"toto",
+		"titi",
+	}, "\n")
+
+	entries, err := parseSrt(bytes.NewBuffer([]byte(document)))
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedEntries := []Entry{
+		{
+			Start: "00:00:00,000",
+			End:   "00:00:14,000",
+			Line1: "toto",
+			Line2: "titi",
+		},
+	}
+	if !reflect.DeepEqual(entries, expectedEntries) {
+		t.Fatalf("unexpected entries: %v", entries)
+	}
+}
+
+func TestParseMultipleEntries(t *testing.T) {
+	document := strings.Join([]string{
+		"1",
+		"00:00:00,000 --> 00:00:14,000",
+		"toto",
+		"titi",
+		"",
+		"2",
+		"00:01:00,000 --> 00:02:14,000",
+		"tutu",
+		"tata",
+	}, "\n")
+	entries, err := parseSrt(bytes.NewBuffer([]byte(document)))
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedEntries := []Entry{
+		{
+			Start: "00:00:00,000",
+			End:   "00:00:14,000",
+			Line1: "toto",
+			Line2: "titi",
+		},
+		{
+			Start: "00:01:00,000",
+			End:   "00:02:14,000",
+			Line1: "tutu",
+			Line2: "tata",
+		},
+	}
+	if !reflect.DeepEqual(entries, expectedEntries) {
+		t.Fatalf("unexpected entries: %v", entries)
+	}
+}
+
+func TestParseInvalidTimestamp(t *testing.T) {
+	document := strings.Join([]string{
+		"1",
+		"00:00:00.000 --> 00:00:14.000",
+		"toto",
+		"titi",
+	}, "\n")
+	_, err := parseSrt(bytes.NewBuffer([]byte(document)))
+	if err == nil {
+		t.Fatal("expected error")
+	} else if eStr := err.Error(); eStr != "invalid timestamp line: \"00:00:00.000 --> 00:00:14.000\"" {
+		t.Fatalf("unexpected error: %q", eStr)
+	}
+}
+
+func TestParseSingleLine(t *testing.T) {
+	document := strings.Join([]string{
+		"1",
+		"00:00:00,000 --> 00:00:14,000",
+		"toto",
+	}, "\n")
+	entries, err := parseSrt(bytes.NewBuffer([]byte(document)))
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedEntries := []Entry{
+		{
+			Start: "00:00:00,000",
+			End:   "00:00:14,000",
+			Line1: "toto",
+			Line2: "",
+		},
+	}
+	if !reflect.DeepEqual(entries, expectedEntries) {
+		t.Fatalf("unexpected entries: %v", entries)
+	}
+}
+
+func TestParseMultipleEntriesSingleLine(t *testing.T) {
+	document := strings.Join([]string{
+		"1",
+		"00:00:00,000 --> 00:00:14,000",
+		"toto",
+		"",
+		"2",
+		"00:01:00,000 --> 00:02:14,000",
+		"tata",
+	}, "\n")
+	entries, err := parseSrt(bytes.NewBuffer([]byte(document)))
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedEntries := []Entry{
+		{
+			Start: "00:00:00,000",
+			End:   "00:00:14,000",
+			Line1: "toto",
+			Line2: "",
+		},
+		{
+			Start: "00:01:00,000",
+			End:   "00:02:14,000",
+			Line1: "tata",
+			Line2: "",
+		},
+	}
+	if !reflect.DeepEqual(entries, expectedEntries) {
+		t.Fatalf("unexpected entries: %v", entries)
+	}
+}