Small fixes

Author: warmans

cmd/tools/tools.go

@@ -3,12 +3,17 @@ package tools
 import (
 	"fmt"
 	"github.com/spf13/cobra"
+	"github.com/warmans/tvgif/pkg/model"
+	"github.com/warmans/tvgif/pkg/srt"
 	"github.com/warmans/tvgif/pkg/util"
 	"io"
 	"log/slog"
+	"math"
 	"os"
 	"regexp"
 	"strings"
+	"time"
+	"unicode"
 )
 
 var (
@@ -25,10 +30,131 @@ func NewToolsCommand(logger *slog.Logger) *cobra.Command {
 	cmd.Flags().StringVar(&metadataPath, "o", "./var/metadata", "output metadata to this path")
 
 	cmd.AddCommand(NewFixNameCommand())
+	cmd.AddCommand(NewRepairSRTCommand())
 
 	return cmd
 }
 
+func NewRepairSRTCommand() *cobra.Command {
+	cmd := &cobra.Command{
+		Use:   "repair-srt [file]",
+		Short: "repair an .srt file in place (e.g. fix incorrect text case such as all upper-case dialog)",
+		Args:  cobra.ExactArgs(1),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			fileName := args[0]
+
+			f, err := os.Open(fileName)
+			if err != nil {
+				return fmt.Errorf("failed to open file '%s': %w", fileName, err)
+			}
+
+			// Parse with no duration limit and without eliminating gaps so that
+			// the original timestamps are preserved as closely as possible.
+			dialog, err := srt.Read(f, false, time.Duration(math.MaxInt64))
+			if err != nil {
+				_ = f.Close()
+				return fmt.Errorf("failed to parse srt: %w", err)
+			}
+			if err := f.Close(); err != nil {
+				return err
+			}
+
+			for i := range dialog {
+				dialog[i].Content = fixCase(dialog[i].Content)
+			}
+
+			if err := os.WriteFile(fileName, []byte(renderSRT(dialog)), 0644); err != nil {
+				return fmt.Errorf("failed to write repaired srt: %w", err)
+			}
+
+			_, err = fmt.Fprintf(os.Stdout, "repaired %d lines of dialog in %s\n", len(dialog), fileName)
+			return err
+		},
+	}
+
+	return cmd
+}
+
+// renderSRT serializes dialog back into the SRT file format.
+func renderSRT(dialog []model.Dialog) string {
+	sb := &strings.Builder{}
+	for _, d := range dialog {
+		fmt.Fprintf(sb, "%d\n", d.Pos)
+		fmt.Fprintf(sb, "%s --> %s\n", formatTimestamp(d.StartTimestamp), formatTimestamp(d.EndTimestamp))
+		sb.WriteString(d.Content)
+		sb.WriteString("\n\n")
+	}
+	return sb.String()
+}
+
+// formatTimestamp renders a duration as an SRT timestamp (HH:MM:SS,mmm).
+func formatTimestamp(d time.Duration) string {
+	hours := d / time.Hour
+	d -= hours * time.Hour
+	minutes := d / time.Minute
+	d -= minutes * time.Minute
+	seconds := d / time.Second
+	d -= seconds * time.Second
+	millis := d / time.Millisecond
+	return fmt.Sprintf("%02d:%02d:%02d,%03d", hours, minutes, seconds, millis)
+}
+
+var standaloneI = regexp.MustCompile(`\bi(?:('[a-z]+)|\b)`)
+
+// fixCase corrects the case of a line of dialog. If the text is "shouty"
+// (i.e. contains letters but none of them are lower-case) it is converted
+// to sentence case. Otherwise it is left untouched.
+func fixCase(s string) string {
+	if !isShouty(s) {
+		return s
+	}
+	return toSentenceCase(s)
+}
+
+// isShouty reports whether a string is predominantly upper-case, allowing for
+// a few stray lower-case letters. A minimum number of letters is required to
+// avoid mis-classifying very short strings (e.g. a lone "A" or "I").
+func isShouty(s string) bool {
+	var upper, lower int
+	for _, r := range s {
+		switch {
+		case unicode.IsUpper(r):
+			upper++
+		case unicode.IsLower(r):
+			lower++
+		}
+	}
+	total := upper + lower
+	if total < 3 {
+		return false
+	}
+	return float64(upper)/float64(total) >= 0.6
+}
+
+// toSentenceCase lower-cases the input and then capitalizes the first letter
+// of each sentence, as well as the standalone pronoun "i".
+func toSentenceCase(s string) string {
+	runes := []rune(strings.ToLower(s))
+	capitalizeNext := true
+	for i, r := range runes {
+		if unicode.IsLetter(r) {
+			if capitalizeNext {
+				runes[i] = unicode.ToUpper(r)
+				capitalizeNext = false
+			}
+			continue
+		}
+		switch r {
+		case '.', '!', '?', '\n':
+			capitalizeNext = true
+		}
+	}
+	result := string(runes)
+	return standaloneI.ReplaceAllStringFunc(result, func(match string) string {
+		return "I" + match[1:]
+	})
+}
+
 func NewFixNameCommand() *cobra.Command {
 	cmd := &cobra.Command{
 		Use:   "fix-name",

pkg/discord/bot.go

@@ -523,7 +523,7 @@ func (b *Bot) updatePreview(s *discordgo.Session, i *discordgo.InteractionCreate
 		return
 	}
 	go func() {
-		interactionResponse, err = b.buildInteractionResponseForPreview(
+		interactionResponse, err := b.buildInteractionResponseForPreview(
 			dialogWithContext,
 			sta,
 			responseWithUsername(username),
@@ -606,7 +606,7 @@ func (b *Bot) createPreview(
 
 	// update with the gif
 	go func() {
-		interactionResponse, err = b.buildInteractionResponseForPreview(
+		interactionResponse, err := b.buildInteractionResponseForPreview(
 			dialogWithContext,
 			state,
 			responseWithUsername(username),