From 3732c346d936d79e75e2ad04644fa1a757744c21 Mon Sep 17 00:00:00 2001
From: Aleksei Besogonov <b@alex.net>
Date: Sun, 8 Dec 2024 19:12:59 -0800
Subject: [PATCH] Add an option (on by default) to strip audio files

Kobo doesn't support audio files or audio overlays in EPUBs, and they
blow up the file sizes tremendously. Strip them by default.
---
 cmd/kepubify/kepubify.go |  4 ++++
 kepub/convert.go         |  8 ++++++++
 kepub/kepub.go           | 10 ++++++++++
 kepub/transform.go       | 10 ++++++++++
 4 files changed, 32 insertions(+)

diff --git a/cmd/kepubify/kepubify.go b/cmd/kepubify/kepubify.go
index 2193959..d682008 100644
--- a/cmd/kepubify/kepubify.go
+++ b/cmd/kepubify/kepubify.go
@@ -52,6 +52,7 @@ func main() {
 	noadddummytitlepage := pflag.Bool("no-add-dummy-titlepage", false, "Force-disables the dummy titlepage")
 	replace := pflag.StringArrayP("replace", "r", nil, "Find and replace on all html files (repeat any number of times) (format: find|replace)")
 	charset := pflag.String("charset", "utf-8", "Override the HTML charset (use \"auto\" to detect it from the content)")
+	leaveaudio := pflag.Bool("leave-audio", false, "Leave audio files in the book")
 
 	for _, flag := range []string{"smarten-punctuation", "css", "hyphenate", "no-hyphenate", "fullscreen-reading-fixes", "add-dummy-titlepage", "no-add-dummy-titlepage", "replace", "charset"} {
 		pflag.CommandLine.SetAnnotation(flag, "category", []string{"3.Conversion Options"})
@@ -108,6 +109,9 @@ func main() {
 	} else if *noadddummytitlepage {
 		opts = append(opts, kepub.ConverterOptionDummyTitlepage(false))
 	}
+	if *leaveaudio {
+		opts = append(opts, kepub.ConverterOptionLeaveAudio(true))
+	}
 	for _, r := range *replace {
 		spl := strings.SplitN(r, "|", 2)
 		if len(spl) != 2 {
diff --git a/kepub/convert.go b/kepub/convert.go
index 279c91a..f2120ec 100644
--- a/kepub/convert.go
+++ b/kepub/convert.go
@@ -137,6 +137,14 @@ func (c *Converter) Convert(ctx context.Context, w io.Writer, r fs.FS) error {
 		}
 	}
 
+	if !c.leaveAudio {
+		for i, f := range files {
+			if c.IsAudioFile(f.Name) {
+				fileAct[i] = FileActionIgnore
+			}
+		}
+	}
+
 	// we'll manually create the mimetype file
 	if i, ok := fileIdx["mimetype"]; ok {
 		fileAct[i] = FileActionIgnore
diff --git a/kepub/kepub.go b/kepub/kepub.go
index 2b6cd08..00d7a0e 100644
--- a/kepub/kepub.go
+++ b/kepub/kepub.go
@@ -25,6 +25,9 @@ type Converter struct {
 
 	// charset override
 	charset string // "auto" for auto-detection
+
+	// do not strip audio files (audiobook data)
+	leaveAudio bool
 }
 
 // ConverterOption configures a Converter.
@@ -44,6 +47,13 @@ func NewConverterWithOptions(opts ...ConverterOption) *Converter {
 	return c
 }
 
+// ConverterOptionLeaveAudio enables or disables leaving audio files in the book.
+func ConverterOptionLeaveAudio(leave bool) ConverterOption {
+	return func(c *Converter) {
+		c.leaveAudio = leave
+	}
+}
+
 // ConverterOptionSmartypants enables smart punctuation.
 func ConverterOptionSmartypants() ConverterOption {
 	return func(c *Converter) {
diff --git a/kepub/transform.go b/kepub/transform.go
index 08e760a..32670f3 100644
--- a/kepub/transform.go
+++ b/kepub/transform.go
@@ -50,6 +50,16 @@ func (c *Converter) TransformFileFilter(fn string) bool {
 	return false
 }
 
+func (c *Converter) IsAudioFile(fn string) bool {
+	switch path.Ext(fn) {
+	case ".mp3", ".m4a", ".m4b", ".m4p", ".m4r", ".m4v", ".mp4", ".aac", ".flac", ".ogg", ".oga", ".opus", ".webm", ".wav":
+		return true
+	case ".smil": // The media overlay
+		return true
+	}
+	return false
+}
+
 // TransformOPF transforms the OPF document for a KEPUB.
 //
 //  * [mandatory] add the cover-image property to the cover.