🎨 Improve EPUB asset file content parsing siyuan-note/siyuan#9072

.gitignore

@@ -0,0 +1,15 @@
+# Binaries for programs and plugins
+*.exe
+*.exe~
+*.dll
+*.so
+*.dylib
+
+# Test binary, built with `go test -c`
+*.test
+
+# Output of the go coverage tool, specifically when used with LiteIDE
+*.out
+
+# IDE
+.idea

book.go

@@ -5,8 +5,8 @@ import (
 	"encoding/xml"
 	"errors"
 	"io"
-	"io/ioutil"
 	"path"
+	"strings"
 )
 
 type Book struct {
@@ -55,7 +55,7 @@ func (p *Book) readBytes(n string) ([]byte, error) {
 	}
 	defer fd.Close()
 
-	return ioutil.ReadAll(fd)
+	return io.ReadAll(fd)
 
 }
 
@@ -65,5 +65,21 @@ func (p *Book) open(n string) (io.ReadCloser, error) {
 			return f.Open()
 		}
 	}
-	return nil, errors.New(n + " not found!")
+
+	if !strings.ContainsAny(n, "-_") {
+		return nil, errors.New(n + " not found")
+	}
+
+	// Improve EPUB asset file content parsing https://github.com/siyuan-note/siyuan/issues/9072
+	nn := strings.ReplaceAll(n, "-", "")
+	nn = strings.ReplaceAll(nn, "_", "")
+	for _, f := range p.fd.File {
+		fn := strings.ReplaceAll(f.Name, "-", "")
+		fn = strings.ReplaceAll(fn, "_", "")
+
+		if fn == nn {
+			return f.Open()
+		}
+	}
+	return nil, errors.New(n + " not found")
 }

epub.go

@@ -3,10 +3,8 @@ package epub
 import (
 	"archive/zip"
 	"bytes"
-	"io"
-	"io/ioutil"
-
 	"github.com/wmentor/html"
+	"io"
 )
 
 func Open(fn string) (*Book, error) {
@@ -58,36 +56,31 @@ func Reader(filename string, onChapter func(chapter string, data []byte) bool) e
 			return nil, err
 		}
 		defer fd.Close()
-
-		return ioutil.ReadAll(fd)
+		return io.ReadAll(fd)
 	}
 
 	for _, pt := range bk.Ncx.Points {
 		for _, np := range pt.Points {
 
 			name := np.Text
-
 			data, err := readerF(np.Content.Src)
 			if err != nil {
-				return err
+				// Improve EPUB asset file content parsing https://github.com/siyuan-note/siyuan/issues/9072
+				// Ignore error
+				return nil
 			}
 
 			if !onChapter(name, data) {
 				return nil
 			}
-
 		}
 	}
-
 	return nil
 }
 
 func ToTxt(filename string, output io.Writer) error {
-
 	notFirst := false
-
 	return Reader(filename, func(chapter string, data []byte) bool {
-
 		parser := html.New()
 		parser.Parse(bytes.NewReader(data))
 
@@ -98,7 +91,6 @@ func ToTxt(filename string, output io.Writer) error {
 		}
 
 		output.Write(parser.Text())
-
 		return true
 	})
 }

epub_test.go

@@ -6,7 +6,9 @@ import (
 )
 
 func TestReader(t *testing.T) {
-	bk, err := Open("./data/test.epub")
+	fn := "./data/test.epub"
+
+	bk, err := Open(fn)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -19,7 +21,7 @@ func TestReader(t *testing.T) {
 
 	i := 0
 
-	Reader("./data/test.epub", func(n string, data []byte) bool {
+	Reader(fn, func(n string, data []byte) bool {
 		i++
 		if data == nil {
 			t.Fatal("reader failed")
@@ -33,7 +35,7 @@ func TestReader(t *testing.T) {
 
 	buf := bytes.NewBuffer(nil)
 
-	ToTxt("./data/test.epub", buf)
+	ToTxt(fn, buf)
 
 	if buf.Len() == 0 {
 		t.Fatal("ToTxt failed")