initial commit

2025-12-16 20:11:00 +01:00 · 2025-10-15 00:54:19 +02:00
parent 97c7383cf1
commit 0d4c44ee11
14 changed files with 1242 additions and 0 deletions
--- a/pkg/epub/book.go
+++ b/pkg/epub/book.go
@@ -0,0 +1,73 @@
+package epub
+
+import (
+	"archive/zip"
+	"encoding/xml"
+	"fmt"
+	"io"
+	"path"
+)
+
+// Book epub book
+type Book struct {
+	Ncx       Ncx       `json:"ncx"`
+	Opf       Opf       `json:"opf"`
+	Container Container `json:"-"`
+	Mimetype  string    `json:"-"`
+
+	fd *zip.ReadCloser
+}
+
+// Open open resource file
+func (p *Book) Open(n string) (io.ReadCloser, error) {
+	return p.open(p.filename(n))
+}
+
+// Files list resource files
+func (p *Book) Files() []string {
+	var fns []string
+	for _, f := range p.fd.File {
+		fns = append(fns, f.Name)
+	}
+	return fns
+}
+
+// Close close file reader
+func (p *Book) Close() {
+	p.fd.Close()
+}
+
+// -----------------------------------------------------------------------------
+func (p *Book) filename(n string) string {
+	return path.Join(path.Dir(p.Container.Rootfile.Path), n)
+}
+
+func (p *Book) readXML(n string, v interface{}) error {
+	fd, err := p.open(n)
+	if err != nil {
+		return nil
+	}
+	defer fd.Close()
+	dec := xml.NewDecoder(fd)
+	return dec.Decode(v)
+}
+
+func (p *Book) readBytes(n string) ([]byte, error) {
+	fd, err := p.open(n)
+	if err != nil {
+		return nil, nil
+	}
+	defer fd.Close()
+
+	return io.ReadAll(fd)
+
+}
+
+func (p *Book) open(n string) (io.ReadCloser, error) {
+	for _, f := range p.fd.File {
+		if f.Name == n {
+			return f.Open()
+		}
+	}
+	return nil, fmt.Errorf("file %s not exist", n)
+}
--- a/pkg/epub/container.go
+++ b/pkg/epub/container.go
@@ -0,0 +1,12 @@
+package epub
+
+//Container META-INF/container.xml file
+type Container struct {
+	Rootfile Rootfile `xml:"rootfiles>rootfile" json:"rootfile"`
+}
+
+//Rootfile root file
+type Rootfile struct {
+	Path string `xml:"full-path,attr" json:"path"`
+	Type string `xml:"media-type,attr" json:"type"`
+}
--- a/pkg/epub/epub_test.go
+++ b/pkg/epub/epub_test.go
@@ -0,0 +1,28 @@
+package epub_test
+
+import (
+	"testing"
+
+	"github.com/kapmahc/epub"
+)
+
+func TestEpub(t *testing.T) {
+	bk, err := open(t, "test.epub")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer bk.Close()
+}
+
+func open(t *testing.T, f string) (*epub.Book, error) {
+	bk, err := epub.Open(f)
+	if err != nil {
+		return nil, err
+	}
+	defer bk.Close()
+
+	t.Logf("files: %+v", bk.Files())
+	t.Logf("book: %+v", bk)
+
+	return bk, nil
+}
--- a/pkg/epub/ncx.go
+++ b/pkg/epub/ncx.go
@@ -0,0 +1,71 @@
+package epub
+
+import (
+	"encoding/xml"
+	"regexp"
+	"strings"
+)
+
+var (
+	cleantitle    = regexp.MustCompile(`(?s)<head>.*</head>`)
+	cleanmarkup   = regexp.MustCompile(`<[^<>]+>`)
+	cleanentities = regexp.MustCompile(`&.+;`)
+	cleancomments = regexp.MustCompile(`/*.*/`)
+	cleanspace    = regexp.MustCompile(`^\s*`)
+	cleanh1       = regexp.MustCompile(`<h[1-6].*</h[1-6]>`)
+)
+
+// Ncx OPS/toc.ncx
+type Ncx struct {
+	Points []*NavPoint `xml:"navMap>navPoint" json:"points"`
+}
+
+// NavPoint nav point
+type NavPoint struct {
+	Text    string     `xml:"navLabel>text" json:"text"`
+	Content Content    `xml:"content" json:"content"`
+	Points  []NavPoint `xml:"navPoint" json:"points"`
+}
+
+type Title struct {
+	Content string `xml:"head>title"`
+}
+
+// Content nav-point content
+type Content struct {
+	Src   string `xml:"src,attr" json:"src"`
+	Empty bool
+	Body  string
+	Title string
+	XML   []byte
+}
+
+func (c *Content) String(content []byte) error {
+	title := Title{}
+
+	err := xml.Unmarshal(content, &title)
+	if err != nil {
+		if !strings.HasPrefix(err.Error(), "XML syntax error") {
+			return err
+		}
+	}
+
+	c.Title = title.Content
+
+	txt := cleantitle.ReplaceAllString(string(content), "")
+	txt = cleanh1.ReplaceAllString(txt, "")
+	txt = cleanmarkup.ReplaceAllString(txt, "")
+	txt = cleanentities.ReplaceAllString(txt, " ")
+	txt = cleancomments.ReplaceAllString(txt, "")
+
+	txt = strings.TrimSpace(txt)
+
+	c.Body = cleanspace.ReplaceAllString(txt, "")
+	c.XML = content
+
+	if len(c.Body) == 0 {
+		c.Empty = true
+	}
+
+	return nil
+}
--- a/pkg/epub/open.go
+++ b/pkg/epub/open.go
@@ -0,0 +1,53 @@
+package epub
+
+import (
+	"archive/zip"
+)
+
+// Open open a epub file
+func Open(fn string) (*Book, error) {
+	fd, err := zip.OpenReader(fn)
+	if err != nil {
+		return nil, err
+	}
+
+	defer fd.Close()
+
+	bk := Book{fd: fd}
+	mt, err := bk.readBytes("mimetype")
+	if err != nil {
+		return &bk, err
+	}
+
+	bk.Mimetype = string(mt)
+
+	err = bk.readXML("META-INF/container.xml", &bk.Container)
+	if err != nil {
+		return &bk, err
+	}
+
+	err = bk.readXML(bk.Container.Rootfile.Path, &bk.Opf)
+	if err != nil {
+		return &bk, err
+	}
+
+	for _, mf := range bk.Opf.Manifest {
+		if mf.ID == bk.Opf.Spine.Toc {
+			err = bk.readXML(bk.filename(mf.Href), &bk.Ncx)
+			break
+		}
+	}
+
+	for _, ncx := range bk.Ncx.Points {
+		content, err := bk.readBytes(bk.filename(ncx.Content.Src))
+		if err != nil {
+			return &bk, err
+		}
+
+		if err := ncx.Content.String(content); err != nil {
+			return &bk, err
+		}
+	}
+
+	return &bk, nil
+}
--- a/pkg/epub/opf.go
+++ b/pkg/epub/opf.go
@@ -0,0 +1,80 @@
+package epub
+
+//Opf content.opf
+type Opf struct {
+	Metadata Metadata   `xml:"metadata" json:"metadata"`
+	Manifest []Manifest `xml:"manifest>item" json:"manifest"`
+	Spine    Spine      `xml:"spine" json:"spine"`
+}
+
+//Metadata metadata
+type Metadata struct {
+	Title       []string     `xml:"title" json:"title"`
+	Language    []string     `xml:"language" json:"language"`
+	Identifier  []Identifier `xml:"identifier" json:"identifier"`
+	Creator     []Author     `xml:"creator" json:"creator"`
+	Subject     []string     `xml:"subject" json:"subject"`
+	Description []string     `xml:"description" json:"description"`
+	Publisher   []string     `xml:"publisher" json:"publisher"`
+	Contributor []Author     `xml:"contributor" json:"contributor"`
+	Date        []Date       `xml:"date" json:"date"`
+	Type        []string     `xml:"type" json:"type"`
+	Format      []string     `xml:"format" json:"format"`
+	Source      []string     `xml:"source" json:"source"`
+	Relation    []string     `xml:"relation" json:"relation"`
+	Coverage    []string     `xml:"coverage" json:"coverage"`
+	Rights      []string     `xml:"rights" json:"rights"`
+	Meta        []Metafield  `xml:"meta" json:"meta"`
+}
+
+// Identifier identifier
+type Identifier struct {
+	Data   string `xml:",chardata" json:"data"`
+	ID     string `xml:"id,attr" json:"id"`
+	Scheme string `xml:"scheme,attr" json:"scheme"`
+}
+
+// Author author
+type Author struct {
+	Data   string `xml:",chardata" json:"author"`
+	FileAs string `xml:"file-as,attr" json:"file_as"`
+	Role   string `xml:"role,attr" json:"role"`
+}
+
+// Date date
+type Date struct {
+	Data  string `xml:",chardata" json:"data"`
+	Event string `xml:"event,attr" json:"event"`
+}
+
+// Metafield metafield
+type Metafield struct {
+	Name    string `xml:"name,attr" json:"name"`
+	Content string `xml:"content,attr" json:"content"`
+}
+
+//Manifest manifest
+type Manifest struct {
+	ID           string `xml:"id,attr" json:"id"`
+	Href         string `xml:"href,attr" json:"href"`
+	MediaType    string `xml:"media-type,attr" json:"type"`
+	Fallback     string `xml:"media-fallback,attr" json:"fallback"`
+	Properties   string `xml:"properties,attr" json:"properties"`
+	MediaOverlay string `xml:"media-overlay,attr" json:"overlay"`
+}
+
+// Spine spine
+type Spine struct {
+	ID              string      `xml:"id,attr" json:"id"`
+	Toc             string      `xml:"toc,attr" json:"toc"`
+	PageProgression string      `xml:"page-progression-direction,attr" json:"progression"`
+	Items           []SpineItem `xml:"itemref" json:"items"`
+}
+
+// SpineItem spine item
+type SpineItem struct {
+	IDref      string `xml:"idref,attr" json:"id_ref"`
+	Linear     string `xml:"linear,attr" json:"linear"`
+	ID         string `xml:"id,attr" json:"id"`
+	Properties string `xml:"properties,attr" json:"properties"`
+}