feat: read epub file (#2)

.github/workflows/build.yml

@@ -2,8 +2,6 @@ name: Build
 
 on:
   push:
-    branches:
-      - 'main'
 
 jobs:
   test:
@@ -24,6 +22,7 @@ jobs:
       run: go test -v ./...
   release:
     needs: [test]
+    if: github.ref == 'refs/heads/main'
     name: create release
     runs-on: ubuntu-latest
     steps:

.github/workflows/codeql-analysis.yml

@@ -2,10 +2,6 @@ name: "CodeQL"
 
 on:
   push:
-  pull_request:
-    branches:
-      - 'dev'
-      - 'main'
   schedule:
     - cron: '42 15 * * 1'
 jobs:

README.md

@@ -0,0 +1,21 @@
+# Epub parser
+
+This parser only supports epub version 3.0 so far.
+This parser also only reads the MetaData.
+
+### How to use?
+```go
+zipReader, err := zip.OpenReader("./test_epub_v3_0.epub")
+if err != nil {
+    t.Log(err.Error())
+    t.Fail()
+}
+defer zipReader.Close()
+
+book, err := OpenBook(zipReader)
+if err != nil {
+    t.Log(err.Error())
+    t.Fail()
+}
+```
+After that, all metadata can be found in the book Object

book.go

@@ -0,0 +1,58 @@
+package epub
+
+import (
+	"archive/zip"
+	"encoding/xml"
+	"fmt"
+	"io"
+	"path"
+)
+
+type Book struct {
+	Opf       Package
+	Container Container
+	ZipReader *zip.ReadCloser
+}
+
+func OpenBook(reader *zip.ReadCloser) (*Book, error) {
+	book := &Book{ZipReader: reader}
+	err := book.readXML("META-INF/container.xml", &book.Container)
+	if err != nil {
+		return nil, err
+	}
+	err = book.readXML(book.Container.Rootfile.Path, &book.Opf)
+	if err != nil {
+		return nil, err
+	}
+	if book.Opf.Version != "3.0" {
+		return nil, fmt.Errorf("%s not supported yet!", book.Opf.Version)
+	}
+	return book, nil
+}
+
+func (book *Book) Open(fileName string) (io.ReadCloser, error) {
+	return book.open(book.getFileFromRootPath(fileName))
+}
+
+func (book *Book) getFileFromRootPath(fileName string) string {
+	return path.Join(path.Dir(book.Container.Rootfile.Path), fileName)
+}
+
+func (book *Book) readXML(fileName string, targetStruct interface{}) error {
+	reader, err := book.open(fileName)
+	if err != nil {
+		return err
+	}
+	defer reader.Close()
+	dec := xml.NewDecoder(reader)
+	return dec.Decode(targetStruct)
+}
+
+func (book *Book) open(fileName string) (io.ReadCloser, error) {
+	for _, file := range book.ZipReader.File {
+		if file.Name == fileName {
+			return file.Open()
+		}
+	}
+	return nil, fmt.Errorf("file %s not exist", fileName)
+}

container.go

@@ -0,0 +1,10 @@
+package epub
+
+type Container struct {
+	Rootfile Rootfile `xml:"rootfiles>rootfile"`
+}
+
+type Rootfile struct {
+	Path string `xml:"full-path,attr"`
+	Type string `xml:"media-type,attr"`
+}

parser_test.go

@@ -1,26 +1,26 @@
 package epub
 
 import (
-	"io/ioutil"
-	"os"
+	"archive/zip"
 	"strconv"
 	"testing"
 )
 
 func Test_parse_epub_3_0_opf(t *testing.T) {
-	b, err := os.Open("./test_opf.xml")
+	zipReader, err := zip.OpenReader("./test_epub_v3_0.epub")
 	if err != nil {
 		t.Log(err.Error())
 		t.Fail()
 	}
-	defer b.Close()
-	reader := ioutil.NopCloser(b)
-	defer reader.Close()
-	opf, err := ReadOpf(reader)
+	defer zipReader.Close()
+
+	book, err := OpenBook(zipReader)
 	if err != nil {
 		t.Log(err.Error())
 		t.Fail()
 	}
+	opf := book.Opf
+
 	assertEquals(t, opf.Lang, "us")
 	assertEquals(t, opf.UniqueIdentifier, "p1234")
 	assertEquals(t, opf.Version, "3.0")