fix XML parsing (#2)

- Use antchfx/xmlquery for easier XML parsing. No more regexp wrangling and the result is much more reliable over a variety of ebooks. Much good.
- fix chapter selection, look for `<?xml[...]` which is much more reliable
- add option `-x` to dump the XML ebook source for debugging

cmd/config.go

@@ -16,7 +16,7 @@ import (
 )
 
 const (
-	Version string = `v0.0.2`
+	Version string = `v0.0.3`
 	Usage   string = `Usage epuppy [options] <epub file>
 
 Options:
@@ -25,6 +25,7 @@ Options:
 -n --line-numbers        add line numbers
 -c --config <file>       use config <file>
 -t --txt                 dump readable content to STDOUT
+-x --xml                 dump source xml to STDOUT
 -d --debug               enable debugging
 -h --help                show help message
 -v --version             show program version`
@@ -37,6 +38,7 @@ type Config struct {
 	Darkmode      bool         `koanf:"dark"`           // -D
 	LineNumbers   bool         `koanf:"line-numbers"`   // -n
 	Dump          bool         `koanf:"txt"`            // -t
+	XML           bool         `koanf:"xml"`            // -x
 	Config        string       `koanf:"config"`         // -c
 	ColorDark     ColorSetting `koanf:"colordark"`      // comes from config file only
 	ColorLight    ColorSetting `koanf:"colorlight"`     // comes from config file only
@@ -66,6 +68,7 @@ func InitConfig(output io.Writer) (*Config, error) {
 	flagset.BoolP("store-progress", "s", false, "store reading progress")
 	flagset.BoolP("line-numbers", "n", false, "add line numbers")
 	flagset.BoolP("txt", "t", false, "dump readable content to STDOUT")
+	flagset.BoolP("xml", "x", false, "dump xml to STDOUT")
 	flagset.StringP("config", "c", "", "read config from file")
 
 	if err := flagset.Parse(os.Args[1:]); err != nil {

cmd/view.go

@@ -2,7 +2,6 @@ package cmd
 
 import (
 	"fmt"
-	"log"
 	"os"
 	"path/filepath"
 	"strings"
@@ -33,17 +32,11 @@ func ViewText(conf *Config) (int, error) {
 }
 
 func ViewEpub(conf *Config) (int, error) {
-	book, err := epub.Open(conf.Document)
+	book, err := epub.Open(conf.Document, conf.XML)
 	if err != nil {
 		return 0, err
 	}
 
-	defer func() {
-		if err := book.Close(); err != nil {
-			log.Fatal(err)
-		}
-	}()
-
 	buf := strings.Builder{}
 	head := strings.Builder{}
 
@@ -59,9 +52,20 @@ func ViewEpub(conf *Config) (int, error) {
 		head.WriteString(" ")
 	}
 
-	// FIXME:  since the  switch to  book.Files() in  epub.Open() this
-	// returns invalid chapter numbering
+	fetchByContent(conf, &buf, book)
+
+	if conf.Dump {
+		return fmt.Println(buf.String())
+	}
+
+	return Pager(conf, head.String(), buf.String())
+}
+
+// FIXME:  since the  switch to  book.Files() in  epub.Open() this
+// returns invalid chapter numbering
+func fetchByContent(conf *Config, buf *strings.Builder, book *epub.Book) bool {
 	chapter := 1
+	var gotbody bool
 
 	for _, content := range book.Content {
 		if len(content.Body) > 0 {
@@ -81,13 +85,10 @@ func ViewEpub(conf *Config) (int, error) {
 
 			buf.WriteString("\r\n\r\n\r\n\r\n")
 			chapter++
+
+			gotbody = true
 		}
-
 	}
 
-	if conf.Dump {
-		return fmt.Println(buf.String())
-	}
-
-	return Pager(conf, head.String(), buf.String())
+	return gotbody
 }