Print 7za/unrar output on failure

arch.go

@@ -48,7 +48,8 @@ func Compress(archPath, contentDir string) error {
 	)
 
 	log.Printf("compressing %s...", archPath)
-	if err := toolCmd.Run(); err != nil {
+	if output, err := toolCmd.CombinedOutput(); err != nil {
+		fmt.Println(string(output))
 		return errors.Join(fmt.Errorf("compression of %s failed", archPath), err)
 	}
 
@@ -91,7 +92,8 @@ func Decompress(archPath string, allocator *TempDirAllocator) (string, error) {
 	toolCmd.Dir = contentDir
 
 	log.Printf("decompressing %s...", archPath)
-	if err := toolCmd.Run(); err != nil {
+	if output, err := toolCmd.CombinedOutput(); err != nil {
+		fmt.Println(string(output))
 		return "", errors.Join(fmt.Errorf("decompression of %s failed", archPath), err)
 	}
 

media.go

@@ -2,19 +2,39 @@ package mex
 
 import (
 	"bytes"
+	"crypto/sha256"
+	_ "embed"
 	"errors"
 	"fmt"
 	"html/template"
-	"log"
+	"io"
 	"math"
 	"os"
 	"path/filepath"
 	"regexp"
+	"sort"
 	"strconv"
 	"strings"
 	"sync"
 )
 
+//go:embed regexp.txt
+var volumeExpStrs string
+
+func parseVolumeIndex(path string) *int {
+	for _, expStr := range strings.Split(volumeExpStrs, "\n") {
+		exp := regexp.MustCompile(expStr)
+		if matches := exp.FindStringSubmatch(filepath.Base(path)); len(matches) >= 2 {
+			if index, err := strconv.ParseInt(matches[1], 10, 32); err == nil {
+				indexInt := int(index)
+				return &indexInt
+			}
+		}
+	}
+
+	return nil
+}
+
 func isImagePath(path string) bool {
 	switch strings.ToLower(filepath.Ext(path)) {
 	case ".jpg", ".jpeg", ".png", ".webp", ".bmp", ".gif":
@@ -92,23 +112,13 @@ type Volume struct {
 	Book  *Book
 	Pages []*Page
 	Index int
-}
 
-func (self *Volume) AveragePageSize() int {
-	if len(self.Pages) == 0 {
-		return 0
-	}
-
-	var totalSize int
-	for _, page := range self.Pages {
-		totalSize += int(page.Node.Info.Size())
-	}
-
-	return totalSize / len(self.Pages)
+	avgSize int
+	hash    []byte
 }
 
 func (self *Volume) export(path string, config ExportConfig, allocator *TempDirAllocator) error {
-	name, err := buildTemplatedName(config.VolumeTemplate, stripExt(self.Node.Name), self.Index, self.Book.MaxVolume)
+	name, err := buildTemplatedName(config.VolumeTemplate, stripExt(self.Node.Name), self.Index, self.Book.VolumeCount)
 	if err != nil {
 		return err
 	}
@@ -145,24 +155,28 @@ func (self *Volume) export(path string, config ExportConfig, allocator *TempDirA
 	return nil
 }
 
-func (self *Volume) supercedes(other *Volume) bool {
+func (self *Volume) compare(other *Volume) int {
 	if len(self.Pages) > len(other.Pages) {
-		log.Printf("picking %s over %s because it has more pages", self.Node.Name, other.Node.Name)
-		return true
+		return 1
+	} else if len(self.Pages) < len(other.Pages) {
+		return -1
 	}
 
-	if self.AveragePageSize() > other.AveragePageSize() {
-		log.Printf("picking %s over %s because it has larger pages", self.Node.Name, other.Node.Name)
-		return true
+	if self.avgSize > other.avgSize {
+		return 1
+	} else if self.avgSize < other.avgSize {
+		return -1
 	}
 
-	return false
+	return bytes.Compare(self.hash, other.hash)
 }
 
 type Book struct {
 	Node        *Node
 	Volumes     map[int]*Volume
-	MaxVolume int
+	VolumeCount int
+
+	orphans []*Volume
 }
 
 func (self *Book) Export(path string, config ExportConfig, allocator *TempDirAllocator) error {
@@ -230,17 +244,42 @@ func (self *Book) Export(path string, config ExportConfig, allocator *TempDirAll
 	return nil
 }
 
-func (self *Book) addVolume(volume *Volume) {
-	currVolume, _ := self.Volumes[volume.Index]
-	if currVolume == nil || volume.supercedes(currVolume) {
-		self.Volumes[volume.Index] = volume
+func (self *Book) addVolume(newVolume *Volume) {
+	insert := func(v *Volume) {
+		self.Volumes[v.Index] = v
+		if v.Index >= self.VolumeCount {
+			self.VolumeCount = v.Index + 1
 		}
 	}
 
-func (self *Book) parseVolumes(node *Node) {
-	if !node.Info.IsDir() {
+	currVolume, _ := self.Volumes[newVolume.Index]
+	if currVolume == nil {
+		insert(newVolume)
+	} else {
+		switch currVolume.compare(newVolume) {
+		case 1:
+			self.addOrphan(newVolume)
+		case -1:
+			self.addOrphan(currVolume)
+			insert(newVolume)
+		}
+	}
+}
+
+func (self *Book) addOrphan(newVolume *Volume) {
+	for _, volume := range self.orphans {
+		if volume.compare(newVolume) == 0 {
 			return
 		}
+	}
+
+	self.orphans = append(self.orphans, newVolume)
+}
+
+func (self *Book) parseVolumes(node *Node) error {
+	if !node.Info.IsDir() {
+		return nil
+	}
 
 	volume := &Volume{
 		Node: node,
@@ -250,29 +289,55 @@ func (self *Book) parseVolumes(node *Node) {
 	var pageIndex int
 	for _, child := range node.Children {
 		if child.Info.IsDir() {
-			self.parseVolumes(child)
+			if err := self.parseVolumes(child); err != nil {
+				return err
+			}
 		} else if isImagePath(child.Name) {
 			volume.Pages = append(volume.Pages, &Page{child, volume, pageIndex})
 			pageIndex++
 		}
 	}
 
-	if len(volume.Pages) > 0 {
-		exp := regexp.MustCompile(`(\d+)\D*$`)
-		if matches := exp.FindStringSubmatch(node.Name); len(matches) >= 2 {
-			index, err := strconv.ParseInt(matches[1], 10, 32)
+	if len(volume.Pages) == 0 {
+		return nil
+	}
+
+	sort.Slice(volume.Pages, func(i, j int) bool {
+		return strings.Compare(volume.Pages[i].Node.Name, volume.Pages[j].Node.Name) < 0
+	})
+
+	var (
+		hasher    = sha256.New()
+		totalSize = 0
+	)
+
+	for _, page := range volume.Pages {
+		fp, err := os.Open(page.Node.Path)
 		if err != nil {
-				panic(err)
+			return err
 		}
 
-			volume.Index = int(index)
-			if volume.Index > self.MaxVolume {
-				self.MaxVolume = volume.Index
+		size, err := io.Copy(hasher, fp)
+		fp.Close()
+
+		if err != nil {
+			return err
 		}
 
+		totalSize += int(size)
+	}
+
+	volume.avgSize = totalSize / len(volume.Pages)
+	volume.hash = hasher.Sum(nil)
+
+	if index := parseVolumeIndex(node.Name); index != nil {
+		volume.Index = *index
 		self.addVolume(volume)
+	} else {
+		self.addOrphan(volume)
 	}
-	}
+
+	return nil
 }
 
 func ParseBook(node *Node) (*Book, error) {
@@ -283,6 +348,19 @@ func ParseBook(node *Node) (*Book, error) {
 
 	book.parseVolumes(node)
 
+	if len(book.orphans) > 0 {
+		sort.Slice(book.orphans, func(i, j int) bool {
+			return strings.Compare(book.orphans[i].Node.Name, book.orphans[j].Node.Name) < 0
+		})
+
+		for _, volume := range book.orphans {
+			volume.Index = book.VolumeCount
+			book.addVolume(volume)
+		}
+
+		book.orphans = nil
+	}
+
 	if len(book.Volumes) == 0 {
 		return nil, errors.New("no volumes found")
 	}

regexp.txt

@@ -0,0 +1,4 @@
+第\s*(\d+)\s*巻
+(?i)vo?l?u?m?e?[.\s]*(\d+)
+(?i)cha?p?t?e?r?[.\s]*(\d+)
+(\d+)\D*$