From d5971a2819780296869fbf2163c949fa156db071 Mon Sep 17 00:00:00 2001
From: lapwat <quentin@lapw.at>
Date: Sun, 10 Oct 2021 22:02:39 +0200
Subject: [PATCH] add threads option

---
 README.md        |  3 ++-
 book/progress.go |  2 +-
 book/scraper.go  | 17 ++++++++++++++---
 cmd/get.go       | 12 ++++++++++--
 cmd/root.go      |  1 +
 cmd/utils.go     |  5 -----
 6 files changed, 28 insertions(+), 12 deletions(-)
 delete mode 100644 cmd/utils.go

diff --git a/README.md b/README.md
index 734998c..8e144ae 100644
--- a/README.md
+++ b/README.md
@@ -75,8 +75,9 @@ Flags:
   -o, --offset int        skip first chapters, in recursive mode
       --output string     output file
   -r, --recursive         create one chapter per natigation item
-  -s, --selector string   table of content CSS selector
+  -s, --selector string   table of content CSS selector, in resursive mode
       --stdout            print to standard output
+  -t, --threads int       download concurrency, in recursive mode (default -1)
 
 Use "papeer [command] --help" for more information about a command.
 ```
diff --git a/book/progress.go b/book/progress.go
index f642092..1f30ee6 100644
--- a/book/progress.go
+++ b/book/progress.go
@@ -19,8 +19,8 @@ func NewProgress(links []link) progress {
 		return fmt.Sprintf("Chapters %d / %d", b.Current(), len(links))
 	})
 
-	individuals := []*uiprogress.Bar{}
 	// hide individual bars if more than 50 chapters
+	individuals := []*uiprogress.Bar{}
 	if len(links) <= 50 {
 		for index, link := range links {
 			bar := uiprogress.AddBar(1)
diff --git a/book/scraper.go b/book/scraper.go
index 6791e50..0bd74ed 100644
--- a/book/scraper.go
+++ b/book/scraper.go
@@ -14,9 +14,9 @@ import (
 	colly "github.com/gocolly/colly/v2"
 )
 
-func NewBookFromURL(url, selector string, recursive, include, images bool, limit, offset, delay int) book {
+func NewBookFromURL(url, selector string, recursive, include, images bool, limit, offset, delay, threads int) book {
 	if recursive {
-		chapters := tableOfContent(url, selector, limit, offset, delay, include, images)
+		chapters := tableOfContent(url, selector, limit, offset, delay, threads, include, images)
 
 		b := New(chapters[0].Name(), chapters[0].Author())
 		for _, c := range chapters {
@@ -56,7 +56,7 @@ func NewChapterFromURL(url string, images bool) chapter {
 	return chapter{article.Title, article.Byline, content}
 }
 
-func tableOfContent(url, selector string, limit, offset, delay int, include, images bool) []chapter {
+func tableOfContent(url, selector string, limit, offset, delay, threads int, include, images bool) []chapter {
 	base, err := urllib.Parse(url)
 	if err != nil {
 		log.Fatal(err)
@@ -71,6 +71,7 @@ func tableOfContent(url, selector string, limit, offset, delay int, include, ima
 	progress := NewProgress(links)
 
 	if delay >= 0 {
+		// synchronous mode
 
 		for index, link := range links {
 			// and then use it to parse relative URLs
@@ -91,10 +92,19 @@ func tableOfContent(url, selector string, limit, offset, delay int, include, ima
 		}
 
 	} else {
+		// asynchronous mode
 		var wg sync.WaitGroup
+
+		if threads == -1 {
+			threads = len(links)
+		}
+		semaphore := make(chan bool, threads)
+
 		for index, l := range links {
 
 			wg.Add(1)
+			semaphore <- true
+
 			go func(index int, l link) {
 				defer wg.Done()
 
@@ -107,6 +117,7 @@ func tableOfContent(url, selector string, limit, offset, delay int, include, ima
 				chapters[index] = NewChapterFromURL(u.String(), images)
 				progress.Incr(index)
 
+				<-semaphore
 			}(index, l)
 		}
 		wg.Wait()
diff --git a/cmd/get.go b/cmd/get.go
index dd8e54e..1772899 100644
--- a/cmd/get.go
+++ b/cmd/get.go
@@ -17,7 +17,7 @@ import (
 
 var stdout, recursive, include, images bool
 var format, output, selector string
-var limit, offset, delay int
+var limit, offset, delay, threads int
 
 var getCmd = &cobra.Command{
 	Use:   "get",
@@ -68,11 +68,19 @@ var getCmd = &cobra.Command{
 			return errors.New("cannot use delay option if not in recursive mode")
 		}
 
+		if cmd.Flags().Changed("threads") && recursive == false {
+			return errors.New("cannot use threads option if not in recursive mode")
+		}
+
+		if cmd.Flags().Changed("delay") && cmd.Flags().Changed("threads") {
+			return errors.New("cannot use delay and threads options at the same time")
+		}
+
 		return nil
 	},
 	Run: func(cmd *cobra.Command, args []string) {
 		url := args[0]
-		b := book.NewBookFromURL(url, selector, recursive, include, images, limit, offset, delay)
+		b := book.NewBookFromURL(url, selector, recursive, include, images, limit, offset, delay, threads)
 
 		if len(output) == 0 {
 			// set default output
diff --git a/cmd/root.go b/cmd/root.go
index 251f211..eba67e2 100644
--- a/cmd/root.go
+++ b/cmd/root.go
@@ -33,6 +33,7 @@ func init() {
 	rootCmd.PersistentFlags().IntVarP(&limit, "limit", "l", -1, "limit number of chapters, in recursive mode")
 	rootCmd.PersistentFlags().IntVarP(&offset, "offset", "o", 0, "skip first chapters, in recursive mode")
 	rootCmd.PersistentFlags().IntVarP(&delay, "delay", "d", -1, "time to wait before downloading next chapter, in milliseconds")
+	rootCmd.PersistentFlags().IntVarP(&threads, "threads", "t", -1, "download concurrency, in recursive mode")
 
 	rootCmd.AddCommand(getCmd)
 	rootCmd.AddCommand(listCmd)
diff --git a/cmd/utils.go b/cmd/utils.go
deleted file mode 100644
index 5411e9e..0000000
--- a/cmd/utils.go
+++ /dev/null
@@ -1,5 +0,0 @@
-package cmd
-
-func getTableOfContent() {
-
-}