[get] print url option

Add arm64 build actions for darwin/macOS and linux.

.github/workflows/release.yml

@@ -9,10 +9,14 @@ jobs:
     strategy:
       matrix:
         goos: [linux, darwin, windows]
+        goarch: [amd64, arm64]
+        exclude:
+            - goarch: arm64
+              goos: windows
     steps:
     - uses: actions/checkout@v3
-    - uses: wangyoucao577/go-release-action@v1.30
+    - uses: wangyoucao577/go-release-action@v1.40
       with:
         github_token: ${{ secrets.GITHUB_TOKEN }}
         goos: ${{ matrix.goos }}
-        goarch: amd64
+        goarch: ${{ matrix.goarch }}

README.md

@@ -40,7 +40,9 @@ Download [latest release](https://github.com/lapwat/papeer/releases/latest) for
 
 ## MOBI support
 
-Install kindlegen to convert websites, Linux only.
+> Kindle e-readers now support EPUB format
+
+Install kindlegen to export websites to Kindle compatible ebooks, Linux only.
 
 ```sh
 TMPDIR=$(mktemp -d -t papeer-XXXXX)

book/chapter.go

@@ -1,6 +1,7 @@
 package book
 
 type chapter struct {
+	url         string
 	body        string
 	name        string
 	author      string
@@ -10,11 +11,11 @@ type chapter struct {
 }
 
 func NewEmptyChapter() chapter {
-	return chapter{"", "", "", "", []chapter{}, NewScrapeConfigNoInclude()}
+	return chapter{"", "", "", "", "", []chapter{}, NewScrapeConfigNoInclude()}
 }
 
-func NewChapter(body, name, author, content string, subChapters []chapter, config *ScrapeConfig) chapter {
-	return chapter{body, name, author, content, subChapters, config}
+func (c chapter) URL() string {
+	return c.url
 }
 
 func (c chapter) Body() string {

book/format.go

@@ -30,6 +30,11 @@ func ToMarkdownString(c chapter) string {
 		markdown += fmt.Sprintf("%s\n", c.Name())
 		markdown += fmt.Sprintf("%s\n\n", strings.Repeat("=", len(c.Name())))
 
+		// url
+		if c.config.PrintURL {
+			markdown += fmt.Sprintf("_Source: %s_\n\n", c.URL())
+		}
+
 		// convert content to markdown
 		content, err := md.NewConverter("", true, nil).ConvertString(c.Content())
 		if err != nil {
@@ -72,7 +77,15 @@ func ToHtmlString(c chapter) string {
 
 	// chapter content
 	if c.config.Include {
-		html += fmt.Sprintf("<h1>%s</h1>", c.Name())
+		// title
+		html += fmt.Sprintf("<h1>%s</h1>\n", c.Name())
+
+		// url
+		if c.config.PrintURL {
+			html += fmt.Sprintf("<p><i>Source: %s</i></p>\n", c.URL())
+		}
+
+		// content
 		html += c.Content()
 	}
 
@@ -114,19 +127,6 @@ func ToEpub(c chapter, filename string) string {
 	e := epub.NewEpub(c.Name())
 	e.SetAuthor(c.Author())
 
-	AppendToEpub(e, c)
-
-	err := e.Write(filename)
-	if err != nil {
-		log.Fatal(err)
-	}
-
-	return filename
-}
-
-func AppendToEpub(e *epub.Epub, c chapter) {
-	content := ""
-
 	// append table of content
 	if len(c.SubChapters()) > 1 {
 		html := "<h1>Table of Contents</h1>"
@@ -143,6 +143,19 @@ func AppendToEpub(e *epub.Epub, c chapter) {
 		}
 	}
 
+	AppendToEpub(e, c)
+
+	err := e.Write(filename)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	return filename
+}
+
+func AppendToEpub(e *epub.Epub, c chapter) {
+	content := ""
+
 	// chapter content
 	if c.config.Include {
 
@@ -173,8 +186,15 @@ func AppendToEpub(e *epub.Epub, c chapter) {
 		html := ""
 		// add title only if ImagesOnly = false
 		if c.config.ImagesOnly == false {
-			html += fmt.Sprintf("<h1>%s</h1>", c.Name())
+			html += fmt.Sprintf("<h1>%s</h1>\n", c.Name())
 		}
+
+		// url
+		if c.config.PrintURL {
+			html += fmt.Sprintf("<p><i>Source: %s</i></p>\n", c.URL())
+		}
+
+		// content
 		html += content
 
 		//  write to epub file

book/link.go

@@ -1,10 +1,13 @@
 package book
 
+import "time"
+
 type link struct {
-	Href string `json:"url"`
-	Text string `json:"name"`
+	Href string     `json:"url"`
+	Text string     `json:"name"`
+	Date *time.Time `json:"date"`
 }
 
-func NewLink(href, text string) link {
-	return link{href, text}
+func NewLink(href, text string, date *time.Time) link {
+	return link{href, text, date}
 }

book/scraper.go

@@ -30,14 +30,19 @@ type ScrapeConfig struct {
 	Include     bool
 	ImagesOnly  bool
 	UseLinkName bool
+	PrintURL    bool
 }
 
 func NewScrapeConfig() *ScrapeConfig {
-	return &ScrapeConfig{0, "", false, -1, 0, false, -1, -1, true, false, false}
+	return &ScrapeConfig{0, "", false, -1, 0, false, -1, -1, true, false, false, false}
+}
+
+func NewScrapeConfigQuiet() *ScrapeConfig {
+	return &ScrapeConfig{0, "", true, -1, 0, false, -1, -1, true, false, false, false}
 }
 
 func NewScrapeConfigNoInclude() *ScrapeConfig {
-	return &ScrapeConfig{0, "", false, -1, 0, false, -1, -1, false, false, false}
+	return &ScrapeConfig{0, "", false, -1, 0, false, -1, -1, false, false, false, false}
 }
 
 func NewScrapeConfigs(selectors []string) []*ScrapeConfig {
@@ -252,7 +257,7 @@ func NewChapterFromURL(url, linkName string, configs []*ScrapeConfig, index int,
 
 	}
 
-	return chapter{string(body), name, article.Byline, content, subchapters, config}
+	return chapter{url, string(body), name, article.Byline, content, subchapters, config}
 }
 
 func tableOfContent(url string, config *ScrapeConfig, subConfig *ScrapeConfig, quiet bool) ([]chapter, chapter) {
@@ -370,7 +375,7 @@ func GetLinks(url *urllib.URL, selector string, limit, offset int, reverse, incl
 				log.Fatal(err)
 			}
 
-			links = append(links, NewLink(u.String(), item.Title))
+			links = append(links, NewLink(u.String(), item.Title, item.PublishedParsed))
 		}
 
 		pathMax = "RSS"
@@ -405,7 +410,7 @@ func GetLinks(url *urllib.URL, selector string, limit, offset int, reverse, incl
 				// if selector is set, we use the selector specified by the user
 
 				key = selector
-				pathLinks[key] = append(pathLinks[key], NewLink(href, text))
+				pathLinks[key] = append(pathLinks[key], NewLink(href, text, &time.Time{}))
 				pathCount[key] += 1
 				pathMax = key
 
@@ -419,7 +424,7 @@ func GetLinks(url *urllib.URL, selector string, limit, offset int, reverse, incl
 
 				// we count this key if the link text is not empty
 				if text != "" {
-					pathLinks[key] = append(pathLinks[key], NewLink(href, text))
+					pathLinks[key] = append(pathLinks[key], NewLink(href, text, &time.Time{}))
 					pathCount[key] += len(text)
 
 					if pathCount[key] > pathCount[pathMax] {
@@ -449,7 +454,7 @@ func GetLinks(url *urllib.URL, selector string, limit, offset int, reverse, incl
 
 	// include home page
 	if include {
-		l := NewLink(url.String(), home.Name())
+		l := NewLink(url.String(), home.Name(), &time.Time{})
 		links = append([]link{l}, links...)
 	}
 

cmd/get.go

@@ -1,6 +1,7 @@
 package cmd
 
 import (
+	"encoding/json"
 	"errors"
 	"fmt"
 	"io/ioutil"
@@ -32,6 +33,7 @@ type GetOptions struct {
 	threads     int
 	include     bool
 	useLinkName bool
+	printURL    bool
 }
 
 var getOpts *GetOptions
@@ -41,10 +43,11 @@ func init() {
 
 	getCmd.Flags().StringVarP(&getOpts.name, "name", "n", "", "book name (default: page title)")
 	getCmd.Flags().StringVarP(&getOpts.author, "author", "a", "", "book author")
-	getCmd.Flags().StringVarP(&getOpts.Format, "format", "f", "md", "file format [md, html, epub, mobi]")
+	getCmd.Flags().StringVarP(&getOpts.Format, "format", "f", "md", "file format [md, json, html, epub, mobi]")
 	getCmd.Flags().StringVarP(&getOpts.output, "output", "", "", "file name (default: book name)")
 	getCmd.Flags().BoolVarP(&getOpts.stdout, "stdout", "", false, "print to standard output")
 	getCmd.Flags().BoolVarP(&getOpts.images, "images", "", false, "retrieve images only")
+	getCmd.Flags().BoolVarP(&getOpts.printURL, "print-url", "", false, "print url after chapter title")
 	getCmd.Flags().BoolVarP(&getOpts.quiet, "quiet", "q", false, "hide progress bar")
 
 	// common with list command
@@ -73,6 +76,7 @@ var getCmd = &cobra.Command{
 		// check provided format is in list
 		formatEnum := map[string]bool{
 			"md":   true,
+			"json": true,
 			"html": true,
 			"epub": true,
 			"mobi": true,
@@ -145,6 +149,7 @@ var getCmd = &cobra.Command{
 			config.ImagesOnly = getOpts.images
 			config.Include = getOpts.include
 			config.UseLinkName = getOpts.useLinkName
+			config.PrintURL = getOpts.printURL
 
 			// do not use link name for root level as there is not parent link
 			if index == 0 {
@@ -182,6 +187,26 @@ var getCmd = &cobra.Command{
 			}
 		}
 
+		if getOpts.Format == "json" {
+			filename := book.ToMarkdown(c, getOpts.output)
+
+			bytesRead, err := ioutil.ReadFile(filename)
+			if err != nil {
+				log.Fatal(err)
+			}
+
+			book := make(map[string]interface{})
+			book["name"] = c.Name()
+			book["content"] = string(bytesRead)
+
+			bookJson, err := json.Marshal(book)
+			if err != nil {
+				log.Fatal(err)
+			}
+
+			fmt.Println(string(bookJson))
+		}
+
 		if getOpts.Format == "html" {
 			filename := book.ToHtml(c, getOpts.output)
 

cmd/version.go

@@ -14,6 +14,6 @@ var versionCmd = &cobra.Command{
 	Use:   "version",
 	Short: "Print the version number of papeer",
 	Run: func(cmd *cobra.Command, args []string) {
-		fmt.Println("papeer v0.7.1")
+		fmt.Println("papeer v0.8.2")
 	},
 }