@@ -7,16 +7,15 @@ import (
77 "html/template"
88 "io"
99 "log"
10+ "math"
1011 "net/url"
1112 "strings"
13+ "time"
1214
1315 "github.com/everywall/ladder/proxychain"
14-
16+ "github.com/markusmobius/go-trafilatura"
1517 "golang.org/x/net/html"
1618 "golang.org/x/net/html/atom"
17-
18- //"github.com/go-shiori/dom"
19- "github.com/markusmobius/go-trafilatura"
2019)
2120
2221//go:embed vendor/generate_readable_outline.html
@@ -63,19 +62,24 @@ func GenerateReadableOutline() proxychain.ResponseModification {
6362 html .Render (& b , extract .ContentNode )
6463 distilledHTML := b .String ()
6564
65+ siteName := strings .Split (extract .Metadata .Sitename , ";" )[0 ]
66+ title := strings .Split (extract .Metadata .Title , "|" )[0 ]
67+ fmtDate := createWikipediaDateLink (extract .Metadata .Date )
68+ readingTime := formatDuration (estimateReadingTime (extract .ContentText ))
69+
6670 // populate template parameters
6771 data := map [string ]interface {}{
6872 "Success" : true ,
6973 "Image" : extract .Metadata .Image ,
7074 "Description" : extract .Metadata .Description ,
71- "Sitename" : strings . Split ( extract . Metadata . Sitename , ";" )[ 0 ] ,
75+ "Sitename" : siteName ,
7276 "Hostname" : extract .Metadata .Hostname ,
7377 "Url" : "/" + chain .Request .URL .String (),
74- "Title" : extract . Metadata . Title , // todo: modify CreateReadableDocument so we don't have <h1> titles duplicated?
75- "Date" : extract . Metadata . Date . String () ,
76- "Author" : createWikipediaSearchLinks (extract .Metadata .Author ),
77- //"Author ": extract.Metadata.Author ,
78- "Body " : distilledHTML ,
78+ "Title" : title ,
79+ "Date" : fmtDate ,
80+ "Author" : createDDGFeelingLuckyLinks (extract .Metadata .Author , extract . Metadata . Hostname ),
81+ "Body " distilledHTML ,
82+ "ReadingTime " : readingTime ,
7983 }
8084
8185 // ============================================================================
@@ -157,9 +161,20 @@ func rewriteHrefLinks(n *html.Node, baseURL string, apiPath string) {
157161 recurse (n )
158162}
159163
160- // createWikipediaSearchLinks takes in comma or semicolon separated terms,
161- // then turns them into <a> links searching for the term.
162- func createWikipediaSearchLinks (searchTerms string ) string {
164+ // createWikipediaDateLink takes in a date
165+ // and returns an <a> link pointing to the current events page for that day
166+ func createWikipediaDateLink (t time.Time ) string {
167+ url := fmt .Sprintf ("https://en.wikipedia.org/wiki/Portal:Current_events#%s" , t .Format ("2006_January_2" ))
168+ date := t .Format ("January 2, 2006" )
169+ return fmt .Sprintf ("<a rel=\" noreferrer\" href=\" %s\" >%s</a>" , url , date )
170+ }
171+
172+ // createDDGFeelingLuckyLinks takes in comma or semicolon separated terms,
173+ // then turns them into <a> links searching for the term using DuckDuckGo's I'm
174+ // feeling lucky feature. It will redirect the user immediately to the first search result.
175+ func createDDGFeelingLuckyLinks (searchTerms string , siteHostname string ) string {
176+
177+ siteHostname = strings .TrimSpace (siteHostname )
163178 semiColonSplit := strings .Split (searchTerms , ";" )
164179
165180 var links []string
@@ -171,11 +186,13 @@ func createWikipediaSearchLinks(searchTerms string) string {
171186 continue
172187 }
173188
174- encodedTerm := url . QueryEscape ( trimmedTerm )
189+ ddgQuery := fmt . Sprintf ( ` site:%s intitle:"%s"` , strings . TrimPrefix ( siteHostname , "www." ), trimmedTerm )
175190
176- wikiURL := fmt .Sprintf ("https://en.wikipedia.org/w/index.php?search=%s" , encodedTerm )
191+ encodedTerm := `\%s:` + url .QueryEscape (ddgQuery )
192+ //ddgURL := `https://html.duckduckgo.com/html/?q=` + encodedTerm
193+ ddgURL := `https://www.duckduckgo.com/?q=` + encodedTerm
177194
178- link := fmt .Sprintf ("<a href=\" %s\" >%s</a>" , wikiURL , trimmedTerm )
195+ link := fmt .Sprintf ("<a rel= \" noreferrer \" href=\" %s\" >%s</a>" , ddgURL , trimmedTerm )
179196 links = append (links , link )
180197 }
181198
@@ -187,3 +204,66 @@ func createWikipediaSearchLinks(searchTerms string) string {
187204
188205 return strings .Join (links , " " )
189206}
207+
208+ // estimateReadingTime estimates how long the given text will take to read using the given configuration.
209+ func estimateReadingTime (text string ) time.Duration {
210+ if len (text ) == 0 {
211+ return 0
212+ }
213+
214+ // Init options with default values.
215+ WordsPerMinute := 200
216+ WordBound := func (b byte ) bool {
217+ return b == ' ' || b == '\n' || b == '\r' || b == '\t'
218+ }
219+
220+ words := 0
221+ start := 0
222+ end := len (text ) - 1
223+
224+ // Fetch bounds.
225+ for WordBound (text [start ]) {
226+ start ++
227+ }
228+ for WordBound (text [end ]) {
229+ end --
230+ }
231+
232+ // Calculate the number of words.
233+ for i := start ; i <= end ; {
234+ for i <= end && ! WordBound (text [i ]) {
235+ i ++
236+ }
237+
238+ words ++
239+
240+ for i <= end && WordBound (text [i ]) {
241+ i ++
242+ }
243+ }
244+
245+ // Reading time stats.
246+ minutes := math .Ceil (float64 (words ) / float64 (WordsPerMinute ))
247+ duration := time .Duration (math .Ceil (minutes ) * float64 (time .Minute ))
248+
249+ return duration
250+
251+ }
252+
253+ func formatDuration (d time.Duration ) string {
254+ // Check if the duration is less than one minute
255+ if d < time .Minute {
256+ seconds := int (d .Seconds ())
257+ return fmt .Sprintf ("%d seconds" , seconds )
258+ }
259+
260+ // Convert the duration to minutes
261+ minutes := int (d .Minutes ())
262+
263+ // Format the string for one or more minutes
264+ if minutes == 1 {
265+ return "1 minute"
266+ } else {
267+ return fmt .Sprintf ("%d minutes" , minutes )
268+ }
269+ }
0 commit comments