okay, now with the imgs

This commit is contained in:
Ted Unangst 2022-07-03 16:17:18 -04:00
parent a452ef3188
commit 578deeb6e9
1 changed files with 8 additions and 1 deletions

View File

@ -31,7 +31,7 @@ import (
var tweetsel = cascadia.MustCompile("div[itemProp=articleBody]") var tweetsel = cascadia.MustCompile("div[itemProp=articleBody]")
var linksel = cascadia.MustCompile("a time") var linksel = cascadia.MustCompile("a time")
var replyingto = cascadia.MustCompile(".ReplyingToContextBelowAuthor") var replyingto = cascadia.MustCompile(".ReplyingToContextBelowAuthor")
var imgsel = cascadia.MustCompile("div.js-adaptive-photo img") var imgsel = cascadia.MustCompile("div[data-testid=tweetPhoto] img")
var authorregex = regexp.MustCompile("twitter.com/([^/]+)") var authorregex = regexp.MustCompile("twitter.com/([^/]+)")
var re_hoots = regexp.MustCompile(`hoot: ?https://\S+`) var re_hoots = regexp.MustCompile(`hoot: ?https://\S+`)
@ -84,9 +84,16 @@ func hootextractor(r io.Reader, url string, seen map[string]bool) string {
continue continue
} }
author := authormatch[1] author := authormatch[1]
if wanted == "" {
wanted = author
}
if author != wanted { if author != wanted {
continue continue
} }
for _, img := range imgsel.MatchAll(twp) {
img.Parent.RemoveChild(img)
div.AppendChild(img)
}
text := htf.NodeText(div) text := htf.NodeText(div)
text = strings.Replace(text, "\n", " ", -1) text = strings.Replace(text, "\n", " ", -1)
fmt.Fprintf(&buf, "> @%s: %s\n", author, text) fmt.Fprintf(&buf, "> @%s: %s\n", author, text)