mirror of
https://github.com/httpjamesm/AnonymousOverflow.git
synced 2025-04-19 13:29:15 +00:00
e82646635e
* feat: replace stackoverflow and exchange links * fix: replace stackoverflow.com links with path * feat: run stack overflow link replacer on process * feat: process HTML on comment text
45 lines
No EOL
1.3 KiB
Go
45 lines
No EOL
1.3 KiB
Go
package utils
|
|
|
|
import (
|
|
"net/url"
|
|
"regexp"
|
|
"strings"
|
|
)
|
|
|
|
// stackOverflowLinkQualifierRegex matches all anchor elements that meet the following conditions:
|
|
// * must be an anchor element
|
|
// * the anchor element must have a pathname beginning with /q or /questions
|
|
// * if there is a host, it must be stackoverflow.com or a subdomain
|
|
var stackOverflowLinkQualifierRegex = regexp.MustCompile(`<a\s[^>]*href="(?:https?://(?:www\.)?(?:\w+\.)*(?:stackoverflow|stackexchange)\.com)?/(?:q|questions)/[^"]*"[^>]*>.*?</a>`)
|
|
|
|
func ReplaceStackOverflowLinks(html string) string {
|
|
return stackOverflowLinkQualifierRegex.ReplaceAllStringFunc(html, func(match string) string {
|
|
// Extract the href attribute value from the anchor tag
|
|
hrefRegex := regexp.MustCompile(`href="([^"]*)"`)
|
|
hrefMatch := hrefRegex.FindStringSubmatch(match)
|
|
if len(hrefMatch) < 2 {
|
|
return match
|
|
}
|
|
href := hrefMatch[1]
|
|
|
|
// Parse the URL
|
|
url, err := url.Parse(href)
|
|
if err != nil {
|
|
return match
|
|
}
|
|
|
|
newUrl := url.String()
|
|
|
|
// Check if the host is a subdomain
|
|
parts := strings.Split(url.Host, ".")
|
|
if len(parts) > 2 {
|
|
// Prepend the subdomain to the path
|
|
url.Path = "/exchange/" + parts[0] + url.Path
|
|
}
|
|
|
|
newUrl = url.Path + url.RawQuery + url.Fragment
|
|
|
|
// Replace the href attribute value in the anchor tag
|
|
return strings.Replace(match, hrefMatch[1], newUrl, 1)
|
|
})
|
|
} |