oss/miniflux-v2
oss/miniflux-v2
1
0
Fork 0
mirror of https://github.com/miniflux/v2.git synced 2025-08-01 17:38:37 +00:00
Code Wiki Activity

Make sure icon URLs are always absolute

Browse source 
Regression introduced in #1907
This commit is contained in:
Frédéric Guillot 2023-09-09 14:18:39 -07:00
parent 48f6885f44
commit 3b94217fb7
4 changed files with 106 additions and 40 deletions
Download patch file Download diff file Expand all files Collapse all files

89
internal/reader/icon/finder.go
View file

@ -21,48 +21,73 @@ import (
)
// FindIcon try to find the website's icon.
func FindIcon(websiteURL, iconURL, userAgent string, fetchViaProxy, allowSelfSignedCertificates bool) (*model.Icon, error) {
if iconURL == "" {
rootURL := urllib.RootURL(websiteURL)
logger.Debug("[FindIcon] Trying to find an icon: rootURL=%q websiteURL=%q userAgent=%q", rootURL, websiteURL, userAgent)
clt := client.NewClientWithConfig(rootURL, config.Opts)
clt.WithUserAgent(userAgent)
clt.AllowSelfSignedCertificates = allowSelfSignedCertificates
if fetchViaProxy {
clt.WithProxy()
}
response, err := clt.Get()
if err != nil {
return nil, fmt.Errorf("icon: unable to download website index page: %v", err)
}
if response.HasServerFailure() {
return nil, fmt.Errorf("icon: unable to download website index page: status=%d", response.StatusCode)
}
iconURL, err = parseDocument(rootURL, response.Body)
func FindIcon(websiteURL, feedIconURL, userAgent string, fetchViaProxy, allowSelfSignedCertificates bool) (icon *model.Icon, err error) {
if feedIconURL == "" {
feedIconURL, err = fetchHTMLDocumentAndFindIconURL(websiteURL, userAgent, fetchViaProxy, allowSelfSignedCertificates)
if err != nil {
return nil, err
}
}
if strings.HasPrefix(iconURL, "data:") {
return parseImageDataURL(iconURL)
if strings.HasPrefix(feedIconURL, "data:") {
return parseImageDataURL(feedIconURL)
}
logger.Debug("[FindIcon] Fetching icon => %s", iconURL)
icon, err := downloadIcon(iconURL, userAgent, fetchViaProxy, allowSelfSignedCertificates)
feedIconURL, err = generateIconURL(websiteURL, feedIconURL)
if err != nil {
return nil, err
}
if icon, err = downloadIcon(feedIconURL, userAgent, fetchViaProxy, allowSelfSignedCertificates); err != nil {
return nil, err
}
return icon, nil
}
func parseDocument(websiteURL string, data io.Reader) (string, error) {
func generateIconURL(websiteURL, feedIconURL string) (iconURL string, err error) {
feedIconURL = strings.TrimSpace(feedIconURL)
if feedIconURL == "" {
iconURL, err = urllib.JoinBaseURLAndPath(urllib.RootURL(websiteURL), "favicon.ico")
if err != nil {
return "", fmt.Errorf(`icon: unable to join base URL and path: %w`, err)
}
} else {
iconURL, err = urllib.AbsoluteURL(websiteURL, feedIconURL)
if err != nil {
return "", fmt.Errorf(`icon: unable to convert icon URL to absolute URL: %w`, err)
}
}
return iconURL, nil
}
func fetchHTMLDocumentAndFindIconURL(websiteURL, userAgent string, fetchViaProxy, allowSelfSignedCertificates bool) (string, error) {
rootURL := urllib.RootURL(websiteURL)
logger.Debug("[FindIcon] Find icon from HTML webpage: %s", rootURL)
clt := client.NewClientWithConfig(rootURL, config.Opts)
clt.WithUserAgent(userAgent)
clt.AllowSelfSignedCertificates = allowSelfSignedCertificates
if fetchViaProxy {
clt.WithProxy()
}
response, err := clt.Get()
if err != nil {
return "", fmt.Errorf("icon: unable to download website index page: %v", err)
}
if response.HasServerFailure() {
return "", fmt.Errorf("icon: unable to download website index page: status=%d", response.StatusCode)
}
return findIconURLFromHTMLDocument(response.Body)
}
func findIconURLFromHTMLDocument(body io.Reader) (string, error) {
queries := []string{
"link[rel='shortcut icon']",
"link[rel='Shortcut Icon']",
@ -70,7 +95,7 @@ func parseDocument(websiteURL string, data io.Reader) (string, error) {
"link[rel='icon']",
}
doc, err := goquery.NewDocumentFromReader(data)
doc, err := goquery.NewDocumentFromReader(body)
if err != nil {
return "", fmt.Errorf("icon: unable to read document: %v", err)
}
@ -88,12 +113,6 @@ func parseDocument(websiteURL string, data io.Reader) (string, error) {
}
}
if iconURL == "" {
iconURL = urllib.RootURL(websiteURL) + "favicon.ico"
} else {
iconURL, _ = urllib.AbsoluteURL(websiteURL, iconURL)
}
return iconURL, nil
}