miniflux-v2/reader/subscription/finder.go

128 lines
3.4 KiB
Go
Raw Normal View History

2017-11-20 06:10:04 +01:00
// Copyright 2017 Frédéric Guillot. All rights reserved.
// Use of this source code is governed by the Apache 2.0
// license that can be found in the LICENSE file.
2018-08-25 06:51:50 +02:00
package subscription // import "miniflux.app/reader/subscription"
2017-11-20 06:10:04 +01:00
import (
"bytes"
"fmt"
"io"
"time"
2018-08-25 06:51:50 +02:00
"miniflux.app/errors"
"miniflux.app/http/client"
"miniflux.app/logger"
"miniflux.app/reader/feed"
"miniflux.app/timer"
"miniflux.app/url"
2017-11-20 06:10:04 +01:00
"github.com/PuerkitoBio/goquery"
)
var (
errConnectionFailure = "Unable to open this link: %v"
errUnreadableDoc = "Unable to analyze this page: %v"
errEmptyBody = "This web page is empty"
errNotAuthorized = "You are not authorized to access this resource (invalid username/password)"
errServerFailure = "Unable to fetch this resource (Status Code = %d)"
2017-11-20 06:10:04 +01:00
)
// FindSubscriptions downloads and try to find one or more subscriptions from an URL.
func FindSubscriptions(websiteURL, userAgent, username, password string) (Subscriptions, error) {
2018-01-03 04:15:08 +01:00
defer timer.ExecutionTime(time.Now(), fmt.Sprintf("[FindSubscriptions] url=%s", websiteURL))
2017-11-20 06:10:04 +01:00
2018-04-28 19:51:07 +02:00
clt := client.New(websiteURL)
clt.WithCredentials(username, password)
clt.WithUserAgent(userAgent)
2018-04-28 19:51:07 +02:00
response, err := clt.Get()
2017-11-20 06:10:04 +01:00
if err != nil {
2018-02-09 03:16:54 +01:00
if _, ok := err.(errors.LocalizedError); ok {
return nil, err
}
2017-11-20 06:10:04 +01:00
return nil, errors.NewLocalizedError(errConnectionFailure, err)
}
if response.IsNotAuthorized() {
return nil, errors.NewLocalizedError(errNotAuthorized)
}
if response.HasServerFailure() {
return nil, errors.NewLocalizedError(errServerFailure, response.StatusCode)
}
// Content-Length = -1 when no Content-Length header is sent
if response.ContentLength == 0 {
return nil, errors.NewLocalizedError(errEmptyBody)
}
body, err := response.NormalizeBodyEncoding()
if err != nil {
return nil, err
}
2017-11-20 06:10:04 +01:00
var buffer bytes.Buffer
size, _ := io.Copy(&buffer, body)
if size == 0 {
return nil, errors.NewLocalizedError(errEmptyBody)
}
2017-11-20 06:10:04 +01:00
reader := bytes.NewReader(buffer.Bytes())
if format := feed.DetectFeedFormat(reader); format != feed.FormatUnknown {
var subscriptions Subscriptions
subscriptions = append(subscriptions, &Subscription{
Title: response.EffectiveURL,
URL: response.EffectiveURL,
Type: format,
})
return subscriptions, nil
}
reader.Seek(0, io.SeekStart)
return parseDocument(response.EffectiveURL, bytes.NewReader(buffer.Bytes()))
}
func parseDocument(websiteURL string, data io.Reader) (Subscriptions, error) {
var subscriptions Subscriptions
queries := map[string]string{
"link[type='application/rss+xml']": "rss",
"link[type='application/atom+xml']": "atom",
"link[type='application/json']": "json",
}
doc, err := goquery.NewDocumentFromReader(data)
if err != nil {
return nil, errors.NewLocalizedError(errUnreadableDoc, err)
}
for query, kind := range queries {
doc.Find(query).Each(func(i int, s *goquery.Selection) {
subscription := new(Subscription)
subscription.Type = kind
if title, exists := s.Attr("title"); exists {
subscription.Title = title
} else {
subscription.Title = "Feed"
}
if feedURL, exists := s.Attr("href"); exists {
2017-12-02 07:29:18 +01:00
subscription.URL, _ = url.AbsoluteURL(websiteURL, feedURL)
2017-11-20 06:10:04 +01:00
}
if subscription.Title == "" {
subscription.Title = subscription.URL
}
if subscription.URL != "" {
2017-12-16 03:55:57 +01:00
logger.Debug("[FindSubscriptions] %s", subscription)
2017-11-20 06:10:04 +01:00
subscriptions = append(subscriptions, subscription)
}
})
}
return subscriptions, nil
}