Fix search cursor
This commit is contained in:
parent
0533b6e909
commit
9a85dfa062
4 changed files with 36 additions and 3 deletions
|
|
@ -19,8 +19,8 @@ func SearchTweets(ctx context.Context, query string, maxTweetsNbr int) <-chan *R
|
||||||
// FetchSearchTweets gets tweets for a given search query, via the Twitter frontend API
|
// FetchSearchTweets gets tweets for a given search query, via the Twitter frontend API
|
||||||
func (s *Scraper) FetchSearchTweets(query string, maxTweetsNbr int, cursor string) ([]*Tweet, string, error) {
|
func (s *Scraper) FetchSearchTweets(query string, maxTweetsNbr int, cursor string) ([]*Tweet, string, error) {
|
||||||
query = url.PathEscape(query)
|
query = url.PathEscape(query)
|
||||||
if maxTweetsNbr > 200 {
|
if maxTweetsNbr > 100 {
|
||||||
maxTweetsNbr = 200
|
maxTweetsNbr = 100
|
||||||
}
|
}
|
||||||
|
|
||||||
req, err := s.newRequest("GET", "https://twitter.com/i/api/2/search/adaptive.json")
|
req, err := s.newRequest("GET", "https://twitter.com/i/api/2/search/adaptive.json")
|
||||||
|
|
|
||||||
|
|
@ -5,10 +5,28 @@ import (
|
||||||
"testing"
|
"testing"
|
||||||
)
|
)
|
||||||
|
|
||||||
|
func TestFetchSearchCursor(t *testing.T) {
|
||||||
|
scraper := New()
|
||||||
|
maxTweetsNbr := 150
|
||||||
|
tweetsNbr := 0
|
||||||
|
nextCursor := ""
|
||||||
|
for tweetsNbr < maxTweetsNbr {
|
||||||
|
tweets, cursor, err := scraper.FetchSearchTweets("twitter", maxTweetsNbr, nextCursor)
|
||||||
|
if err != nil {
|
||||||
|
t.Fatal(err)
|
||||||
|
}
|
||||||
|
if cursor == "" {
|
||||||
|
t.Fatal("Expected search cursor is not empty")
|
||||||
|
}
|
||||||
|
tweetsNbr += len(tweets)
|
||||||
|
nextCursor = cursor
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
func TestGetSearchTweets(t *testing.T) {
|
func TestGetSearchTweets(t *testing.T) {
|
||||||
count := 0
|
count := 0
|
||||||
maxTweetsNbr := 250
|
maxTweetsNbr := 250
|
||||||
for tweet := range SearchTweets(context.Background(), "twitter scraper data -filter:retweets", maxTweetsNbr) {
|
for tweet := range SearchTweets(context.Background(), "twitter -filter:retweets", maxTweetsNbr) {
|
||||||
if tweet.Error != nil {
|
if tweet.Error != nil {
|
||||||
t.Error(tweet.Error)
|
t.Error(tweet.Error)
|
||||||
} else {
|
} else {
|
||||||
|
|
|
||||||
12
types.go
12
types.go
|
|
@ -169,6 +169,18 @@ type (
|
||||||
} `json:"content"`
|
} `json:"content"`
|
||||||
} `json:"entry"`
|
} `json:"entry"`
|
||||||
} `json:"pinEntry,omitempty"`
|
} `json:"pinEntry,omitempty"`
|
||||||
|
ReplaceEntry struct {
|
||||||
|
Entry struct {
|
||||||
|
Content struct {
|
||||||
|
Operation struct {
|
||||||
|
Cursor struct {
|
||||||
|
Value string `json:"value"`
|
||||||
|
CursorType string `json:"cursorType"`
|
||||||
|
} `json:"cursor"`
|
||||||
|
} `json:"operation"`
|
||||||
|
} `json:"content"`
|
||||||
|
} `json:"entry"`
|
||||||
|
} `json:"replaceEntry,omitempty"`
|
||||||
} `json:"instructions"`
|
} `json:"instructions"`
|
||||||
} `json:"timeline"`
|
} `json:"timeline"`
|
||||||
}
|
}
|
||||||
|
|
|
||||||
3
util.go
3
util.go
|
|
@ -225,6 +225,9 @@ func parseTimeline(timeline *timeline) ([]*Tweet, string) {
|
||||||
cursor = entry.Content.Operation.Cursor.Value
|
cursor = entry.Content.Operation.Cursor.Value
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
if instruction.ReplaceEntry.Entry.Content.Operation.Cursor.CursorType == "Bottom" {
|
||||||
|
cursor = instruction.ReplaceEntry.Entry.Content.Operation.Cursor.Value
|
||||||
|
}
|
||||||
}
|
}
|
||||||
if pinnedTweet != nil && len(orderedTweets) > 0 {
|
if pinnedTweet != nil && len(orderedTweets) > 0 {
|
||||||
orderedTweets = append([]*Tweet{pinnedTweet}, orderedTweets...)
|
orderedTweets = append([]*Tweet{pinnedTweet}, orderedTweets...)
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue