twitter-scrapper/tweets_test.go

173 lines
6.2 KiB
Go
Raw Normal View History

2021-12-07 10:18:01 +02:00
package twitterscraper_test
2019-09-21 12:18:34 +03:00
import (
"context"
2019-09-21 12:18:34 +03:00
"testing"
2021-03-09 10:40:22 +02:00
"time"
"github.com/google/go-cmp/cmp"
"github.com/google/go-cmp/cmp/cmpopts"
2021-12-07 10:18:01 +02:00
twitterscraper "github.com/n0madic/twitter-scraper"
2019-09-21 12:18:34 +03:00
)
2021-07-30 14:46:38 +03:00
var cmpOptions = cmp.Options{
2021-12-07 10:18:01 +02:00
cmpopts.IgnoreFields(twitterscraper.Tweet{}, "Likes"),
cmpopts.IgnoreFields(twitterscraper.Tweet{}, "Replies"),
cmpopts.IgnoreFields(twitterscraper.Tweet{}, "Retweets"),
2021-07-30 14:46:38 +03:00
}
2019-09-21 12:18:34 +03:00
func TestGetTweets(t *testing.T) {
count := 0
maxTweetsNbr := 300
dupcheck := make(map[string]bool)
2022-05-04 11:55:12 +03:00
scraper := twitterscraper.New()
for tweet := range scraper.GetTweets(context.Background(), "Twitter", maxTweetsNbr) {
2019-09-21 12:18:34 +03:00
if tweet.Error != nil {
t.Error(tweet.Error)
} else {
count++
if tweet.ID == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet ID is empty")
} else {
if dupcheck[tweet.ID] {
t.Errorf("Detect duplicated tweet ID: %s", tweet.ID)
} else {
dupcheck[tweet.ID] = true
}
}
if tweet.UserID == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet UserID is empty")
2019-09-21 12:18:34 +03:00
}
2020-06-15 16:38:51 +03:00
if tweet.Username == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet Username is empty")
2020-06-15 16:38:51 +03:00
}
2019-09-21 12:18:34 +03:00
if tweet.PermanentURL == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet PermanentURL is empty")
2019-09-21 12:18:34 +03:00
}
if tweet.Text == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet Text is empty")
2019-09-21 12:18:34 +03:00
}
if tweet.TimeParsed.IsZero() {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet TimeParsed is zero")
2019-09-21 12:18:34 +03:00
}
if tweet.Timestamp == 0 {
t.Error("Expected tweet Timestamp is greater than zero")
}
for _, video := range tweet.Videos {
if video.ID == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet video ID is empty")
}
if video.Preview == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet video Preview is empty")
}
if video.URL == "" {
2021-04-22 20:35:33 +03:00
t.Error("Expected tweet video URL is empty")
}
}
2019-09-21 12:18:34 +03:00
}
}
if count != maxTweetsNbr {
t.Errorf("Expected tweets count=%v, got: %v", maxTweetsNbr, count)
2019-09-21 12:18:34 +03:00
}
}
2021-03-09 10:40:22 +02:00
func TestGetTweet(t *testing.T) {
2021-12-07 10:18:01 +02:00
sample := twitterscraper.Tweet{
2021-03-09 10:40:22 +02:00
HTML: "That thing you didnt Tweet but wanted to but didnt but got so close but then were like nah. <br><br>We have a place for that now—Fleets! <br><br>Rolling out to everyone starting today. <br><a href=\"https://t.co/auQAHXZMfH\"><img src=\"https://pbs.twimg.com/amplify_video_thumb/1328684333599756289/img/cP5KwbIXbGunNSBy.jpg\"/></a>",
ID: "1328684389388185600",
PermanentURL: "https://twitter.com/Twitter/status/1328684389388185600",
2023-05-10 03:18:34 -07:00
Photos: nil,
2021-03-09 10:40:22 +02:00
Text: "That thing you didnt Tweet but wanted to but didnt but got so close but then were like nah. \n\nWe have a place for that now—Fleets! \n\nRolling out to everyone starting today. https://t.co/auQAHXZMfH",
TimeParsed: time.Date(2020, 11, 17, 13, 0, 18, 0, time.FixedZone("UTC", 0)),
Timestamp: 1605618018,
UserID: "783214",
Username: "Twitter",
2021-12-07 10:18:01 +02:00
Videos: []twitterscraper.Video{{
2021-03-09 10:40:22 +02:00
ID: "1328684333599756289",
Preview: "https://pbs.twimg.com/amplify_video_thumb/1328684333599756289/img/cP5KwbIXbGunNSBy.jpg",
2021-09-13 17:04:23 +03:00
URL: "https://video.twimg.com/amplify_video/1328684333599756289/vid/960x720/PcL8yv8KhgQ48Qpt.mp4?tag=13",
2021-03-09 10:40:22 +02:00
}},
}
2022-05-04 11:55:12 +03:00
scraper := twitterscraper.New()
tweet, err := scraper.GetTweet("1328684389388185600")
2021-03-09 10:40:22 +02:00
if err != nil {
t.Error(err)
} else {
if diff := cmp.Diff(sample, *tweet, cmpOptions...); diff != "" {
t.Error("Resulting tweet does not match the sample", diff)
}
}
}
2021-03-28 12:25:55 +03:00
func TestQuotedAndReply(t *testing.T) {
2021-12-07 10:18:01 +02:00
sample := &twitterscraper.Tweet{
HTML: "The Easiest Problem Everyone Gets Wrong <br><br>[new video] --&gt; <a href=\"https://youtu.be/ytfCdqWhmdg\">https://t.co/YdaeDYmPAU</a> <br><a href=\"https://t.co/iKu4Xs6o2V\"><img src=\"https://pbs.twimg.com/media/ESsZa9AXgAIAYnF.jpg\"/></a>",
ID: "1237110546383724547",
2021-08-04 11:00:12 +03:00
Likes: 485,
PermanentURL: "https://twitter.com/VsauceTwo/status/1237110546383724547",
2023-05-10 03:18:34 -07:00
Photos: []twitterscraper.Photo{{
ID: "1237110473486729218",
URL: "https://pbs.twimg.com/media/ESsZa9AXgAIAYnF.jpg",
}},
Replies: 12,
Retweets: 18,
Text: "The Easiest Problem Everyone Gets Wrong \n\n[new video] --&gt; https://t.co/YdaeDYmPAU https://t.co/iKu4Xs6o2V",
TimeParsed: time.Date(2020, 0o3, 9, 20, 18, 33, 0, time.FixedZone("UTC", 0)),
Timestamp: 1583785113,
URLs: []string{"https://youtu.be/ytfCdqWhmdg"},
UserID: "978944851",
Username: "VsauceTwo",
}
2022-05-04 11:55:12 +03:00
scraper := twitterscraper.New()
tweet, err := scraper.GetTweet("1237110897597976576")
if err != nil {
t.Error(err)
} else {
2021-07-16 12:39:11 +03:00
if !tweet.IsQuoted {
t.Error("IsQuoted must be True")
}
2021-08-09 14:02:35 +03:00
if diff := cmp.Diff(sample, tweet.QuotedStatus, cmpOptions...); diff != "" {
t.Error("Resulting quote does not match the sample", diff)
}
}
2022-05-04 11:55:12 +03:00
tweet, err = scraper.GetTweet("1237111868445134850")
if err != nil {
t.Error(err)
} else {
2021-07-16 12:39:11 +03:00
if !tweet.IsReply {
t.Error("IsReply must be True")
}
2021-08-09 14:02:35 +03:00
if diff := cmp.Diff(sample, tweet.InReplyToStatus, cmpOptions...); diff != "" {
t.Error("Resulting reply does not match the sample", diff)
}
}
}
2021-03-28 12:25:55 +03:00
func TestRetweet(t *testing.T) {
2021-12-07 10:18:01 +02:00
sample := &twitterscraper.Tweet{
2021-07-16 11:37:43 +03:00
HTML: "Weve seen an increase in attacks against Asian communities and individuals around the world. Its important to know that this isnt new; throughout history, Asians have experienced violence and exclusion. However, their diverse lived experiences have largely been overlooked.",
ID: "1359151057872580612",
2021-07-23 17:04:23 +03:00
Likes: 6683,
2021-07-16 11:37:43 +03:00
PermanentURL: "https://twitter.com/TwitterTogether/status/1359151057872580612",
2021-07-23 17:04:23 +03:00
Replies: 456,
2021-07-16 11:37:43 +03:00
Retweets: 1495,
Text: "Weve seen an increase in attacks against Asian communities and individuals around the world. Its important to know that this isnt new; throughout history, Asians have experienced violence and exclusion. However, their diverse lived experiences have largely been overlooked.",
TimeParsed: time.Date(2021, 02, 9, 14, 43, 58, 0, time.FixedZone("UTC", 0)),
Timestamp: 1612881838,
UserID: "773578328498372608",
Username: "TwitterTogether",
2021-03-28 12:25:55 +03:00
}
2022-05-04 11:55:12 +03:00
scraper := twitterscraper.New()
tweet, err := scraper.GetTweet("1362849141248974853")
2021-03-28 12:25:55 +03:00
if err != nil {
t.Error(err)
} else {
2021-07-16 12:39:11 +03:00
if !tweet.IsRetweet {
t.Error("IsRetweet must be True")
}
2021-07-30 14:46:38 +03:00
if diff := cmp.Diff(sample, tweet.RetweetedStatus, cmpOptions...); diff != "" {
2021-03-28 12:25:55 +03:00
t.Error("Resulting retweet does not match the sample", diff)
}
}
}