twitter-scrapper/timeline_v2.go

397 lines
12 KiB
Go
Raw Normal View History

package twitterscraper
import (
"strconv"
2024-08-01 17:39:06 +03:00
"strings"
)
2024-04-25 23:23:29 +03:00
type tweet struct {
Core struct {
UserResults struct {
Result struct {
IsBlueVerified bool `json:"is_blue_verified"`
Legacy legacyUser `json:"legacy"`
} `json:"result"`
} `json:"user_results"`
} `json:"core"`
Views struct {
Count string `json:"count"`
} `json:"views"`
NoteTweet struct {
NoteTweetResults struct {
Result struct {
Text string `json:"text"`
} `json:"result"`
} `json:"note_tweet_results"`
} `json:"note_tweet"`
QuotedStatusResult struct {
Result *result `json:"result"`
} `json:"quoted_status_result"`
Legacy legacyTweet `json:"legacy"`
}
2024-04-25 23:23:29 +03:00
type result struct {
Typename string `json:"__typename"`
tweet
Tweet tweet `json:"tweet"`
}
func (result *result) parse() *Tweet {
if result.NoteTweet.NoteTweetResults.Result.Text != "" {
result.Legacy.FullText = result.NoteTweet.NoteTweetResults.Result.Text
}
2024-04-25 23:48:41 +03:00
var legacy *legacyTweet = &result.Legacy
var user *legacyUser = &result.Core.UserResults.Result.Legacy
2024-04-25 23:23:29 +03:00
if result.Typename == "TweetWithVisibilityResults" {
2024-04-25 23:48:41 +03:00
legacy = &result.Tweet.Legacy
2024-04-25 23:23:29 +03:00
user = &result.Tweet.Core.UserResults.Result.Legacy
}
2024-04-25 23:48:41 +03:00
tw := parseLegacyTweet(user, legacy)
2023-07-03 14:57:29 +03:00
if tw == nil {
return nil
}
if tw.Views == 0 && result.Views.Count != "" {
tw.Views, _ = strconv.Atoi(result.Views.Count)
}
if result.QuotedStatusResult.Result != nil {
tw.QuotedStatus = result.QuotedStatusResult.Result.parse()
}
2024-11-08 21:57:56 +02:00
tw.HTML = expandURLs(tw.HTML, legacy.Entities.URLs, legacy.ExtendedEntities.Media)
tw.HTML = expandURLs(tw.Text, legacy.Entities.URLs, legacy.ExtendedEntities.Media)
return tw
}
2024-02-03 14:53:11 +01:00
type userResult struct {
Typename string `json:"__typename"`
ID string `json:"id"`
RestID string `json:"rest_id"`
AffiliatesHighlightedLabel struct{} `json:"affiliates_highlighted_label"`
HasGraduatedAccess bool `json:"has_graduated_access"`
IsBlueVerified bool `json:"is_blue_verified"`
ProfileImageShape string `json:"profile_image_shape"`
Legacy legacyUserV2 `json:"legacy"`
LegacyExtendedProfile legacyExtendedProfile `json:"legacy_extended_profile"`
IsProfileTranslatable bool `json:"is_profile_translatable"`
VerificationInfo verificationInfo `json:"verification_info"`
HighlightsInfo highlightsInfo `json:"highlights_info"`
UserSeedTweetCount int `json:"user_seed_tweet_count"`
PremiumGiftingEligible bool `json:"premium_gifting_eligible"`
CreatorSubscriptionsCount int `json:"creator_subscriptions_count"`
2024-02-03 14:53:11 +01:00
}
func (result *userResult) parse() Profile {
return parseProfileV2(*result)
}
2024-02-13 04:45:08 +03:00
type item struct {
2024-08-01 17:39:06 +03:00
EntryID string `json:"entryId"`
Item struct {
2024-02-13 04:45:08 +03:00
ItemContent struct {
2024-08-01 17:39:06 +03:00
ItemType string `json:"itemType"`
2024-02-13 04:45:08 +03:00
TweetDisplayType string `json:"tweetDisplayType"`
TweetResults struct {
Result result `json:"result"`
} `json:"tweet_results"`
2024-08-01 17:39:06 +03:00
CursorType string `json:"cursorType"`
Value string `json:"value"`
2024-02-13 04:45:08 +03:00
} `json:"itemContent"`
} `json:"item"`
}
type entry struct {
Content struct {
2024-02-13 04:45:08 +03:00
CursorType string `json:"cursorType"`
Value string `json:"value"`
Items []item `json:"items"`
ItemContent struct {
2024-08-01 17:39:06 +03:00
ItemType string `json:"itemType"`
2023-06-05 13:49:56 +03:00
TweetDisplayType string `json:"tweetDisplayType"`
TweetResults struct {
Result result `json:"result"`
} `json:"tweet_results"`
2023-07-03 14:38:46 +03:00
UserDisplayType string `json:"userDisplayType"`
UserResults struct {
2024-02-03 14:53:11 +01:00
Result userResult `json:"result"`
2023-07-03 14:38:46 +03:00
} `json:"user_results"`
2024-08-01 17:39:06 +03:00
CursorType string `json:"cursorType"`
Value string `json:"value"`
} `json:"itemContent"`
} `json:"content"`
}
// timeline v2 JSON object
type timelineV2 struct {
Data struct {
User struct {
Result struct {
TimelineV2 struct {
Timeline struct {
Instructions []struct {
2024-02-13 04:45:08 +03:00
ModuleItems []item `json:"moduleItems"`
Entries []entry `json:"entries"`
Entry entry `json:"entry"`
Type string `json:"type"`
} `json:"instructions"`
} `json:"timeline"`
} `json:"timeline_v2"`
2024-02-03 14:53:11 +01:00
Timeline struct {
Timeline struct {
Instructions []struct {
Entries []entry `json:"entries"`
Entry entry `json:"entry"`
Type string `json:"type"`
} `json:"instructions"`
} `json:"timeline"`
} `json:"timeline"`
} `json:"result"`
} `json:"user"`
} `json:"data"`
}
func (timeline *timelineV2) parseTweets() ([]*Tweet, string) {
var cursor string
var tweets []*Tweet
for _, instruction := range timeline.Data.User.Result.TimelineV2.Timeline.Instructions {
for _, entry := range instruction.Entries {
if entry.Content.CursorType == "Bottom" {
cursor = entry.Content.Value
continue
}
2024-04-25 23:48:41 +03:00
if entry.Content.ItemContent.TweetResults.Result.Typename == "Tweet" || entry.Content.ItemContent.TweetResults.Result.Typename == "TweetWithVisibilityResults" {
if tweet := entry.Content.ItemContent.TweetResults.Result.parse(); tweet != nil {
tweets = append(tweets, tweet)
}
}
2024-02-13 04:45:08 +03:00
if len(entry.Content.Items) > 0 {
for _, item := range entry.Content.Items {
if tweet := item.Item.ItemContent.TweetResults.Result.parse(); tweet != nil {
tweets = append(tweets, tweet)
}
}
}
}
if len(instruction.ModuleItems) > 0 {
for _, entry := range instruction.ModuleItems {
2024-04-25 23:48:41 +03:00
if entry.Item.ItemContent.TweetResults.Result.Typename == "Tweet" || entry.Item.ItemContent.TweetResults.Result.Typename == "TweetWithVisibilityResults" {
2024-02-13 04:45:08 +03:00
if tweet := entry.Item.ItemContent.TweetResults.Result.parse(); tweet != nil {
tweets = append(tweets, tweet)
}
}
}
}
}
return tweets, cursor
}
2024-02-21 04:41:51 +03:00
type bookmarksTimelineV2 struct {
Data struct {
Bookmarks struct {
Timeline struct {
Instructions []struct {
Entries []entry `json:"entries"`
Type string `json:"type"`
} `json:"instructions"`
} `json:"timeline"`
} `json:"bookmark_timeline_v2"`
} `json:"data"`
}
func (timeline *bookmarksTimelineV2) parseTweets() ([]*Tweet, string) {
var cursor string
var tweets []*Tweet
for _, instruction := range timeline.Data.Bookmarks.Timeline.Instructions {
for _, entry := range instruction.Entries {
if entry.Content.CursorType == "Bottom" {
cursor = entry.Content.Value
continue
}
if entry.Content.ItemContent.TweetResults.Result.Typename == "Tweet" {
if tweet := entry.Content.ItemContent.TweetResults.Result.parse(); tweet != nil {
tweets = append(tweets, tweet)
}
}
}
}
return tweets, cursor
}
2024-08-05 17:57:18 +03:00
type retweetersTimelineV2 struct {
Data struct {
RetweetersTimeline struct {
Timeline struct {
Instructions []struct {
Type string `json:"type"`
Entries []entry `json:"entries"`
} `json:"instructions"`
} `json:"timeline"`
} `json:"retweeters_timeline"`
} `json:"data"`
}
func (timeline *retweetersTimelineV2) parseUsers() ([]*Profile, string) {
var cursor string
var users []*Profile
for _, instruction := range timeline.Data.RetweetersTimeline.Timeline.Instructions {
for _, entry := range instruction.Entries {
if entry.Content.CursorType == "Bottom" {
cursor = entry.Content.Value
continue
}
if entry.Content.ItemContent.UserResults.Result.Typename == "User" {
user := entry.Content.ItemContent.UserResults.Result.parse()
users = append(users, &user)
}
}
}
return users, cursor
}
2024-02-03 14:53:11 +01:00
func (timeline *timelineV2) parseUsers() ([]*Profile, string) {
var cursor string
var users []*Profile
for _, instruction := range timeline.Data.User.Result.Timeline.Timeline.Instructions {
for _, entry := range instruction.Entries {
if entry.Content.CursorType == "Bottom" {
cursor = entry.Content.Value
continue
}
if entry.Content.ItemContent.UserResults.Result.Typename == "User" {
user := entry.Content.ItemContent.UserResults.Result.parse()
users = append(users, &user)
}
}
}
return users, cursor
}
type threadedConversation struct {
Data struct {
ThreadedConversationWithInjectionsV2 struct {
Instructions []struct {
2024-08-01 17:39:06 +03:00
Type string `json:"type"`
Entry entry `json:"entry"`
Entries []entry `json:"entries"`
ModuleItems []item `json:"moduleItems"`
} `json:"instructions"`
} `json:"threaded_conversation_with_injections_v2"`
} `json:"data"`
}
2024-08-01 17:39:06 +03:00
func (conversation *threadedConversation) parse(focalTweetID string) ([]*Tweet, []*ThreadCursor) {
var tweets []*Tweet
2024-08-01 17:39:06 +03:00
var cursors []*ThreadCursor
for _, instruction := range conversation.Data.ThreadedConversationWithInjectionsV2.Instructions {
for _, entry := range instruction.Entries {
2024-04-25 23:23:29 +03:00
if entry.Content.ItemContent.TweetResults.Result.Typename == "Tweet" || entry.Content.ItemContent.TweetResults.Result.Typename == "TweetWithVisibilityResults" {
if tweet := entry.Content.ItemContent.TweetResults.Result.parse(); tweet != nil {
2023-06-05 13:49:56 +03:00
if entry.Content.ItemContent.TweetDisplayType == "SelfThread" {
tweet.IsSelfThread = true
}
tweets = append(tweets, tweet)
}
}
2024-08-01 17:39:06 +03:00
if entry.Content.ItemContent.CursorType != "" && entry.Content.ItemContent.Value != "" {
cursors = append(cursors, &ThreadCursor{
FocalTweetID: focalTweetID,
ThreadID: focalTweetID,
Cursor: entry.Content.ItemContent.Value,
CursorType: entry.Content.ItemContent.CursorType,
})
}
for _, item := range entry.Content.Items {
2024-04-25 23:23:29 +03:00
if item.Item.ItemContent.TweetResults.Result.Typename == "Tweet" || item.Item.ItemContent.TweetResults.Result.Typename == "TweetWithVisibilityResults" {
if tweet := item.Item.ItemContent.TweetResults.Result.parse(); tweet != nil {
2023-06-05 13:49:56 +03:00
if item.Item.ItemContent.TweetDisplayType == "SelfThread" {
tweet.IsSelfThread = true
}
tweets = append(tweets, tweet)
}
}
2024-08-01 17:39:06 +03:00
if item.Item.ItemContent.CursorType != "" && item.Item.ItemContent.Value != "" {
threadID := ""
entryId := strings.Split(item.EntryID, "-")
if len(entryId) > 1 && entryId[0] == "conversationthread" {
if i, _ := strconv.Atoi(entryId[1]); i != 0 {
threadID = entryId[1]
}
}
cursors = append(cursors, &ThreadCursor{
FocalTweetID: focalTweetID,
ThreadID: threadID,
Cursor: item.Item.ItemContent.Value,
CursorType: item.Item.ItemContent.CursorType,
})
}
}
}
for _, item := range instruction.ModuleItems {
if item.Item.ItemContent.TweetResults.Result.Typename == "Tweet" || item.Item.ItemContent.TweetResults.Result.Typename == "TweetWithVisibilityResults" {
if tweet := item.Item.ItemContent.TweetResults.Result.parse(); tweet != nil {
if item.Item.ItemContent.TweetDisplayType == "SelfThread" {
tweet.IsSelfThread = true
}
tweets = append(tweets, tweet)
}
}
if item.Item.ItemContent.CursorType != "" && item.Item.ItemContent.Value != "" {
threadID := ""
entryId := strings.Split(item.EntryID, "-")
if len(entryId) > 1 && entryId[0] == "conversationthread" {
if i, _ := strconv.Atoi(entryId[1]); i != 0 {
threadID = entryId[1]
}
}
cursors = append(cursors, &ThreadCursor{
FocalTweetID: focalTweetID,
ThreadID: threadID,
Cursor: item.Item.ItemContent.Value,
CursorType: item.Item.ItemContent.CursorType,
})
}
}
}
2024-08-01 17:39:06 +03:00
for _, tweet := range tweets {
if tweet.InReplyToStatusID != "" {
for _, parentTweet := range tweets {
if parentTweet.ID == tweet.InReplyToStatusID {
tweet.InReplyToStatus = parentTweet
break
}
}
}
2023-06-05 13:49:56 +03:00
if tweet.IsSelfThread && tweet.ConversationID == tweet.ID {
for _, childTweet := range tweets {
if childTweet.IsSelfThread && childTweet.ID != tweet.ID {
tweet.Thread = append(tweet.Thread, childTweet)
}
}
if len(tweet.Thread) == 0 {
tweet.IsSelfThread = false
}
}
}
2024-08-01 17:39:06 +03:00
return tweets, cursors
}
2024-03-08 18:43:26 +03:00
type tweetResult struct {
Data struct {
TweetResult struct {
Result result `json:"result"`
} `json:"tweetResult"`
} `json:"data"`
}
func (tweetResult *tweetResult) parse() *Tweet {
return tweetResult.Data.TweetResult.Result.parse()
}