Add Notifications parser

pkg/scraper/api_types.go

@@ -260,6 +260,37 @@ type APIUser struct {
 	DoesntExist          bool
 }
 
+type APINotification struct {
+	ID          string `json:"id"`
+	TimestampMs int64  `json:"timestampMs,string"`
+	Message     struct {
+		Text     string `json:"text"`
+		Entities []struct {
+			FromIndex int `json:"fromIndex"`
+			ToIndex   int `json:"toIndex"`
+			Ref       struct {
+				User struct {
+					ID int `json:"id,string"`
+				} `json:"user"`
+			} `json:"ref"`
+		} `json:"entities"`
+	} `json:"message"`
+	Template struct {
+		AggregateUserActionsV1 struct {
+			TargetObjects []struct {
+				Tweet struct {
+					ID int `json:"id,string"`
+				} `json:"tweet"`
+			} `json:"targetObjects"`
+			FromUsers []struct {
+				User struct {
+					ID int `json:"id,string"`
+				} `json:"user"`
+			} `json:"fromUsers"`
+		} `json:"aggregateUserActionsV1"`
+	} `json:"template"`
+}
+
 type UserResponse struct {
 	Data struct {
 		User struct {
@@ -326,7 +357,15 @@ type Entry struct {
 				Tweet struct {
 					ID int64 `json:"id,string"`
 				} `json:"tweet"`
+				Notification struct {
+					ID           string     `json:"id"`
+					FromUsers    Int64Slice `json:"fromUsers"`
+					TargetTweets Int64Slice `json:"targetTweets"`
+				} `json:"notification"`
 			} `json:"content"`
+			ClientEventInfo struct {
+				Element string `json:"element"`
+			} `json:"clientEventInfo"`
 		} `json:"item"`
 		Operation struct {
 			Cursor struct {
@@ -350,6 +389,7 @@ type TweetResponse struct {
 	GlobalObjects struct {
 		Tweets        map[string]APITweet        `json:"tweets"`
 		Users         map[string]APIUser         `json:"users"`
+		Notifications map[string]APINotification `json:"notifications"`
 	} `json:"globalObjects"`
 	Timeline struct {
 		Instructions []struct {
@@ -458,13 +498,15 @@ func (t *TweetResponse) HandleTombstones() []UserHandle {
 }
 
 func (t *TweetResponse) GetCursor() string {
-	entries := t.Timeline.Instructions[0].AddEntries.Entries
-	if len(entries) > 0 {
-		last_entry := entries[len(entries)-1]
+	// TODO: is this function used anywhere other than Notifications?
+	for _, instr := range t.Timeline.Instructions {
+		if len(instr.AddEntries.Entries) > 0 {
+			last_entry := instr.AddEntries.Entries[len(instr.AddEntries.Entries)-1]
 			if strings.Contains(last_entry.EntryID, "cursor") {
 				return last_entry.Content.Operation.Cursor.Value
 			}
 		}
+	}
 
 	// Next, try the other format ("replaceEntry")
 	instructions := t.Timeline.Instructions
@@ -523,6 +565,10 @@ func (t *TweetResponse) ToTweetTrove() (TweetTrove, error) {
 		}
 		ret.Users[new_user.ID] = new_user
 	}
+	for _, n := range t.GlobalObjects.Notifications {
+		new_notification := ParseSingleNotification(n)
+		ret.Notifications[new_notification.ID] = new_notification
+	}
 	return ret, nil
 }
 

pkg/scraper/api_types_notifications.go

@@ -2,6 +2,9 @@ package scraper
 
 import (
 	"net/url"
+	"strings"
+	"regexp"
+	"sort"
 )
 
 func (api API) GetNotifications(cursor string) (TweetResponse, error) {
@@ -19,3 +22,103 @@ func (api API) GetNotifications(cursor string) (TweetResponse, error) {
 
 	return result, err
 }
+
+func (t *TweetResponse) ToTweetTroveAsNotifications() (TweetTrove, error) {
+	ret, err := t.ToTweetTrove()
+	if err != nil {
+		return TweetTrove{}, err
+	}
+
+	// Find the "addEntries" instruction
+	for _, instr := range t.Timeline.Instructions {
+		sort.Sort(instr.AddEntries.Entries)
+		for _, entry := range instr.AddEntries.Entries {
+			id_re := regexp.MustCompile(`notification-([\w-]+)`)
+			matches := id_re.FindStringSubmatch(entry.EntryID)
+			if matches == nil || len(matches) == 1 {
+				// Not a notification entry
+				continue
+			}
+			notification_id := matches[1]
+			notification, is_ok := ret.Notifications[NotificationID(notification_id)]
+			if !is_ok {
+				// Tweet entry (e.g., someone replied to you)
+				notification = Notification{ID: NotificationID(notification_id)}
+			}
+			notification.SortIndex = entry.SortIndex
+			if strings.Contains(entry.Content.Item.ClientEventInfo.Element, "replied") {
+				notification.Type = 4
+			} else if strings.Contains(entry.Content.Item.ClientEventInfo.Element, "recommended") {
+				notification.Type = 11
+			}
+			if entry.Content.Item.Content.Tweet.ID != 0 {
+				notification.ActionTweetID = TweetID(entry.Content.Item.Content.Tweet.ID)
+				notification.ActionUserID = UserID(ret.Tweets[notification.ActionTweetID].UserID)
+			}
+
+			if entry.Content.Item.Content.Notification.ID != "" {
+				notification.UserIDs = []UserID{}
+				for _, u_id := range entry.Content.Item.Content.Notification.FromUsers {
+					notification.UserIDs = append(notification.UserIDs, UserID(u_id))
+					notification.ActionUserID = UserID(u_id)
+				}
+
+				notification.TweetIDs = []TweetID{}
+				for _, t_id := range entry.Content.Item.Content.Notification.TargetTweets {
+					notification.TweetIDs = append(notification.TweetIDs, TweetID(t_id))
+					notification.ActionTweetID = TweetID(t_id)
+				}
+			}
+			ret.Notifications[notification.ID] = notification
+		}
+	}
+	return ret, err
+}
+
+func ParseSingleNotification(n APINotification) Notification {
+	ret := Notification{}
+	ret.ID = NotificationID(n.ID)
+
+	for i := len(n.Message.Entities) - 1; i >= 0; i -= 1 {
+		from := n.Message.Entities[i].FromIndex
+		to := n.Message.Entities[i].ToIndex
+
+		runetext := []rune(n.Message.Text)
+
+		n.Message.Text = string(runetext[0:from]) + string(runetext[to:])
+	}
+	// t.Entities.ReplyMentions = strings.TrimSpace(string([]rune(t.FullText)[0:t.DisplayTextRange[0]]))
+
+	if strings.HasSuffix(n.Message.Text, "followed you") {
+		ret.Type = 5
+	} else if strings.Contains(n.Message.Text, "liked") {
+		ret.Type = 1
+	} else if strings.Contains(n.Message.Text, "reposted") {
+		ret.Type = 2
+	} else if strings.Contains(n.Message.Text, "There was a login to your account") {
+		ret.Type = 9
+	} else {
+		// TODO: more types?
+	}
+
+	ret.SentAt = TimestampFromUnixMilli(n.TimestampMs)
+	// TODO: caller should set ret.UserID
+	ret.UserIDs = []UserID{}
+	for _, u := range n.Template.AggregateUserActionsV1.FromUsers {
+		ret.UserIDs = append(ret.UserIDs, UserID(u.User.ID))
+	}
+
+	target_objs := n.Template.AggregateUserActionsV1.TargetObjects
+	if len(target_objs) > 0 {
+		if strings.HasSuffix(n.Message.Text, "liked your repost") {
+			// Retweet
+			ret.ActionRetweetID = TweetID(target_objs[0].Tweet.ID)
+		} else {
+			// Normal tweet
+			ret.ActionTweetID = TweetID(target_objs[0].Tweet.ID)
+			ret.TweetIDs = []TweetID{TweetID(target_objs[0].Tweet.ID)}
+		}
+	}
+
+	return ret
+}

pkg/scraper/api_types_notifications_test.go

@@ -0,0 +1,77 @@
+package scraper_test
+
+import (
+	"testing"
+
+	"encoding/json"
+	"os"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	. "gitlab.com/offline-twitter/twitter_offline_engine/pkg/scraper"
+)
+
+func TestParseNotificationsPage(t *testing.T) {
+	assert := assert.New(t)
+	require := require.New(t)
+	data, err := os.ReadFile("test_responses/notifications/notifications_response_first_page.json")
+	require.NoError(err)
+
+	var resp TweetResponse
+	err = json.Unmarshal(data, &resp)
+	require.NoError(err)
+
+	tweet_trove, err := resp.ToTweetTroveAsNotifications()
+	require.NoError(err)
+
+	notif1, is_ok := tweet_trove.Notifications["FKncQJGVgAQAAAABSQ3bEYsN6BFN3re-ZsU"]
+	assert.True(is_ok)
+	assert.Equal(9, notif1.Type) // login
+
+	notif2, is_ok := tweet_trove.Notifications["FKncQJGVgAQAAAABSQ3bEYsN6BFaOkNV8aw"]
+	assert.True(is_ok)
+	assert.Equal(2, notif2.Type) // retweet
+	assert.Equal(UserID(1458284524761075714), notif2.ActionUserID)
+	assert.Equal(TweetID(1824915465275392037), notif2.ActionTweetID)
+	assert.Equal(TimestampFromUnixMilli(1723928739342), notif2.SentAt)
+
+	notif3, is_ok := tweet_trove.Notifications["FKncQJGVgAQAAAABSQ3bEYsN6BE-OY688aw"]
+	assert.True(is_ok)
+	assert.Equal(1, notif3.Type) // like
+	assert.Equal(UserID(1458284524761075714), notif3.ActionUserID)
+	assert.Equal(TweetID(1824915465275392037), notif3.ActionTweetID)
+
+	notif4, is_ok := tweet_trove.Notifications["FKncQJGVgAQAAAABSQ3bEYsN6BGLlh8UIQs"]
+	assert.True(is_ok)
+	assert.Equal(11, notif4.Type) // recommended
+
+	notif5, is_ok := tweet_trove.Notifications["FKncQJGVgAQAAAABSQ3bEYsN6BHS11EvITw"]
+	assert.True(is_ok)
+	assert.Equal(5, notif5.Type) // followed you
+	assert.Equal(UserID(28815778), notif5.ActionUserID)
+
+	notif6, is_ok := tweet_trove.Notifications["FKncQJGVgAQAAAABSQ3bEYsN6BE5ujkCepo"]
+	assert.True(is_ok)
+	assert.Equal(1, notif6.Type)
+	assert.Equal(UserID(1458284524761075714), notif6.ActionUserID)
+	assert.Equal(TweetID(1826778617705115868), notif6.ActionTweetID)
+	assert.Contains(notif6.UserIDs, UserID(1458284524761075714))
+	assert.Contains(notif6.UserIDs, UserID(2694459866))
+
+	// Check users
+	for _, u_id := range []UserID{1458284524761075714, 28815778} {
+		_, is_ok := tweet_trove.Users[u_id]
+		assert.True(is_ok)
+	}
+
+	// Check tweets
+	for _, t_id := range []TweetID{1824915465275392037, 1826778617705115868} {
+		_, is_ok := tweet_trove.Tweets[t_id]
+		assert.True(is_ok)
+	}
+
+	// Test cursor-bottom
+	bottom_cursor := resp.GetCursor()
+	assert.Equal("DAACDAABCgABFKncQJGVgAQIAAIAAAABCAADSQ3bEQgABIsN6BEACwACAAAAC0FaRkxRSXFNLTJJAAA", bottom_cursor)
+}

pkg/scraper/notification.go

@@ -0,0 +1,18 @@
+package scraper
+
+type NotificationID string
+
+type Notification struct {
+	ID        NotificationID
+	Type      int
+	SentAt    Timestamp
+	SortIndex int64
+	UserID    UserID // recipient of the notification
+
+	ActionUserID    UserID
+	ActionTweetID   TweetID
+	ActionRetweetID TweetID
+
+	TweetIDs []TweetID
+	UserIDs  []UserID
+}

pkg/scraper/tweet_trove.go

@@ -14,6 +14,7 @@ type TweetTrove struct {
 	Spaces        map[SpaceID]Space
 	Likes         map[LikeSortID]Like
 	Bookmarks     map[BookmarkSortID]Bookmark
+	Notifications map[NotificationID]Notification
 
 	TombstoneUsers []UserHandle
 
@@ -30,6 +31,7 @@ func NewTweetTrove() TweetTrove {
 	ret.Spaces = make(map[SpaceID]Space)
 	ret.Likes = make(map[LikeSortID]Like)
 	ret.Bookmarks = make(map[BookmarkSortID]Bookmark)
+	ret.Notifications = make(map[NotificationID]Notification)
 	ret.TombstoneUsers = []UserHandle{}
 	ret.Rooms = make(map[DMChatRoomID]DMChatRoom)
 	ret.Messages = make(map[DMMessageID]DMMessage)
@@ -70,6 +72,9 @@ func (t1 *TweetTrove) MergeWith(t2 TweetTrove) {
 	for id, val := range t2.Bookmarks {
 		t1.Bookmarks[id] = val
 	}
+	for id, val := range t2.Notifications {
+		t1.Notifications[id] = val
+	}
 
 	t1.TombstoneUsers = append(t1.TombstoneUsers, t2.TombstoneUsers...)