Add 'fetch_dm' subcommand

cmd/tests.sh

@@ -377,6 +377,8 @@ tw fetch_inbox
 test $(sqlite3 twitter.db "select count(*) from chat_rooms") -ge "1"
 test $(sqlite3 twitter.db "select count(*) from chat_messages where chat_room_id = '1458284524761075714-1488963321701171204'") -ge "5"
 
+# Test fetch a DM conversation
+tw fetch_dm "1458284524761075714-1488963321701171204"
 
 # TODO: Maybe this file should be broken up into multiple test scripts
 

cmd/twitter/main.go

@@ -161,6 +161,8 @@ func main() {
 		start_webserver(*addr)
 	case "fetch_inbox":
 		fetch_inbox(*how_many)
+	case "fetch_dm":
+		fetch_dm(target, *how_many)
 	default:
 		die(fmt.Sprintf("Invalid operation: %s", operation), true, 3)
 	}
@@ -403,3 +405,14 @@ func fetch_inbox(how_many int) {
 	profile.SaveDMTrove(trove)
 	happy_exit(fmt.Sprintf("Saved %d messages from %d chats", len(trove.Messages), len(trove.Rooms)))
 }
+
+func fetch_dm(id string, how_many int) {
+	room, err := profile.GetChatRoom(scraper.DMChatRoomID(id))
+	if err != nil {
+		panic(err)
+	}
+	max_id := scraper.DMMessageID(^uint(0) >> 1)
+	trove := scraper.GetConversation(room.ID, max_id, how_many)
+	profile.SaveDMTrove(trove)
+	happy_exit(fmt.Sprintf("Saved %d messages from %d chats", len(trove.Messages), len(trove.Rooms)))
+}

pkg/scraper/api_types_dms.go

@@ -67,8 +67,9 @@ type APIInbox struct {
 }
 
 type APIDMResponse struct {
-	InboxInitialState APIInbox `json:"inbox_initial_state"`
-	InboxTimeline     APIInbox `json:"inbox_timeline"`
+	InboxInitialState    APIInbox `json:"inbox_initial_state"`
+	InboxTimeline        APIInbox `json:"inbox_timeline"`
+	ConversationTimeline APIInbox `json:"conversation_timeline"`
 }
 
 func (r APIInbox) ToDMTrove() DMTrove {
@@ -212,3 +213,61 @@ func (api *API) GetInboxTrusted(oldest_id int) (APIInbox, error) {
 	err = api.do_http(url.String(), "", &result)
 	return result.InboxTimeline, err
 }
+
+func (api *API) GetDMConversation(id DMChatRoomID, max_id DMMessageID) (APIInbox, error) {
+	url, err := url.Parse("https://twitter.com/i/api/1.1/dm/conversation/" + string(id) + ".json")
+	if err != nil {
+		panic(err)
+	}
+	query := url.Query()
+	query.Add("max_id", fmt.Sprint(max_id))
+	query.Add("context", "FETCH_DM_CONVERSATION_HISTORY")
+	query.Add("include_profile_interstitial_type", "1")
+	query.Add("include_blocking", "1")
+	query.Add("include_blocked_by", "1")
+	query.Add("include_followed_by", "1")
+	query.Add("include_want_retweets", "1")
+	query.Add("include_mute_edge", "1")
+	query.Add("include_can_dm", "1")
+	query.Add("include_can_media_tag", "1")
+	query.Add("include_ext_has_nft_avatar", "1")
+	query.Add("include_ext_is_blue_verified", "1")
+	query.Add("include_ext_verified_type", "1")
+	query.Add("include_ext_profile_image_shape", "1")
+	query.Add("skip_status", "1")
+	query.Add("dm_secret_conversations_enabled", "false")
+	query.Add("krs_registration_enabled", "true")
+	query.Add("cards_platform", "Web-12")
+	query.Add("include_cards", "1")
+	query.Add("include_ext_alt_text", "true")
+	query.Add("include_ext_limited_action_results", "true")
+	query.Add("include_quote_count", "true")
+	query.Add("include_reply_count", "1")
+	query.Add("tweet_mode", "extended")
+	query.Add("include_ext_views", "true")
+	query.Add("dm_users", "false")
+	query.Add("include_groups", "true")
+	query.Add("include_inbox_timelines", "true")
+	query.Add("include_ext_media_color", "true")
+	query.Add("supports_reactions", "true")
+	query.Add("include_conversation_info", "true")
+	query.Add("ext", strings.Join([]string{
+		"mediaColor",
+		"altText",
+		"mediaStats",
+		"highlightedLabel",
+		"hasNftAvatar",
+		"voiceInfo",
+		"birdwatchPivot",
+		"enrichments",
+		"superFollowMetadata",
+		"unmentionInfo",
+		"editControl",
+		"vibe",
+	}, ","))
+	url.RawQuery = query.Encode()
+
+	var result APIDMResponse
+	err = api.do_http(url.String(), "", &result)
+	return result.ConversationTimeline, err
+}

pkg/scraper/api_types_dms_test.go

@@ -116,3 +116,37 @@ func TestParseInbox(t *testing.T) {
 	assert.True(is_ok)
 	assert.Equal(room.ID, room_id)
 }
+
+func TestParseDMRoomResponse(t *testing.T) {
+	assert := assert.New(t)
+	data, err := os.ReadFile("test_responses/dms/dm_conversation_response.json")
+	require.NoError(t, err)
+
+	var inbox APIDMResponse
+	err = json.Unmarshal(data, &inbox)
+	require.NoError(t, err)
+
+	trove := inbox.ConversationTimeline.ToDMTrove()
+
+	for _, id := range []DMMessageID{
+		1663623062195957773,
+		1663623203644751885,
+		1665922180176044037,
+		1665936253483614212,
+		1726009944393372005,
+	} {
+		m, is_ok := trove.Messages[id]
+		assert.True(is_ok, "Message with ID %d not in the trove!")
+		assert.Equal(m.ID, id)
+	}
+	for _, id := range []UserID{1458284524761075714, 1488963321701171204} {
+		u, is_ok := trove.TweetTrove.Users[id]
+		assert.True(is_ok, "User with ID %d not in the trove!")
+		assert.Equal(u.ID, id)
+	}
+	room_id := DMChatRoomID("1458284524761075714-1488963321701171204")
+	room, is_ok := trove.Rooms[room_id]
+	assert.True(is_ok)
+	assert.Equal(room.ID, room_id)
+	assert.Equal(trove.GetOldestMessage(room_id), DMMessageID(1663623062195957773))
+}

pkg/scraper/dm_trove.go

@@ -28,6 +28,16 @@ func (t1 *DMTrove) MergeWith(t2 DMTrove) {
 	t1.TweetTrove.MergeWith(t2.TweetTrove)
 }
 
+func (t DMTrove) GetOldestMessage(id DMChatRoomID) DMMessageID {
+	oldest := DMMessageID(^uint(0) >> 1) // Max integer
+	for _, m := range t.Messages {
+		if m.ID < oldest && m.DMChatRoomID == id {
+			oldest = m.ID
+		}
+	}
+	return oldest
+}
+
 // Returns a DMTrove and the cursor for the next update
 func GetInbox(how_many int) (DMTrove, string) {
 	if !the_api.IsAuthenticated {
@@ -53,3 +63,27 @@ func GetInbox(how_many int) (DMTrove, string) {
 
 	return trove, cursor
 }
+
+func GetConversation(id DMChatRoomID, max_id DMMessageID, how_many int) DMTrove {
+	if !the_api.IsAuthenticated {
+		log.Fatalf("Fetching DMs can only be done when authenticated.  Please provide `--session [user]`")
+	}
+	dm_response, err := the_api.GetDMConversation(id, max_id)
+	if err != nil {
+		panic(err)
+	}
+
+	trove := dm_response.ToDMTrove()
+	oldest := trove.GetOldestMessage(id)
+	for len(trove.Messages) < how_many && dm_response.Status != "AT_END" {
+		dm_response, err = the_api.GetDMConversation(id, oldest)
+		if err != nil {
+			panic(err)
+		}
+		next_trove := dm_response.ToDMTrove()
+		oldest = next_trove.GetOldestMessage(id)
+		trove.MergeWith(next_trove)
+	}
+
+	return trove
+}