From cac87ec973810d45fac59422b04067708f425b65 Mon Sep 17 00:00:00 2001
From: Alessio <alex@playfulpachyderm.com>
Date: Mon, 10 Jun 2024 15:47:13 -0700
Subject: [PATCH] Save tweets with unknown user as <UNKNOWN USER> instead of
 UserID 0, which causes foreign key errors

---
 pkg/scraper/api_types_v2.go      | 10 +++++++---
 pkg/scraper/api_types_v2_test.go | 11 ++++++++++-
 2 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/pkg/scraper/api_types_v2.go b/pkg/scraper/api_types_v2.go
index a9d0c4c..45c9b41 100644
--- a/pkg/scraper/api_types_v2.go
+++ b/pkg/scraper/api_types_v2.go
@@ -522,7 +522,7 @@ func (e APIV2Entry) ToTweetTrove() TweetTrove {
 		}
 		ret, err := e.Content.ItemContent.TweetResults.ToTweetTrove()
 
-		// Handle tombstones in parent reply thread
+		// Handle tombstones in the focused tweet and parent reply thread
 		if errors.Is(err, ErrorIsTombstone) {
 			ret = NewTweetTrove() // clear the result just in case there is a TweetID(0) in it
 			tombstoned_tweet := APITweet{}
@@ -547,6 +547,10 @@ func (e APIV2Entry) ToTweetTrove() TweetTrove {
 			if err != nil {
 				panic(err)
 			}
+
+			fake_user := GetUnknownUser()
+			ret.Users[fake_user.ID] = fake_user
+			parsed_tombstone_tweet.UserID = fake_user.ID
 			ret.Tweets[parsed_tombstone_tweet.ID] = parsed_tombstone_tweet
 		} else if err != nil {
 			if e.Content.ItemContent.APIV2UserResult.UserResults.Result.ID != 0 {
@@ -764,9 +768,9 @@ func (api_response APIV2Response) ToTweetTrove() (TweetTrove, error) {
 
 		// Fill out the replied tweet's UserID using this tweet's "in_reply_to_user_id".
 		// If this tweet doesn't have it (i.e., this tweet is also a tombstone), create a fake user instead, and add it to the tweet trove.
-		if replied_tweet.UserID == 0 {
+		if replied_tweet.UserID == 0 || replied_tweet.UserID == GetUnknownUser().ID {
 			replied_tweet.UserID = tweet.in_reply_to_user_id
-			if replied_tweet.UserID == 0 {
+			if replied_tweet.UserID == 0 || replied_tweet.UserID == GetUnknownUser().ID {
 				fake_user := GetUnknownUser()
 				ret.Users[fake_user.ID] = fake_user
 				replied_tweet.UserID = fake_user.ID
diff --git a/pkg/scraper/api_types_v2_test.go b/pkg/scraper/api_types_v2_test.go
index 33010ac..ce4e827 100644
--- a/pkg/scraper/api_types_v2_test.go
+++ b/pkg/scraper/api_types_v2_test.go
@@ -584,12 +584,21 @@ func TestAPIV2TombstoneEntry(t *testing.T) {
 	trove := entry.ToTweetTrove()
 	assert.NoError(err)
 	assert.Len(trove.Tweets, 1)
-	assert.Len(trove.Users, 0)
+	assert.Len(trove.Users, 1)
 	assert.Len(trove.Retweets, 0)
 
 	tweet, is_ok := trove.Tweets[1454515503242829830]
 	assert.True(is_ok)
+	assert.True(tweet.IsStub)
+	assert.Equal("hidden", tweet.TombstoneType)
 	assert.Equal(tweet.ID, TweetID(1454515503242829830))
+
+	expected_user := GetUnknownUser()
+	assert.Equal(expected_user.ID, tweet.UserID)
+
+	assert.Len(trove.Users, 1)
+	_, is_ok = trove.Users[expected_user.ID]
+	assert.True(is_ok)
 }
 
 func TestAPIV2UserFeedTombstoneEntry(t *testing.T) {