From 134e704577a093a532b6eecb0805c8b0a4669dc5 Mon Sep 17 00:00:00 2001
From: Parker-Kasiewicz <parkerkasiewicz@gmail.com>
Date: Sat, 2 Nov 2024 19:42:02 -0700
Subject: [PATCH] updated schema & null behavior for hashtags & urls to
 biqquery

---
 pkg/hydrator/hydrator.go       |  2 ++
 pkg/output/bq/schema/schema.go | 30 ++++++++++++++++++++++++++++++
 2 files changed, 32 insertions(+)

diff --git a/pkg/hydrator/hydrator.go b/pkg/hydrator/hydrator.go
index 472707b..aa7ad47 100644
--- a/pkg/hydrator/hydrator.go
+++ b/pkg/hydrator/hydrator.go
@@ -251,6 +251,8 @@ func (h *Hydrator) flattenFullProfile(profile *bsky.ActorDefs_ProfileViewDetaile
 }
 
 func (h *Hydrator) flattenFacets(facets []*bsky.RichtextFacet) (hashtags []string, urls []string) {
+	hashtags = []string{}
+	urls = []string{}
 	if facets != nil {
 			for _, facet := range facets {
 					if facet != nil {
diff --git a/pkg/output/bq/schema/schema.go b/pkg/output/bq/schema/schema.go
index 31007c3..7f6d72c 100644
--- a/pkg/output/bq/schema/schema.go
+++ b/pkg/output/bq/schema/schema.go
@@ -276,6 +276,16 @@ func GetSchema() bigquery.Schema {
 				"name": "RepostCount",
 				"type": "INTEGER"
 			},
+			{
+				"mode": "REPEATED",
+				"name": "Hashtags",
+				"type": "STRING"
+			},
+			{
+				"mode": "REPEATED",
+				"name": "URLs",
+				"type": "STRING"
+			},
 			{
 				"name": "Text",
 				"type": "STRING"
@@ -399,6 +409,16 @@ func GetSchema() bigquery.Schema {
 				"name": "ReplyParentCID",
 				"type": "STRING"
 			},
+			{
+				"mode": "REPEATED",
+				"name": "Hashtags",
+				"type": "STRING"
+			},
+			{
+				"mode": "REPEATED",
+				"name": "URLs",
+				"type": "STRING"
+			},
 			{
 				"name": "Text",
 				"type": "STRING"
@@ -566,6 +586,16 @@ func GetSchema() bigquery.Schema {
 				"name": "RepostCount",
 				"type": "INTEGER"
 			},
+			{
+				"mode": "REPEATED",
+				"name": "Hashtags",
+				"type": "STRING"
+			},
+			{
+				"mode": "REPEATED",
+				"name": "URLs",
+				"type": "STRING"
+			},
 			{
 				"name": "Text",
 				"type": "STRING"