From 3218765413e5bd6ab5c602a0ad19d5eb5f5902f0 Mon Sep 17 00:00:00 2001
From: Hassan Syyid <hassansyyid@gmail.com>
Date: Thu, 10 Oct 2024 00:37:29 +0200
Subject: [PATCH] HG-3499: parse empty strings as booleans

---
 tap_hubspot_beta/client_base.py | 28 +++++++++++++++++++++++++++-
 tap_hubspot_beta/client_v1.py   |  1 +
 tap_hubspot_beta/client_v2.py   |  1 +
 tap_hubspot_beta/client_v3.py   |  3 +++
 4 files changed, 32 insertions(+), 1 deletion(-)

diff --git a/tap_hubspot_beta/client_base.py b/tap_hubspot_beta/client_base.py
index f6e5d4b..1ea98f0 100644
--- a/tap_hubspot_beta/client_base.py
+++ b/tap_hubspot_beta/client_base.py
@@ -331,7 +331,33 @@ def stream_maps(self) -> List[StreamMap]:
                 )
             ]
         return self._stream_maps
-    
+
+    def process_row_types(self,row) -> Dict[str, Any]:
+        schema = self.schema['properties']
+        # If the row is null we ignore
+        if row is None:
+            return row
+
+        for field, value in row.items():
+            if field not in schema:
+                # Skip fields not found in the schema
+                continue
+
+            field_info = schema[field]
+            field_type = field_info.get("type", ["null"])[0]
+
+            if field_type == "boolean":
+                if value is None:
+                    row[field] = False
+                elif not isinstance(value, bool):
+                    # Attempt to cast to boolean
+                    if value.lower() == "true":
+                        row[field] = True
+                    elif value == "" or value.lower() == "false":
+                        row[field] = False
+
+        return row
+
     def is_first_sync(self):
         if self.stream_state.get("replication_key"):
             return False
diff --git a/tap_hubspot_beta/client_v1.py b/tap_hubspot_beta/client_v1.py
index b579c46..aeeb3a2 100644
--- a/tap_hubspot_beta/client_v1.py
+++ b/tap_hubspot_beta/client_v1.py
@@ -73,4 +73,5 @@ def post_process(self, row: dict, context: Optional[dict]) -> dict:
                         dt_field = datetime.fromtimestamp(int(row[field]) / 1000)
                         dt_field = dt_field.replace(tzinfo=None)
                         row[field] = dt_field.isoformat()
+        row = self.process_row_types(row)
         return row
diff --git a/tap_hubspot_beta/client_v2.py b/tap_hubspot_beta/client_v2.py
index ef4059d..fffd14f 100644
--- a/tap_hubspot_beta/client_v2.py
+++ b/tap_hubspot_beta/client_v2.py
@@ -44,6 +44,7 @@ def post_process(self, row: dict, context: Optional[dict]) -> dict:
         row["updatedAt"] = row["hs_lastmodifieddate"]
         row["createdAt"] = row["createdate"]
         row["archived"] = False
+        row = self.process_row_types(row)
         return row
 
     def request_records(self, context: Optional[dict]) -> Iterable[dict]:
diff --git a/tap_hubspot_beta/client_v3.py b/tap_hubspot_beta/client_v3.py
index 1d6585d..81559f4 100644
--- a/tap_hubspot_beta/client_v3.py
+++ b/tap_hubspot_beta/client_v3.py
@@ -116,6 +116,7 @@ def post_process(self, row: dict, context: Optional[dict]) -> dict:
             del row["properties"]
         # store archived value in _hg_archived
         row["_hg_archived"] = False
+        row = self.process_row_types(row)
         return row
 
     def _sync_records(  # noqa C901  # too complex
@@ -241,6 +242,7 @@ def post_process(self, row: dict, context: Optional[dict]) -> dict:
             for name, value in row["properties"].items():
                 row[name] = value
             del row["properties"]
+        row = self.process_row_types(row)
         return row
 
 
@@ -296,6 +298,7 @@ def post_process(self, row: dict, context: Optional[dict]) -> dict:
             for name, value in row["properties"].items():
                 row[name] = value
             del row["properties"]
+        row = self.process_row_types(row)
         return row
     
 class hubspotHistoryV3Stream(hubspotV3Stream):