Fix the script

catalog/dags/providers/provider_api_scripts/justtakeitfree.py

@@ -31,7 +31,7 @@ def get_next_query_params(self, prev_query_params: dict | None, **kwargs) -> dic
         # On the first request, `prev_query_params` will be `None`. We can detect this
         # and return our default params.
         if not prev_query_params:
-            return {"page": 0, "key": Variable.get("API_KEY_JUSTTAKEITFREE")}
+            return {"page": 1, "key": Variable.get("API_KEY_JUSTTAKEITFREE")}
         else:
             return {
                 **prev_query_params,
@@ -40,13 +40,14 @@ def get_next_query_params(self, prev_query_params: dict | None, **kwargs) -> dic
 
     def get_batch_data(self, response_json):
         if response_json and (data := response_json.get("data")):
-            return data[0]
+            return data
         return None
 
     def get_media_type(self, record: dict):
         return IMAGE
 
-    def get_record_data(self, data: dict) -> dict | list[dict] | None:
+    def get_record_data(self, data: list[dict]) -> dict | None:
+        data = data[0]
         if not (foreign_identifier := data.get("page_link", "").split("/")[-2]):
             logger.debug("Skipping record with missing foreign_identifier")
             return None
@@ -71,7 +72,9 @@ def get_record_data(self, data: dict) -> dict | list[dict] | None:
         creator_url = "https://justtakeitfree.com"
         raw_tags = data.get("tags")
 
-        return {
+        filesize = self.get_file_info(url)
+
+        record_data = {
             "foreign_landing_url": foreign_landing_url,
             "url": url,
             "license_info": license_info,
@@ -80,7 +83,16 @@ def get_record_data(self, data: dict) -> dict | list[dict] | None:
             "creator": creator,
             "creator_url": creator_url,
             "raw_tags": raw_tags,
+            "filesize": filesize,
         }
+        return record_data
+
+    def get_file_info(self, url):
+        """Get the image size in bytes."""
+        resp = self.delayed_requester.head(url)
+        if resp:
+            filesize = int(resp.headers.get("Content-Length", 0))
+            return filesize if filesize != 0 else None
 
 
 def main():