refactor code

Author: darynaishchenko

airbyte_cdk/sources/file_based/config/file_based_stream_config.py

@@ -74,9 +74,9 @@ class FileBasedStreamConfig(BaseModel):
         default=None,
         gt=0,
     )
-    use_first_found_file_for_schema_discovery: Optional[bool] = Field(
+    use_first_found_file_for_schema_discovery: bool = Field(
         title="Use first found file for schema discovery",
-        description="When enable, the source will use the first found file for schema discovery. Helps to avoid long discovery step",
+        description="When enabled, the source will use the first found file for schema discovery. Helps to avoid long discovery step",
         default=False,
     )
 

airbyte_cdk/sources/file_based/stream/default_file_based_stream.py

@@ -274,8 +274,10 @@ def _get_raw_json_schema(self) -> JsonSchema:
         elif self.config.schemaless:
             return schemaless_schema
         elif self.config.use_first_found_file_for_schema_discovery:
-            self.logger.info(msg=f"Using only first found file for schema discovery.")
-            files = [next(iter(self.get_files()))]
+            self.logger.info(
+                msg=f"Using only first found file for schema discovery for stream {self.name} due to limitation in config."
+            )
+            files = list(itertools.islice(self.get_files(), 1))
             first_n_files = len(files)
         else:
             files = self.list_files()

unit_tests/sources/file_based/stream/test_default_file_based_stream.py

@@ -239,6 +239,8 @@ def test_use_first_found_file_for_schema_discovery(self) -> None:
         self._stream_reader.get_matching_files.return_value = files
 
         schema = self._stream.get_json_schema()
+        assert self._parser.infer_schema.call_count == 1
+        assert self._parser.infer_schema.call_args[0][1].uri == "file0"
         assert schema == {
             "properties": {
                 "_ab_source_file_last_modified": {"type": "string"},