create an instance of each necessary validator when handling schema i…

…nstead to run it in the loop (#19)

target_bigquery/processhandler.py

@@ -12,7 +12,7 @@
 from google.cloud.bigquery import WriteDisposition
 from google.cloud.bigquery.job import SourceFormat
 from google.cloud.exceptions import NotFound
-from jsonschema import validate
+from jsonschema.validators import validator_for
 
 from target_bigquery.encoders import DecimalEncoder
 from target_bigquery.schema import build_schema, cleanup_record, format_record_to_schema
@@ -45,6 +45,7 @@ def __init__(self, logger, **kwargs):
 
         self.tables = {}
         self.schemas = {}
+        self.validators = {}
         self.key_properties = {}
         self.bq_schemas = {}
         self.bq_schema_dicts = {}
@@ -84,6 +85,9 @@ def handle_schema_message(self, msg):
             self.table_prefix, msg.stream, self.table_suffix
         )
         self.schemas[msg.stream] = msg.schema
+        validator_cls = validator_for(msg.schema)
+        validator_cls.check_schema(msg.schema)
+        self.validators[msg.stream] = validator_cls(msg.schema)
         self.key_properties[msg.stream] = msg.key_properties
 
         validate_json_schema_completeness(self.schemas[msg.stream])
@@ -194,11 +198,12 @@ def handle_record_message(self, msg):
         # schema validation may fail if data doesn't match schema in terms of data types
         # in this case, we validate schema again on data which has been forced to match schema
         # nr is based on msg.record, but all data from msg.record has been forced to match schema
+        validator = self.validators[stream]
         if self.validate_records:
             try:
-                validate(msg.record, schema)
+                validator.validate(msg.record, schema)
             except Exception as e:
-                validate(nr, schema)
+                validator.validate(nr, schema)
 
         if self.add_metadata_columns:
             nr["_time_extracted"] = msg.time_extracted.isoformat() \