From dd91d7975ec11ca1d8912050e68f67f99b47387b Mon Sep 17 00:00:00 2001
From: jochen <jochen.christ@innoq.com>
Date: Thu, 16 May 2024 17:32:12 +0200
Subject: [PATCH] Remove jsonschema dependency (we use fastjsonschema)

---
 datacontract/imports/jsonschema_importer.py   | 18 ++++++++++++++----
 pyproject.toml                                |  1 -
 tests/fixtures/import/orders-datacontract.yml |  1 -
 tests/fixtures/import/orders.json             |  3 +--
 tests/test_import_jsonschema.py               |  1 +
 5 files changed, 16 insertions(+), 8 deletions(-)

diff --git a/datacontract/imports/jsonschema_importer.py b/datacontract/imports/jsonschema_importer.py
index e185eaa5..c8ef2977 100644
--- a/datacontract/imports/jsonschema_importer.py
+++ b/datacontract/imports/jsonschema_importer.py
@@ -1,8 +1,10 @@
+import json
+
+import fastjsonschema
+
 from datacontract.model.data_contract_specification import \
     DataContractSpecification, Model, Field, Definition
 from datacontract.model.exceptions import DataContractException
-import json
-import jsonschema
 
 
 def convert_json_schema_properties(properties, is_definition=False):
@@ -72,8 +74,8 @@ def import_jsonschema(data_contract_specification: DataContractSpecification, so
     try:
         with open(source, "r") as file:
             json_schema = json.loads(file.read())
-            validator = jsonschema.Draft7Validator({})
-            validator.check_schema(json_schema)
+            validator = fastjsonschema.compile({})
+            validator(json_schema)
 
             model = Model(
                 description=json_schema.get('description'),
@@ -127,6 +129,14 @@ def import_jsonschema(data_contract_specification: DataContractSpecification, so
                     definition = Definition(name=def_name, **definition_kwargs)
                     data_contract_specification.definitions[def_name] = definition
 
+    except fastjsonschema.JsonSchemaException as e:
+        raise DataContractException(
+            type="schema",
+            name="Parse json schema",
+            reason=f"Failed to parse json schema from {source}: {e}",
+            engine="datacontract"
+        )
+
     except Exception as e:
         raise DataContractException(
             type="schema",
diff --git a/pyproject.toml b/pyproject.toml
index 59dffb99..590240a6 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -41,7 +41,6 @@ dependencies = [
   "deltalake~=0.17.0",
   "boto3>=1.34.41,<1.34.99",
   "botocore>=1.34.41,<1.34.99",
-  "jsonschema>=4.22",
   "jinja_partials >= 0.2.1"
 ]
 
diff --git a/tests/fixtures/import/orders-datacontract.yml b/tests/fixtures/import/orders-datacontract.yml
index c51be0f2..93b38fc6 100644
--- a/tests/fixtures/import/orders-datacontract.yml
+++ b/tests/fixtures/import/orders-datacontract.yml
@@ -12,7 +12,6 @@ models:
       order_id:
         title: Order ID
         type: string
-        format: uuid
         required: true
         description: Unique identifier for the order
       order_timestamp:
diff --git a/tests/fixtures/import/orders.json b/tests/fixtures/import/orders.json
index 1e4ed27a..03d12050 100644
--- a/tests/fixtures/import/orders.json
+++ b/tests/fixtures/import/orders.json
@@ -1,12 +1,11 @@
 {
-  "$schema": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema",
   "title": "OrderSchema",
   "description": "Schema for order details",
   "type": "object",
   "properties": {
     "order_id": {
       "type": "string",
-      "format": "uuid",
       "title": "Order ID",
       "description": "Unique identifier for the order"
     },
diff --git a/tests/test_import_jsonschema.py b/tests/test_import_jsonschema.py
index 9748b941..0273d8d9 100644
--- a/tests/test_import_jsonschema.py
+++ b/tests/test_import_jsonschema.py
@@ -34,6 +34,7 @@ def test_import_json_schema_orders():
     assert yaml.safe_load(result.to_yaml()) == yaml.safe_load(expected)
     assert DataContract(data_contract_str=expected).lint(enabled_linters="none").has_passed()
 
+
 def test_import_json_schema_football():
     result = DataContract().import_from_source("jsonschema", "fixtures/import/football.json")