Add "test --examples"

datacontract/cli.py

@@ -96,13 +96,7 @@ def test(
     Run schema and quality tests on configured servers.
     """
     print(f"Testing {location}")
-    if examples:
-        print(f"Using examples")
-        run = DataContract(data_contract_file=location).testExample()
-        _handle_result(run)
-        return
-
-    run = DataContract(data_contract_file=location, publish_url=publish).test()
+    run = DataContract(data_contract_file=location, publish_url=publish, examples=examples).test()
     _handle_result(run)
 
 

datacontract/data_contract.py

@@ -30,13 +30,15 @@ def __init__(
         data_contract_str: str = None,
         data_contract: DataContractSpecification = None,
         server: str = None,
+        examples: bool = False,
         publish_url: str = None,
         spark: str = None,
     ):
         self._data_contract_file = data_contract_file
         self._data_contract_str = data_contract_str
         self._data_contract = data_contract
         self._server = server
+        self._examples = examples
         self._publish_url = publish_url
         self._spark = spark
 
@@ -86,20 +88,27 @@ def test(self) -> Run:
 
             check_that_datacontract_contains_valid_server_configuration(run, data_contract, self._server)
             # TODO check yaml contains models
-            server_name = list(data_contract.servers.keys())[0]
-            server = data_contract.servers.get(server_name)
-            run.log_info(f"Running tests for data contract {data_contract.id} with server {server_name}")
-            run.dataContractId = data_contract.id
-            run.dataContractVersion = data_contract.info.version
-            run.dataProductId = server.dataProductId
-            run.outputPortId = server.outputPortId
-            run.server = server_name
 
-            # 5. check server is supported type
-            # 6. check server credentials are complete
-            if server.format == "json":
-                check_jsonschema(run, data_contract, server)
-            check_soda_execute(run, data_contract, server, self._spark)
+            with tempfile.TemporaryDirectory(prefix="datacontract-cli") as tmp_dir:
+                if self._examples:
+                    server_name = "examples"
+                    server = self._get_examples_server(data_contract, run, tmp_dir)
+                else:
+                    server_name = list(data_contract.servers.keys())[0]
+                    server = data_contract.servers.get(server_name)
+
+                run.log_info(f"Running tests for data contract {data_contract.id} with server {server_name}")
+                run.dataContractId = data_contract.id
+                run.dataContractVersion = data_contract.info.version
+                run.dataProductId = server.dataProductId
+                run.outputPortId = server.outputPortId
+                run.server = server_name
+
+                # 5. check server is supported type
+                # 6. check server credentials are complete
+                if server.format == "json":
+                    check_jsonschema(run, data_contract, server)
+                check_soda_execute(run, data_contract, server, self._spark)
 
         except DataContractException as e:
             run.checks.append(Check(
@@ -129,77 +138,6 @@ def test(self) -> Run:
 
         return run
 
-    def testExample(self) -> Run:
-        run = Run.create_run()
-        try:
-            run.log_info(f"Testing data contract")
-            data_contract = resolve.resolve_data_contract(self._data_contract_file, self._data_contract_str,
-                                                          self._data_contract)
-
-            # TODO check yaml contains models
-            run.log_info(f"Running tests for data contract {data_contract.id} ({data_contract.info.version}) against examples")
-            run.dataContractId = data_contract.id
-            run.dataContractVersion = data_contract.info.version
-            run.server = "examples"
-
-            with tempfile.TemporaryDirectory(prefix="datacontract-cli") as tmp_dir:
-                run.log_info(f"Copying examples to files in temporary directory {tmp_dir}")
-                format = "json"
-
-                for example in data_contract.examples:
-                    format = example.type
-                    p = f"{tmp_dir}/{example.model}.{format}"
-                    run.log_info(f"Creating example file {p}")
-                    with open(p, "w") as f:
-                        content = ""
-                        if format == "json" and type(example.data) is list:
-                            content = json.dumps(example.data)
-                        elif format == "json" and type(example.data) is str:
-                            content = example.data
-                        elif format == "yaml" and type(example.data) is list:
-                            content = yaml.dump(example.data)
-                        elif format == "yaml" and type(example.data) is str:
-                            content = example.data
-                        elif format == "csv":
-                            content = example.data
-                        logging.debug(f"Content: {content}")
-                        f.write(content)
-                path = f"{tmp_dir}" + "/{model}." + format
-                delimiter = "array"
-
-                server = Server(
-                    type="local",
-                    path=path,
-                    format=format,
-                    delimiter=delimiter,
-                )
-                print(server)
-                run.log_info(f"Using {server} for testing the examples")
-
-                # 5. check server is supported type
-                # 6. check server credentials are complete
-                if server.format == "json":
-                    check_jsonschema(run, data_contract, server)
-                check_soda_execute(run, data_contract, server, self._spark)
-
-        except DataContractException as e:
-            run.checks.append(Check(
-                type=e.type,
-                result=e.result,
-                name=e.name,
-                reason=e.reason,
-                engine=e.engine,
-                details=""
-            ))
-            run.log_error(str(e))
-
-
-        run.finish()
-
-        if self._publish_url is not None:
-            publish_datamesh_manager(run, self._publish_url)
-
-        return run
 
     def diff(self, other):
         pass
@@ -216,3 +154,35 @@ def export(self, export_format) -> str:
         else:
             print(f"Export format {export_format} not supported.")
             return ""
+
+    def _get_examples_server(self, data_contract, run, tmp_dir):
+        run.log_info(f"Copying examples to files in temporary directory {tmp_dir}")
+        format = "json"
+        for example in data_contract.examples:
+            format = example.type
+            p = f"{tmp_dir}/{example.model}.{format}"
+            run.log_info(f"Creating example file {p}")
+            with open(p, "w") as f:
+                content = ""
+                if format == "json" and type(example.data) is list:
+                    content = json.dumps(example.data)
+                elif format == "json" and type(example.data) is str:
+                    content = example.data
+                elif format == "yaml" and type(example.data) is list:
+                    content = yaml.dump(example.data)
+                elif format == "yaml" and type(example.data) is str:
+                    content = example.data
+                elif format == "csv":
+                    content = example.data
+                logging.debug(f"Content of example file {p}: {content}")
+                f.write(content)
+        path = f"{tmp_dir}" + "/{model}." + format
+        delimiter = "array"
+        server = Server(
+            type="local",
+            path=path,
+            format=format,
+            delimiter=delimiter,
+        )
+        run.log_info(f"Using {server} for testing the examples")
+        return server

tests/examples/local-json-simple/datacontract_json_inline.yaml → tests/examples/examples/datacontract_inline.yaml

@@ -25,8 +25,8 @@ models:
           - "endgueltig"
 
 examples:
-  - type: json
-    description: Example entry for CPI data
+  - description: Example entry for CPI data
+    type: json # TODO should be inline
     model: verbraucherpreisindex
     data:
       - wert: 99

tests/examples/examples/datacontract_json.yaml

@@ -0,0 +1,35 @@
+dataContractSpecification: 0.9.2
+id: "61111-0002"
+info:
+  title: "Verbraucherpreisindex: Deutschland, Monate"
+  description: A data contract for the distribution and use of the German Consumer Price Index data.
+  version: 1.0.0
+  owner: my-domain-team
+models:
+  verbraucherpreisindex:
+    description: Model representing the Consumer Price Index for Germany
+    fields:
+      wert:
+          description: Value of the Consumer Price Index
+          type: integer
+          required: true
+      jahrMonat:
+          description: Year and month of the data
+          type: string
+          required: true
+      qualitaet:
+          description: Quality of the data
+          type: string
+          enum:
+          - "vorlaeufig"
+          - "endgueltig"
+
+examples:
+  - type: json
+    description: Example entry for CPI data
+    model: verbraucherpreisindex
+    data: |-
+      [{ "wert": 99, "jahrMonat": "2022-00" },
+      { "wert": 100, "jahrMonat": "2022-01" },
+      { "wert": 101, "jahrMonat": "2022-02", "qualitaet": "vorlaeufig" }]
+

tests/test_examples_local_json_example.py → tests/test_examples_examples_csv.py

@@ -11,13 +11,13 @@
 logging.basicConfig(level=logging.DEBUG, force=True)
 
 def test_cli():
-    result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract_json_inline.yaml"])
+    result = runner.invoke(app, ["test", "--examples", "./examples/examples/datacontract_csv.yaml"])
     assert result.exit_code == 0
 
 
-def test_local_json():
-    data_contract = DataContract(data_contract_file="examples/local-json-simple/datacontract_json_inline.yaml")
-    run = data_contract.testExample()
+def test_csv():
+    data_contract = DataContract(data_contract_file="examples/examples/datacontract_csv.yaml", examples=True)
+    run = data_contract.test()
     print(run)
     print(run.result)
     assert run.result == "passed"

tests/test_examples_examples_inline.py

@@ -0,0 +1,25 @@
+import logging
+
+import pytest
+from typer.testing import CliRunner
+
+from datacontract.cli import app
+from datacontract.data_contract import DataContract
+
+runner = CliRunner()
+
+logging.basicConfig(level=logging.DEBUG, force=True)
+
+def test_cli():
+    result = runner.invoke(app, ["test", "--examples", "./examples/examples/datacontract_inline.yaml"])
+    assert result.exit_code == 0
+
+
+def test_json_inline():
+    data_contract = DataContract(data_contract_file="examples/examples/datacontract_inline.yaml", examples=True)
+    run = data_contract.test()
+    print(run)
+    print(run.result)
+    assert run.result == "passed"
+
+

tests/test_examples_local_json_example_csv.py → tests/test_examples_examples_json.py

@@ -1,3 +1,5 @@
+import logging
+
 import pytest
 from typer.testing import CliRunner
 
@@ -6,15 +8,16 @@
 
 runner = CliRunner()
 
+logging.basicConfig(level=logging.DEBUG, force=True)
 
 def test_cli():
-    result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract_csv.yaml"])
+    result = runner.invoke(app, ["test", "--examples", "./examples/examples/datacontract_json.yaml"])
     assert result.exit_code == 0
 
 
-def test_local_json():
-    data_contract = DataContract(data_contract_file="examples/local-json-simple/datacontract_csv.yaml")
-    run = data_contract.testExample()
+def test_json():
+    data_contract = DataContract(data_contract_file="examples/examples/datacontract_json.yaml", examples=True)
+    run = data_contract.test()
     print(run)
     print(run.result)
     assert run.result == "passed"