From 46f3309568a6ea5923b540c73bb3fb561536f674 Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 14:29:33 +0100
Subject: [PATCH 01/11] Add "test --examples"

---
 datacontract/cli.py                           | 10 ++-
 datacontract/data_contract.py                 | 63 ++++++++++++++++++-
 .../fastjsonschema/check_jsonschema.py        |  2 +-
 .../local-json-simple/datacontract.yaml       | 39 ++++++++++++
 tests/test_examples_local_json_example.py     | 22 +++++++
 5 files changed, 133 insertions(+), 3 deletions(-)
 create mode 100644 tests/examples/local-json-simple/datacontract.yaml
 create mode 100644 tests/test_examples_local_json_example.py

diff --git a/datacontract/cli.py b/datacontract/cli.py
index 2a7a4724..235425c9 100644
--- a/datacontract/cli.py
+++ b/datacontract/cli.py
@@ -87,13 +87,21 @@ def test(
              "Use the key of the server object in the data contract yaml file "
              "to refer to a server, e.g., `production`, or `all` for all "
              "servers (default).")] = "all",
+    examples: Annotated[bool, typer.Option(
+        help="Run the schema and quality tests on the example data within the data contract.")] = None,
     publish: Annotated[str, typer.Option(
-        help="")] = None,
+        help="The url to publish the results after the test")] = None,
 ):
     """
     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()
     _handle_result(run)
 
diff --git a/datacontract/data_contract.py b/datacontract/data_contract.py
index a1744413..64809cc5 100644
--- a/datacontract/data_contract.py
+++ b/datacontract/data_contract.py
@@ -1,5 +1,7 @@
 import json
 import logging
+import tempfile
+import yaml
 
 from datacontract.engines.datacontract.check_that_datacontract_contains_valid_servers_configuration import \
     check_that_datacontract_contains_valid_server_configuration
@@ -13,7 +15,7 @@
 from datacontract.lint import resolve
 from datacontract.lint.linters.example_model_linter import ExampleModelLinter
 from datacontract.model.data_contract_specification import \
-    DataContractSpecification
+    DataContractSpecification, Server
 from datacontract.model.exceptions import DataContractException
 from datacontract.model.run import \
     Run, Check
@@ -125,6 +127,65 @@ 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} 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"Writing example data to {tmp_dir} to be used as a local server")
+                path = f"{tmp_dir}" + "/{model}.json"
+                format = "json"
+                delimiter = "array"
+                for example in data_contract.examples:
+                    format = example.type
+                    p = f"{tmp_dir}/{example.model}.json"
+                    run.log_info(f"Creating example file {p}")
+                    with open(p, "w") as f:
+                        f.write(json.dumps(example.data))
+
+                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
 
diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index 4a32f61d..d24f52f4 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -40,7 +40,7 @@ def read_json_lines_from_file(file):
 
 
 def read_json_array(file):
-    data = json.loads(file)
+    data = json.load(file)
     for item in data:
         yield item
 
diff --git a/tests/examples/local-json-simple/datacontract.yaml b/tests/examples/local-json-simple/datacontract.yaml
new file mode 100644
index 00000000..3563fd0f
--- /dev/null
+++ b/tests/examples/local-json-simple/datacontract.yaml
@@ -0,0 +1,39 @@
+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"
+
diff --git a/tests/test_examples_local_json_example.py b/tests/test_examples_local_json_example.py
new file mode 100644
index 00000000..4c544773
--- /dev/null
+++ b/tests/test_examples_local_json_example.py
@@ -0,0 +1,22 @@
+import pytest
+from typer.testing import CliRunner
+
+from datacontract.cli import app
+from datacontract.data_contract import DataContract
+
+runner = CliRunner()
+
+
+def test_cli():
+    result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract.yaml"])
+    assert result.exit_code == 0
+
+
+def test_local_json():
+    data_contract = DataContract(data_contract_file="examples/local-json-simple/datacontract.yaml")
+    run = data_contract.testExample()
+    print(run)
+    print(run.result)
+    assert run.result == "passed"
+
+

From 022c775f625b3aa9980d593475ee5b06f37dff6e Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 14:49:21 +0100
Subject: [PATCH 02/11] Add "test --examples"

---
 datacontract/data_contract.py                 | 22 +++++++++---
 .../local-json-simple/datacontract_csv.yaml   | 35 +++++++++++++++++++
 ...act.yaml => datacontract_json_inline.yaml} |  0
 tests/test_examples_local_json_example.py     |  4 +--
 tests/test_examples_local_json_example_csv.py | 22 ++++++++++++
 5 files changed, 77 insertions(+), 6 deletions(-)
 create mode 100644 tests/examples/local-json-simple/datacontract_csv.yaml
 rename tests/examples/local-json-simple/{datacontract.yaml => datacontract_json_inline.yaml} (100%)
 create mode 100644 tests/test_examples_local_json_example_csv.py

diff --git a/datacontract/data_contract.py b/datacontract/data_contract.py
index 64809cc5..403d7d0f 100644
--- a/datacontract/data_contract.py
+++ b/datacontract/data_contract.py
@@ -1,6 +1,8 @@
 import json
 import logging
 import tempfile
+from typing import List
+
 import yaml
 
 from datacontract.engines.datacontract.check_that_datacontract_contains_valid_servers_configuration import \
@@ -142,15 +144,27 @@ def testExample(self) -> Run:
 
             with tempfile.TemporaryDirectory(prefix="datacontract-cli") as tmp_dir:
                 run.log_info(f"Writing example data to {tmp_dir} to be used as a local server")
-                path = f"{tmp_dir}" + "/{model}.json"
                 format = "json"
-                delimiter = "array"
+
                 for example in data_contract.examples:
                     format = example.type
-                    p = f"{tmp_dir}/{example.model}.json"
+                    p = f"{tmp_dir}/{example.model}.{format}"
                     run.log_info(f"Creating example file {p}")
                     with open(p, "w") as f:
-                        f.write(json.dumps(example.data))
+                        content = ""
+                        if format == "json" and example.data is List:
+                            content = json.dumps(example.data)
+                        elif format == "json" and example.data is str:
+                            content = example.data
+                        elif format == "yaml" and example.data is List:
+                            content = yaml.dump(example.data)
+                        elif format == "yaml" and example.data is str:
+                            content = example.data
+                        elif format == "csv":
+                            content = example.data
+                        f.write(content)
+                path = f"{tmp_dir}" + "/{model}." + format
+                delimiter = "array"
 
                 server = Server(
                     type="local",
diff --git a/tests/examples/local-json-simple/datacontract_csv.yaml b/tests/examples/local-json-simple/datacontract_csv.yaml
new file mode 100644
index 00000000..77569d37
--- /dev/null
+++ b/tests/examples/local-json-simple/datacontract_csv.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: bigint # integer
+          required: true
+      jahrMonat:
+          description: Year and month of the data
+          type: varchar # string
+          required: true
+      qualitaet:
+          description: Quality of the data
+          type: varchar # string
+          enum:
+          - "vorlaeufig"
+          - "endgueltig"
+examples:
+  - type: csv
+    description: Example entry for CPI data
+    model: verbraucherpreisindex
+    data: |-
+      wert, jahrMonat, qualitaet
+      99, "2022-00",
+      100, "2022-01",
+      101, "2022-02", "vorlaeufig"
+
diff --git a/tests/examples/local-json-simple/datacontract.yaml b/tests/examples/local-json-simple/datacontract_json_inline.yaml
similarity index 100%
rename from tests/examples/local-json-simple/datacontract.yaml
rename to tests/examples/local-json-simple/datacontract_json_inline.yaml
diff --git a/tests/test_examples_local_json_example.py b/tests/test_examples_local_json_example.py
index 4c544773..e48c6857 100644
--- a/tests/test_examples_local_json_example.py
+++ b/tests/test_examples_local_json_example.py
@@ -8,12 +8,12 @@
 
 
 def test_cli():
-    result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract.yaml"])
+    result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract_json_inline.yaml"])
     assert result.exit_code == 0
 
 
 def test_local_json():
-    data_contract = DataContract(data_contract_file="examples/local-json-simple/datacontract.yaml")
+    data_contract = DataContract(data_contract_file="examples/local-json-simple/datacontract_json_inline.yaml")
     run = data_contract.testExample()
     print(run)
     print(run.result)
diff --git a/tests/test_examples_local_json_example_csv.py b/tests/test_examples_local_json_example_csv.py
new file mode 100644
index 00000000..ce626cc7
--- /dev/null
+++ b/tests/test_examples_local_json_example_csv.py
@@ -0,0 +1,22 @@
+import pytest
+from typer.testing import CliRunner
+
+from datacontract.cli import app
+from datacontract.data_contract import DataContract
+
+runner = CliRunner()
+
+
+def test_cli():
+    result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract_csv.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()
+    print(run)
+    print(run.result)
+    assert run.result == "passed"
+
+

From 3a682a9ead00869795f13e1201941e9f79adceaa Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 19:36:49 +0100
Subject: [PATCH 03/11] Add "test --examples"

---
 .../engines/fastjsonschema/check_jsonschema.py         | 10 +++++-----
 .../engines/fastjsonschema/s3/s3_read_files.py         |  4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index d24f52f4..628f86ce 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -40,7 +40,7 @@ def read_json_lines_from_file(file):
 
 
 def read_json_array(file):
-    data = json.load(file)
+    data = json.loads(file)
     for item in data:
         yield item
 
@@ -90,13 +90,13 @@ def process_s3_file(server, model_name, validate):
         s3_location = s3_location.format(model = model_name)
     json_stream = None
 
-    for file_content in yield_s3_files(s3_endpoint_url, s3_location):
+    for file in yield_s3_files(s3_endpoint_url, s3_location):
         if server.delimiter == "new_line":
-            json_stream = read_json_lines(file_content)
+            json_stream = read_json_lines(file)
         elif server.delimiter == "array":
-            json_stream = read_json_array(file_content)
+            json_stream = read_json_array(file)
         else:
-            json_stream = read_json_file(file_content)
+            json_stream = read_json_file(file)
 
     if json_stream is None:
         raise DataContractException(
diff --git a/datacontract/engines/fastjsonschema/s3/s3_read_files.py b/datacontract/engines/fastjsonschema/s3/s3_read_files.py
index 2172b1d3..814028b6 100644
--- a/datacontract/engines/fastjsonschema/s3/s3_read_files.py
+++ b/datacontract/engines/fastjsonschema/s3/s3_read_files.py
@@ -9,8 +9,8 @@ def yield_s3_files(s3_endpoint_url, s3_location):
     files = fs.glob(s3_location)
     for file in files:
         with fs.open(file) as f:
-            logging.info(f"Reading file {file}")
-            yield f.read()
+            logging.info(f"Opening file {file}")
+            yield f
 
 
 def s3_fs(s3_endpoint_url):

From 9131a9e6ec44befdda5953ca4a86c41bafdfa1cf Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 19:38:34 +0100
Subject: [PATCH 04/11] Add "test --examples"

---
 datacontract/engines/fastjsonschema/check_jsonschema.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index 628f86ce..7419742e 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -29,7 +29,8 @@ def validate_json_stream(model_name, validate, json_stream):
         )
 
 
-def read_json_lines(file_content: str):
+def read_json_lines(file):
+    file_content = file.read()
     for line in file_content.splitlines():
         yield json.loads(line)
 
@@ -40,13 +41,13 @@ def read_json_lines_from_file(file):
 
 
 def read_json_array(file):
-    data = json.loads(file)
+    data = json.load(file)
     for item in data:
         yield item
 
 
 def read_json_file(file):
-    yield json.loads(file.read())
+    yield json.load(file)
 
 
 def process_json_file(run, model_name, validate, file, delimiter):

From a842045a25276dcecc3a2ea08c979f25e3c89f7c Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 19:39:18 +0100
Subject: [PATCH 05/11] Add "test --examples"

---
 datacontract/engines/fastjsonschema/check_jsonschema.py | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index 7419742e..ab95f3a4 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -35,11 +35,6 @@ def read_json_lines(file):
         yield json.loads(line)
 
 
-def read_json_lines_from_file(file):
-    for line in file:
-        yield json.loads(line)
-
-
 def read_json_array(file):
     data = json.load(file)
     for item in data:
@@ -52,7 +47,7 @@ def read_json_file(file):
 
 def process_json_file(run, model_name, validate, file, delimiter):
     if delimiter == "new_line":
-        json_stream = read_json_lines_from_file(file)
+        json_stream = read_json_lines(file)
     elif delimiter == "array":
         json_stream = read_json_array(file)
     else:

From b1d64f38f483c79e00e0b541de71d496048d4f0d Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 20:52:57 +0100
Subject: [PATCH 06/11] Add "test --examples"

---
 datacontract/data_contract.py                       | 13 +++++++------
 .../engines/fastjsonschema/check_jsonschema.py      |  1 +
 tests/test_examples_local_json_example.py           |  3 +++
 3 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/datacontract/data_contract.py b/datacontract/data_contract.py
index 403d7d0f..c19a252a 100644
--- a/datacontract/data_contract.py
+++ b/datacontract/data_contract.py
@@ -137,13 +137,13 @@ def testExample(self) -> Run:
                                                           self._data_contract)
 
             # TODO check yaml contains models
-            run.log_info(f"Running tests for data contract {data_contract.id} against examples")
+            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"Writing example data to {tmp_dir} to be used as a local server")
+                run.log_info(f"Copying examples to files in temporary directory {tmp_dir}")
                 format = "json"
 
                 for example in data_contract.examples:
@@ -152,16 +152,17 @@ def testExample(self) -> Run:
                     run.log_info(f"Creating example file {p}")
                     with open(p, "w") as f:
                         content = ""
-                        if format == "json" and example.data is List:
+                        if format == "json" and type(example.data) is list:
                             content = json.dumps(example.data)
-                        elif format == "json" and example.data is str:
+                        elif format == "json" and type(example.data) is str:
                             content = example.data
-                        elif format == "yaml" and example.data is List:
+                        elif format == "yaml" and type(example.data) is list:
                             content = yaml.dump(example.data)
-                        elif format == "yaml" and example.data is str:
+                        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"
diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index ab95f3a4..07ff664d 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -63,6 +63,7 @@ def process_local_file(run, server, model_name, validate):
     if os.path.isdir(path):
         return process_directory(run, path, server, model_name, validate)
     else:
+        logging.info(f"Processing file {path}")
         with open(path, 'r') as file:
             process_json_file(run, model_name, validate, file, server.delimiter)
 
diff --git a/tests/test_examples_local_json_example.py b/tests/test_examples_local_json_example.py
index e48c6857..cbd73d9b 100644
--- a/tests/test_examples_local_json_example.py
+++ b/tests/test_examples_local_json_example.py
@@ -1,3 +1,5 @@
+import logging
+
 import pytest
 from typer.testing import CliRunner
 
@@ -6,6 +8,7 @@
 
 runner = CliRunner()
 
+logging.basicConfig(level=logging.DEBUG, force=True)
 
 def test_cli():
     result = runner.invoke(app, ["test", "--examples", "./examples/local-json-simple/datacontract_json_inline.yaml"])

From 0be064d987f74567b8c6017a98b929f7da9a7611 Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 21:03:13 +0100
Subject: [PATCH 07/11] Add "test --examples"

---
 .../fastjsonschema/check_jsonschema.py        | 25 ++++++++++++++-----
 .../fastjsonschema/s3/s3_read_files.py        |  4 +--
 2 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index 07ff664d..f663ee71 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -35,16 +35,29 @@ def read_json_lines(file):
         yield json.loads(line)
 
 
+def read_json_lines_content(file_content):
+    for line in file_content.splitlines():
+        yield json.loads(line)
+
+
 def read_json_array(file):
     data = json.load(file)
     for item in data:
         yield item
 
+def read_json_array_content(file_content):
+    data = json.loads(file_content)
+    for item in data:
+        yield item
+
 
 def read_json_file(file):
     yield json.load(file)
 
 
+def read_json_file_content(file_content):
+    yield json.loads(file_content)
+
 def process_json_file(run, model_name, validate, file, delimiter):
     if delimiter == "new_line":
         json_stream = read_json_lines(file)
@@ -58,7 +71,7 @@ def process_json_file(run, model_name, validate, file, delimiter):
 def process_local_file(run, server, model_name, validate):
     path = server.path
     if "{model}" in path:
-        path = path.format(model = model_name)
+        path = path.format(model=model_name)
 
     if os.path.isdir(path):
         return process_directory(run, path, server, model_name, validate)
@@ -84,16 +97,16 @@ def process_s3_file(server, model_name, validate):
     s3_endpoint_url = server.endpointUrl
     s3_location = server.location
     if "{model}" in s3_location:
-        s3_location = s3_location.format(model = model_name)
+        s3_location = s3_location.format(model=model_name)
     json_stream = None
 
-    for file in yield_s3_files(s3_endpoint_url, s3_location):
+    for file_content in yield_s3_files(s3_endpoint_url, s3_location):
         if server.delimiter == "new_line":
-            json_stream = read_json_lines(file)
+            json_stream = read_json_lines_content(file_content)
         elif server.delimiter == "array":
-            json_stream = read_json_array(file)
+            json_stream = read_json_array_content(file_content)
         else:
-            json_stream = read_json_file(file)
+            json_stream = read_json_file_content(file_content)
 
     if json_stream is None:
         raise DataContractException(
diff --git a/datacontract/engines/fastjsonschema/s3/s3_read_files.py b/datacontract/engines/fastjsonschema/s3/s3_read_files.py
index 814028b6..fb0ae4b2 100644
--- a/datacontract/engines/fastjsonschema/s3/s3_read_files.py
+++ b/datacontract/engines/fastjsonschema/s3/s3_read_files.py
@@ -9,8 +9,8 @@ def yield_s3_files(s3_endpoint_url, s3_location):
     files = fs.glob(s3_location)
     for file in files:
         with fs.open(file) as f:
-            logging.info(f"Opening file {file}")
-            yield f
+            logging.info(f"Downloading file {file}")
+            yield f.read()
 
 
 def s3_fs(s3_endpoint_url):

From ac93083dd820806388cbba93dce5dffa8d9e7613 Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Tue, 13 Feb 2024 21:03:43 +0100
Subject: [PATCH 08/11] Add "test --examples"

---
 datacontract/engines/fastjsonschema/check_jsonschema.py | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/datacontract/engines/fastjsonschema/check_jsonschema.py b/datacontract/engines/fastjsonschema/check_jsonschema.py
index f663ee71..f85ea732 100644
--- a/datacontract/engines/fastjsonschema/check_jsonschema.py
+++ b/datacontract/engines/fastjsonschema/check_jsonschema.py
@@ -35,7 +35,7 @@ def read_json_lines(file):
         yield json.loads(line)
 
 
-def read_json_lines_content(file_content):
+def read_json_lines_content(file_content: str):
     for line in file_content.splitlines():
         yield json.loads(line)
 
@@ -45,7 +45,8 @@ def read_json_array(file):
     for item in data:
         yield item
 
-def read_json_array_content(file_content):
+
+def read_json_array_content(file_content: str):
     data = json.loads(file_content)
     for item in data:
         yield item
@@ -55,9 +56,10 @@ def read_json_file(file):
     yield json.load(file)
 
 
-def read_json_file_content(file_content):
+def read_json_file_content(file_content: str):
     yield json.loads(file_content)
 
+
 def process_json_file(run, model_name, validate, file, delimiter):
     if delimiter == "new_line":
         json_stream = read_json_lines(file)

From fb99eb306957308e0a3a1272594102d5249dbafa Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Wed, 14 Feb 2024 09:35:42 +0100
Subject: [PATCH 09/11] Add "test --examples"

---
 datacontract/cli.py                           |   8 +-
 datacontract/data_contract.py                 | 138 +++++++-----------
 .../datacontract_csv.yaml                     |   0
 .../datacontract_inline.yaml}                 |   4 +-
 .../examples/examples/datacontract_json.yaml  |  35 +++++
 ...ample.py => test_examples_examples_csv.py} |   8 +-
 tests/test_examples_examples_inline.py        |  25 ++++
 ..._csv.py => test_examples_examples_json.py} |  11 +-
 8 files changed, 128 insertions(+), 101 deletions(-)
 rename tests/examples/{local-json-simple => examples}/datacontract_csv.yaml (100%)
 rename tests/examples/{local-json-simple/datacontract_json_inline.yaml => examples/datacontract_inline.yaml} (92%)
 create mode 100644 tests/examples/examples/datacontract_json.yaml
 rename tests/{test_examples_local_json_example.py => test_examples_examples_csv.py} (64%)
 create mode 100644 tests/test_examples_examples_inline.py
 rename tests/{test_examples_local_json_example_csv.py => test_examples_examples_json.py} (56%)

diff --git a/datacontract/cli.py b/datacontract/cli.py
index 235425c9..b9bea568 100644
--- a/datacontract/cli.py
+++ b/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)
 
 
diff --git a/datacontract/data_contract.py b/datacontract/data_contract.py
index c19a252a..a762b72c 100644
--- a/datacontract/data_contract.py
+++ b/datacontract/data_contract.py
@@ -30,6 +30,7 @@ def __init__(
         data_contract_str: str = None,
         data_contract: DataContractSpecification = None,
         server: str = None,
+        examples: bool = False,
         publish_url: str = None,
         spark: str = None,
     ):
@@ -37,6 +38,7 @@ def __init__(
         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
diff --git a/tests/examples/local-json-simple/datacontract_csv.yaml b/tests/examples/examples/datacontract_csv.yaml
similarity index 100%
rename from tests/examples/local-json-simple/datacontract_csv.yaml
rename to tests/examples/examples/datacontract_csv.yaml
diff --git a/tests/examples/local-json-simple/datacontract_json_inline.yaml b/tests/examples/examples/datacontract_inline.yaml
similarity index 92%
rename from tests/examples/local-json-simple/datacontract_json_inline.yaml
rename to tests/examples/examples/datacontract_inline.yaml
index 3563fd0f..f0b3ae79 100644
--- a/tests/examples/local-json-simple/datacontract_json_inline.yaml
+++ b/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
diff --git a/tests/examples/examples/datacontract_json.yaml b/tests/examples/examples/datacontract_json.yaml
new file mode 100644
index 00000000..b73b2b24
--- /dev/null
+++ b/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" }]
+
diff --git a/tests/test_examples_local_json_example.py b/tests/test_examples_examples_csv.py
similarity index 64%
rename from tests/test_examples_local_json_example.py
rename to tests/test_examples_examples_csv.py
index cbd73d9b..e47bf44f 100644
--- a/tests/test_examples_local_json_example.py
+++ b/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"
diff --git a/tests/test_examples_examples_inline.py b/tests/test_examples_examples_inline.py
new file mode 100644
index 00000000..325a8ce8
--- /dev/null
+++ b/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"
+
+
diff --git a/tests/test_examples_local_json_example_csv.py b/tests/test_examples_examples_json.py
similarity index 56%
rename from tests/test_examples_local_json_example_csv.py
rename to tests/test_examples_examples_json.py
index ce626cc7..ace33729 100644
--- a/tests/test_examples_local_json_example_csv.py
+++ b/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"

From 5eed62491509064ad52fdc3f23dffa3616a50a28 Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Wed, 14 Feb 2024 09:48:41 +0100
Subject: [PATCH 10/11] Add "test --examples"

---
 README.md      | 3 +++
 pyproject.toml | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 781a9e7f..c3ebcb08 100644
--- a/README.md
+++ b/README.md
@@ -48,6 +48,9 @@ $ datacontract lint datacontract.yaml
 # execute schema and quality checks
 $ datacontract test datacontract.yaml
 
+# execute schema and quality checks on the examples within the contract
+$ datacontract test --examples datacontract.yaml
+
 # find differences between to data contracts (Coming Soon)
 $ datacontract diff datacontract-v1.yaml datacontract-v2.yaml
 
diff --git a/pyproject.toml b/pyproject.toml
index 956e076a..ecf25aec 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -25,7 +25,7 @@ dependencies = [
   "soda-core-spark[databricks]~=3.1.5",
   "soda-core-spark-df~=3.1.5",
   "snowflake-connector-python[pandas]>=3.6,<3.8",
-  "duckdb>=0.9.3.dev3920",
+  "duckdb==0.9.2",
   "fastjsonschema~=2.19.1",
   "python-dotenv~=1.0.0",
   "s3fs==2024.2.0",

From 13eb776375e87c38d42a7aef85bffc58797111c7 Mon Sep 17 00:00:00 2001
From: Simon Harrer <simon.harrer@gmail.com>
Date: Wed, 14 Feb 2024 10:00:07 +0100
Subject: [PATCH 11/11] Add "test --examples"

---
 pyproject.toml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pyproject.toml b/pyproject.toml
index ecf25aec..f1121c58 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -25,7 +25,7 @@ dependencies = [
   "soda-core-spark[databricks]~=3.1.5",
   "soda-core-spark-df~=3.1.5",
   "snowflake-connector-python[pandas]>=3.6,<3.8",
-  "duckdb==0.9.2",
+  "duckdb==0.9.3.dev3920",
   "fastjsonschema~=2.19.1",
   "python-dotenv~=1.0.0",
   "s3fs==2024.2.0",