Add spec validation

Author: basepi

.flake8

@@ -0,0 +1,7 @@
+[flake8]
+exclude=
+    tests/**,
+    conftest.py,
+    setup.py
+max-line-length=120
+ignore=E731,W503,E203,BLK100,B301

tests/conftest.py

@@ -0,0 +1,68 @@
+import datetime
+import json
+import os
+
+import pytest
+
+
+class ValidationError(Exception):
+    pass
+
+
+@pytest.fixture
+def spec_validator():
+    with open(
+        os.path.join(os.path.dirname(__file__), "resources", "spec.json"), "r"
+    ) as fh:
+        spec = json.load(fh)
+
+    def validator(data_json):
+        """
+        Throws a ValidationError if anything doesn't match the spec.
+
+        Returns the original json (pass-through)
+        """
+        fields = spec["fields"]
+        data = json.loads(data_json)
+        for k, v in fields.items():
+            if v.get("required"):
+                found = False
+                if k in data:
+                    found = True
+                elif "." in k:
+                    # Dotted keys could be nested, like ecs.version
+                    subkeys = k.split(".")
+                    subval = data
+                    for subkey in subkeys:
+                        subval = subval.get(subkey, {})
+                    if subval:
+                        found = True
+                if not found:
+                    raise ValidationError("Missing required key {}".format(k))
+            if k in data:
+                if v["type"] == "string" and not isinstance(data[k], str):
+                    raise ValidationError(
+                        "Value {0} for key {1} should be string".format(data[k], k)
+                    )
+                if v["type"] == "datetime":
+                    try:
+                        datetime.datetime.fromisoformat(data[k].rstrip("Z"))
+                    except ValueError:
+                        raise ValidationError(
+                            "Value {0} for key {1} doesn't parse as an ISO datetime".format(
+                                data[k], k
+                            )
+                        )
+            if v.get("index"):
+                index = v.get("index")
+                key = json.loads(
+                    data_json.split(",")[index].split(":")[0].strip().lstrip("{")
+                )
+                if key != k:
+                    raise ValidationError(
+                        "Key {0} is not at index {1}".format(k, index)
+                    )
+
+        return data_json
+
+    return validator

tests/test_apm.py

@@ -10,7 +10,7 @@
 from .compat import StringIO
 
 
-def test_elasticapm_structlog_log_correlation_ecs_fields():
+def test_elasticapm_structlog_log_correlation_ecs_fields(spec_validator):
     apm = elasticapm.Client({"SERVICE_NAME": "apm-service", "DISABLE_SEND": True})
     stream = StringIO()
     logger = structlog.PrintLogger(stream)
@@ -30,7 +30,7 @@ def test_elasticapm_structlog_log_correlation_ecs_fields():
     finally:
         apm.end_transaction("test-transaction")
 
-    ecs = json.loads(stream.getvalue().rstrip())
+    ecs = json.loads(spec_validator(stream.getvalue().rstrip()))
     ecs.pop("@timestamp")
     assert ecs == {
         "ecs": {"version": "1.6.0"},

tests/test_stdlib_formatter.py

@@ -35,25 +35,25 @@ def make_record():
     return record
 
 
-def test_record_formatted():
+def test_record_formatted(spec_validator):
     formatter = ecs_logging.StdlibFormatter(exclude_fields=["process"])
 
-    assert formatter.format(make_record()) == (
+    assert spec_validator(formatter.format(make_record())) == (
         '{"@timestamp":"2020-03-20T14:12:46.123Z","log.level":"debug","message":"1: hello","ecs":{"version":"1.6.0"},'
         '"log":{"logger":"logger-name","origin":{"file":{"line":10,"name":"file.py"},"function":"test_function"},'
         '"original":"1: hello"}}'
     )
 
 
-def test_can_be_overridden():
+def test_can_be_overridden(spec_validator):
     class CustomFormatter(ecs_logging.StdlibFormatter):
         def format_to_ecs(self, record):
             ecs_dict = super(CustomFormatter, self).format_to_ecs(record)
             ecs_dict["custom"] = "field"
             return ecs_dict
 
     formatter = CustomFormatter(exclude_fields=["process"])
-    assert formatter.format(make_record()) == (
+    assert spec_validator(formatter.format(make_record())) == (
         '{"@timestamp":"2020-03-20T14:12:46.123Z","log.level":"debug","message":"1: hello",'
         '"custom":"field","ecs":{"version":"1.6.0"},"log":{"logger":"logger-name","origin":'
         '{"file":{"line":10,"name":"file.py"},"function":"test_function"},"original":"1: hello"}}'

tests/test_structlog_formatter.py

@@ -9,19 +9,19 @@ def make_event_dict():
 
 
 @mock.patch("time.time")
-def test_event_dict_formatted(time):
+def test_event_dict_formatted(time, spec_validator):
     time.return_value = 1584720997.187709
 
     formatter = ecs_logging.StructlogFormatter()
-    assert formatter(None, "debug", make_event_dict()) == (
+    assert spec_validator(formatter(None, "debug", make_event_dict())) == (
         '{"@timestamp":"2020-03-20T16:16:37.187Z","log.level":"debug",'
         '"message":"test message","ecs":{"version":"1.6.0"},'
         '"log":{"logger":"logger-name"}}'
     )
 
 
 @mock.patch("time.time")
-def test_can_be_set_as_processor(time):
+def test_can_be_set_as_processor(time, spec_validator):
     time.return_value = 1584720997.187709
 
     stream = StringIO()
@@ -35,7 +35,7 @@ def test_can_be_set_as_processor(time):
     logger = structlog.get_logger("logger-name")
     logger.debug("test message", custom="key", **{"dot.ted": 1})
 
-    assert stream.getvalue() == (
+    assert spec_validator(stream.getvalue()) == (
         '{"@timestamp":"2020-03-20T16:16:37.187Z","log.level":"debug",'
         '"message":"test message","custom":"key","dot":{"ted":1},'
         '"ecs":{"version":"1.6.0"}}\n'