From c9ddc2e84e72144f06deb4b312f9cabe3e79be90 Mon Sep 17 00:00:00 2001
From: Finn van Krieken <fvankrieken@planning.nyc.gov>
Date: Fri, 20 Dec 2024 13:35:58 -0500
Subject: [PATCH] chk test validate

---
 dcpy/test/lifecycle/ingest/shared.py        | 37 +++++++++++++
 dcpy/test/lifecycle/ingest/test_validate.py | 57 ++++++++++++++++++++-
 2 files changed, 92 insertions(+), 2 deletions(-)

diff --git a/dcpy/test/lifecycle/ingest/shared.py b/dcpy/test/lifecycle/ingest/shared.py
index b6995ba01..98aef2cbf 100644
--- a/dcpy/test/lifecycle/ingest/shared.py
+++ b/dcpy/test/lifecycle/ingest/shared.py
@@ -1,20 +1,30 @@
+from datetime import datetime
 from pathlib import Path
 
 from dcpy.models.connectors.edm.publishing import GisDataset
+from dcpy.models.connectors.edm.recipes import Dataset
+from dcpy.models import file, library
 from dcpy.models.connectors import socrata, web
 from dcpy.models.lifecycle.ingest import (
     LocalFileSource,
     ScriptSource,
     S3Source,
     DEPublished,
+    DatasetAttributes,
+    ArchivalMetadata,
+    Ingestion,
+    Config,
 )
+from dcpy.utils.metadata import get_run_details
 from dcpy.test.conftest import RECIPES_BUCKET
 
 RESOURCES = Path(__file__).parent / "resources"
 TEMPLATE_DIR = RESOURCES / "templates"
 TEST_DATA_DIR = "test_data"
+TEST_OUTPUT = RESOURCES / TEST_DATA_DIR / "output.parquet"
 TEST_DATASET_NAME = "test_dataset"
 FAKE_VERSION = "20240101"
+TEST_DATASET = Dataset(id=TEST_DATASET_NAME, version=FAKE_VERSION)
 
 
 class Sources:
@@ -39,6 +49,33 @@ class Sources:
     )
 
 
+BASIC_CONFIG = Config(
+    id=TEST_DATASET_NAME,
+    version=FAKE_VERSION,
+    attributes=DatasetAttributes(name=TEST_DATASET_NAME),
+    archival=ArchivalMetadata(
+        archival_timestamp=datetime(2024, 1, 1),
+        raw_filename="dummy.txt",
+        acl="public-read",
+    ),
+    ingestion=Ingestion(source=Sources.local_file, file_format=file.Csv(type="csv")),
+    run_details=get_run_details(),
+)
+
+BASIC_LIBRARY_CONFIG = library.Config(
+    dataset=library.DatasetDefinition(
+        name=TEST_DATASET_NAME,
+        version=FAKE_VERSION,
+        acl="public-read",
+        source=library.DatasetDefinition.SourceSection(),
+        destination=library.DatasetDefinition.DestinationSection(
+            geometry=library.GeometryType(SRS="NONE", type="NONE")
+        ),
+    ),
+    execution_details=get_run_details(),
+)
+
+
 SOURCE_FILENAMES = [
     (Sources.local_file, "dummy.txt"),
     (Sources.gis, f"{TEST_DATASET_NAME}.zip"),
diff --git a/dcpy/test/lifecycle/ingest/test_validate.py b/dcpy/test/lifecycle/ingest/test_validate.py
index 70fbc2261..13e74eb4a 100644
--- a/dcpy/test/lifecycle/ingest/test_validate.py
+++ b/dcpy/test/lifecycle/ingest/test_validate.py
@@ -1,15 +1,68 @@
+from io import BytesIO
+import json
 import pytest
 import yaml
 
+from dcpy.test.conftest import RECIPES_BUCKET
 from dcpy.models.lifecycle.ingest import Template
-from dcpy.lifecycle.ingest import configure, transform
+from dcpy.utils import s3
+from dcpy.connectors.edm import recipes
+from dcpy.lifecycle.ingest import configure, transform, validate
+
+from .shared import (
+    TEST_DATASET,
+    TEST_OUTPUT,
+    BASIC_CONFIG,
+    BASIC_LIBRARY_CONFIG,
+)
 
 
 @pytest.mark.parametrize("dataset", [t.name for t in configure.TEMPLATE_DIR.glob("*")])
-def test_validate_all_datasets(dataset):
+def test_validate_all_templates(dataset):
     with open(configure.TEMPLATE_DIR / dataset, "r") as f:
         s = yaml.safe_load(f)
     template = Template(**s)
     transform.validate_processing_steps(
         template.id, template.ingestion.processing_steps
     )
+
+
+class TestValidateAgainstExistingVersions:
+    def test_new(self, create_buckets):
+        assert (
+            validate.validate_against_existing_versions(TEST_DATASET, TEST_OUTPUT)
+            == validate.ArchiveAction.push
+        )
+
+    def test_existing_library(self, create_buckets):
+        ds = BASIC_LIBRARY_CONFIG.sparse_dataset
+        config_str = json.dumps(BASIC_LIBRARY_CONFIG.model_dump(mode="json"))
+        s3.upload_file_obj(
+            BytesIO(config_str.encode()),
+            RECIPES_BUCKET,
+            f"{recipes.s3_folder_path(ds)}/config.json",
+            BASIC_LIBRARY_CONFIG.dataset.acl,
+        )
+        assert recipes.exists(ds)
+        assert (
+            validate.validate_against_existing_versions(ds, TEST_OUTPUT)
+            == validate.ArchiveAction.do_nothing
+        )
+
+    def test_existing(self, create_buckets):
+        ds = BASIC_CONFIG.dataset
+        recipes.archive_dataset(BASIC_CONFIG, TEST_OUTPUT)
+        assert recipes.exists(ds)
+        assert (
+            validate.validate_against_existing_versions(ds, TEST_OUTPUT)
+            == validate.ArchiveAction.update_freshness
+        )
+
+    def test_existing_data_diffs(self, create_buckets):
+        ds = BASIC_CONFIG.dataset
+        recipes.archive_dataset(BASIC_CONFIG, TEST_OUTPUT)
+        assert recipes.exists(ds)
+        with pytest.raises(FileExistsError):
+            validate.validate_against_existing_versions(
+                ds, TEST_OUTPUT.parent / "test.parquet"
+            )