diff options
Diffstat (limited to 'tests/incite')
| -rw-r--r-- | tests/incite/collections/test_df_collection_item_thl_web.py | 273 | ||||
| -rw-r--r-- | tests/incite/mergers/foundations/test_enriched_session.py | 52 | ||||
| -rw-r--r-- | tests/incite/mergers/foundations/test_enriched_task_adjust.py | 38 | ||||
| -rw-r--r-- | tests/incite/mergers/foundations/test_enriched_wall.py | 73 | ||||
| -rw-r--r-- | tests/incite/mergers/test_merge_collection.py | 53 | ||||
| -rw-r--r-- | tests/incite/mergers/test_merge_collection_item.py | 25 | ||||
| -rw-r--r-- | tests/incite/mergers/test_pop_ledger.py | 109 | ||||
| -rw-r--r-- | tests/incite/mergers/test_ym_survey_merge.py | 55 | ||||
| -rw-r--r-- | tests/incite/schemas/test_admin_responses.py | 36 | ||||
| -rw-r--r-- | tests/incite/schemas/test_thl_web.py | 8 | ||||
| -rw-r--r-- | tests/incite/test_collection_base.py | 62 | ||||
| -rw-r--r-- | tests/incite/test_collection_base_item.py | 74 | ||||
| -rw-r--r-- | tests/incite/test_interval_idx.py | 6 |
13 files changed, 503 insertions, 361 deletions
diff --git a/tests/incite/collections/test_df_collection_item_thl_web.py b/tests/incite/collections/test_df_collection_item_thl_web.py index 8038d3b..edf90f7 100644 --- a/tests/incite/collections/test_df_collection_item_thl_web.py +++ b/tests/incite/collections/test_df_collection_item_thl_web.py @@ -5,12 +5,12 @@ from datetime import UTC, datetime, timedelta from itertools import product as iter_product from os.path import join as pjoin from pathlib import Path, PurePath -from typing import TYPE_CHECKING from uuid import uuid4 import dask.dataframe as dd import pandas as pd import pytest +from dask.distributed import Client as DaskClient from distributed import Client, Scheduler, Worker # noinspection PyUnresolvedReferences @@ -21,20 +21,19 @@ from faker import Faker from pandera.pandas import DataFrameSchema from pydantic import FilePath -from generalresearch.incite.base import CollectionItemBase +from generalresearch.incite.base import CollectionItemBase, GRLDatasets from generalresearch.incite.collections import ( + DFCollection, DFCollectionItem, DFCollectionType, ) from generalresearch.incite.schemas import ARCHIVE_AFTER +from generalresearch.managers.thl.ledger_manager.thl_ledger import ThlLedgerManager from generalresearch.models.thl.product import Product from generalresearch.models.thl.user import User from generalresearch.pg_helper import PostgresConfig from generalresearch.sql_helper import PostgresDsn -if TYPE_CHECKING: - from generalresearch.incite.base import GRLDatasets - fake = Faker() df_collections = [ @@ -72,7 +71,12 @@ class TestDFCollectionItemBase: ) class TestDFCollectionItemProperties: - def test_filename(self, df_collection_data_type, df_collection, offset: str): + def test_filename( + self, + df_collection_data_type: DFCollectionType, + df_collection: DFCollection, + offset: str, + ): for i in df_collection.items: assert isinstance(i.filename, str) @@ -89,37 +93,59 @@ class TestDFCollectionItemProperties: ) class TestDFCollectionItemPropertiesBase: - def test_name(self, df_collection_data_type, offset: str, df_collection): + def test_name( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.name, str) - def test_finish(self, df_collection_data_type, offset: str, df_collection): + def test_finish( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.finish, datetime) - def test_interval(self, df_collection_data_type, offset: str, df_collection): + def test_interval( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.interval, pd.Interval) def test_partial_filename( - self, df_collection_data_type, offset: str, df_collection + self, + df_collection: DFCollection, ): for i in df_collection.items: assert isinstance(i.partial_filename, str) - def test_empty_filename(self, df_collection_data_type, offset: str, df_collection): + def test_empty_filename( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.empty_filename, str) - def test_path(self, df_collection_data_type, offset: str, df_collection): + def test_path( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.path, FilePath) - def test_partial_path(self, df_collection_data_type, offset: str, df_collection): + def test_partial_path( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.partial_path, FilePath) - def test_empty_path(self, df_collection_data_type, offset: str, df_collection): + def test_empty_path( + self, + df_collection: DFCollection, + ): for i in df_collection.items: assert isinstance(i.empty_path, FilePath) @@ -138,11 +164,8 @@ class TestDFCollectionItemMethod: def test_has_mysql( self, - df_collection, + df_collection: DFCollection, thl_web_rr: PostgresConfig, - offset: str, - duration: timedelta, - df_collection_data_type, delete_df_collection: Callable[..., None], ): delete_df_collection(coll=df_collection) @@ -168,12 +191,6 @@ class TestDFCollectionItemMethod: @pytest.mark.skip def test_update_partial_archive( self, - df_collection, - offset: str, - duration: timedelta, - thl_web_rw: PostgresConfig, - df_collection_data_type, - delete_df_collection: Callable[..., None], ): # for i in collection.items: # assert i.update_partial_archive() @@ -183,28 +200,12 @@ class TestDFCollectionItemMethod: @pytest.mark.skip def test_create_partial_archive( self, - df_collection, - offset: str, - duration: str, - create_main_accounts: Callable[..., None], - thl_web_rw: PostgresConfig, - thl_lm, - df_collection_data_type, - user_factory: Callable[..., User], - product: product: Product, - client_no_amm, - incite_item_factory, - delete_df_collection: Callable[..., None], - mnt_filepath: GRLDatasets, ): assert 1 + 1 == 2 def test_dict( self, - df_collection_data_type, - offset: str, - duration: timedelta, - df_collection, + df_collection: DFCollection, delete_df_collection: Callable[..., None], ): delete_df_collection(coll=df_collection) @@ -225,15 +226,15 @@ class TestDFCollectionItemMethod: def test_from_mysql( self, - df_collection_data_type, - df_collection, + df_collection_data_type: DFCollectionType, + df_collection: DFCollection, offset: str, duration: timedelta, create_main_accounts: Callable[..., None], thl_web_rw: PostgresConfig, user_factory: Callable[..., User], - product: product: Product, - incite_item_factory, + product: Product, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], ): @@ -253,12 +254,14 @@ class TestDFCollectionItemMethod: if df_collection.data_type == DFCollectionType.LEDGER: assert df is None else: + assert isinstance(df, pd.DataFrame) assert df.empty assert set(df.columns) == set(df_collection._schema.columns.keys()) incite_item_factory(user=u1, item=item) df = item.from_mysql() + assert isinstance(df, pd.DataFrame) assert not df.empty assert set(df.columns) == set(df_collection._schema.columns.keys()) if df_collection.data_type == DFCollectionType.LEDGER: @@ -270,13 +273,13 @@ class TestDFCollectionItemMethod: def test_from_mysql_standard( self, - df_collection_data_type, - df_collection, + df_collection_data_type: DFCollectionType, + df_collection: DFCollection, offset: str, duration: timedelta, user_factory: Callable[..., User], - product: product: Product, - incite_item_factory, + product: Product, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], ): @@ -293,7 +296,7 @@ class TestDFCollectionItemMethod: # We're using parametrize, so this If statement is just to # confirm other Item Types will always raise an assertion with pytest.raises(expected_exception=AssertionError) as cm: - res = item.from_mysql_standard() + _ = item.from_mysql_standard() assert ( "Can't call from_mysql_standard for Ledger DFCollectionItem" in str(cm.value) @@ -304,32 +307,34 @@ class TestDFCollectionItemMethod: # Unlike .from_mysql_ledger(), .from_mysql_standard() will return # back and empty df with the correct columns in place df = item.from_mysql_standard() + assert isinstance(df, pd.DataFrame) assert df.empty assert set(df.columns) == set(df_collection._schema.columns.keys()) incite_item_factory(user=u1, item=item) df = item.from_mysql_standard() + assert isinstance(df, pd.DataFrame) assert not df.empty assert set(df.columns) == set(df_collection._schema.columns.keys()) assert df.shape[0] > 0 def test_from_mysql_ledger( self, - df_collection, + df_collection: DFCollection, user: User, create_main_accounts: Callable[..., None], offset: str, duration: timedelta, thl_web_rw: PostgresConfig, - thl_lm, - df_collection_data_type, + thl_ledger_manager: ThlLedgerManager, + df_collection_data_type: DFCollectionType, user_factory: Callable[..., User], - product: product: Product, - client_no_amm, - incite_item_factory, + product: Product, + client_no_amm: DaskClient, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], - mnt_filepath, + mnt_filepath: GRLDatasets, ): if df_collection.data_type != DFCollectionType.LEDGER: @@ -370,17 +375,17 @@ class TestDFCollectionItemMethod: def test_to_archive( self, - df_collection, + df_collection: DFCollection, user: User, offset: str, duration: timedelta, - df_collection_data_type, + df_collection_data_type: DFCollectionType, user_factory: Callable[..., User], - product: product: Product, - client_no_amm, - incite_item_factory, + product: Product, + client_no_amm: DaskClient, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], - mnt_filepath, + mnt_filepath: GRLDatasets, ): if df_collection.data_type in unsupported_mock_types: @@ -407,17 +412,17 @@ class TestDFCollectionItemMethod: def test__to_archive( self, - df_collection_data_type, - df_collection, + df_collection_data_type: DFCollectionType, + df_collection: DFCollection, user_factory: Callable[..., User], - product: product: Product, + product: Product, offset: str, duration: timedelta, - client_no_amm, + client_no_amm: DaskClient, user: User, - incite_item_factory, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], - mnt_filepath, + mnt_filepath: GRLDatasets, ): """We already have a test for the "non-private" version of this, which primarily just uses the respective Client to determine if @@ -480,19 +485,19 @@ class TestDFCollectionItemMethod: @pytest.mark.skip def test_to_archive_numbered_partial( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_initial_load( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_clear_corrupt_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @@ -505,34 +510,40 @@ class TestDFCollectionItemMethodBase: @pytest.mark.skip def test_path_exists( - self, df_collection_data_type, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_next_numbered_path( - self, df_collection_data_type, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_search_highest_numbered_path( - self, df_collection_data_type, offset: str, duration: timedelta + self, + df_collection_data_type: DFCollectionType, + offset: str, + duration: timedelta, ): pass @pytest.mark.skip def test_tmp_filename( - self, df_collection_data_type, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip - def test_tmp_path(self, df_collection_data_type, offset: str, duration: timedelta): + def test_tmp_path( + self, + ): pass def test_is_empty( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): """ test_has_empty was merged into this because item.has_empty is @@ -549,7 +560,8 @@ class TestDFCollectionItemMethodBase: assert item.has_empty() def test_has_partial_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): for item in df_collection.items: assert not item.has_partial_archive() @@ -557,7 +569,8 @@ class TestDFCollectionItemMethodBase: assert item.has_partial_archive() def test_has_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): for item in df_collection.items: # (1) Originally, nothing exists... so let's just make a file and @@ -594,7 +607,8 @@ class TestDFCollectionItemMethodBase: assert item.has_archive(include_empty=True) def test_delete_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): for item in df_collection.items: item: DFCollectionItem @@ -617,7 +631,8 @@ class TestDFCollectionItemMethodBase: assert not item.partial_path.exists() def test_should_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): schema: DataFrameSchema = df_collection._schema aa = schema.metadata[ARCHIVE_AFTER] @@ -635,12 +650,13 @@ class TestDFCollectionItemMethodBase: @pytest.mark.skip def test_set_empty( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass def test_valid_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): # Originally, nothing has been saved or anything.. so confirm it # always comes back as None @@ -664,18 +680,19 @@ class TestDFCollectionItemMethodBase: @pytest.mark.skip def test_validate_df( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_from_archive( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass def test__to_dict( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): for item in df_collection.items: @@ -694,19 +711,19 @@ class TestDFCollectionItemMethodBase: @pytest.mark.skip def test_delete_partial( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_cleanup_partials( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @pytest.mark.skip def test_delete_dangling_partials( - self, df_collection_data_type, df_collection, offset: str, duration: timedelta + self, ): pass @@ -726,7 +743,9 @@ async def test_client(client, s, worker): ) @gen_cluster(client=True, nthreads=[("127.0.0.1", 1)]) @pytest.mark.anyio -async def test_client_parametrize(c, s, w, df_collection_data_type, offset: str): +async def test_client_parametrize( + c, s, w, df_collection_data_type: DFCollectionType, offset: str +): """c,s,a are all required - the secondary Worker (b) is not required""" assert isinstance(c, Client), f"c is not Client, it's {type(c)}" @@ -750,17 +769,12 @@ class TestDFCollectionItemFunctionalTest: def test_to_archive_and_ddf( self, - df_collection_data_type, - offset: str, - duration: timedelta, - client_no_amm, - df_collection, - user: User, + client_no_amm: DaskClient, + df_collection: DFCollection, user_factory: Callable[..., User], - product: product: Product, - incite_item_factory, + product: Product, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], - mnt_filepath: GRLDatasets, ): if df_collection.data_type in unsupported_mock_types: @@ -799,17 +813,11 @@ class TestDFCollectionItemFunctionalTest: def test_filesize_estimate( self, - df_collection, - user: User, - offset: str, - duration: timedelta, - client_no_amm, + df_collection: DFCollection, user_factory: Callable[..., User], - product: product: Product, - df_collection_data_type, - incite_item_factory, + product: Product, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], - mnt_filepath: GRLDatasets, ): """A functional test to write some Parquet files for the DFCollection and then confirm that the files get written @@ -846,16 +854,12 @@ class TestDFCollectionItemFunctionalTest: def test_to_archive_client( self, - client_no_amm, - df_collection, + client_no_amm: DaskClient, + df_collection: DFCollection, user_factory: Callable[..., User], - product: product: Product, - offset: str, - duration: timedelta, - df_collection_data_type, - incite_item_factory, + product: Product, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], - mnt_filepath: GRLDatasets, ): delete_df_collection(coll=df_collection) @@ -885,7 +889,8 @@ class TestDFCollectionItemFunctionalTest: @pytest.mark.skip def test_get_items( - self, df_collection, product: product: Product, offset: str, duration: timedelta + self, + df_collection: DFCollection, ): with pytest.warns(expected_warning=ResourceWarning) as cm: df_collection.get_items_last365() @@ -898,16 +903,11 @@ class TestDFCollectionItemFunctionalTest: def test_saving_protections( self, - client_no_amm, - df_collection_data_type, - df_collection, - incite_item_factory, + df_collection: DFCollection, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], user_factory: Callable[..., User], - product: product: Product, - offset: str, - duration: timedelta, - mnt_filepath: GRLDatasets, + product: Product, ): """Don't allow creating an archive for data that will likely be overwritten or updated @@ -939,15 +939,8 @@ class TestDFCollectionItemFunctionalTest: def test_empty_item( self, - client_no_amm, - df_collection_data_type, - df_collection, - incite_item_factory, + df_collection: DFCollection, delete_df_collection: Callable[..., None], - user: User, - offset: str, - duration: timedelta, - mnt_filepath: GRLDatasets, ): delete_df_collection(coll=df_collection) @@ -967,16 +960,12 @@ class TestDFCollectionItemFunctionalTest: def test_file_touching( self, - client_no_amm, - df_collection_data_type, - df_collection, - incite_item_factory, + client_no_amm: DaskClient, + df_collection: DFCollection, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], user_factory: Callable[..., User], - product: product: Product, - offset: str, - duration: timedelta, - mnt_filepath, + product: Product, ): delete_df_collection(coll=df_collection) diff --git a/tests/incite/mergers/foundations/test_enriched_session.py b/tests/incite/mergers/foundations/test_enriched_session.py index 8254d81..2a161e4 100644 --- a/tests/incite/mergers/foundations/test_enriched_session.py +++ b/tests/incite/mergers/foundations/test_enriched_session.py @@ -1,3 +1,6 @@ +from __future__ import annotations + +from collections.abc import Callable from datetime import UTC, datetime, timedelta from decimal import Decimal from itertools import product @@ -5,10 +8,24 @@ from itertools import product import dask.dataframe as dd import pandas as pd import pytest +from dask.distributed import Client as DaskClient +from generalresearch.incite.collections.thl_web import ( + SessionDFCollection, + WallDFCollection, +) +from generalresearch.incite.mergers.foundations.enriched_session import ( + EnrichedSessionMerge, +) from generalresearch.incite.schemas.admin_responses import ( AdminPOPSessionSchema, ) +from generalresearch.models.admin.request import ( + ReportRequest, +) +from generalresearch.models.thl.product import Product +from generalresearch.models.thl.session import Session +from generalresearch.models.thl.user import User from generalresearch.pg_helper import PostgresConfig @@ -25,21 +42,20 @@ class TestEnrichedSession: def test_base( self, - client_no_amm, + client_no_amm: DaskClient, product: Product, user_factory: Callable[..., User], - wall_collection, - session_collection, - enriched_session_merge, + wall_collection: WallDFCollection, + session_collection: SessionDFCollection, + enriched_session_merge: EnrichedSessionMerge, thl_web_rr: PostgresConfig, delete_df_collection: Callable[..., None], - incite_item_factory, + incite_item_factory: Callable[..., None], ): - from generalresearch.models.thl.user import User delete_df_collection(coll=session_collection) - u1: User = user_factory(product=product: Product, created=session_collection.start) + u1: User = user_factory(product=product, created=session_collection.start) for item in session_collection.items: incite_item_factory(item=item, user=u1) @@ -52,7 +68,7 @@ class TestEnrichedSession: client=client_no_amm, wall_coll=wall_collection, session_coll=session_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) # -- @@ -85,16 +101,16 @@ class TestEnrichedSessionAdmin: def test_to_admin_response( self, - event_report_request, - enriched_session_merge, - client_no_amm, - wall_collection, - session_collection, + event_report_request: ReportRequest, + enriched_session_merge: EnrichedSessionMerge, + client_no_amm: DaskClient, + wall_collection: WallDFCollection, + session_collection: SessionDFCollection, thl_web_rr: PostgresConfig, - session_report_request, + session_report_request: ReportRequest, user_factory: Callable[..., User], - start, - session_factory, + start: datetime, + session_factory: Callable[..., Session], product_factory: Callable[..., Product], delete_df_collection: Callable[..., None], ): @@ -107,7 +123,7 @@ class TestEnrichedSessionAdmin: for p in [p1, p2]: u = user_factory(product=p) for i in range(50): - s = session_factory( + _ = session_factory( user=u, wall_count=1, wall_req_cpi=Decimal("1.00"), @@ -120,7 +136,7 @@ class TestEnrichedSessionAdmin: client=client_no_amm, session_coll=session_collection, wall_coll=wall_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) df = enriched_session_merge.to_admin_response( diff --git a/tests/incite/mergers/foundations/test_enriched_task_adjust.py b/tests/incite/mergers/foundations/test_enriched_task_adjust.py index a33a55a..0606b6f 100644 --- a/tests/incite/mergers/foundations/test_enriched_task_adjust.py +++ b/tests/incite/mergers/foundations/test_enriched_task_adjust.py @@ -1,9 +1,28 @@ +from __future__ import annotations + +from collections.abc import Callable from datetime import timedelta from itertools import product as iter_product import dask.dataframe as dd import pandas as pd import pytest +from dask.distributed import Client as DaskClient + +from generalresearch.incite.collections.thl_web import ( + SessionDFCollection, + TaskAdjustmentDFCollection, + WallDFCollection, +) +from generalresearch.incite.mergers.foundations.enriched_task_adjust import ( + EnrichedTaskAdjustMerge, +) +from generalresearch.incite.mergers.foundations.enriched_wall import ( + EnrichedWallMerge, +) +from generalresearch.models.thl.product import Product +from generalresearch.models.thl.user import User +from generalresearch.pg_helper import PostgresConfig @pytest.mark.parametrize( @@ -20,19 +39,18 @@ class TestEnrichedTaskAdjust: @pytest.mark.skip def test_base( self, - client_no_amm, + client_no_amm: DaskClient, user_factory: Callable[..., User], product: Product, - task_adj_collection, - wall_collection, - session_collection, - enriched_wall_merge, - enriched_task_adjust_merge, - incite_item_factory, + task_adj_collection: TaskAdjustmentDFCollection, + wall_collection: WallDFCollection, + session_collection: SessionDFCollection, + enriched_wall_merge: EnrichedWallMerge, + enriched_task_adjust_merge: EnrichedTaskAdjustMerge, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], thl_web_rr: PostgresConfig, ): - from generalresearch.models.thl.user import User # -- Build & Setup delete_df_collection(coll=session_collection) @@ -48,14 +66,14 @@ class TestEnrichedTaskAdjust: client=client_no_amm, session_coll=session_collection, wall_coll=wall_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) enriched_task_adjust_merge.build( client=client_no_amm, task_adjust_coll=task_adj_collection, enriched_wall=enriched_wall_merge, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) # -- diff --git a/tests/incite/mergers/foundations/test_enriched_wall.py b/tests/incite/mergers/foundations/test_enriched_wall.py index a0ca4dd..0cb8f60 100644 --- a/tests/incite/mergers/foundations/test_enriched_wall.py +++ b/tests/incite/mergers/foundations/test_enriched_wall.py @@ -1,3 +1,4 @@ +from collections.abc import Callable from datetime import UTC, datetime, timedelta from decimal import Decimal from itertools import product as iter_product @@ -5,11 +6,23 @@ from itertools import product as iter_product import dask.dataframe as dd import pandas as pd import pytest +from dask.distributed import Client as DaskClient + +from generalresearch.incite.collections.thl_web import ( + SessionDFCollection, + WallDFCollection, +) # noinspection PyUnresolvedReferences from generalresearch.incite.mergers.foundations.enriched_wall import ( + EnrichedWallMerge, EnrichedWallMergeItem, ) +from generalresearch.models.admin.request import ReportRequest +from generalresearch.models.thl.product import Product +from generalresearch.models.thl.session import Session +from generalresearch.models.thl.user import User +from generalresearch.pg_helper import PostgresConfig @pytest.mark.parametrize( @@ -20,22 +33,21 @@ class TestEnrichedWall: def test_base( self, - client_no_amm, + client_no_amm: DaskClient, product: Product, user_factory: Callable[..., User], - wall_collection, + wall_collection: WallDFCollection, thl_web_rr: PostgresConfig, - session_collection, - enriched_wall_merge, + session_collection: SessionDFCollection, + enriched_wall_merge: EnrichedWallMerge, delete_df_collection: Callable[..., None], - incite_item_factory, + incite_item_factory: Callable[..., None], ): - from generalresearch.models.thl.user import User # -- Build & Setup delete_df_collection(coll=session_collection) delete_df_collection(coll=wall_collection) - u1: User = user_factory(product=product: Product, created=session_collection.start) + u1: User = user_factory(product=product, created=session_collection.start) for item in session_collection.items: incite_item_factory(item=item, user=u1) @@ -48,7 +60,7 @@ class TestEnrichedWall: client=client_no_amm, wall_coll=wall_collection, session_coll=session_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) # -- @@ -63,19 +75,19 @@ class TestEnrichedWall: def test_base_item( self, - client_no_amm, + client_no_amm: DaskClient, product: Product, user_factory: Callable[..., User], - wall_collection, - session_collection, - enriched_wall_merge, + wall_collection: WallDFCollection, + session_collection: SessionDFCollection, + enriched_wall_merge: EnrichedWallMerge, delete_df_collection: Callable[..., None], thl_web_rr: PostgresConfig, - incite_item_factory, + incite_item_factory: Callable[..., None], ): # -- Build & Setup delete_df_collection(coll=session_collection) - u = user_factory(product=product: Product, created=session_collection.start) + u = user_factory(product=product, created=session_collection.start) for item in session_collection.items: incite_item_factory(item=item, user=u) @@ -87,7 +99,7 @@ class TestEnrichedWall: client=client_no_amm, wall_coll=wall_collection, session_coll=session_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) # -- @@ -99,14 +111,14 @@ class TestEnrichedWall: try: modified_time1 = path.stat().st_mtime - except Exception: + except OSError: modified_time1 = 0 item.build( client=client_no_amm, wall_coll=wall_collection, session_coll=session_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) modified_time2 = path.stat().st_mtime @@ -150,7 +162,12 @@ class TestEnrichedWallToAdmin: def duration(self) -> timedelta | None: return timedelta(days=5) - def test_empty(self, enriched_wall_merge, client_no_amm, start): + def test_empty( + self, + enriched_wall_merge: EnrichedWallMerge, + client_no_amm: DaskClient, + start: datetime, + ): from generalresearch.models.admin.request import ReportRequest rr = ReportRequest.model_validate({"interval": "5min", "start": start}) @@ -167,18 +184,18 @@ class TestEnrichedWallToAdmin: def test_to_admin_response( self, - event_report_request, - enriched_wall_merge, - client_no_amm, - wall_collection, - session_collection, + event_report_request: ReportRequest, + enriched_wall_merge: EnrichedWallMerge, + client_no_amm: DaskClient, + wall_collection: WallDFCollection, + session_collection: SessionDFCollection, thl_web_rr: PostgresConfig, - user, - session_factory, + user: User, + session_factory: Callable[..., Session], delete_df_collection: Callable[..., None], product_factory: Callable[..., Product], user_factory: Callable[..., User], - start, + start: datetime, ): delete_df_collection(coll=wall_collection) delete_df_collection(coll=session_collection) @@ -189,7 +206,7 @@ class TestEnrichedWallToAdmin: for p in [p1, p2]: u = user_factory(product=p) for i in range(50): - s = session_factory( + _ = session_factory( user=u, wall_count=2, wall_req_cpi=Decimal("1.00"), @@ -203,7 +220,7 @@ class TestEnrichedWallToAdmin: client=client_no_amm, wall_coll=wall_collection, session_coll=session_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) df = enriched_wall_merge.to_admin_response( diff --git a/tests/incite/mergers/test_merge_collection.py b/tests/incite/mergers/test_merge_collection.py index 15fa4db..cf8315f 100644 --- a/tests/incite/mergers/test_merge_collection.py +++ b/tests/incite/mergers/test_merge_collection.py @@ -1,3 +1,5 @@ +from __future__ import annotations + from datetime import UTC, datetime, timedelta from itertools import product @@ -5,12 +7,13 @@ import pandas as pd import pytest from pandera.pandas import DataFrameSchema +from generalresearch.incite.base import GRLDatasets from generalresearch.incite.mergers import ( MergeCollection, MergeType, ) -merge_types = list(e for e in MergeType if e != MergeType.TEST) +merge_types = [e for e in MergeType if e != MergeType.TEST] @pytest.mark.parametrize( @@ -26,7 +29,11 @@ merge_types = list(e for e in MergeType if e != MergeType.TEST) ) class TestMergeCollection: - def test_init(self, mnt_filepath, merge_type, offset, duration, start): + def test_init( + self, + mnt_filepath: GRLDatasets, + merge_type: MergeType, + ): with pytest.raises(expected_exception=ValueError) as cm: MergeCollection(archive_path=mnt_filepath.data_src) assert "Must explicitly provide a merge_type" in str(cm.value) @@ -37,7 +44,14 @@ class TestMergeCollection: ) assert instance.merge_type == merge_type - def test_items(self, mnt_filepath, merge_type, offset, duration, start): + def test_items( + self, + mnt_filepath: GRLDatasets, + merge_type: MergeType, + offset: str, + duration: timedelta, + start: datetime, + ): instance = MergeCollection( merge_type=merge_type, offset=offset, @@ -48,7 +62,14 @@ class TestMergeCollection: assert len(instance.interval_range) == len(instance.items) - def test_progress(self, mnt_filepath, merge_type, offset, duration, start): + def test_progress( + self, + mnt_filepath: GRLDatasets, + merge_type: MergeType, + offset: str, + duration: timedelta, + start: datetime, + ): instance = MergeCollection( merge_type=merge_type, offset=offset, @@ -62,7 +83,11 @@ class TestMergeCollection: assert instance.progress.shape[1] == 7 assert instance.progress["group_by"].isnull().all() - def test_schema(self, mnt_filepath, merge_type, offset, duration, start): + def test_schema( + self, + mnt_filepath: GRLDatasets, + merge_type: MergeType, + ): instance = MergeCollection( merge_type=merge_type, archive_path=mnt_filepath.archive_path(enum_type=merge_type), @@ -70,7 +95,14 @@ class TestMergeCollection: assert isinstance(instance._schema, DataFrameSchema) - def test_load(self, mnt_filepath, merge_type, offset, duration, start): + def test_load( + self, + mnt_filepath: GRLDatasets, + merge_type: MergeType, + offset: str, + duration: timedelta, + start: datetime, + ): instance = MergeCollection( merge_type=merge_type, start=start, @@ -82,7 +114,14 @@ class TestMergeCollection: # Confirm that there are no archives available yet assert instance.progress.has_archive.eq(False).all() - def test_get_items(self, mnt_filepath, merge_type, offset, duration, start): + def test_get_items( + self, + mnt_filepath: GRLDatasets, + merge_type: MergeType, + offset: str, + duration: timedelta, + start: datetime, + ): instance = MergeCollection( start=start, finished=start + duration, diff --git a/tests/incite/mergers/test_merge_collection_item.py b/tests/incite/mergers/test_merge_collection_item.py index 3d0b644..5ca2f6b 100644 --- a/tests/incite/mergers/test_merge_collection_item.py +++ b/tests/incite/mergers/test_merge_collection_item.py @@ -1,10 +1,16 @@ +from __future__ import annotations + from datetime import timedelta from itertools import product from pathlib import PurePath import pytest -from generalresearch.incite.mergers import MergeCollectionItem, MergeType +from generalresearch.incite.mergers import ( + MergeCollection, + MergeCollectionItem, + MergeType, +) @pytest.mark.parametrize( @@ -19,7 +25,10 @@ from generalresearch.incite.mergers import MergeCollectionItem, MergeType ) class TestMergeCollectionItem: - def test_file_naming(self, merge_collection, offset, duration, start): + def test_file_naming( + self, + merge_collection: MergeCollection, + ): assert len(merge_collection.items) == 25 items: list[MergeCollectionItem] = merge_collection.items @@ -34,7 +43,10 @@ class TestMergeCollectionItem: assert i._collection.offset in i.filename assert i.start.strftime("%Y-%m-%d-%H-%M-%S") in i.filename - def test_archives(self, merge_collection, offset, duration, start): + def test_archives( + self, + merge_collection: MergeCollection, + ): assert len(merge_collection.items) == 25 for i in merge_collection.items: @@ -44,10 +56,13 @@ class TestMergeCollectionItem: assert not i.has_partial_archive() assert i.has_archive() == i.path_exists(generic_path=i.path) - res = set([i.should_archive() for i in merge_collection.items]) + res = {i.should_archive() for i in merge_collection.items} assert len(res) == 1 - def test_item_to_archive(self, merge_collection, offset, duration, start): + def test_item_to_archive( + self, + merge_collection: MergeCollection, + ): for item in merge_collection.items: item: MergeCollectionItem assert not item.has_archive() diff --git a/tests/incite/mergers/test_pop_ledger.py b/tests/incite/mergers/test_pop_ledger.py index d054eb6..529a641 100644 --- a/tests/incite/mergers/test_pop_ledger.py +++ b/tests/incite/mergers/test_pop_ledger.py @@ -1,12 +1,25 @@ +from __future__ import annotations + +from collections.abc import Callable from datetime import UTC, datetime, timedelta from itertools import product as iter_product import pandas as pd import pytest +from dask.distributed import Client as DaskClient +from generalresearch.incite.base import GRLDatasets +from generalresearch.incite.collections.thl_web import ( + LedgerDFCollection, + SessionDFCollection, +) +from generalresearch.incite.mergers.pop_ledger import PopLedgerMerge from generalresearch.incite.schemas.mergers.pop_ledger import ( numerical_col_names, ) +from generalresearch.managers.thl.ledger_manager.thl_ledger import ThlLedgerManager +from generalresearch.models.thl.product import Product +from generalresearch.models.thl.user import User @pytest.mark.parametrize( @@ -30,20 +43,20 @@ class TestMergePOPLedger: def test_base( self, - client_no_amm, - ledger_collection, - pop_ledger_merge, + client_no_amm: DaskClient, + ledger_collection: LedgerDFCollection, + pop_ledger_merge: PopLedgerMerge, product: Product, user_factory: Callable[..., User], create_main_accounts: Callable[..., None], - thl_lm, + thl_ledger_manager: ThlLedgerManager, delete_df_collection: Callable[..., None], - incite_item_factory, + incite_item_factory: Callable[..., None], delete_ledger_db: Callable[..., None], ): from generalresearch.models.thl.ledger import LedgerAccount - u = user_factory(product=product: Product, created=ledger_collection.start) + u = user_factory(product=product, created=ledger_collection.start) # -- Build & Setup delete_ledger_db() @@ -73,19 +86,21 @@ class TestMergePOPLedger: # -- - user_wallet_account: LedgerAccount = thl_lm.get_account_or_create_user_wallet( - user=u + user_wallet_account: LedgerAccount = ( + thl_ledger_manager.get_account_or_create_user_wallet(user=u) + ) + cash_account: LedgerAccount = thl_ledger_manager.get_account_cash() + rev_account: LedgerAccount = ( + thl_ledger_manager.get_account_task_complete_revenue() ) - cash_account: LedgerAccount = thl_lm.get_account_cash() - rev_account: LedgerAccount = thl_lm.get_account_task_complete_revenue() item_finishes = [i.finish for i in ledger_collection.items] item_finishes.sort(reverse=True) last_item_finish = item_finishes[0] # Pure SQL based lookups - cash_balance: int = thl_lm.get_account_balance(account=cash_account) - rev_balance: int = thl_lm.get_account_balance(account=rev_account) + cash_balance: int = thl_ledger_manager.get_account_balance(account=cash_account) + rev_balance: int = thl_ledger_manager.get_account_balance(account=rev_account) assert cash_balance > rev_balance # (1) Test Cash Account @@ -123,39 +138,42 @@ class TestMergePOPLedger: def test_pydantic_init( self, - client_no_amm, - ledger_collection, - pop_ledger_merge, - mnt_filepath, + client_no_amm: DaskClient, + ledger_collection: LedgerDFCollection, + pop_ledger_merge: PopLedgerMerge, + mnt_filepath: GRLDatasets, product: Product, user_factory: Callable[..., User], create_main_accounts: Callable[..., None], - offset, - duration, - start, - thl_lm, - incite_item_factory, + offset: str, + duration: timedelta, + start: datetime, + thl_ledger_manager: ThlLedgerManager, + incite_item_factory: Callable[..., None], delete_df_collection: Callable[..., None], delete_ledger_db: Callable[..., None], - session_collection, + session_collection: SessionDFCollection, ): from generalresearch.models.thl.finance import ProductBalances from generalresearch.models.thl.ledger import LedgerAccount from generalresearch.models.thl.product import Product - u = user_factory(product=product: Product, created=session_collection.start) + u = user_factory(product=product, created=session_collection.start) assert ledger_collection.finished is not None - assert isinstance(u.product: Product, Product) + assert isinstance(u.product, Product) delete_ledger_db() - create_main_accounts(), + create_main_accounts() + delete_df_collection(coll=ledger_collection) - bp_account: LedgerAccount = thl_lm.get_account_or_create_bp_wallet( + bp_account: LedgerAccount = thl_ledger_manager.get_account_or_create_bp_wallet( product=u.product ) - cash_account: LedgerAccount = thl_lm.get_account_cash() - rev_account: LedgerAccount = thl_lm.get_account_task_complete_revenue() + cash_account: LedgerAccount = thl_ledger_manager.get_account_cash() + rev_account: LedgerAccount = ( + thl_ledger_manager.get_account_task_complete_revenue() + ) for item in ledger_collection.items: incite_item_factory(item=item, user=u) @@ -185,8 +203,10 @@ class TestMergePOPLedger: assert instance.payout == instance.net == instance.bp_payment_credit assert instance.available_balance < instance.net assert instance.available_balance + instance.retainer == instance.net - assert instance.balance == thl_lm.get_account_balance(bp_account) - assert df["bp_payment.CREDIT"].sum() == thl_lm.get_account_balance(bp_account) + assert instance.balance == thl_ledger_manager.get_account_balance(bp_account) + assert df["bp_payment.CREDIT"].sum() == thl_ledger_manager.get_account_balance( + bp_account + ) # (2) Filter by the Cash Account ddf = pop_ledger_merge.ddf( @@ -199,7 +219,7 @@ class TestMergePOPLedger: ) df: pd.DataFrame = client_no_amm.compute(collections=ddf, sync=True) - cash_balance: int = thl_lm.get_account_balance(account=cash_account) + cash_balance: int = thl_ledger_manager.get_account_balance(account=cash_account) assert df["bp_payment.CREDIT"].sum() == 0 assert cash_balance > 0 assert df["mp_payment.CREDIT"].sum() == 0 @@ -216,7 +236,7 @@ class TestMergePOPLedger: ) df: pd.DataFrame = client_no_amm.compute(collections=ddf, sync=True) - rev_balance: int = thl_lm.get_account_balance(account=rev_account) + rev_balance: int = thl_ledger_manager.get_account_balance(account=rev_account) assert rev_balance == 0 assert df["bp_payment.CREDIT"].sum() == 0 assert df["mp_payment.DEBIT"].sum() == 0 @@ -224,27 +244,28 @@ class TestMergePOPLedger: def test_resample( self, - client_no_amm, - ledger_collection, - pop_ledger_merge, - mnt_filepath, + client_no_amm: DaskClient, + ledger_collection: LedgerDFCollection, + pop_ledger_merge: PopLedgerMerge, + mnt_filepath: GRLDatasets, user_factory: Callable[..., User], product: Product, create_main_accounts: Callable[..., None], - offset, - duration, - start, - thl_lm, + offset: str, + duration: timedelta, + start: datetime, + thl_ledger_manager: ThlLedgerManager, delete_df_collection: Callable[..., None], - incite_item_factory, + incite_item_factory: Callable[..., None], ): - from generalresearch.models.thl.user import User assert ledger_collection.finished is not None delete_df_collection(coll=ledger_collection) u1: User = user_factory(product=product) - bp_account = thl_lm.get_account_or_create_bp_wallet(product=u1.product) + bp_account = thl_ledger_manager.get_account_or_create_bp_wallet( + product=u1.product + ) for item in ledger_collection.items: incite_item_factory(user=u1, item=item) @@ -274,7 +295,7 @@ class TestMergePOPLedger: assert isinstance(df.index, pd.Index) assert isinstance(df.index, pd.DatetimeIndex) - bp_account_balance = thl_lm.get_account_balance(account=bp_account) + bp_account_balance = thl_ledger_manager.get_account_balance(account=bp_account) # Initial sum initial_sum = df.sum().sum() diff --git a/tests/incite/mergers/test_ym_survey_merge.py b/tests/incite/mergers/test_ym_survey_merge.py index a0b8b87..8a4897b 100644 --- a/tests/incite/mergers/test_ym_survey_merge.py +++ b/tests/incite/mergers/test_ym_survey_merge.py @@ -1,8 +1,24 @@ +from __future__ import annotations + +from collections.abc import Callable from datetime import UTC, datetime, timedelta from itertools import product import pandas as pd import pytest +from dask.distributed import Client as DaskClient + +from generalresearch.incite.collections.thl_web import ( + SessionDFCollection, + WallDFCollection, +) +from generalresearch.incite.mergers.foundations.enriched_session import ( + EnrichedSessionMerge, +) +from generalresearch.incite.mergers.ym_survey_wall import YMSurveyWallMerge +from generalresearch.models.thl.product import Product +from generalresearch.models.thl.user import User +from generalresearch.pg_helper import PostgresConfig # noinspection PyUnresolvedReferences @@ -27,21 +43,20 @@ class TestYMSurveyMerge: def test_base( self, - client_no_amm, + client_no_amm: DaskClient, user_factory: Callable[..., User], product: Product, - ym_survey_wall_merge, - wall_collection, - session_collection, - enriched_session_merge, + ym_survey_wall_merge: YMSurveyWallMerge, + wall_collection: WallDFCollection, + session_collection: SessionDFCollection, + enriched_session_merge: EnrichedSessionMerge, delete_df_collection: Callable[..., None], - incite_item_factory, + incite_item_factory: Callable[..., None], thl_web_rr: PostgresConfig, ): - from generalresearch.models.thl.user import User delete_df_collection(coll=session_collection) - user: User = user_factory(product=product: Product, created=session_collection.start) + user: User = user_factory(product=product, created=session_collection.start) # -- Build & Setup assert ym_survey_wall_merge.start is None @@ -61,15 +76,15 @@ class TestYMSurveyMerge: client=client_no_amm, session_coll=session_collection, wall_coll=wall_collection, - pg_config=thl_web_rr: PostgresConfig, + pg_config=thl_web_rr, ) assert enriched_session_merge.progress.has_archive.eq(True).all() ddf = enriched_session_merge.ddf() - df: pd.DataFrame = client_no_amm.compute(collections=ddf, sync=True) + df1: pd.DataFrame | None = client_no_amm.compute(collections=ddf, sync=True) - assert isinstance(df, pd.DataFrame) - assert not df.empty + assert isinstance(df1, pd.DataFrame) + assert not df1.empty # -- @@ -83,18 +98,18 @@ class TestYMSurveyMerge: # -- ddf = ym_survey_wall_merge.ddf() - df: pd.DataFrame = client_no_amm.compute(collections=ddf, sync=True) + df2: pd.DataFrame | None = client_no_amm.compute(collections=ddf, sync=True) - assert isinstance(df, pd.DataFrame) - assert not df.empty + assert isinstance(df2, pd.DataFrame) + assert not df2.empty # -- - assert df.product_id.nunique() == 1 - assert df.team_id.nunique() == 1 - assert df.source.nunique() > 1 + assert df2.product_id.nunique() == 1 + assert df2.team_id.nunique() == 1 + assert df2.source.nunique() > 1 - started_min_ts = df.started.min() - started_max_ts = df.started.max() + started_min_ts = df2.started.min() + started_max_ts = df2.started.max() assert type(started_min_ts) is pd.Timestamp assert type(started_max_ts) is pd.Timestamp diff --git a/tests/incite/schemas/test_admin_responses.py b/tests/incite/schemas/test_admin_responses.py index e98eecd..d2658ea 100644 --- a/tests/incite/schemas/test_admin_responses.py +++ b/tests/incite/schemas/test_admin_responses.py @@ -1,8 +1,11 @@ +from __future__ import annotations + from datetime import UTC, datetime, timedelta from random import sample import numpy as np import pandas as pd +import pandera as pa import pytest from generalresearch.incite.schemas import empty_dataframe_from_schema @@ -16,12 +19,14 @@ from generalresearch.locales import Localelator class TestAdminPOPSchema: schema_df = empty_dataframe_from_schema(AdminPOPSchema) countries = list(Localelator().get_all_countries())[:5] - dates = [datetime(year=2024, month=1, day=i, tzinfo=None) for i in range(1, 10)] + dates = [ + datetime(year=2024, month=1, day=i, tzinfo=None) for i in range(1, 10) # noqa + ] @classmethod def assign_valid_vals(cls, df: pd.DataFrame) -> pd.DataFrame: for c in df.columns: - check_attrs: dict = AdminPOPSchema.columns[c].checks[0].statistics + check_attrs = AdminPOPSchema.columns[c].checks[0].statistics df[c] = np.random.randint( check_attrs["min_value"], check_attrs["max_value"], df.shape[0] ) @@ -29,7 +34,7 @@ class TestAdminPOPSchema: return df def test_empty(self): - with pytest.raises(Exception): + with pytest.raises(pa.errors.SchemaError): AdminPOPSchema.validate(pd.DataFrame()) def test_new_empty_df(self): @@ -42,7 +47,7 @@ class TestAdminPOPSchema: def test_valid(self): # (1) Works with raw naive datetime dates = [ - datetime(year=2024, month=1, day=i, tzinfo=None).isoformat() + datetime(year=2024, month=1, day=i, tzinfo=None).isoformat() # noqa for i in range(1, 10) ] df = pd.DataFrame( @@ -57,7 +62,10 @@ class TestAdminPOPSchema: assert isinstance(df, pd.DataFrame) # (2) Works with isoformat naive datetime - dates = [datetime(year=2024, month=1, day=i, tzinfo=None) for i in range(1, 10)] + dates = [ + datetime(year=2024, month=1, day=i, tzinfo=None) # noqa + for i in range(1, 10) + ] df = pd.DataFrame( index=pd.MultiIndex.from_product( iterables=[dates, self.countries], names=["index0", "index1"] @@ -84,12 +92,12 @@ class TestAdminPOPSchema: # Initially, they're all set with a timezone timestmaps: list[pd.Timestamp] = [i for i in df.index.get_level_values(0)] - assert all([ts.tz == UTC for ts in timestmaps]) + assert all(ts.tz == UTC for ts in timestmaps) # After validation, the timezone is removed df = AdminPOPSchema.validate(df) timestmaps: list[pd.Timestamp] = [i for i in df.index.get_level_values(0)] - assert all([ts.tz is None for ts in timestmaps]) + assert all(ts.tz is None for ts in timestmaps) def test_index_tz_no_future_beyond_one_year(self): now = datetime.now(tz=UTC) @@ -123,12 +131,12 @@ class TestAdminPOPSchema: df = self.assign_valid_vals(df) vals = [i for i in df.index.get_level_values(1)] - assert all([isinstance(v, float) for v in vals]) + assert all(isinstance(v, float) for v in vals) df = AdminPOPSchema.validate(df, lazy=True) vals = [i for i in df.index.get_level_values(1)] - assert all([isinstance(v, str) for v in vals]) + assert all(isinstance(v, str) for v in vals) # --- int to str --- @@ -142,12 +150,12 @@ class TestAdminPOPSchema: df = self.assign_valid_vals(df) vals = [i for i in df.index.get_level_values(1)] - assert all([isinstance(v, int) for v in vals]) + assert all(isinstance(v, int) for v in vals) df = AdminPOPSchema.validate(df, lazy=True) vals = [i for i in df.index.get_level_values(1)] - assert all([isinstance(v, str) for v in vals]) + assert all(isinstance(v, str) for v in vals) # a = 1 assert isinstance(df, pd.DataFrame) @@ -170,7 +178,7 @@ class TestAdminPOPSchema: assert isinstance(df, pd.DataFrame) timestmaps: list[pd.Timestamp] = [i for i in df.index.get_level_values(0)] - assert all([ts.tz is None for ts in timestmaps]) + assert all(ts.tz is None for ts in timestmaps) # (2) Timezones are removed dates = [ @@ -187,12 +195,12 @@ class TestAdminPOPSchema: # Has tz before validation, and none after timestmaps: list[pd.Timestamp] = [i for i in df.index.get_level_values(0)] - assert all([ts.tz is UTC for ts in timestmaps]) + assert all(ts.tz is UTC for ts in timestmaps) df = AdminPOPSchema.validate(df, lazy=True) timestmaps: list[pd.Timestamp] = [i for i in df.index.get_level_values(0)] - assert all([ts.tz is None for ts in timestmaps]) + assert all(ts.tz is None for ts in timestmaps) def test_clipping(self): df = pd.DataFrame( diff --git a/tests/incite/schemas/test_thl_web.py b/tests/incite/schemas/test_thl_web.py index 7f4434b..9b34ce0 100644 --- a/tests/incite/schemas/test_thl_web.py +++ b/tests/incite/schemas/test_thl_web.py @@ -16,7 +16,7 @@ class TestWallSchema: df = pd.DataFrame(columns=THLWallSchema.columns.keys()) - with pytest.raises(SchemaError) as cm: + with pytest.raises(SchemaError): THLWallSchema.validate(df) def test_no_rows(self): @@ -24,7 +24,7 @@ class TestWallSchema: df = pd.DataFrame(index=["uuid"], columns=THLWallSchema.columns.keys()) - with pytest.raises(SchemaError) as cm: + with pytest.raises(SchemaError): THLWallSchema.validate(df) def test_new_empty_df(self): @@ -50,7 +50,7 @@ class TestSessionSchema: df = pd.DataFrame(columns=THLSessionSchema.columns.keys()) df.set_index("uuid", inplace=True) - with pytest.raises(SchemaError) as cm: + with pytest.raises(SchemaError): THLSessionSchema.validate(df) def test_no_rows(self): @@ -58,7 +58,7 @@ class TestSessionSchema: df = pd.DataFrame(index=["id"], columns=THLSessionSchema.columns.keys()) - with pytest.raises(SchemaError) as cm: + with pytest.raises(SchemaError): THLSessionSchema.validate(df) def test_new_empty_df(self): diff --git a/tests/incite/test_collection_base.py b/tests/incite/test_collection_base.py index 7e1577a..d6ce2b1 100644 --- a/tests/incite/test_collection_base.py +++ b/tests/incite/test_collection_base.py @@ -1,3 +1,5 @@ +from __future__ import annotations + from datetime import UTC, datetime, timedelta, timezone from os.path import exists as pexists from os.path import join as pjoin @@ -9,7 +11,7 @@ import pandas as pd import pytest from _pytest._code.code import ExceptionInfo -from generalresearch.incite.base import CollectionBase +from generalresearch.incite.base import CollectionBase, GRLDatasets AGO_15min = (datetime.now(tz=UTC) - timedelta(minutes=15)).replace(microsecond=0) AGO_1HR = (datetime.now(tz=UTC) - timedelta(hours=1)).replace(microsecond=0) @@ -17,11 +19,11 @@ AGO_2HR = (datetime.now(tz=UTC) - timedelta(hours=2)).replace(microsecond=0) class TestCollectionBase: - def test_init(self, mnt_filepath): + def test_init(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) assert instance.df.empty is True - def test_init_df(self, mnt_filepath): + def test_init_df(self, mnt_filepath: GRLDatasets): # Only an empty pd.DataFrame can ever be provided instance = CollectionBase( df=pd.DataFrame({}), archive_path=mnt_filepath.data_src @@ -43,7 +45,7 @@ class TestCollectionBase: ) assert "Do not provide a pd.DataFrame" in str(cm.value) - def test_init_start(self, mnt_filepath): + def test_init_start(self, mnt_filepath: GRLDatasets): with pytest.raises(expected_exception=ValueError) as cm: cm: ExceptionInfo CollectionBase( @@ -74,7 +76,7 @@ class TestCollectionBase: cm.value ) - def test_init_archive_path(self, mnt_filepath): + def test_init_archive_path(self, mnt_filepath: GRLDatasets): """DirectoryPath is apparently smart enough to confirm that the directory path exists. """ @@ -99,7 +101,7 @@ class TestCollectionBase: CollectionBase(archive_path=new_path) assert "Path does not point to a directory" in str(cm.value) - def test_init_offset(self, mnt_filepath): + def test_init_offset(self, mnt_filepath: GRLDatasets): with pytest.raises(expected_exception=ValueError) as cm: cm: ExceptionInfo CollectionBase(offset="1:X", archive_path=mnt_filepath.data_src) @@ -118,14 +120,14 @@ class TestCollectionBase: class TestCollectionBaseProperties: - def test_items(self, mnt_filepath): + def test_items(self, mnt_filepath: GRLDatasets): with pytest.raises(expected_exception=NotImplementedError) as cm: cm: ExceptionInfo instance = CollectionBase(archive_path=mnt_filepath.data_src) - x = instance.items + _ = instance.items assert "Must override" in str(cm.value) - def test_interval_range(self, mnt_filepath): + def test_interval_range(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) # Private method requires the end parameter with pytest.raises(expected_exception=AssertionError) as cm: @@ -147,7 +149,7 @@ class TestCollectionBaseProperties: assert res.is_monotonic_increasing assert res.is_unique - def test_interval_range2(self, mnt_filepath): + def test_interval_range2(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) assert isinstance(instance.interval_range, list) @@ -166,16 +168,16 @@ class TestCollectionBaseProperties: ) assert len(instance.interval_range) == 2 - def test_progress(self, mnt_filepath): + def test_progress(self, mnt_filepath: GRLDatasets): with pytest.raises(expected_exception=NotImplementedError) as cm: cm: ExceptionInfo instance = CollectionBase( start=AGO_15min, offset="3min", archive_path=mnt_filepath.data_src ) - x = instance.progress + _ = instance.progress assert "Must override" in str(cm.value) - def test_progress2(self, mnt_filepath): + def test_progress2(self, mnt_filepath: GRLDatasets): instance = CollectionBase( start=AGO_2HR, offset="15min", @@ -184,10 +186,10 @@ class TestCollectionBaseProperties: assert instance.df.empty with pytest.raises(expected_exception=NotImplementedError) as cm: - df = instance.progress + _ = instance.progress assert "Must override" in str(cm.value) - def test_items2(self, mnt_filepath): + def test_items2(self, mnt_filepath: GRLDatasets): """There can't be a test for this because the Items need a path whic isn't possible in the generic form """ @@ -197,7 +199,7 @@ class TestCollectionBaseProperties: with pytest.raises(expected_exception=NotImplementedError) as cm: cm: ExceptionInfo - items = instance.items + _ = instance.items assert "Must override" in str(cm.value) # item = items[-3] @@ -208,19 +210,19 @@ class TestCollectionBaseProperties: # assert str(df.product_id.dtype) == "object" # assert str(ddf.product_id.dtype) == "string" - def test_items3(self, mnt_filepath): + def test_items3(self, mnt_filepath: GRLDatasets): instance = CollectionBase( start=AGO_2HR, offset="15min", archive_path=mnt_filepath.data_src, ) with pytest.raises(expected_exception=NotImplementedError) as cm: - item = instance.items[0] + _ = instance.items[0] assert "Must override" in str(cm.value) class TestCollectionBaseMethodsCleanup: - def test_fetch_force_rr_latest(self, mnt_filepath): + def test_fetch_force_rr_latest(self, mnt_filepath: GRLDatasets): coll = CollectionBase(archive_path=mnt_filepath.data_src) with pytest.raises(expected_exception=Exception) as cm: @@ -228,7 +230,7 @@ class TestCollectionBaseMethodsCleanup: coll.fetch_force_rr_latest(sources=[]) assert "Must override" in str(cm.value) - def test_fetch_all_paths(self, mnt_filepath): + def test_fetch_all_paths(self, mnt_filepath: GRLDatasets): coll = CollectionBase(archive_path=mnt_filepath.data_src) with pytest.raises(expected_exception=NotImplementedError) as cm: @@ -242,16 +244,16 @@ class TestCollectionBaseMethodsCleanup: class TestCollectionBaseMethodsCleanup: @pytest.mark.skip - def test_cleanup_partials(self, mnt_filepath): + def test_cleanup_partials(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) assert instance.cleanup_partials() is None # it doesn't return anything - def test_clear_tmp_archives(self, mnt_filepath): + def test_clear_tmp_archives(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) assert instance.clear_tmp_archives() is None # it doesn't return anything @pytest.mark.skip - def test_clear_corrupt_archives(self, mnt_filepath): + def test_clear_corrupt_archives(self, mnt_filepath: GRLDatasets): """TODO: expand this so it actually has corrupt archives that we check to see if they're removed """ @@ -259,14 +261,14 @@ class TestCollectionBaseMethodsCleanup: assert instance.clear_corrupt_archives() is None # it doesn't return anything @pytest.mark.skip - def test_rebuild_symlinks(self, mnt_filepath): + def test_rebuild_symlinks(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) assert instance.rebuild_symlinks() is None class TestCollectionBaseMethodsSourceTiming: - def test_get_item(self, mnt_filepath): + def test_get_item(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) i = pd.Interval(left=1, right=2, closed="left") @@ -274,7 +276,7 @@ class TestCollectionBaseMethodsSourceTiming: instance.get_item(interval=i) assert "Must override" in str(cm.value) - def test_get_item_start(self, mnt_filepath): + def test_get_item_start(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) dt = datetime.now(tz=UTC) @@ -284,7 +286,7 @@ class TestCollectionBaseMethodsSourceTiming: instance.get_item_start(start=start) assert "Must override" in str(cm.value) - def test_get_items(self, mnt_filepath): + def test_get_items(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) dt = datetime.now(tz=UTC) @@ -293,21 +295,21 @@ class TestCollectionBaseMethodsSourceTiming: instance.get_items(since=dt) assert "Must override" in str(cm.value) - def test_get_items_from_year(self, mnt_filepath): + def test_get_items_from_year(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) with pytest.raises(expected_exception=NotImplementedError) as cm: instance.get_items_from_year(year=2020) assert "Must override" in str(cm.value) - def test_get_items_last90(self, mnt_filepath): + def test_get_items_last90(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) with pytest.raises(expected_exception=NotImplementedError) as cm: instance.get_items_last90() assert "Must override" in str(cm.value) - def test_get_items_last365(self, mnt_filepath): + def test_get_items_last365(self, mnt_filepath: GRLDatasets): instance = CollectionBase(archive_path=mnt_filepath.data_src) with pytest.raises(expected_exception=NotImplementedError) as cm: diff --git a/tests/incite/test_collection_base_item.py b/tests/incite/test_collection_base_item.py index 7a0a581..e09f54a 100644 --- a/tests/incite/test_collection_base_item.py +++ b/tests/incite/test_collection_base_item.py @@ -1,3 +1,5 @@ +from __future__ import annotations + from datetime import UTC, datetime from os.path import join as pjoin from pathlib import Path @@ -8,7 +10,7 @@ import pandas as pd import pytest from pydantic import ValidationError -from generalresearch.incite.base import CollectionItemBase +from generalresearch.incite.base import CollectionItemBase, GRLDatasets class TestCollectionItemBase: @@ -40,20 +42,20 @@ class TestCollectionItemBaseProperties: def test_finish(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.finish + with pytest.raises(expected_exception=AttributeError): + _ = instance.finish def test_interval(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.interval + with pytest.raises(expected_exception=AttributeError): + _ = instance.interval def test_filename(self): instance = CollectionItemBase() with pytest.raises(expected_exception=NotImplementedError) as cm: - res = instance.filename + _ = instance.filename assert "Do not use CollectionItemBase directly" in str(cm.value) @@ -61,7 +63,7 @@ class TestCollectionItemBaseProperties: instance = CollectionItemBase() with pytest.raises(expected_exception=NotImplementedError) as cm: - res = instance.filename + _ = instance.filename assert "Do not use CollectionItemBase directly" in str(cm.value) @@ -69,27 +71,27 @@ class TestCollectionItemBaseProperties: instance = CollectionItemBase() with pytest.raises(expected_exception=NotImplementedError) as cm: - res = instance.filename + _ = instance.filename assert "Do not use CollectionItemBase directly" in str(cm.value) def test_path(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.path + with pytest.raises(expected_exception=AttributeError): + _ = instance.path def test_partial_path(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.partial_path + with pytest.raises(expected_exception=AttributeError): + _ = instance.partial_path def test_empty_path(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.empty_path + with pytest.raises(expected_exception=AttributeError): + _ = instance.empty_path class TestCollectionItemBaseMethods: @@ -106,41 +108,41 @@ class TestCollectionItemBaseMethods: instance = CollectionItemBase() with pytest.raises(expected_exception=NotImplementedError) as cm: - res = instance.tmp_filename() + _ = instance.tmp_filename() assert "Do not use CollectionItemBase directly" in str(cm.value) def test_tmp_path(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.tmp_path() + with pytest.raises(expected_exception=AttributeError): + instance.tmp_path() def test_is_empty(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.is_empty() + with pytest.raises(expected_exception=AttributeError): + instance.is_empty() def test_has_empty(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.has_empty() + with pytest.raises(expected_exception=AttributeError): + instance.has_empty() def test_has_partial_archive(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.has_partial_archive() + with pytest.raises(expected_exception=AttributeError): + instance.has_partial_archive() @pytest.mark.parametrize("include_empty", [True, False]) - def test_has_archive(self, include_empty): + def test_has_archive(self, include_empty: bool): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.has_archive(include_empty=include_empty) + with pytest.raises(expected_exception=AttributeError): + instance.has_archive(include_empty=include_empty) - def test_delete_archive_file(self, mnt_filepath): + def test_delete_archive_file(self, mnt_filepath: GRLDatasets): path1 = Path(pjoin(mnt_filepath.data_src, f"{uuid4().hex}.zip")) # Confirm it doesn't exist, and that delete_archive() doesn't throw @@ -155,7 +157,7 @@ class TestCollectionItemBaseMethods: CollectionItemBase.delete_archive(generic_path=path1) assert not path1.exists() - def test_delete_archive_dir(self, mnt_filepath): + def test_delete_archive_dir(self, mnt_filepath: GRLDatasets): path1 = Path(pjoin(mnt_filepath.data_src, f"{uuid4().hex}")) # Confirm it doesn't exist, and that delete_archive() doesn't throw @@ -174,20 +176,20 @@ class TestCollectionItemBaseMethods: def test_should_archive(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.should_archive() + with pytest.raises(expected_exception=AttributeError): + _ = instance.should_archive() def test_set_empty(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.set_empty() + with pytest.raises(expected_exception=AttributeError): + _ = instance.set_empty() def test_valid_archive(self): instance = CollectionItemBase() - with pytest.raises(expected_exception=AttributeError) as cm: - res = instance.valid_archive(generic_path=None, sample=None) + with pytest.raises(expected_exception=AttributeError): + _ = instance.valid_archive(generic_path=None, sample=None) class TestCollectionItemBaseMethodsORM: @@ -197,11 +199,11 @@ class TestCollectionItemBaseMethodsORM: pass @pytest.mark.parametrize("is_partial", [True, False]) - def test_to_archive(self, is_partial): + def test_to_archive(self, is_partial: bool): instance = CollectionItemBase() with pytest.raises(expected_exception=NotImplementedError) as cm: - res = instance.to_archive( + _ = instance.to_archive( ddf=dd.from_pandas(data=pd.DataFrame()), is_partial=is_partial ) assert "Must override" in str(cm.value) diff --git a/tests/incite/test_interval_idx.py b/tests/incite/test_interval_idx.py index 3034c21..03d29ea 100644 --- a/tests/incite/test_interval_idx.py +++ b/tests/incite/test_interval_idx.py @@ -1,4 +1,4 @@ -from datetime import datetime +from datetime import UTC, datetime import pandas as pd @@ -6,8 +6,8 @@ import pandas as pd class TestIntervalIndex: def test_init(self): - start = datetime(year=2000, month=1, day=1) - end = datetime(year=2000, month=1, day=10) + start = datetime(year=2000, month=1, day=1, tzinfo=UTC) + end = datetime(year=2000, month=1, day=10, tzinfo=UTC) iv_r: pd.IntervalIndex = pd.interval_range( start=start, end=end, freq="1d", closed="left" |