Fix 2 file staging unit tests

nmdc_automation/jgi_file_staging/models.py

@@ -1,21 +1,21 @@
 from pydantic import BaseModel
 from datetime import datetime
-from typing import Optional
+from typing import Optional, List
 
 
 class Sample(BaseModel):
-    project: str
     apGoldId: str
     studyId: str
     itsApId: str
+    projects: str
     biosample_id: str
     seq_id: str
     file_name: str
     file_status: str
     file_size: int
     jdp_file_id: str
-    md5sum: Optional[str]
+    md5sum: Optional[str] = None
     analysis_project_id: str
     create_date: datetime = datetime.now()
-    update_date: Optional[datetime]
-    request_id: Optional[str]
+    update_date: Optional[datetime] = None
+    request_id: Optional[str] = None

tests/test_jgi_file_staging/conftest.py

@@ -35,6 +35,7 @@ def grow_analysis_df():
             "apGoldId",
             "studyId",
             "itsApId",
+            "projects",
             "biosample_id",
             "seq_id",
             "file_name",
@@ -45,5 +46,5 @@ def grow_analysis_df():
             "analysis_project_id",
         ]
     ]
-    grow_analysis_df["project"] = "test_project"
+    # grow_analysis_df["project"] = "test_project"
     return grow_analysis_df

tests/test_jgi_file_staging/fixtures/grow_analysis_projects.csv

@@ -1,11 +1,11 @@
 apGoldId,studyId,itsApId,projects,biosample_id,seq_id,file_name,file_status,file_size,jdp_file_id,md5sum,analysis_project_id
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,52614.1.394702.GCACTAAC-CCAAGACT.filtered-report.txt,RESTORED,3645,6190d7d30de2fc3298da6f7a,fcd87248b5922a8bd0d530bcb23bffae,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,52614.1.394702.GCACTAAC-CCAAGACT.filter_cmd-METAGENOME.sh,RESTORED,6151,6190d7d30de2fc3298da6f7c,892bfb0ad0f44ce133530e07d24ab37f,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,Ga0499978_imgap.info,RESTORED,411,61a9d6ee8277d7ede604d0f6,852f507c44a0743e08cc3cc0de9575d2,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,Ga0499978_proteins.supfam.domtblout,RESTORED,7424934505,61a9d6ef8277d7ede604d105,c09cc12998669c5d4ec3973ff4d27580,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,Ga0499978_ko.tsv,RESTORED,44433089,61a9d6ef8277d7ede604d0f8,0ad4c6ca9deab065699e6f431d939cdf,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,Ga0499978_proteins.faa,RESTORED,417512094,61a9d6ef8277d7ede604d101,bd36edd00d12188b1d45a9b1c942bbb4,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,pairedMapped_sorted.bam.cov,RESTORED,80122313,61a9d6f18277d7ede604d116,30c9b60aab947deba0ffaa6e21755964,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,Table_8_-_3300049478.taxonomic_composition.txt,RESTORED,16996,61a9d6ed8277d7ede604d0e4,ee3efca7c81b6dbf837be5e91c5bbc78,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,Ga0499978_annotation_config.yaml,RESTORED,3015,61a9d6ee8277d7ede604d0e5,a7950724aa003bea5f23f2edbd24c596,1323348
-Ga0499978,Gs0149396,1323348,['Gp0587070'],Gb0305643,1323445,rqc-stats.pdf,RESTORED,290034,619d6f9850d56abc0a99a4f4,100954063b2bd8a0bafb2488f6f07bdd,1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,52614.1.394702.GCACTAAC-CCAAGACT.filtered-report.txt,RESTORED,3645,6190d7d30de2fc3298da6f7a,fcd87248b5922a8bd0d530bcb23bffae,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,52614.1.394702.GCACTAAC-CCAAGACT.filter_cmd-METAGENOME.sh,RESTORED,6151,6190d7d30de2fc3298da6f7c,892bfb0ad0f44ce133530e07d24ab37f,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,Ga0499978_imgap.info,RESTORED,411,61a9d6ee8277d7ede604d0f6,852f507c44a0743e08cc3cc0de9575d2,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,Ga0499978_proteins.supfam.domtblout,RESTORED,7424934505,61a9d6ef8277d7ede604d105,c09cc12998669c5d4ec3973ff4d27580,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,Ga0499978_ko.tsv,RESTORED,44433089,61a9d6ef8277d7ede604d0f8,0ad4c6ca9deab065699e6f431d939cdf,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,Ga0499978_proteins.faa,RESTORED,417512094,61a9d6ef8277d7ede604d101,bd36edd00d12188b1d45a9b1c942bbb4,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,pairedMapped_sorted.bam.cov,RESTORED,80122313,61a9d6f18277d7ede604d116,30c9b60aab947deba0ffaa6e21755964,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,Table_8_-_3300049478.taxonomic_composition.txt,RESTORED,16996,61a9d6ed8277d7ede604d0e4,ee3efca7c81b6dbf837be5e91c5bbc78,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,Ga0499978_annotation_config.yaml,RESTORED,3015,61a9d6ee8277d7ede604d0e5,a7950724aa003bea5f23f2edbd24c596,p1323348
+Ga0499978,Gs0149396,a1323348,['Gp0587070'],Gb0305643,s1323445,rqc-stats.pdf,RESTORED,290034,619d6f9850d56abc0a99a4f4,100954063b2bd8a0bafb2488f6f07bdd,p1323348

tests/test_jgi_file_staging/test_file_metadata.py

@@ -1,13 +1,18 @@
+import mongomock
 import pandas as pd
 from pathlib import Path
 import pytest
 
+
 from nmdc_automation.jgi_file_staging.jgi_file_metadata import (
     get_access_token,
+    get_mongo_db,
     check_access_token,
     get_sequence_id,
     get_analysis_projects_from_proposal_id,
+    insert_samples_into_mongodb,
 )
+from nmdc_automation.jgi_file_staging.models import Sample
 
 
 FIXTURE_DIR = Path(__file__).parent / "fixtures"
@@ -84,34 +89,34 @@ def test_get_analysis_projects_from_proposal_id(mock_get):
         "projects": "['Gp0503551']",
     }
 
-# TODO: fix this test.  Data fixtures raise ValidationError from Sample model
-# def test_sample_model_instance_creation(monkeypatch, grow_analysis_df):
-#     sample_dict = grow_analysis_df.to_dict("records")[0]
-#     sample_model = Sample(**sample_dict)
-    # assert sample_model.apGoldId == "Ga0499978"
-    # assert sample_model.studyId == "Gs0149396"
-    # assert sample_model.itsApId == 1323348
-    # assert sample_model.projects == "['Gp0587070']"
-    # assert sample_model.biosample_id == "Ga0499978"
-    # assert sample_model.seq_id == "Ga0499978"
-    # assert sample_model.file_name == "Ga0499978.fna.gz"
-    # assert sample_model.file_status == "uploaded"
-    # assert sample_model.file_size == 1000
-    # assert sample_model.jdp_file_id == 123456
-    # assert sample_model.md5sum == "1234567890abcdef"
-    # assert sample_model.analysis_project_id == 123456
-    # assert sample_model.project == "test_project"
+
+def test_sample_model_instance_creation(monkeypatch, grow_analysis_df):
+    sample_dict = grow_analysis_df.to_dict("records")[0]
+    sample_model = Sample(**sample_dict)
+    assert sample_model.apGoldId == "Ga0499978"
+    assert sample_model.studyId == "Gs0149396"
+    assert sample_model.itsApId == "a1323348"
+    assert sample_model.projects == "['Gp0587070']"
+    assert sample_model.biosample_id == "Gb0305643"
+    assert sample_model.seq_id == "s1323445"
+    assert sample_model.file_name == "52614.1.394702.GCACTAAC-CCAAGACT.filtered-report.txt"
+    assert sample_model.file_status == "RESTORED"
+    assert sample_model.file_size == 3645
+    assert sample_model.jdp_file_id == "6190d7d30de2fc3298da6f7a"
+    assert sample_model.md5sum == "fcd87248b5922a8bd0d530bcb23bffae"
+    assert sample_model.analysis_project_id == "p1323348"
+
 
 
 # TODO: fix this test.  Data fixtures are raising ValidationError from
 # the pydantic Sample model
-# @mongomock.patch(servers=(("localhost", 27017),), on_new="create")
-# def test_insert_samples_into_mongodb(monkeypatch, grow_analysis_df):
-#     monkeypatch.setenv("MONGO_DBNAME", "test_db")
-#     client = get_mongo_db()
-#     mdb = client["test_db"]
-#
-#     insert_samples_into_mongodb(grow_analysis_df.to_dict("records"), mdb)
-#     mdb = get_mongo_db()
-#     sample = mdb.samples.find_one({"apGoldId": "Ga0499978"})
-#     assert sample["studyId"] == "Gs0149396"
+@mongomock.patch(servers=(("localhost", 27017),), on_new="create")
+def test_insert_samples_into_mongodb(monkeypatch, grow_analysis_df):
+    monkeypatch.setenv("MONGO_DBNAME", "test_db")
+    client = get_mongo_db()
+    mdb = client["test_db"]
+
+    insert_samples_into_mongodb(grow_analysis_df.to_dict("records"))
+    mdb = get_mongo_db()
+    sample = mdb.samples.find_one({"apGoldId": "Ga0499978"})
+    assert sample["studyId"] == "Gs0149396"