From b88d63ac06f157a97cc88bee0ea702949a5a0c64 Mon Sep 17 00:00:00 2001
From: Alexander Kabui
Date: Tue, 30 Mar 2021 23:58:02 +0300
Subject: refactor retrieve trait sample data and tests

---
 gn3/api/traits.py                        |  2 +-
 gn3/computations/datasets.py             | 62 +++++++++++++++++++++++++++-----
 tests/unit/computations/test_datasets.py |  8 ++++-
 3 files changed, 61 insertions(+), 11 deletions(-)

diff --git a/gn3/api/traits.py b/gn3/api/traits.py
index b3ddcc0..a9ff906 100644
--- a/gn3/api/traits.py
+++ b/gn3/api/traits.py
@@ -20,7 +20,7 @@ def home():
 def create_trait(trait_name, dataset_name):
     """endpoints for creating trait first should\
     call the endpoint for creating the trait only\
-    trait  data is the  sample data"""
+    also acts as endpoints for fetching trait data"""
 
     trait_dataset = mock.Mock()  # xtodo should replace this with calling epoints
     trait_dataset.name = dataset_name
diff --git a/gn3/computations/datasets.py b/gn3/computations/datasets.py
index 92a7bdf..6df5777 100644
--- a/gn3/computations/datasets.py
+++ b/gn3/computations/datasets.py
@@ -5,24 +5,30 @@ from typing import Optional
 from typing import List
 
 
-def retrieve_trait_sample_data(dataset_id, dataset_type: str, trait_name: str) -> List:
+def retrieve_trait_sample_data(dataset,
+                               trait_name: str,
+                               group_species_id=None,) -> List:
     """given the dataset id and trait_name fetch the\
     sample_name,value from the dataset"""
 
     # should pass the db as arg all  do a setup
 
-    _func_args = (dataset_id, dataset_type, trait_name)
+    (dataset_name, dataset_id, dataset_type) = (dataset.get("name"), dataset.get(
+        "id"), dataset.get("type"))
+
     dataset_query = get_query_for_dataset_sample(dataset_type)
+    results = []
+    sample_query_values = {
+        "Publish": (trait_name, dataset_id),
+        "Geno": (group_species_id, trait_name, dataset_name),
+        "ProbeSet": (trait_name, dataset_name)
+    }
 
     if dataset_query:
-        if dataset_type == "Publish":
-            formatted_query = dataset_query % (trait_name, dataset_id)
-            results = fetch_from_db_sample_data(formatted_query, mock.Mock())
-            return results
-
-        return []
+        formatted_query = dataset_query % sample_query_values[dataset_type]
+        results = fetch_from_db_sample_data(formatted_query, mock.Mock())
 
-    return []
+    return results
 
 
 def fetch_from_db_sample_data(formatted_query: str, database_instance) -> List:
@@ -60,7 +66,45 @@ def get_query_for_dataset_sample(dataset_type) -> Optional[str]:
                 Order BY
                         Strain.Name
                 """
+    geno_query = """
+                SELECT
+                        Strain.Name, GenoData.value, GenoSE.error, "N/A", Strain.Name2
+                FROM
+                        (GenoData, GenoFreeze, Strain, Geno, GenoXRef)
+                left join GenoSE on
+                        (GenoSE.DataId = GenoData.Id AND GenoSE.StrainId = GenoData.StrainId)
+                WHERE
+                        Geno.SpeciesId = %s AND Geno.Name = %s AND GenoXRef.GenoId = Geno.Id AND
+                        GenoXRef.GenoFreezeId = GenoFreeze.Id AND
+                        GenoFreeze.Name = %s AND
+                        GenoXRef.DataId = GenoData.Id AND
+                        GenoData.StrainId = Strain.Id
+                Order BY
+                        Strain.Name
+                """
+
+    probeset_query = """
+                SELECT
+                        Strain.Name, ProbeSetData.value, ProbeSetSE.error, NStrain.count, Strain.Name2
+                FROM
+                        (ProbeSetData, ProbeSetFreeze, Strain, ProbeSet, ProbeSetXRef)
+                left join ProbeSetSE on
+                        (ProbeSetSE.DataId = ProbeSetData.Id AND ProbeSetSE.StrainId = ProbeSetData.StrainId)
+                left join NStrain on
+                        (NStrain.DataId = ProbeSetData.Id AND
+                        NStrain.StrainId = ProbeSetData.StrainId)
+                WHERE
+                        ProbeSet.Name = '%s' AND ProbeSetXRef.ProbeSetId = ProbeSet.Id AND
+                        ProbeSetXRef.ProbeSetFreezeId = ProbeSetFreeze.Id AND
+                        ProbeSetFreeze.Name = '%s' AND
+                        ProbeSetXRef.DataId = ProbeSetData.Id AND
+                        ProbeSetData.StrainId = Strain.Id
+                Order BY
+                        Strain.Name
+                """
 
     dataset_query["Publish"] = pheno_query
+    dataset_query["Geno"] = geno_query
+    dataset_query["ProbeSet"] = probeset_query
 
     return dataset_query.get(dataset_type)
diff --git a/tests/unit/computations/test_datasets.py b/tests/unit/computations/test_datasets.py
index 2cd58b2..408f13b 100644
--- a/tests/unit/computations/test_datasets.py
+++ b/tests/unit/computations/test_datasets.py
@@ -18,12 +18,18 @@ class TestDatasets(TestCase):
         dataset_id = "HC_M2_0606_P&"
         dataset_type = "Publish"
 
+        dataset = {
+            "id": dataset_id,
+            "type": dataset_type,
+            "name": dataset_id
+        }
+
         fetch_results = [('BXD32', 8.001, None, None, 'BXD32')]
 
         mock_fetch_sample_results.return_value = fetch_results
 
         results = retrieve_trait_sample_data(
-            dataset_id, dataset_type, trait_name)
+            dataset, trait_name)
         self.assertEqual(mock_fetch_sample_results.call_count, 1)
         self.assertEqual(results, fetch_results)
 
-- 
cgit v1.2.3