Merge pull request #243 from Clinical-Genomics/crud-separation

Continuation crud separation

Author: karlnyr

microSALT/cli.py

@@ -315,14 +315,6 @@ def finish(
         if os.path.isdir(f"{input}/{subfolder}"):
             pool.append(subfolder)
 
-    run_settings = {
-        "input": input,
-        "track": track,
-        "dry": dry,
-        "email": config.regex.mail_recipient,
-        "skip_update": skip_update,
-    }
-
     sampleinfo = review_sampleinfo(sampleinfo_file)
     ext_refs = Referencer(
         log=logger,

microSALT/store/db_manipulator.py

@@ -279,6 +279,18 @@ def delete_sample(self, cg_id: str) -> None:
         self.session.commit()
         self.logger.info(f"Removed sample {cg_id} and its child rows")
 
+    def delete_sample_results(self, cg_id: str) -> None:
+        """Delete only the analysis result rows for a sample (seq_types, resistances, expacs)
+        without removing the Samples row itself."""
+        for obj in self.session.query(Expacs).filter(Expacs.CG_ID_sample == cg_id).all():
+            self.session.delete(obj)
+        for obj in self.session.query(Seq_types).filter(Seq_types.CG_ID_sample == cg_id).all():
+            self.session.delete(obj)
+        for obj in self.session.query(Resistances).filter(Resistances.CG_ID_sample == cg_id).all():
+            self.session.delete(obj)
+        self.session.commit()
+        self.logger.info(f"Cleared analysis results for sample {cg_id}")
+
     def delete_project(self, name: str) -> None:
         """Delete all samples (and their child rows) belonging to a project."""
         for obj in self.session.query(Expacs).filter(Expacs.CG_ID_sample.like(f"{name}%")).all():
@@ -413,16 +425,29 @@ def read_columns(self, tablename: str):
         table = _resolve_orm_table(tablename)
         return dict.fromkeys(table.__table__.columns.keys())
 
-    def read_exists(self, table: str, item: dict[str, str]):
-        """Takes a k-v pair and checks for the entrys existence in the given table"""
-        orm_table = _resolve_orm_table(table)
-        filter_clauses = [getattr(orm_table, k) == v for k, v in item.items()]
-        entry = self.session.query(orm_table).filter(and_(*filter_clauses)).scalar()
-        return entry is not None
+    def get_projects_by_cg_id_project(self, cg_id_project_name: str) -> Projects | None:
+        """Fetch a Projects record by CG_ID_project."""
+        return (
+            self.session.query(Projects)
+            .filter(Projects.CG_ID_project == cg_id_project_name)
+            .scalar()
+        )
+
+    def get_collection_by_id(self, collection_id: str) -> Collections | None:
+        return (
+            self.session.query(Collections)
+            .filter(Collections.ID_collection == collection_id)
+            .scalar()
+        )
+
+    def get_sample_by_cg_id_sample(self, cg_id_sample: str) -> Samples | None:
+        return self.session.query(Samples).filter(Samples.CG_ID_sample == cg_id_sample).scalar()
 
     def read_version(self, name: str):
         """Gets the version from a given name. Should be generalized to return any value for any input"""
-        version = self.session.query(Versions).filter(Versions.name == name).scalar()
+        version: Versions | None = (
+            self.session.query(Versions).filter(Versions.name == name).scalar()
+        )
         if version is None:
             return "0"
         else:

microSALT/utils/job_creator.py

@@ -526,12 +526,10 @@ def create_snpsection(self):
 
     def create_collection(self):
         """Creates collection entry in database"""
-        if self.db_pusher.read_exists("Collections", {"ID_collection": self.name}):
+        if self.db_pusher.get_collection_by_id(self.name):
             self.db_pusher.delete_collection(self.name)
             for sample in self.pool:
-                self.db_pusher.add_rec(
-                    {"ID_collection": self.name, "CG_ID_sample": sample}, "Collections"
-                )
+                self.db_pusher.add_collection(ID_collectiuon=self.name, CG_ID_sample=sample)
 
         addedprojs = []
         for sample in self.pool:
@@ -541,8 +539,8 @@ def create_collection(self):
                 self.create_project(lims_project)
                 addedprojs.append(lims_project)
 
-    def create_project(self, name: str):
-        """Creates project in database"""
+    def create_project(self, name: str) -> None:
+        """Creates or updates a project in the database."""
         if not self.sample:
             raise JobCreationError(
                 "No sample information provided. Cannot create project in database."
@@ -552,12 +550,15 @@ def create_project(self, name: str):
             "Customer_ID_project": self.sample["Customer_ID_project"],
             "Customer_ID": self.sample["Customer_ID"],
         }
-        self.db_pusher.add_to_session(self.db_pusher.add_project(**project_data))
-        self.db_pusher.commit_session()
-
-    def create_sample(self):
-        """Creates sample in database"""
+        if self.db_pusher.get_projects_by_cg_id_project(name):
+            update_data = {k: v for k, v in project_data.items() if k != "CG_ID_project"}
+            self.db_pusher.update_project({"CG_ID_project": name}, update_data)
+        else:
+            self.db_pusher.add_to_session(self.db_pusher.add_project(**project_data))
+            self.db_pusher.commit_session()
 
+    def create_sample(self) -> None:
+        """Creates or updates a sample in the database."""
         try:
             if not self.sample:
                 raise JobCreationError(
@@ -581,8 +582,13 @@ def create_sample(self):
                 "method_libprep": self.sample["method_libprep"],
                 "method_sequencing": self.sample["method_sequencing"],
             }
-            self.db_pusher.add_to_session(self.db_pusher.add_sample(**sample_data))
-            self.db_pusher.commit_session()
+            cg_id = self.sample["CG_ID_sample"]
+            if self.db_pusher.get_sample_by_cg_id_sample(cg_id):
+                update_data = {k: v for k, v in sample_data.items() if k != "CG_ID_sample"}
+                self.db_pusher.update_sample({"CG_ID_sample": cg_id}, update_data)
+            else:
+                self.db_pusher.add_to_session(self.db_pusher.add_sample(**sample_data))
+                self.db_pusher.commit_session()
         except JobCreationError as e:
             self.logger.error(f"Unable to add sample {self.name} to database: {e}")
         except KeyError as e:

microSALT/utils/scraper.py

@@ -108,9 +108,7 @@ def scrape_project(self, project=None):
         if project is None:
             project = self.name
         self.db_pusher.delete_project(project)
-        if not self.db_pusher.read_exists("Projects", {"CG_ID_project": project}):
-            self.logger.warning(f"Replacing project {project}")
-            self.job_fallback.create_project(project)
+        self.job_fallback.create_project(project)
 
         # Scrape order matters a lot!
         for item in os.listdir(self.infolder):
@@ -140,17 +138,9 @@ def scrape_sample(self, sample=None):
         """Scrapes a sample folder for information"""
         if sample is None:
             sample = self.name
-        self.db_pusher.delete_sample(sample)
-
-        if not self.db_pusher.read_exists(
-            "Projects", {"CG_ID_project": self.sample.get("CG_ID_project")}
-        ):
-            self.logger.warning(f"Replacing project {self.sample.get('CG_ID_project')}")
-            self.job_fallback.create_project(self.sample.get("CG_ID_project"))
-
-        if not self.db_pusher.read_exists("Samples", {"CG_ID_sample": sample}):
-            self.logger.info(f"Replacing sample {sample}")
-            self.job_fallback.create_sample(sample)
+        self.db_pusher.delete_sample_results(sample)
+        self.job_fallback.create_project(self.sample.get("CG_ID_project"))
+        self.job_fallback.create_sample()
 
         # Scrape order matters a lot!
         self.sampledir = self.infolder

tests/store/test_database.py

@@ -267,17 +267,6 @@ def test_purge_rec(sysexit, caplog, dbm):
     dbm.delete_collection("UPD1234A1")
 
 
-def test_top_index(dbm):
-    dbm.add_to_session(dbm.add_sample(CG_ID_sample="Uniq_ID_123", total_reads=100))
-    dbm.add_to_session(dbm.add_sample(CG_ID_sample="Uniq_ID_321", total_reads=100))
-    dbm.commit_session()
-    ti_returned = dbm.read_top_index("Samples", {"total_reads": "100"}, "total_reads")
-    assert ti_returned == 100
-
-    ti_missing = dbm.read_top_index("Samples", {"total_reads": "99999"}, "total_reads")
-    assert ti_missing == -1
-
-
 def test_query_rec(dbm):
     dbm.add_to_session(dbm.add_sample(CG_ID_sample="QRY_001"))
     dbm.add_to_session(dbm.add_sample(CG_ID_sample="QRY_002"))
@@ -301,12 +290,20 @@ def test_get_columns(dbm):
     assert "organism" in cols
 
 
-def test_exists(dbm):
+def test_get_sample_by_cg_id(dbm: DB_Manipulator):
     dbm.add_to_session(dbm.add_sample(CG_ID_sample="EXS_001"))
     dbm.commit_session()
 
-    assert dbm.read_exists("Samples", {"CG_ID_sample": "EXS_001"}) is True
-    assert dbm.read_exists("Samples", {"CG_ID_sample": "DOES_NOT_EXIST"}) is False
+    assert dbm.get_sample_by_cg_id_sample("EXS_001")
+    assert dbm.get_sample_by_cg_id_sample("DOES_NOT_EXIST") is None
+
+
+def test_get_collection_by_id(dbm: DB_Manipulator):
+    dbm.add_to_session(dbm.add_collection(CG_ID_sample="EXS_001", ID_collection="COLL_001"))
+    dbm.commit_session()
+
+    assert dbm.get_collection_by_id("COLL_001")
+    assert dbm.get_collection_by_id("DOES_NOT_EXIST") is None
 
 
 def test_resolve_orm_table_unknown():