Set flowcell_id based on central regex group

Author: ssjunnebo

dataflow_transfer/run_classes/element_runs.py

@@ -21,7 +21,6 @@ class AVITIRun(ElementRun):
 
     def __init__(self, run_dir, configuration):
         super().__init__(run_dir, configuration)
-        self.flowcell_id = self.run_id.split("_")[-1][1:]  # 2507535225
 
 
 # TODO: Add Teton run class

dataflow_transfer/run_classes/generic_runs.py

@@ -38,6 +38,7 @@ def __init__(self, run_dir, configuration):
         self.remote_destination = self.sequencer_config.get("remote_destination")
         self.db = StatusdbSession(self.configuration.get("statusdb"))
         self.run_id_format = self._resolve_run_id_format()
+        self.flowcell_id = re.match(self.run_id_format, self.run_id).group("flowcell_id") if self.run_id_format else None
 
     def _resolve_run_id_format(self):
         """Resolve the run ID regex from central config."""
@@ -181,10 +182,15 @@ def has_status(self, status_name):
 
     def update_statusdb(self, status, additional_info=None):
         """Update the statusdb document for this run with the given status."""
-        db_doc = (
-            self.db.get_db_doc(ddoc="lookup", view="runfolder_id", run_id=self.run_id)
-            or {}
+        doc_id = self.db.get_doc_id(
+            ddoc="lookup", view="runfolder_id", run_id=self.run_id
         )
+        if doc_id:
+            db_doc = self.db.get_document(
+                db=self.db.db_name, doc_id=doc_id
+            )
+        else:
+            db_doc = {}
 
         statuses_to_only_update_once = [
             "sequencing_started",

dataflow_transfer/run_classes/illumina_runs.py

@@ -11,7 +11,6 @@ class IlluminaRun(Run):
     def __init__(self, run_dir, configuration):
         super().__init__(run_dir, configuration)
         self.final_file = "CopyComplete.txt"
-        self.flowcell_id = self.run_id.split("_")[-1]
 
 
 @register_run_class
@@ -22,7 +21,6 @@ class NovaSeqXPlusRun(IlluminaRun):
 
     def __init__(self, run_dir, configuration):
         super().__init__(run_dir, configuration)
-        self.flowcell_id = self.run_id.split("_")[-1][1:]  # 22CVHTLT1
 
 
 @register_run_class
@@ -53,4 +51,3 @@ class MiSeqi100Run(IlluminaRun):
 
     def __init__(self, run_dir, configuration):
         super().__init__(run_dir, configuration)
-        self.flowcell_id = self.run_id.split("_")[-1][1:]  # SC2150561-SC3

dataflow_transfer/run_classes/ont_runs.py

@@ -11,7 +11,6 @@ class ONTRun(Run):
     def __init__(self, run_dir, configuration):
         super().__init__(run_dir, configuration)
         self.final_file = "final_summary.txt"
-        self.flowcell_id = self.run_id.split("_")[-2]
 
 
 @register_run_class

dataflow_transfer/utils/statusdb.py

@@ -55,19 +55,6 @@ def _retry_call(self, func):
         # re-raise last exception for caller to handle
         raise last_exception
 
-    def get_db_doc(self, ddoc, view, run_id):
-        """Retrieve a document from the database via retried call."""
-        doc_id = self.get_doc_id(
-            ddoc, view, run_id
-        )  # TODO: refactor to use get_document
-        if doc_id:
-            return self._retry_call(
-                lambda: self.connection.get_document(
-                    db=self.db_name, doc_id=doc_id
-                ).get_result()
-            )
-        return None
-
     def get_document(self, db, doc_id):
         """Retrieve a document from any database via retried call."""
         return self._retry_call(