Refactor of caching & data storage part 7 #118

core/data_store.py

@@ -1,4 +1,5 @@
 import logging
+import re
 from pathlib import Path
 
 import diskcache
@@ -23,6 +24,7 @@ class DataStore:
         self.spots = None
         self.callsigns = None
         self.dxcc_data = None
+        self.dxcc_lookup_by_call_regex = []
         self.sigrefs = None
         self.status_data = None
         self._status = None
@@ -46,6 +48,7 @@ class DataStore:
         # Standard disk cache for static reference and SIG ref data. Separate provider threads will repopulate these on
         # a regular basis but there's no need for a TTL since old data is better than no data.
         self.dxcc_data = diskcache.Cache(CACHE_DIR + "dxcc_data")
+        self.regenerateCallRegexToDXCCEntityMap()
 
         # For SIG reference data specifically, we need to key on both SIG *and* reference, and trying to do two layers
         # of dict in diskcache absolutely destroys performance with unpickling huge dicts, so we have an ugly "SIG:ref"
@@ -72,6 +75,15 @@ class DataStore:
                                     snapshot_interval_sec=self._SPOT_ALERT_SNAPSHOT_INTERVAL_SEC)
         logging.info(f"Loaded %d alerts from a previous run.", len(self.alerts.keys()))
 
+    def regenerateCallRegexToDXCCEntityMap(self):
+        """DXCC entity data from K0SWE includes a regex which we can use to match a callsign, and determine which DXCC
+        entity it belongs to. But getting every DXCC entity data object out of DiskCache, iterating, compiling its regex
+        and testing the callsign every time is expensive. So instead we build a separate in-memory lookup of compiled
+        regex against DXCC entity code, as a list of tuples we can iterate through."""
+
+        for entry in [DATA_STORE.dxcc_data[key] for key in DATA_STORE.dxcc_data]:
+            self.dxcc_lookup_by_call_regex.append((re.compile(entry["prefixRegex"]), entry["entityCode"]))
+
     def close(self):
         self.spots.close()
         self.alerts.close()