Fix extra duplicates showing for spots and alerts

data/spot.py

@@ -279,7 +279,8 @@ class Spot:
                     # If so, add that to the sig_refs list for this spot.
                     ref_regex = get_ref_regex_for_sig(found_sig)
                     if ref_regex:
-                        ref_matches = re.finditer(r"(^|\W)" + found_sig + r"([ -])(" + ref_regex + r")($|\W)", self.comment,
+                        ref_matches = re.finditer(r"(^|\W)" + found_sig + r"([ -])(" + ref_regex + r")($|\W)",
+                                                  self.comment,
                                                   re.IGNORECASE)
                         for ref_match in ref_matches:
                             self._append_sig_ref_if_missing(SIGRef(id=ref_match.group(3).upper(), sig=found_sig))
@@ -361,18 +362,15 @@ class Spot:
             if self.comment and not self.qrt:
                 self.qrt = "QRT" in self.comment.upper()
 
-            # Always create an ID based on a hash of every parameter *except* received_time. This is used as the index
-            # to a map, which as a byproduct avoids us having multiple duplicate copies of the object that are identical
-            # apart from that they were retrieved from the API at different times. Note that the simple Python hash()
-            # function includes a seed randomly generated at runtime; this is therefore not consistent between runs. But we
-            # use diskcache to store our data between runs, so we use SHA256 which does not include this random element.
-            # The ID is computed before the online lookups below so that it is stable regardless of whether credentials
-            # are provided, allowing the enriched API response to be matched to the stored spot by ID.
+            # Create an ID based on the source and source ID if possible, as these guaranee uniqueness. If there is no
+            # source ID, use a combination of callsign and spot time. Spot time is down to the second or even the
+            # millisecond, so we can be reasonably sure two spots that match are the same spot.
             if not self.id:
-                self_copy = copy.deepcopy(self)
-                self_copy.received_time = 0
-                self_copy.received_time_iso = ""
-                self.id = hashlib.sha256(str(self_copy).encode("utf-8")).hexdigest()
+                if self.source and self.source_id:
+                    self.id = hashlib.sha256(str({"s": self.source, "sid": self.source_id}).encode("utf-8")).hexdigest()
+                else:
+                    self.id = hashlib.sha256(
+                        str({"s": self.source, "c": self.dx_call, "t": self.time}).encode("utf-8")).hexdigest()
 
             # DX operator details lookup. This should be the last resort compared to taking the data from the actual
             # spotting service, e.g. we don't want to accidentally use a user's QRZ.com home lat/lon or DXCC lat/lon