Switch to using i64/u64 internally for hash operations

Author: jb3

bot/exts/filtering/_filter_lists/image_hash.py

@@ -16,7 +16,6 @@
 log = get_logger(__name__)
 
 _MAX_IMAGE_SIZE = 5_000_000
-_U64_MASK = (1 << 64) - 1
 
 
 class ImageHashesList(FilterList[ImageHashFilter]):
@@ -65,7 +64,7 @@ async def actions_for(
                 log.exception("Timed out getting image hash")
                 continue
 
-            image_hashes.append(image_hash & _U64_MASK)
+            image_hashes.append(image_hash)
 
         if not image_hashes:
             return None, [], {}
@@ -78,15 +77,9 @@ async def actions_for(
         actions = self[ListType.DENY].merge_actions(triggers)
         messages = []
         for filter_ in triggers:
-            distance = self._closest_distance(filter_, image_hashes)
+            distance = ctx.filter_info.get(filter_, "?")
             messages.append(
                 f"{filter_.id} (`{filter_.content}` distance `{distance}`)"
                 f" - {filter_.description or '*No description*'}"
             )
         return actions, messages, {ListType.DENY: triggers}
-
-    @staticmethod
-    def _closest_distance(filter_: Filter, image_hashes: list[int]) -> int:
-        """Return the closest Hamming distance between a filter hash and any uploaded image hash."""
-        candidate_hash = int(filter_.content, 16)
-        return min(int.bit_count(image_hash ^ candidate_hash) for image_hash in image_hashes)

bot/exts/filtering/_filters/image_hash.py

@@ -4,7 +4,7 @@
 
 from bot.exts.filtering._filter_context import FilterContext
 from bot.exts.filtering._filters.filter import Filter
-from bot.exts.filtering._image_hash import HASH_DISTANCE_THRESHOLD
+from bot.exts.filtering._image_hash import HASH_DISTANCE_THRESHOLD, signed_i64_to_u64
 
 _HEX_RE = re.compile(r"^(?:0x)?([0-9a-fA-F]{1,16})$")
 
@@ -19,8 +19,11 @@ async def triggered_on(self, ctx: FilterContext) -> bool:
         candidate_hash = int(self.content, 16)
 
         for image_hash in ctx.content:
-            if int.bit_count(image_hash ^ candidate_hash) <= HASH_DISTANCE_THRESHOLD:
-                ctx.matches.append(f"{image_hash:016x}")
+            normalized_image_hash = signed_i64_to_u64(image_hash)
+            distance = int.bit_count(normalized_image_hash ^ candidate_hash)
+            if distance <= HASH_DISTANCE_THRESHOLD:
+                ctx.matches.append(f"{normalized_image_hash:016x}")
+                ctx.filter_info[self] = str(distance)
                 return True
         return False
 

bot/exts/filtering/_image_hash.py

@@ -25,16 +25,14 @@ async def get_image_hash(image_url: str) -> int:
             raise RhodiumAPIError(f"Rhodium API returned status code {response.status}: {contents}")
 
         response_data = await response.json()
-        hash_hex = response_data.get("hex")
-        if not hash_hex:
-            raise RhodiumAPIError("Rhodium API response did not include a hex hash.")
-
-        unsigned = int(str(hash_hex).removeprefix("0x"), 16)
-        if unsigned >= (1 << 63):
-            return unsigned - (1 << 64)
-        return unsigned
+        return response_data["i64"]
 
 
 def signed_i64_to_hex(value: int) -> str:
     """Convert a signed 64-bit integer to a normalized lowercase 16-char hexadecimal string."""
     return f"{value & ((1 << 64) - 1):016x}"
+
+
+def signed_i64_to_u64(value: int) -> int:
+    """Convert a signed 64-bit integer into its unsigned 64-bit representation."""
+    return value & ((1 << 64) - 1)