cache, remove old references, drop py 3.8

.github/workflows/build.yml

@@ -8,7 +8,6 @@ jobs:
     strategy:
       matrix:
         python:
-          - "3.8"
           - "3.9"
           - "3.10"
           - "3.11"

fingerprints/__init__.py

@@ -1,5 +1,6 @@
 from fingerprints.fingerprint import fingerprint
 from fingerprints.cleanup import clean_entity_prefix
+from fingerprints.cleanup import clean_brackets
 from fingerprints.cleanup import clean_name_light
 from fingerprints.cleanup import clean_name_ascii
 from fingerprints.types import remove_types, replace_types
@@ -10,6 +11,7 @@
     "fingerprint",
     "generate",
     "clean_entity_prefix",
+    "clean_brackets",
     "clean_name_light",
     "clean_name_ascii",
     "remove_types",

fingerprints/cleanup.py

@@ -1,6 +1,7 @@
 import re
 import logging
 from typing import Optional
+from functools import lru_cache
 from normality import collapse_spaces, ascii_text, category_replace
 
 from fingerprints.constants import WS, BRACKETED
@@ -58,15 +59,22 @@ def clean_brackets(text: str) -> str:
     return BRACKETED.sub(WS, text)
 
 
+@lru_cache(maxsize=2000)
 def clean_name_ascii(text: Optional[str]) -> Optional[str]:
     """Super-hardcore string scrubbing."""
     # transliterate to ascii
     text = ascii_text(text)
     if text is None:
         return None
-    return clean_name_light(text)
+    # replace punctuation and symbols
+    text = CHARACTERS_REMOVE_RE.sub("", text)
+    text = text.lower()
+    cleaned = category_replace(text)
+    cleaned = collapse_spaces(cleaned)
+    return cleaned
 
 
+@lru_cache(maxsize=2000)
 def clean_name_light(text: str) -> Optional[str]:
     """Clean up a name for comparison, but don't convert to ASCII/Latin."""
     # replace punctuation and symbols

fingerprints/types/check.py

@@ -1,7 +1,5 @@
-import os
 import yaml
 from typing import Set
-from normality import slugify
 
 from fingerprints.types.common import TYPES_PATH, TypesList
 

fingerprints/types/replacer.py

@@ -5,7 +5,7 @@
 
 from fingerprints.types.data import TYPES
 from fingerprints.constants import WS
-from fingerprints.cleanup import clean_strict
+from fingerprints.cleanup import clean_name_ascii
 
 log = logging.getLogger(__name__)
 NormFunc = Callable[[Optional[str]], Optional[str]]
@@ -61,7 +61,9 @@ def normalize_replacements(norm_func: NormFunc) -> Dict[str, str]:
 
 
 @lru_cache(maxsize=None)
-def get_replacer(clean: NormFunc = clean_strict, remove: bool = False) -> ReplaceFunc:
+def get_replacer(
+    clean: NormFunc = clean_name_ascii, remove: bool = False
+) -> ReplaceFunc:
     replacements = normalize_replacements(clean)
     return Replacer(replacements, remove=remove)