refactor(utils): replace URI_replace_settings.yml with clean_uri.toml

... and move the `clean_uri.toml` to the `utils` folder

Closes: #572

apis_core/utils/clean_uri.toml

@@ -0,0 +1,7 @@
+[geonames_org]
+regex = '^https?://(?:[^.]*[.])?geonames[.]org/([0-9]+)'
+replace = 'https://sws.geonames.org/{}/'
+
+[d-nb_info]
+regex = '^https?://(?:[^.]*[.])?d-nb.info/gnd/([0-9A-Za-u\-]+)'
+replace = 'https://d-nb.info/gnd/{}'

apis_core/utils/normalize.py

@@ -2,19 +2,16 @@
 # SPDX-License-Identifier: MIT
 
 import re
+import tomllib
 from apis_core.utils.settings import clean_uri_mapping_file
-from yaml import safe_load
 
 
 def clean_uri(uri: str) -> str:
-    if uri:
-        settings = safe_load(clean_uri_mapping_file().read_text())
-        for mapping in settings.get("mappings", []):
-            domain = mapping["domain"]
-            regex = mapping["regex"]
-            replace = mapping["replace"]
-            if domain in uri:
-                m = re.match(regex, uri)
-                if m:
-                    uri = replace.format(m.group(1))
+    settings = tomllib.loads(clean_uri_mapping_file().read_text())
+    if uri is not None:
+        for entry in settings.values():
+            regex = entry["regex"]
+            replace = entry["replace"]
+            if m := re.match(regex, uri):
+                uri = replace.format(m.group(1))
     return uri

apis_core/utils/settings.py

@@ -1,6 +1,7 @@
 # SPDX-FileCopyrightText: 2023 Birger Schacht
 # SPDX-License-Identifier: MIT
 
+import os
 from pathlib import Path
 from django.conf import settings
 
@@ -11,7 +12,7 @@ def default_settings() -> Path:
 
 
 def clean_uri_mapping_file() -> Path:
-    default = default_settings() / "URI_replace_settings.yml"
+    default = Path(__file__).parent / "clean_uri.toml"
     mapping_file = getattr(settings, "CLEANURI_MAPPINGS", default)
     return Path(mapping_file)