data: osm_keys_tags: use SQLite instead of JSON

searx/data/__init__.py

@@ -11,12 +11,12 @@ __all__ = [
     'EXTERNAL_URLS',
     'WIKIDATA_UNITS',
     'EXTERNAL_BANGS',
-    'OSM_KEYS_TAGS',
     'LOCALES',
     'ahmia_blacklist_loader',
     'fetch_engine_descriptions',
     'fetch_iso4217_from_user',
     'fetch_name_from_iso4217',
+    'fetch_osm_key_label',
 ]
 
 import re
@@ -109,6 +109,45 @@ def fetch_name_from_iso4217(iso4217: str, language: str) -> Optional[str]:
         return None
 
 
+@lru_cache(100)
+def fetch_osm_key_label(key_name: str, language: str) -> Optional[str]:
+    if key_name.startswith('currency:'):
+        # currency:EUR --> get the name from the CURRENCIES variable
+        # see https://wiki.openstreetmap.org/wiki/Key%3Acurrency
+        # and for example https://taginfo.openstreetmap.org/keys/currency:EUR#values
+        # but there is also currency=EUR (currently not handled)
+        # https://taginfo.openstreetmap.org/keys/currency#values
+        currency = key_name.split(':')
+        if len(currency) > 1:
+            label = fetch_name_from_iso4217(currency[1], language)
+            if label:
+                return label
+            return currency[1]
+
+    language = language.lower()
+    language_short = language.split('-')[0]
+    with sql_connection("osm_keys_tags.db") as conn:
+        res = conn.execute(
+            "SELECT language, label FROM osm_keys WHERE name=? AND language in (?, ?, 'en')",
+            (key_name, language, language_short),
+        )
+        result = {result[0]: result[1] for result in res.fetchall()}
+        return result.get(language) or result.get(language_short) or result.get('en')
+
+
+@lru_cache(100)
+def fetch_osm_tag_label(tag_key: str, tag_value: str, language: str) -> Optional[str]:
+    language = language.lower()
+    language_short = language.split('-')[0]
+    with sql_connection("osm_keys_tags.db") as conn:
+        res = conn.execute(
+            "SELECT language, label FROM osm_tags WHERE tag_key=? AND tag_value=? AND language in (?, ?, 'en')",
+            (tag_key, tag_value, language, language_short),
+        )
+        result = {result[0]: result[1] for result in res.fetchall()}
+        return result.get(language) or result.get(language_short) or result.get('en')
+
+
 def ahmia_blacklist_loader():
     """Load data from `ahmia_blacklist.txt` and return a list of MD5 values of onion
     names.  The MD5 values are fetched by::
@@ -126,6 +165,5 @@ USER_AGENTS = _load('useragents.json')
 EXTERNAL_URLS = _load('external_urls.json')
 WIKIDATA_UNITS = _load('wikidata_units.json')
 EXTERNAL_BANGS = _load('external_bangs.json')
-OSM_KEYS_TAGS = _load('osm_keys_tags.json')
 ENGINE_TRAITS = _load('engine_traits.json')
 LOCALES = _load('locales.json')

searx/engines/apple_maps.py

@@ -6,7 +6,7 @@ from time import time
 from urllib.parse import urlencode
 
 from searx.network import get as http_get
-from searx.engines.openstreetmap import get_key_label
+from searx.data import fetch_osm_key_label
 
 about = {
     "website": 'https://www.apple.com/maps/',
@@ -72,7 +72,7 @@ def response(resp):
             telephone = result['telephone']
             links.append(
                 {
-                    'label': get_key_label('phone', user_language),
+                    'label': fetch_osm_key_label('phone', user_language),
                     'url': 'tel:' + telephone,
                     'url_label': telephone,
                 }
@@ -81,7 +81,7 @@ def response(resp):
             url = result['urls'][0]
             links.append(
                 {
-                    'label': get_key_label('website', user_language),
+                    'label': fetch_osm_key_label('website', user_language),
                     'url': url,
                     'url_label': url,
                 }

searx/engines/openstreetmap.py

@@ -10,7 +10,7 @@ from functools import partial
 
 from flask_babel import gettext
 
-from searx.data import OSM_KEYS_TAGS, fetch_name_from_iso4217
+from searx.data import fetch_osm_tag_label, fetch_osm_key_label
 from searx.utils import searx_useragent
 from searx.external_urls import get_external_url
 from searx.engines.wikidata import send_wikidata_query, sparql_string_escape, get_thumbnail
@@ -187,14 +187,14 @@ def response(resp):
                 'template': 'map.html',
                 'title': title,
                 'address': address,
-                'address_label': get_key_label('addr', user_language),
+                'address_label': fetch_osm_key_label('addr', user_language),
                 'url': url,
                 'osm': osm,
                 'geojson': geojson,
                 'thumbnail': thumbnail,
                 'links': links,
                 'data': data,
-                'type': get_tag_label(result.get('category'), result.get('type', ''), user_language),
+                'type': fetch_osm_tag_label(result.get('category'), result.get('type', ''), user_language),
                 'type_icon': result.get('icon'),
                 'content': '',
                 'longitude': result['lon'],
@@ -367,7 +367,7 @@ def get_links(result, user_language):
             url_label = result.get('wikidata', {}).get('itemLabel') or url_label
         links.append(
             {
-                'label': get_key_label(k, user_language),
+                'label': fetch_osm_key_label(k, user_language),
                 'url': url,
                 'url_label': url_label,
             }
@@ -389,7 +389,7 @@ def get_data(result, user_language, ignore_keys):
             continue
         if get_key_rank(k) is None:
             continue
-        k_label = get_key_label(k, user_language)
+        k_label = fetch_osm_key_label(k, user_language)
         if k_label:
             data.append(
                 {
@@ -412,51 +412,3 @@ def get_key_rank(k):
         # "payment:*" in KEY_ORDER matches "payment:cash", "payment:debit card", etc...
         key_rank = KEY_RANKS.get(k.split(':')[0] + ':*')
     return key_rank
-
-
-def get_label(labels, lang):
-    """Get label from labels in OSM_KEYS_TAGS
-
-    in OSM_KEYS_TAGS, labels have key == '*'
-    """
-    tag_label = labels.get(lang.lower())
-    if tag_label is None:
-        # example: if 'zh-hk' is not found, check 'zh'
-        tag_label = labels.get(lang.split('-')[0])
-    if tag_label is None and lang != 'en':
-        # example: if 'zh' is not found, check 'en'
-        tag_label = labels.get('en')
-    if tag_label is None and len(labels.values()) > 0:
-        # example: if still not found, use the first entry
-        tag_label = labels.values()[0]
-    return tag_label
-
-
-def get_tag_label(tag_category, tag_name, lang):
-    """Get tag label from OSM_KEYS_TAGS"""
-    tag_name = '' if tag_name is None else tag_name
-    tag_labels = OSM_KEYS_TAGS['tags'].get(tag_category, {}).get(tag_name, {})
-    return get_label(tag_labels, lang)
-
-
-def get_key_label(key_name, lang):
-    """Get key label from OSM_KEYS_TAGS"""
-    if key_name.startswith('currency:'):
-        # currency:EUR --> get the name from the CURRENCIES variable
-        # see https://wiki.openstreetmap.org/wiki/Key%3Acurrency
-        # and for example https://taginfo.openstreetmap.org/keys/currency:EUR#values
-        # but there is also currency=EUR (currently not handled)
-        # https://taginfo.openstreetmap.org/keys/currency#values
-        currency = key_name.split(':')
-        if len(currency) > 1:
-            label = fetch_name_from_iso4217(currency[1], lang)
-            if label:
-                return label
-            return currency[1]
-
-    labels = OSM_KEYS_TAGS['keys']
-    for k in key_name.split(':') + ['*']:
-        labels = labels.get(k)
-        if labels is None:
-            return None
-    return get_label(labels, lang)