Merge pull request #3675 from TuringVerified/generic-preprocessors

Add generic preprocessors

docs/customize/Tokenizers.md

@@ -67,7 +67,13 @@ Here is an example configuration file:
 
 ``` yaml
 query-preprocessing:
-    - normalize
+    - step: split_japanese_phrases
+    - step: regex_replace
+      replacements:
+        - pattern: https?://[^\s]* # Filter URLs starting with http or https
+          replace: ''
+    - step: normalize
+
 normalization:
     - ":: lower ()"
     - "ß > 'ss'" # German szet is unambiguously equal to double ss
@@ -88,8 +94,8 @@ token-analysis:
             replacements: ['ä', 'ae']
 ```
 
-The configuration file contains four sections:
-`normalization`, `transliteration`, `sanitizers` and `token-analysis`.
+The configuration file contains five sections:
+`query-preprocessing`, `normalization`, `transliteration`, `sanitizers` and `token-analysis`.
 
 #### Query preprocessing
 
@@ -106,6 +112,19 @@ The following is a list of preprocessors that are shipped with Nominatim.
         heading_level: 6
         docstring_section_style: spacy
 
+##### regex-replace
+
+::: nominatim_api.query_preprocessing.regex_replace
+    options:
+        members: False
+        heading_level: 6
+        docstring_section_style: spacy
+    description: 
+        This option runs any given regex pattern on the input and replaces values accordingly
+    replacements:
+        - pattern: regex pattern
+          replace: string to replace with
+
 
 #### Normalization and Transliteration
 

src/nominatim_api/query_preprocessing/regex_replace.py

@@ -0,0 +1,52 @@
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2025 by the Nominatim developer community.
+# For a full list of authors see the git log.
+"""
+This preprocessor replaces values in a given input based on pre-defined regex rules.
+
+Arguments:
+    pattern: Regex pattern to be applied on the input
+    replace: The string that it is to be replaced with
+"""
+from typing import List
+import re
+
+from .config import QueryConfig
+from .base import QueryProcessingFunc
+from ..search.query import Phrase
+
+
+class _GenericPreprocessing:
+    """Perform replacements to input phrases using custom regex patterns."""
+
+    def __init__(self, config: QueryConfig) -> None:
+        """Initialise the _GenericPreprocessing class with patterns from the ICU config file."""
+        self.config = config
+
+        match_patterns = self.config.get('replacements', 'Key not found')
+        self.compiled_patterns = [
+            (re.compile(item['pattern']), item['replace']) for item in match_patterns
+            ]
+
+    def split_phrase(self, phrase: Phrase) -> Phrase:
+        """This function performs replacements on the given text using regex patterns."""
+        for item in self.compiled_patterns:
+            phrase.text = item[0].sub(item[1], phrase.text)
+
+        return phrase
+
+    def __call__(self, phrases: List[Phrase]) -> List[Phrase]:
+        """
+        Return the final Phrase list.
+        Returns an empty list if there is nothing left after split_phrase.
+        """
+        result = [p for p in map(self.split_phrase, phrases) if p.text.strip()]
+        return result
+
+
+def create(config: QueryConfig) -> QueryProcessingFunc:
+    """ Create a function for generic preprocessing."""
+    return _GenericPreprocessing(config)

test/python/api/query_processing/test_regex_replace.py

@@ -0,0 +1,49 @@
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2025 by the Nominatim developer community.
+# For a full list of authors see the git log.
+'''
+Tests for replacing values in an input using custom regex.
+'''
+import pytest
+
+import nominatim_api.search.query as qmod
+from nominatim_api.query_preprocessing.config import QueryConfig
+from nominatim_api.query_preprocessing import regex_replace
+
+
+def run_preprocessor_on(query):
+    config = QueryConfig()
+    config.set_normalizer(None)
+
+    config['replacements'] = [
+        {'pattern': r'\b(?:\d{1,3}\.){3}\d{1,3}\b', 'replace': ''},  # IPv4
+        {'pattern': r'https?://\S+', 'replace': ''}  # HTTP/HTTPS URLs
+    ]
+
+    proc = regex_replace.create(config)
+    return proc(query)
+
+
+@pytest.mark.parametrize('inp,outp', [
+    (['45.67.89.101'], []),
+    (['198.51.100.23'], []),
+    (['203.0.113.255'], []),
+    (['http://www.openstreetmap.org'], []),
+    (['https://www.openstreetmap.org/edit'], []),
+    (['http://osm.org'], []),
+    (['https://www.openstreetmap.org/user/abc'], []),
+    (['https://tile.openstreetmap.org/12/2048/2048.png'], []),
+    (['Check the map at https://www.openstreetmap.org'], ['Check the map at ']),
+    (['Use 203.0.113.255 for routing'], ['Use  for routing']),
+    (['Find maps at https://osm.org and http://openstreetmap.org'], ['Find maps at  and ']),
+    (['203.0.113.255', 'Some Address'], ['Some Address']),
+    (['https://osm.org', 'Another Place'], ['Another Place']),
+])
+def test_split_phrases(inp, outp):
+    query = [qmod.Phrase(qmod.PHRASE_ANY, text) for text in inp]
+
+    out = run_preprocessor_on(query)
+    assert out == [qmod.Phrase(qmod.PHRASE_ANY, text) for text in outp]