]> git.openstreetmap.org Git - nominatim.git/blobdiff - docs/customize/Tokenizers.md
Merge pull request #3520 from lonvia/restrict-interpolation-range
[nominatim.git] / docs / customize / Tokenizers.md
index 4a9d77bd6c1633344c7cc5e394f401047b7daf72..49e86a5009289cea7f12aea36202abbda1548737 100644 (file)
@@ -17,6 +17,11 @@ they can be configured.
 
 ## Legacy tokenizer
 
+!!! danger
+    The Legacy tokenizer is deprecated and will be removed in Nominatim 5.0.
+    If you still use a database with the legacy tokenizer, you must reimport
+    it using the ICU tokenizer below.
+
 The legacy tokenizer implements the analysis algorithms of older Nominatim
 versions. It uses a special Postgresql module to normalize names and queries.
 This tokenizer is automatically installed and used when upgrading an older
@@ -52,7 +57,7 @@ NOMINATIM_DATABASE_MODULE_PATH=<path to directory where nominatim.so resides>
 ```
 
 This is in particular useful when the database runs on a different server.
-See [Advanced installations](../admin/Advanced-Installations.md#importing-nominatim-to-an-external-postgresql-database) for details.
+See [Advanced installations](../admin/Advanced-Installations.md#using-an-external-postgresql-database) for details.
 
 There are no other configuration options for the legacy tokenizer. All
 normalization functions are hard-coded.
@@ -175,73 +180,65 @@ The following is a list of sanitizers that are shipped with Nominatim.
 
 ##### split-name-list
 
-::: nominatim.tokenizer.sanitizers.split_name_list
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.split_name_list
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 ##### strip-brace-terms
 
-::: nominatim.tokenizer.sanitizers.strip_brace_terms
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.strip_brace_terms
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 ##### tag-analyzer-by-language
 
-::: nominatim.tokenizer.sanitizers.tag_analyzer_by_language
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.tag_analyzer_by_language
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 ##### clean-housenumbers
 
-::: nominatim.tokenizer.sanitizers.clean_housenumbers
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.clean_housenumbers
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 ##### clean-postcodes
 
-::: nominatim.tokenizer.sanitizers.clean_postcodes
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.clean_postcodes
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 ##### clean-tiger-tags
 
-::: nominatim.tokenizer.sanitizers.clean_tiger_tags
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.clean_tiger_tags
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 #### delete-tags
 
-::: nominatim.tokenizer.sanitizers.delete_tags
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.delete_tags
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
 #### tag-japanese
 
-::: nominatim.tokenizer.sanitizers.tag_japanese
-    selection:
-        members: False
+::: nominatim_db.tokenizer.sanitizers.tag_japanese
     options:
+        members: False
         heading_level: 6
         docstring_section_style: spacy
 
@@ -402,7 +399,7 @@ The analyzer cannot be customized.
 ##### Postcode token analyzer
 
 The analyzer `postcodes` is pupose-made to analyze postcodes. It supports
-a 'lookup' varaint of the token, which produces variants with optional
+a 'lookup' variant of the token, which produces variants with optional
 spaces. Use together with the clean-postcodes sanitizer.
 
 The analyzer cannot be customized.