introduce and use analyzer for postcodes

[nominatim.git] / nominatim / tokenizer / icu_tokenizer.py

diff --git a/nominatim/tokenizer/icu_tokenizer.py b/nominatim/tokenizer/icu_tokenizer.py
index 7bc4720ef56ed82b4d8fe45f484ea8a386ade422..e9812ba0430338e6647d459a0c162eadad0d467c 100644 (file)
--- a/nominatim/tokenizer/icu_tokenizer.py
+++ b/nominatim/tokenizer/icu_tokenizer.py
@@ -11,7 +11,6 @@ libICU instead of the PostgreSQL module.
 import itertools
 import json
 import logging
 import itertools
 import json
 import logging

-import re

 from textwrap import dedent
 
 from nominatim.db.connection import connect
 from textwrap import dedent
 
 from nominatim.db.connection import connect


@@ -51,7 +50,7 @@ class LegacyICUTokenizer(AbstractTokenizer):
         """
         self.loader = ICURuleLoader(config)
 
         """
         self.loader = ICURuleLoader(config)
 

-        self._install_php(config.lib_dir.php)
+        self._install_php(config.lib_dir.php, overwrite=True)

         self._save_config()
 
         if init_db:
         self._save_config()
 
         if init_db:


@@ -67,6 +66,8 @@ class LegacyICUTokenizer(AbstractTokenizer):
         with connect(self.dsn) as conn:
             self.loader.load_config_from_db(conn)
 
         with connect(self.dsn) as conn:
             self.loader.load_config_from_db(conn)
 

+        self._install_php(config.lib_dir.php, overwrite=False)
+

 
     def finalize_import(self, config):
         """ Do any required postprocessing to make the tokenizer data ready
 
     def finalize_import(self, config):
         """ Do any required postprocessing to make the tokenizer data ready


@@ -119,12 +120,13 @@ class LegacyICUTokenizer(AbstractTokenizer):
             if not conn.table_exists('search_name'):
                 return
             with conn.cursor(name="hnr_counter") as cur:
             if not conn.table_exists('search_name'):
                 return
             with conn.cursor(name="hnr_counter") as cur:

-                cur.execute("""SELECT word_id, word_token FROM word
+                cur.execute("""SELECT DISTINCT word_id, coalesce(info->>'lookup', word_token)
+                               FROM word

                                WHERE type = 'H'
                                  AND NOT EXISTS(SELECT * FROM search_name
                                                 WHERE ARRAY[word.word_id] && name_vector)
                                WHERE type = 'H'
                                  AND NOT EXISTS(SELECT * FROM search_name
                                                 WHERE ARRAY[word.word_id] && name_vector)

-                                 AND (char_length(word_token) > 6
-                                      OR word_token not similar to '\\d+')
+                                 AND (char_length(coalesce(word, word_token)) > 6
+                                      OR coalesce(word, word_token) not similar to '\\d+')

                             """)
                 candidates = {token: wid for wid, token in cur}
             with conn.cursor(name="hnr_counter") as cur:
                             """)
                 candidates = {token: wid for wid, token in cur}
             with conn.cursor(name="hnr_counter") as cur:


@@ -137,6 +139,7 @@ class LegacyICUTokenizer(AbstractTokenizer):
                     for hnr in row[0].split(';'):
                         candidates.pop(hnr, None)
             LOG.info("There are %s outdated housenumbers.", len(candidates))
                     for hnr in row[0].split(';'):
                         candidates.pop(hnr, None)
             LOG.info("There are %s outdated housenumbers.", len(candidates))

+            LOG.debug("Outdated housenumbers: %s", candidates.keys())

             if candidates:
                 with conn.cursor() as cur:
                     cur.execute("""DELETE FROM word WHERE word_id = any(%s)""",
             if candidates:
                 with conn.cursor() as cur:
                     cur.execute("""DELETE FROM word WHERE word_id = any(%s)""",


@@ -172,16 +175,18 @@ class LegacyICUTokenizer(AbstractTokenizer):
                                      self.loader.make_token_analysis())
 
 
                                      self.loader.make_token_analysis())
 
 

-    def _install_php(self, phpdir):
+    def _install_php(self, phpdir, overwrite=True):

         """ Install the php script for the tokenizer.
         """
         php_file = self.data_dir / "tokenizer.php"
         """ Install the php script for the tokenizer.
         """
         php_file = self.data_dir / "tokenizer.php"

-        php_file.write_text(dedent(f"""\
-            <?php
-            @define('CONST_Max_Word_Frequency', 10000000);
-            @define('CONST_Term_Normalization_Rules', "{self.loader.normalization_rules}");
-            @define('CONST_Transliteration', "{self.loader.get_search_rules()}");
-            require_once('{phpdir}/tokenizer/icu_tokenizer.php');"""))
+
+        if not php_file.exists() or overwrite:
+            php_file.write_text(dedent(f"""\
+                <?php
+                @define('CONST_Max_Word_Frequency', 10000000);
+                @define('CONST_Term_Normalization_Rules', "{self.loader.normalization_rules}");
+                @define('CONST_Transliteration', "{self.loader.get_search_rules()}");
+                require_once('{phpdir}/tokenizer/icu_tokenizer.php');"""), encoding='utf-8')

 
 
     def _save_config(self):
 
 
     def _save_config(self):


@@ -272,8 +277,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
                + [(k, v, part_ids.get(v, None)) for k, v in partial_tokens.items()]
 
 
                + [(k, v, part_ids.get(v, None)) for k, v in partial_tokens.items()]
 
 

-    @staticmethod
-    def normalize_postcode(postcode):
+    def normalize_postcode(self, postcode):

         """ Convert the postcode to a standardized form.
 
             This function must yield exactly the same result as the SQL function
         """ Convert the postcode to a standardized form.
 
             This function must yield exactly the same result as the SQL function


@@ -468,7 +472,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
     def _process_place_address(self, token_info, address):
         for item in address:
             if item.kind == 'postcode':
     def _process_place_address(self, token_info, address):
         for item in address:
             if item.kind == 'postcode':

-                self._add_postcode(item.name)
+                token_info.set_postcode(self._add_postcode(item))

             elif item.kind == 'housenumber':
                 token_info.add_housenumber(*self._compute_housenumber_token(item))
             elif item.kind == 'street':
             elif item.kind == 'housenumber':
                 token_info.add_housenumber(*self._compute_housenumber_token(item))
             elif item.kind == 'street':


@@ -477,7 +481,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
                 if not item.suffix:
                     token_info.add_place(self._compute_partial_tokens(item.name))
             elif not item.kind.startswith('_') and not item.suffix and \
                 if not item.suffix:
                     token_info.add_place(self._compute_partial_tokens(item.name))
             elif not item.kind.startswith('_') and not item.suffix and \

-                 item.kind not in ('country', 'full'):
+                 item.kind not in ('country', 'full', 'inclusion'):

                 token_info.add_address_term(item.kind, self._compute_partial_tokens(item.name))
 
 
                 token_info.add_address_term(item.kind, self._compute_partial_tokens(item.name))
 
 


@@ -588,7 +592,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
                     continue
 
                 with self.conn.cursor() as cur:
                     continue
 
                 with self.conn.cursor() as cur:

-                    cur.execute("SELECT (getorcreate_full_word(%s, %s)).*",
+                    cur.execute("SELECT * FROM getorcreate_full_word(%s, %s)",

                                 (token_id, variants))
                     full, part = cur.fetchone()
 
                                 (token_id, variants))
                     full, part = cur.fetchone()
 


@@ -600,26 +604,36 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         return full_tokens, partial_tokens
 
 
         return full_tokens, partial_tokens
 
 

-    def _add_postcode(self, postcode):
+    def _add_postcode(self, item):

         """ Make sure the normalized postcode is present in the word table.
         """
         """ Make sure the normalized postcode is present in the word table.
         """

-        if re.search(r'[:,;]', postcode) is None:
-            postcode = self.normalize_postcode(postcode)
+        analyzer = self.token_analysis.get_analyzer('@postcode')

 
 

-            if postcode not in self._cache.postcodes:
-                term = self._search_normalized(postcode)
-                if not term:
-                    return
+        if analyzer is None:
+            postcode_name = item.name.strip().upper()
+            variant_base = None
+        else:
+            postcode_name = analyzer.normalize(item.name)
+            variant_base = item.get_attr("variant")

 
 

-                with self.conn.cursor() as cur:
-                    # no word_id needed for postcodes
-                    cur.execute("""INSERT INTO word (word_token, type, word)
-                                   (SELECT %s, 'P', pc FROM (VALUES (%s)) as v(pc)
-                                    WHERE NOT EXISTS
-                                     (SELECT * FROM word
-                                      WHERE type = 'P' and word = pc))
-                                """, (term, postcode))
-                self._cache.postcodes.add(postcode)
+        if variant_base is not None:
+            postcode = f'{postcode_name}@{variant_base}'
+        else:
+            postcode = postcode_name
+
+        if postcode not in self._cache.postcodes:
+            term = self._search_normalized(postcode_name)
+            if not term:
+                return
+
+            variants = {term}
+            if analyzer is not None and variant_base is not None:
+                variants.update(analyzer.get_variants_ascii(variant_base))
+
+            with self.conn.cursor() as cur:
+                cur.execute("SELECT create_postcode_word(%s, %s)",
+                            (postcode, list(variants)))
+            self._cache.postcodes.add(postcode)

 
 
 class _TokenInfo:
 
 
 class _TokenInfo:


@@ -632,6 +646,7 @@ class _TokenInfo:
         self.street_tokens = set()
         self.place_tokens = set()
         self.address_tokens = {}
         self.street_tokens = set()
         self.place_tokens = set()
         self.address_tokens = {}

+        self.postcode = None

 
 
     @staticmethod
 
 
     @staticmethod


@@ -696,6 +711,11 @@ class _TokenInfo:
         if partials:
             self.address_tokens[key] = self._mk_array(partials)
 
         if partials:
             self.address_tokens[key] = self._mk_array(partials)
 

+    def set_postcode(self, postcode):
+        """ Set the postcode to the given one.
+        """
+        self.postcode = postcode
+

 
 class _TokenCache:
     """ Cache for token information to avoid repeated database queries.
 
 class _TokenCache:
     """ Cache for token information to avoid repeated database queries.