restore the tokenizer directory when missing

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

diff --git a/nominatim/tokenizer/icu_tokenizer.py b/nominatim/tokenizer/icu_tokenizer.py
index 3ce4895b901f7253744e4900f3072997f202507b..b553dbc641d708175e8f7281f05cf14cf4673484 100644 (file)
--- a/nominatim/tokenizer/icu_tokenizer.py
+++ b/nominatim/tokenizer/icu_tokenizer.py
@@ -51,7 +51,7 @@ class LegacyICUTokenizer(AbstractTokenizer):
         """
         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:
@@ -67,6 +67,8 @@ class LegacyICUTokenizer(AbstractTokenizer):
         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
@@ -119,12 +121,13 @@ class LegacyICUTokenizer(AbstractTokenizer):
             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)
-                                 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:
@@ -137,6 +140,7 @@ class LegacyICUTokenizer(AbstractTokenizer):
                     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             LOG.info("There are %s outdated housenumbers.", len(candidates))
+            LOG.debug("Outdated housenumbers: %s", candidates.keys())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)""",
@@ -172,16 +176,18 @@ class LegacyICUTokenizer(AbstractTokenizer):
                                      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"
-        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');"""))
 
 
     def _save_config(self):
@@ -485,18 +491,36 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         """ Normalize the housenumber and return the word token and the
             canonical form.
         """
-        norm_name = self._search_normalized(hnr.name)
-        if not norm_name:
-            return None, None
+        analyzer = self.token_analysis.analysis.get('@housenumber')
+        result = None, None
 
-        token = self._cache.housenumbers.get(norm_name)
-        if token is None:
-            with self.conn.cursor() as cur:
-                cur.execute("SELECT getorcreate_hnr_id(%s)", (norm_name, ))
-                token = cur.fetchone()[0]
-                self._cache.housenumbers[norm_name] = token
+        if analyzer is None:
+            # When no custom analyzer is set, simply normalize and transliterate
+            norm_name = self._search_normalized(hnr.name)
+            if norm_name:
+                result = self._cache.housenumbers.get(norm_name, result)
+                if result[0] is None:
+                    with self.conn.cursor() as cur:
+                        cur.execute("SELECT getorcreate_hnr_id(%s)", (norm_name, ))
+                        result = cur.fetchone()[0], norm_name
+                        self._cache.housenumbers[norm_name] = result
+        else:
+            # Otherwise use the analyzer to determine the canonical name.
+            # Per convention we use the first variant as the 'lookup name', the
+            # name that gets saved in the housenumber field of the place.
+            norm_name = analyzer.normalize(hnr.name)
+            if norm_name:
+                result = self._cache.housenumbers.get(norm_name, result)
+                if result[0] is None:
+                    variants = analyzer.get_variants_ascii(norm_name)
+                    if variants:
+                        with self.conn.cursor() as cur:
+                            cur.execute("SELECT create_analyzed_hnr_id(%s, %s)",
+                                        (norm_name, list(variants)))
+                            result = cur.fetchone()[0], variants[0]
+                            self._cache.housenumbers[norm_name] = result
 
-        return token, norm_name
+        return result
 
 
     def _compute_partial_tokens(self, name):
@@ -570,7 +594,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
                     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()