Merge remote-tracking branch 'upstream/master'

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

diff --git a/nominatim/tokenizer/icu_tokenizer.py b/nominatim/tokenizer/icu_tokenizer.py
index 2ece10f2ccd28aecb3a181ceafabeb86c9f289a7..90caec1c9041697f02fec06c62960d08eaf01dcb 100644 (file)
--- a/nominatim/tokenizer/icu_tokenizer.py
+++ b/nominatim/tokenizer/icu_tokenizer.py
@@ -2,7 +2,6 @@
 Tokenizer implementing normalisation as used before Nominatim 4 but using
 libICU instead of the PostgreSQL module.
 """
 Tokenizer implementing normalisation as used before Nominatim 4 but using
 libICU instead of the PostgreSQL module.
 """

-from collections import Counter

 import itertools
 import json
 import logging
 import itertools
 import json
 import logging


@@ -10,7 +9,6 @@ import re
 from textwrap import dedent
 
 from nominatim.db.connection import connect
 from textwrap import dedent
 
 from nominatim.db.connection import connect

-from nominatim.db.properties import set_property, get_property

 from nominatim.db.utils import CopyBuffer
 from nominatim.db.sql_preprocessor import SQLPreprocessor
 from nominatim.indexer.place_info import PlaceInfo
 from nominatim.db.utils import CopyBuffer
 from nominatim.db.sql_preprocessor import SQLPreprocessor
 from nominatim.indexer.place_info import PlaceInfo


@@ -37,7 +35,6 @@ class LegacyICUTokenizer(AbstractTokenizer):
         self.dsn = dsn
         self.data_dir = data_dir
         self.loader = None
         self.dsn = dsn
         self.data_dir = data_dir
         self.loader = None

-        self.term_normalization = None

 
 
     def init_new_db(self, config, init_db=True):
 
 
     def init_new_db(self, config, init_db=True):


@@ -48,8 +45,6 @@ class LegacyICUTokenizer(AbstractTokenizer):
         """
         self.loader = ICURuleLoader(config)
 
         """
         self.loader = ICURuleLoader(config)
 

-        self.term_normalization = config.TERM_NORMALIZATION
-

         self._install_php(config.lib_dir.php)
         self._save_config()
 
         self._install_php(config.lib_dir.php)
         self._save_config()
 


@@ -65,13 +60,15 @@ 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.term_normalization = get_property(conn, DBCFG_TERM_NORMALIZATION)

 
 
 
 

-    def finalize_import(self, _):
+    def finalize_import(self, config):

         """ Do any required postprocessing to make the tokenizer data ready
             for use.
         """
         """ Do any required postprocessing to make the tokenizer data ready
             for use.
         """

+        with connect(self.dsn) as conn:
+            sqlp = SQLPreprocessor(conn, config)
+            sqlp.run_sql_file(conn, 'tokenizer/legacy_tokenizer_indices.sql')

 
 
     def update_sql_functions(self, config):
 
 
     def update_sql_functions(self, config):


@@ -85,12 +82,28 @@ class LegacyICUTokenizer(AbstractTokenizer):
     def check_database(self, config):
         """ Check that the tokenizer is set up correctly.
         """
     def check_database(self, config):
         """ Check that the tokenizer is set up correctly.
         """

+        # Will throw an error if there is an issue.

         self.init_from_project(config)
 
         self.init_from_project(config)
 

-        if self.term_normalization is None:
-            return "Configuration for tokenizer 'icu' are missing."

 
 

-        return None
+    def update_statistics(self):
+        """ Recompute frequencies for all name words.
+        """
+        with connect(self.dsn) as conn:
+            if conn.table_exists('search_name'):
+                with conn.cursor() as cur:
+                    cur.drop_table("word_frequencies")
+                    LOG.info("Computing word frequencies")
+                    cur.execute("""CREATE TEMP TABLE word_frequencies AS
+                                     SELECT unnest(name_vector) as id, count(*)
+                                     FROM search_name GROUP BY id""")
+                    cur.execute("CREATE INDEX ON word_frequencies(id)")
+                    LOG.info("Update word table with recomputed frequencies")
+                    cur.execute("""UPDATE word
+                                   SET info = info || jsonb_build_object('count', count)
+                                   FROM word_frequencies WHERE word_id = id""")
+                    cur.drop_table("word_frequencies")
+            conn.commit()

 
 
     def name_analyzer(self):
 
 
     def name_analyzer(self):


@@ -119,7 +132,7 @@ class LegacyICUTokenizer(AbstractTokenizer):
         php_file.write_text(dedent(f"""\
             <?php
             @define('CONST_Max_Word_Frequency', 10000000);
         php_file.write_text(dedent(f"""\
             <?php
             @define('CONST_Max_Word_Frequency', 10000000);

-            @define('CONST_Term_Normalization_Rules', "{self.term_normalization}");
+            @define('CONST_Term_Normalization_Rules', "{self.loader.normalization_rules}");

             @define('CONST_Transliteration', "{self.loader.get_search_rules()}");
             require_once('{phpdir}/tokenizer/icu_tokenizer.php');"""))
 
             @define('CONST_Transliteration', "{self.loader.get_search_rules()}");
             require_once('{phpdir}/tokenizer/icu_tokenizer.php');"""))
 


@@ -130,7 +143,6 @@ class LegacyICUTokenizer(AbstractTokenizer):
         """
         with connect(self.dsn) as conn:
             self.loader.save_config_to_db(conn)
         """
         with connect(self.dsn) as conn:
             self.loader.save_config_to_db(conn)

-            set_property(conn, DBCFG_TERM_NORMALIZATION, self.term_normalization)

 
 
     def _init_db_tables(self, config):
 
 
     def _init_db_tables(self, config):


@@ -142,45 +154,6 @@ class LegacyICUTokenizer(AbstractTokenizer):
             sqlp.run_sql_file(conn, 'tokenizer/icu_tokenizer_tables.sql')
             conn.commit()
 
             sqlp.run_sql_file(conn, 'tokenizer/icu_tokenizer_tables.sql')
             conn.commit()
 

-            LOG.warning("Precomputing word tokens")
-
-            # get partial words and their frequencies
-            words = self._count_partial_terms(conn)
-
-            # copy them back into the word table
-            with CopyBuffer() as copystr:
-                for term, cnt in words.items():
-                    copystr.add('w', term, json.dumps({'count': cnt}))
-
-                with conn.cursor() as cur:
-                    copystr.copy_out(cur, 'word',
-                                     columns=['type', 'word_token', 'info'])
-                    cur.execute("""UPDATE word SET word_id = nextval('seq_word')
-                                   WHERE word_id is null and type = 'w'""")
-
-            conn.commit()
-
-    def _count_partial_terms(self, conn):
-        """ Count the partial terms from the names in the place table.
-        """
-        words = Counter()
-        name_proc = self.loader.make_token_analysis()
-
-        with conn.cursor(name="words") as cur:
-            cur.execute(""" SELECT v, count(*) FROM
-                              (SELECT svals(name) as v FROM place)x
-                            WHERE length(v) < 75 GROUP BY v""")
-
-            for name, cnt in cur:
-                terms = set()
-                for word in name_proc.get_variants_ascii(name_proc.get_normalized(name)):
-                    if ' ' in word:
-                        terms.update(word.split())
-                for term in terms:
-                    words[term] += cnt
-
-        return words
-

 
 class LegacyICUNameAnalyzer(AbstractAnalyzer):
     """ The legacy analyzer uses the ICU library for splitting names.
 
 class LegacyICUNameAnalyzer(AbstractAnalyzer):
     """ The legacy analyzer uses the ICU library for splitting names.


@@ -209,14 +182,14 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
     def _search_normalized(self, name):
         """ Return the search token transliteration of the given name.
         """
     def _search_normalized(self, name):
         """ Return the search token transliteration of the given name.
         """

-        return self.token_analysis.get_search_normalized(name)
+        return self.token_analysis.search.transliterate(name).strip()

 
 
     def _normalized(self, name):
         """ Return the normalized version of the given name with all
             non-relevant information removed.
         """
 
 
     def _normalized(self, name):
         """ Return the normalized version of the given name with all
             non-relevant information removed.
         """

-        return self.token_analysis.get_normalized(name)
+        return self.token_analysis.normalizer.transliterate(name).strip()

 
 
     def get_word_token_info(self, words):
 
 
     def get_word_token_info(self, words):


@@ -436,16 +409,18 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
     def _process_place_address(self, token_info, address):
         hnrs = []
         addr_terms = []
     def _process_place_address(self, token_info, address):
         hnrs = []
         addr_terms = []

+        streets = []

         for item in address:
             if item.kind == 'postcode':
                 self._add_postcode(item.name)
             elif item.kind in ('housenumber', 'streetnumber', 'conscriptionnumber'):
                 hnrs.append(item.name)
             elif item.kind == 'street':
         for item in address:
             if item.kind == 'postcode':
                 self._add_postcode(item.name)
             elif item.kind in ('housenumber', 'streetnumber', 'conscriptionnumber'):
                 hnrs.append(item.name)
             elif item.kind == 'street':

-                token_info.add_street(self._compute_partial_tokens(item.name))
+                streets.extend(self._retrieve_full_tokens(item.name))

             elif item.kind == 'place':
             elif item.kind == 'place':

-                token_info.add_place(self._compute_partial_tokens(item.name))
-            elif not item.kind.startswith('_') 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'):
                 addr_terms.append((item.kind, self._compute_partial_tokens(item.name)))
 
                  item.kind not in ('country', 'full'):
                 addr_terms.append((item.kind, self._compute_partial_tokens(item.name)))
 


@@ -456,6 +431,10 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         if addr_terms:
             token_info.add_address_terms(addr_terms)
 
         if addr_terms:
             token_info.add_address_terms(addr_terms)
 

+        if streets:
+            token_info.add_street(streets)
+
+

     def _compute_partial_tokens(self, name):
         """ Normalize the given term, split it into partial words and return
             then token list for them.
     def _compute_partial_tokens(self, name):
         """ Normalize the given term, split it into partial words and return
             then token list for them.


@@ -484,6 +463,26 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         return tokens
 
 
         return tokens
 
 

+    def _retrieve_full_tokens(self, name):
+        """ Get the full name token for the given name, if it exists.
+            The name is only retrived for the standard analyser.
+        """
+        norm_name = self._search_normalized(name)
+
+        # return cached if possible
+        if norm_name in self._cache.fulls:
+            return self._cache.fulls[norm_name]
+
+        with self.conn.cursor() as cur:
+            cur.execute("SELECT word_id FROM word WHERE word_token = %s and type = 'W'",
+                        (norm_name, ))
+            full = [row[0] for row in cur]
+
+        self._cache.fulls[norm_name] = full
+
+        return full
+
+

     def _compute_name_tokens(self, names):
         """ Computes the full name and partial name tokens for the given
             dictionary of names.
     def _compute_name_tokens(self, names):
         """ Computes the full name and partial name tokens for the given
             dictionary of names.


@@ -492,19 +491,25 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         partial_tokens = set()
 
         for name in names:
         partial_tokens = set()
 
         for name in names:

+            analyzer_id = name.get_attr('analyzer')

             norm_name = self._normalized(name.name)
             norm_name = self._normalized(name.name)

-            full, part = self._cache.names.get(norm_name, (None, None))
+            if analyzer_id is None:
+                token_id = norm_name
+            else:
+                token_id = f'{norm_name}@{analyzer_id}'
+
+            full, part = self._cache.names.get(token_id, (None, None))

             if full is None:
             if full is None:

-                variants = self.token_analysis.get_variants_ascii(norm_name)
+                variants = self.token_analysis.analysis[analyzer_id].get_variants_ascii(norm_name)

                 if not variants:
                     continue
 
                 with self.conn.cursor() as cur:
                     cur.execute("SELECT (getorcreate_full_word(%s, %s)).*",
                 if not variants:
                     continue
 
                 with self.conn.cursor() as cur:
                     cur.execute("SELECT (getorcreate_full_word(%s, %s)).*",

-                                (norm_name, variants))
+                                (token_id, variants))

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

-                self._cache.names[norm_name] = (full, part)
+                self._cache.names[token_id] = (full, part)

 
             full_tokens.add(full)
             partial_tokens.update(part)
 
             full_tokens.add(full)
             partial_tokens.update(part)


@@ -581,8 +586,7 @@ class _TokenInfo:
     def add_street(self, tokens):
         """ Add addr:street match terms.
         """
     def add_street(self, tokens):
         """ Add addr:street match terms.
         """

-        if tokens:
-            self.data['street'] = self._mk_array(tokens)
+        self.data['street'] = self._mk_array(tokens)

 
 
     def add_place(self, tokens):
 
 
     def add_place(self, tokens):


@@ -611,6 +615,7 @@ class _TokenCache:
     def __init__(self):
         self.names = {}
         self.partials = {}
     def __init__(self):
         self.names = {}
         self.partials = {}

+        self.fulls = {}

         self.postcodes = set()
         self.housenumbers = {}
 
         self.postcodes = set()
         self.housenumbers = {}