icu: switch postcodes to using the pre-formatted one

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

diff --git a/nominatim/tokenizer/icu_tokenizer.py b/nominatim/tokenizer/icu_tokenizer.py
index 9c25b6d7940fc145a2565a326d239463e32227cc..61c47c118805f4638cd5fc022d006564e1ff9b15 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


@@ -282,13 +286,6 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         return postcode.strip().upper()
 
 
         return postcode.strip().upper()
 
 

-    def _make_standard_hnr(self, hnr):
-        """ Create a normalised version of a housenumber.
-
-            This function takes minor shortcuts on transliteration.
-        """
-        return self._search_normalized(hnr)
-

     def update_postcodes_from_db(self):
         """ Update postcode tokens in the word table from the location_postcode
             table.
     def update_postcodes_from_db(self):
         """ Update postcode tokens in the word table from the location_postcode
             table.


@@ -456,14 +453,12 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
             Returns a JSON-serializable structure that will be handed into
             the database via the token_info field.
         """
             Returns a JSON-serializable structure that will be handed into
             the database via the token_info field.
         """

-        token_info = _TokenInfo(self._cache)
+        token_info = _TokenInfo()

 
         names, address = self.sanitizer.process_names(place)
 
         if names:
 
         names, address = self.sanitizer.process_names(place)
 
         if names:

-            fulls, partials = self._compute_name_tokens(names)
-
-            token_info.add_names(fulls, partials)
+            token_info.set_names(*self._compute_name_tokens(names))

 
             if place.is_country():
                 self._add_country_full_names(place.country_code, names)
 
             if place.is_country():
                 self._add_country_full_names(place.country_code, names)


@@ -471,37 +466,59 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
         if address:
             self._process_place_address(token_info, address)
 
         if address:
             self._process_place_address(token_info, address)
 

-        return token_info.data
+        return token_info.to_dict()

 
 
     def _process_place_address(self, token_info, address):
 
 
     def _process_place_address(self, token_info, address):

-        hnrs = set()
-        addr_terms = []
-        streets = []

         for item in address:
             if item.kind == 'postcode':
         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':
             elif item.kind == 'housenumber':

-                norm_name = self._make_standard_hnr(item.name)
-                if norm_name:
-                    hnrs.add(norm_name)
+                token_info.add_housenumber(*self._compute_housenumber_token(item))

             elif item.kind == 'street':
             elif item.kind == 'street':

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

             elif item.kind == 'place':
                 if not item.suffix:
                     token_info.add_place(self._compute_partial_tokens(item.name))
             elif not item.kind.startswith('_') and not item.suffix and \
             elif item.kind == 'place':
                 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', 'inclusion'):
+                token_info.add_address_term(item.kind, self._compute_partial_tokens(item.name))
+

 
 

-        if hnrs:
-            token_info.add_housenumbers(self.conn, hnrs)
+    def _compute_housenumber_token(self, hnr):
+        """ Normalize the housenumber and return the word token and the
+            canonical form.
+        """
+        analyzer = self.token_analysis.analysis.get('@housenumber')
+        result = None, None

 
 

-        if addr_terms:
-            token_info.add_address_terms(addr_terms)
+        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

 
 

-        if streets:
-            token_info.add_street(streets)
+        return result

 
 
     def _compute_partial_tokens(self, name):
 
 
     def _compute_partial_tokens(self, name):


@@ -561,7 +578,8 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
 
         for name in names:
             analyzer_id = name.get_attr('analyzer')
 
         for name in names:
             analyzer_id = name.get_attr('analyzer')

-            norm_name = self._normalized(name.name)
+            analyzer = self.token_analysis.get_analyzer(analyzer_id)
+            norm_name = analyzer.normalize(name.name)

             if analyzer_id is None:
                 token_id = norm_name
             else:
             if analyzer_id is None:
                 token_id = norm_name
             else:


@@ -569,12 +587,12 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer):
 
             full, part = self._cache.names.get(token_id, (None, None))
             if full is None:
 
             full, part = self._cache.names.get(token_id, (None, None))
             if full is None:

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

                 if not variants:
                     continue
 
                 with self.conn.cursor() as cur:
                 if not variants:
                     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()
 


@@ -586,75 +604,120 @@ 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:
     """ Collect token information to be sent back to the database.
     """
 
 
 class _TokenInfo:
     """ Collect token information to be sent back to the database.
     """

-    def __init__(self, cache):
-        self._cache = cache
-        self.data = {}
+    def __init__(self):
+        self.names = None
+        self.housenumbers = set()
+        self.housenumber_tokens = set()
+        self.street_tokens = set()
+        self.place_tokens = set()
+        self.address_tokens = {}
+        self.postcode = None
+

 
     @staticmethod
     def _mk_array(tokens):
 
     @staticmethod
     def _mk_array(tokens):

-        return '{%s}' % ','.join((str(s) for s in tokens))
+        return f"{{{','.join((str(s) for s in tokens))}}}"

 
 
 
 

-    def add_names(self, fulls, partials):
+    def to_dict(self):
+        """ Return the token information in database importable format.
+        """
+        out = {}
+
+        if self.names:
+            out['names'] = self.names
+
+        if self.housenumbers:
+            out['hnr'] = ';'.join(self.housenumbers)
+            out['hnr_tokens'] = self._mk_array(self.housenumber_tokens)
+
+        if self.street_tokens:
+            out['street'] = self._mk_array(self.street_tokens)
+
+        if self.place_tokens:
+            out['place'] = self._mk_array(self.place_tokens)
+
+        if self.address_tokens:
+            out['addr'] = self.address_tokens
+
+        if self.postcode:
+            out['postcode'] = self.postcode
+
+        return out
+
+
+    def set_names(self, fulls, partials):

         """ Adds token information for the normalised names.
         """
         """ Adds token information for the normalised names.
         """

-        self.data['names'] = self._mk_array(itertools.chain(fulls, partials))
+        self.names = self._mk_array(itertools.chain(fulls, partials))

 
 
 
 

-    def add_housenumbers(self, conn, hnrs):
+    def add_housenumber(self, token, hnr):

         """ Extract housenumber information from a list of normalised
             housenumbers.
         """
         """ Extract housenumber information from a list of normalised
             housenumbers.
         """

-        self.data['hnr_tokens'] = self._mk_array(self._cache.get_hnr_tokens(conn, hnrs))
-        self.data['hnr'] = ';'.join(hnrs)
+        if token:
+            self.housenumbers.add(hnr)
+            self.housenumber_tokens.add(token)

 
 
     def add_street(self, tokens):
         """ Add addr:street match terms.
         """
 
 
     def add_street(self, tokens):
         """ Add addr:street match terms.
         """

-        self.data['street'] = self._mk_array(tokens)
+        self.street_tokens.update(tokens)

 
 
     def add_place(self, tokens):
         """ Add addr:place search and match terms.
         """
 
 
     def add_place(self, tokens):
         """ Add addr:place search and match terms.
         """

-        if tokens:
-            self.data['place'] = self._mk_array(tokens)
+        self.place_tokens.update(tokens)

 
 
 
 

-    def add_address_terms(self, terms):
+    def add_address_term(self, key, partials):

         """ Add additional address terms.
         """
         """ Add additional address terms.
         """

-        tokens = {key: self._mk_array(partials)
-                  for key, partials in terms if partials}
+        if partials:
+            self.address_tokens[key] = self._mk_array(partials)

 
 

-        if tokens:
-            self.data['addr'] = tokens
+    def set_postcode(self, postcode):
+        """ Set the postcode to the given one.
+        """
+        self.postcode = postcode

 
 
 class _TokenCache:
 
 
 class _TokenCache:


@@ -669,29 +732,3 @@ class _TokenCache:
         self.fulls = {}
         self.postcodes = set()
         self.housenumbers = {}
         self.fulls = {}
         self.postcodes = set()
         self.housenumbers = {}

-
-
-    def get_hnr_tokens(self, conn, terms):
-        """ Get token ids for a list of housenumbers, looking them up in the
-            database if necessary. `terms` is an iterable of normalized
-            housenumbers.
-        """
-        tokens = []
-        askdb = []
-
-        for term in terms:
-            token = self.housenumbers.get(term)
-            if token is None:
-                askdb.append(term)
-            else:
-                tokens.append(token)
-
-        if askdb:
-            with conn.cursor() as cur:
-                cur.execute("SELECT nr, getorcreate_hnr_id(nr) FROM unnest(%s) as nr",
-                            (askdb, ))
-                for term, tid in cur:
-                    self.housenumbers[term] = tid
-                    tokens.append(tid)
-
-        return tokens