// Check which tokens we have, get the ID numbers
$sSQL = 'SELECT word_id, word_token, type';
$sSQL .= " info->>'cc' as country, info->>'postcode' as postcode,";
- $sSQL .= " info->>'word' as word, info->>'op' as operator,";
- $sSQL .= " info->>'class' as class, info->>'type' as type";
+ $sSQL .= " info->>'op' as operator,";
+ $sSQL .= " info->>'class' as class, info->>'type' as type,";
+ $sSQL .= " info->>'count' as count";
$sSQL .= ' FROM word WHERE word_token in (';
$sSQL .= join(',', $this->oDB->getDBQuotedList($aTokens)).')';
break;
'S': // tokens for classification terms (special phrases)
if ($aWord['class'] === null || $aWord['type'] === null
- || $aWord['word'] === null
- || strpos($sNormQuery, $aWord['word']) === false
) {
continue;
}
$aWord['op'] ? Operator::NEAR : Operator::NONE
);
break;
+ 'W': // full-word tokens
+ $oToken = new Token\Word(
+ $iId,
+ (int) $aWord['count'],
+ substr_count($aWord['word_token'], ' ')
+ );
+ break;
+ 'w': // partial word terms
+ $oToken = new Token\Partial(
+ $iId,
+ $aWord['word_token'],
+ (int) $aWord['count']
+ );
+ break;
default:
continue;
}
-/* if ($aWord['class']) {
- // Special terms need to appear in their normalized form.
- // (postcodes are not normalized in the word table)
- $sNormWord = $this->normalizeString($aWord['word']);
- if ($aWord['word'] && strpos($sNormQuery, $sNormWord) === false) {
- continue;
- }
-
- if ($aWord['class'] == 'place' && $aWord['type'] == 'house') {
- $oToken = new Token\HouseNumber($iId, trim($aWord['word_token']));
- } elseif ($aWord['class'] == 'place' && $aWord['type'] == 'postcode') {
- if ($aWord['word']
- && pg_escape_string($aWord['word']) == $aWord['word']
- ) {
- $oToken = new Token\Postcode(
- $iId,
- $aWord['word'],
- $aWord['country_code']
- );
- }
- } else {
- // near and in operator the same at the moment
- $oToken = new Token\SpecialTerm(
- $iId,
- $aWord['class'],
- $aWord['type'],
- $aWord['operator'] ? Operator::NEAR : Operator::NONE
- );
- }
- } elseif ($aWord['country_code']) {
- // Filter country tokens that do not match restricted countries.
- if (!$this->aCountryRestriction
- || in_array($aWord['country_code'], $this->aCountryRestriction)
- ) {
- $oToken = new Token\Country($iId, $aWord['country_code']);
- }
- } elseif ($aWord['word_token'][0] == ' ') {
- $oToken = new Token\Word(
- $iId,
- (int) $aWord['count'],
- substr_count($aWord['word_token'], ' ')
- );
- } else {
- $oToken = new Token\Partial(
- $iId,
- $aWord['word_token'],
- (int) $aWord['count']
- );
- }*/
$oValidTokens->addToken($aWord['word_token'], $oToken);
}
-- Used when inserting new postcodes on updates.
CREATE INDEX idx_word_postcodes ON word
USING btree((info->>'postcode')) {{db.tablespace.address_index}}
- WHERE type = 'P'
+ WHERE type = 'P';
+-- Used when inserting full words.
+CREATE INDEX idx_word_full_word ON word
+ USING btree((info->>'word')) {{db.tablespace.address_index}}
+ WHERE type = 'W';
+
GRANT SELECT ON word TO "{{config.DATABASE_WEBUSER}}";
DROP SEQUENCE IF EXISTS seq_word;
term_count INTEGER;
BEGIN
SELECT min(word_id) INTO full_token
- FROM word WHERE word = norm_term and class is null and country_code is null;
+ FROM word WHERE info->>'word' = norm_term and type = 'W';
IF full_token IS NULL THEN
full_token := nextval('seq_word');
- INSERT INTO word (word_id, word_token, word, search_name_count)
- SELECT full_token, ' ' || lookup_term, norm_term, 0 FROM unnest(lookup_terms) as lookup_term;
+ INSERT INTO word (word_id, word_token, info)
+ SELECT full_token, lookup_term,
+ json_build_object('word', norm_term, 'count', 0)
+ FROM unnest(lookup_terms) as lookup_term;
END IF;
FOR term IN SELECT unnest(string_to_array(unnest(lookup_terms), ' ')) LOOP
partial_tokens := '{}'::INT[];
FOR term IN SELECT unnest(partial_terms) LOOP
- SELECT min(word_id), max(search_name_count) INTO term_id, term_count
- FROM word WHERE word_token = term and class is null and country_code is null;
+ SELECT min(word_id), max(info->>'count') INTO term_id, term_count
+ FROM word WHERE word_token = term and type = 'w';
IF term_id IS NULL THEN
term_id := nextval('seq_word');
term_count := 0;
- INSERT INTO word (word_id, word_token, search_name_count)
- VALUES (term_id, term, 0);
+ INSERT INTO word (word_id, word_token, info)
+ VALUES (term_id, term, json_build_object('count', term_count));
END IF;
IF term_count < {{ max_word_freq }} THEN
self.max_word_frequency = get_property(conn, DBCFG_MAXWORDFREQ)
- def finalize_import(self, config):
+ def finalize_import(self, _):
""" 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')
+ pass
def update_sql_functions(self, config):
"""
return LegacyICUNameAnalyzer(self.dsn, ICUNameProcessor(self.naming_rules))
- # pylint: disable=missing-format-attribute
+
def _install_php(self, phpdir):
""" Install the php script for the tokenizer.
"""
php_file = self.data_dir / "tokenizer.php"
- php_file.write_text(dedent("""\
+ php_file.write_text(dedent(f"""\
<?php
- @define('CONST_Max_Word_Frequency', {0.max_word_frequency});
- @define('CONST_Term_Normalization_Rules', "{0.term_normalization}");
- @define('CONST_Transliteration', "{0.naming_rules.search_rules}");
- require_once('{1}/tokenizer/legacy_icu_tokenizer.php');
- """.format(self, phpdir)))
+ @define('CONST_Max_Word_Frequency', {self.max_word_frequency});
+ @define('CONST_Term_Normalization_Rules', "{self.term_normalization}");
+ @define('CONST_Transliteration', "{self.naming_rules.search_rules}");
+ require_once('{phpdir}/tokenizer/legacy_icu_tokenizer.php');"""))
def _save_config(self, config):
# copy them back into the word table
with CopyBuffer() as copystr:
- for args in words.items():
- copystr.add(*args)
+ for k, v in words.items():
+ copystr.add('w', k, {'count': v})
with conn.cursor() as cur:
copystr.copy_out(cur, 'word',
- columns=['word_token', 'search_name_count'])
+ columns=['type', 'word_token', 'info'])
cur.execute("""UPDATE word SET word_id = nextval('seq_word')
- WHERE word_id is null""")
+ WHERE word_id is null and type = 'w'""")
conn.commit()
The function is used for testing and debugging only
and not necessarily efficient.
"""
- tokens = {}
+ full_tokens = {}
+ partial_tokens = {}
for word in words:
if word.startswith('#'):
- tokens[word] = ' ' + self.name_processor.get_search_normalized(word[1:])
+ full_tokens[word] = self.name_processor.get_search_normalized(word[1:])
else:
- tokens[word] = self.name_processor.get_search_normalized(word)
+ partial_tokens[word] = self.name_processor.get_search_normalized(word)
with self.conn.cursor() as cur:
- cur.execute("""SELECT word_token, word_id
- FROM word, (SELECT unnest(%s::TEXT[]) as term) t
- WHERE word_token = t.term
- and class is null and country_code is null""",
- (list(tokens.values()), ))
+ cur.execute("""(SELECT word_token, word_id
+ FROM word WHERE word_token = ANY(%s) and type = 'W')
+ UNION
+ (SELECT word_token, word_id
+ FROM word WHERE word_token = ANY(%s) and type = 'w')""",
+ (list(full_tokens.values()),
+ list(partial_tokens.values())))
ids = {r[0]: r[1] for r in cur}
- return [(k, v, ids.get(v, None)) for k, v in tokens.items()]
+ return [(k, v, ids.get(v, None)) for k, v in full_tokens.items()] \
+ + [(k, v, ids.get(v, None)) for k, v in partial_tokens.items()]
@staticmethod
if to_delete:
cur.execute("""DELETE FROM WORD
- WHERE class ='P' and info->>'postcode' = any(%s)
+ WHERE type ='P' and info->>'postcode' = any(%s)
""", (to_delete, ))
copystr.copy_out(cur, 'word',
projects / nominatim.git / commitdiff