add normalized version of special search terms on import

[nominatim.git] / utils / specialphrases.php

diff --git a/utils/specialphrases.php b/utils/specialphrases.php
index 652230fc7dd239f39033aca3126a352b31a0c713..156169761fef71102ff21f47a802375f5796064b 100755 (executable)
--- a/utils/specialphrases.php
+++ b/utils/specialphrases.php
@@ -12,67 +12,60 @@ $aCMDOptions
    array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
    array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
    array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
    array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
    array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
    array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),

-   array('countries', '', 0, 1, 0, 0, 'bool', 'Create import script for country codes and names'),

    array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create import script for search phrases '),
   );
 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
 include(CONST_InstallPath.'/settings/phrase_settings.php');
 
    array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create import script for search phrases '),
   );
 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
 include(CONST_InstallPath.'/settings/phrase_settings.php');
 

-
-if ($aCMDResult['countries']) {
-    echo "select getorcreate_country(make_standard_name('uk'), 'gb');\n";
-    echo "select getorcreate_country(make_standard_name('united states'), 'us');\n";
-    echo "select count(*) from (select getorcreate_country(make_standard_name(country_code), country_code) from country_name where country_code is not null) as x;\n";
-
-    echo "select count(*) from (select getorcreate_country(make_standard_name(get_name_by_language(country_name.name,ARRAY['name'])), country_code) from country_name where get_name_by_language(country_name.name, ARRAY['name']) is not null) as x;\n";
-    foreach ($aLanguageIn as $sLanguage) {
-        echo "select count(*) from (select getorcreate_country(make_standard_name(get_name_by_language(country_name.name,ARRAY['name:".$sLanguage."'])), country_code) from country_name where get_name_by_language(country_name.name, ARRAY['name:".$sLanguage."']) is not null) as x;\n";
-    }
-}
-

 if ($aCMDResult['wiki-import']) {
 if ($aCMDResult['wiki-import']) {

+    $oNormalizer = Transliterator::createFromRules(":: NFD (); [:Nonspacing Mark:] >;  :: lower (); [[:Punctuation:][:Space:]]+ > ' '; :: NFC ();");

     $aPairs = array();
 
     $aPairs = array();
 

-    foreach ($aLanguageIn as $sLanguage) {
+    $sLanguageIn = CONST_Languages ? CONST_Languages :
+        ('af,ar,br,ca,cs,de,en,es,et,eu,fa,fi,fr,gl,hr,hu,'.
+         'ia,is,it,ja,mk,nl,no,pl,ps,pt,ru,sk,sl,sv,uk,vi');
+
+    foreach (explode(',', $sLanguageIn) as $sLanguage) {

         $sURL = 'http://wiki.openstreetmap.org/wiki/Special:Export/Nominatim/Special_Phrases/'.strtoupper($sLanguage);
         $sWikiPageXML = file_get_contents($sURL);
         if (preg_match_all('#\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([\\-YN])#', $sWikiPageXML, $aMatches, PREG_SET_ORDER)) {
             foreach ($aMatches as $aMatch) {
                 $sLabel = trim($aMatch[1]);
         $sURL = 'http://wiki.openstreetmap.org/wiki/Special:Export/Nominatim/Special_Phrases/'.strtoupper($sLanguage);
         $sWikiPageXML = file_get_contents($sURL);
         if (preg_match_all('#\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([\\-YN])#', $sWikiPageXML, $aMatches, PREG_SET_ORDER)) {
             foreach ($aMatches as $aMatch) {
                 $sLabel = trim($aMatch[1]);

+                $sTrans = pg_escape_string($oNormalizer->transliterate($sLabel));

                 $sClass = trim($aMatch[2]);
                 $sType = trim($aMatch[3]);
                 $sClass = trim($aMatch[2]);
                 $sType = trim($aMatch[3]);

-                # hack around a bug where building=yes was imported with
-                # quotes into the wiki
+                // hack around a bug where building=yes was imported with
+                // quotes into the wiki

                 $sType = preg_replace('/"/', '', $sType);
                 $sType = preg_replace('/"/', '', $sType);

-                # sanity check, in case somebody added garbage in the wiki
+                // sanity check, in case somebody added garbage in the wiki

                 if (preg_match('/^\\w+$/', $sClass) < 1
                     || preg_match('/^\\w+$/', $sType) < 1
                 ) {
                     trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
                     exit;
                 }
                 if (preg_match('/^\\w+$/', $sClass) < 1
                     || preg_match('/^\\w+$/', $sType) < 1
                 ) {
                     trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
                     exit;
                 }

-                # blacklisting: disallow certain class/type combinations
+                // blacklisting: disallow certain class/type combinations

                 if (isset($aTagsBlacklist[$sClass]) && in_array($sType, $aTagsBlacklist[$sClass])) {
                 if (isset($aTagsBlacklist[$sClass]) && in_array($sType, $aTagsBlacklist[$sClass])) {

-                    # fwrite(STDERR, "Blacklisted: ".$sClass."/".$sType."\n");
+                    // fwrite(STDERR, "Blacklisted: ".$sClass."/".$sType."\n");

                     continue;
                 }
                     continue;
                 }

-                # whitelisting: if class is in whitelist, allow only tags in the list
+                // whitelisting: if class is in whitelist, allow only tags in the list

                 if (isset($aTagsWhitelist[$sClass]) && !in_array($sType, $aTagsWhitelist[$sClass])) {
                 if (isset($aTagsWhitelist[$sClass]) && !in_array($sType, $aTagsWhitelist[$sClass])) {

-                    # fwrite(STDERR, "Non-Whitelisted: ".$sClass."/".$sType."\n");
+                    // fwrite(STDERR, "Non-Whitelisted: ".$sClass."/".$sType."\n");

                     continue;
                 }
                 $aPairs[$sClass.'|'.$sType] = array($sClass, $sType);
 
                 switch (trim($aMatch[4])) {
                     case 'near':
                     continue;
                 }
                 $aPairs[$sClass.'|'.$sType] = array($sClass, $sType);
 
                 switch (trim($aMatch[4])) {
                     case 'near':

-                        echo "select getorcreate_amenityoperator(make_standard_name('".pg_escape_string($sLabel)."'), '$sClass', '$sType', 'near');\n";
+                        echo "select getorcreate_amenityoperator(make_standard_name('".pg_escape_string($sLabel)."'), '$sTrans', '$sClass', '$sType', 'near');\n";

                         break;
                     case 'in':
                         break;
                     case 'in':

-                        echo "select getorcreate_amenityoperator(make_standard_name('".pg_escape_string($sLabel)."'), '$sClass', '$sType', 'in');\n";
+                        echo "select getorcreate_amenityoperator(make_standard_name('".pg_escape_string($sLabel)."'), '$sTrans', '$sClass', '$sType', 'in');\n";

                         break;
                     default:
                         break;
                     default:

-                        echo "select getorcreate_amenity(make_standard_name('".pg_escape_string($sLabel)."'), '$sClass', '$sType');\n";
+                        echo "select getorcreate_amenity(make_standard_name('".pg_escape_string($sLabel)."'), '$sTrans', '$sClass', '$sType');\n";

                         break;
                 }
             }
                         break;
                 }
             }