merge in the postcode and is_in terms into the address vector for search

[nominatim.git] / utils / specialphrases.php

diff --git a/utils/specialphrases.php b/utils/specialphrases.php
index 81d240a612c1dc77062023967fff7cfb88ea734a..f1a0d0d078035c75d4168922f2e9612cc7516592 100755 (executable)
--- a/utils/specialphrases.php
+++ b/utils/specialphrases.php
@@ -10,44 +10,13 @@
                 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 coutry codes and names'),
+                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);
 
                 array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create import script for search phrases '),
         );
         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 

-       $aLanguageIn = array(
-                       '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',
-                       'sv',
-                       'uk',
-                       'vi',
-               );
+               include(CONST_BasePath.'/settings/phrase_settings.php');
+

 
     if ($aCMDResult['countries']) {
         echo "select getorcreate_country(make_standard_name('uk'), 'gb');\n";
 
     if ($aCMDResult['countries']) {
         echo "select getorcreate_country(make_standard_name('uk'), 'gb');\n";


@@ -73,9 +42,9 @@
                        {
                                foreach($aMatches as $aMatch)
                                {
                        {
                                foreach($aMatches as $aMatch)
                                {

-                                       $sLabel = $aMatch[1];
-                                       $sClass = $aMatch[2];
-                                       $sType = $aMatch[3];
+                                       $sLabel = trim($aMatch[1]);
+                                       $sClass = trim($aMatch[2]);
+                                       $sType = trim($aMatch[3]);

                                        # hack around a bug where building=yes was imported with
                                        # quotes into the wiki
                                        $sType = preg_replace('/"/', '', $sType);
                                        # hack around a bug where building=yes was imported with
                                        # quotes into the wiki
                                        $sType = preg_replace('/"/', '', $sType);


@@ -84,7 +53,17 @@
                                                preg_match('/^\\w+$/', $sType) < 1) {
                                                trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
                                                exit;
                                                preg_match('/^\\w+$/', $sType) < 1) {
                                                trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
                                                exit;

-                                       }       
+                                       }
+                                       # blacklisting: disallow certain class/type combinations
+                                       if (isset($aTagsBlacklist[$sClass]) && in_array($sType, $aTagsBlacklist[$sClass])) {
+                                               # fwrite(STDERR, "Blacklisted: ".$sClass."/".$sType."\n");
+                                               continue;
+                                       }
+                                       # whitelisting: if class is in whitelist, allow only tags in the list
+                                       if (isset($aTagsWhitelist[$sClass])     && !in_array($sType, $aTagsWhitelist[$sClass])) {
+                                               # fwrite(STDERR, "Non-Whitelisted: ".$sClass."/".$sType."\n");
+                                               continue;
+                                       }

                                        $aPairs[$sClass.'|'.$sType] = array($sClass, $sType);
 
                                        switch(trim($aMatch[4]))
                                        $aPairs[$sClass.'|'.$sType] = array($sClass, $sType);
 
                                        switch(trim($aMatch[4]))


@@ -107,11 +86,6 @@
 
                foreach($aPairs as $aPair)
                {
 
                foreach($aPairs as $aPair)
                {

-                       if ($aPair[0] == 'yes') continue;
-                       if ($aPair[1] == 'yes') continue;
-                       if ($aPair[0] == 'highway') continue;
-                       if ($aPair[1] == 'highway') continue;
-

                        echo "create table place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." as ";
                        echo "select place_id as place_id,st_centroid(geometry) as centroid from placex where ";
                        echo "class = '".pg_escape_string($aPair[0])."' and type = '".pg_escape_string($aPair[1])."';\n";
                        echo "create table place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." as ";
                        echo "select place_id as place_id,st_centroid(geometry) as centroid from placex where ";
                        echo "class = '".pg_escape_string($aPair[0])."' and type = '".pg_escape_string($aPair[1])."';\n";