delete more hardcoded rules for special phrases

[nominatim.git] / utils / specialphrases.php

diff --git a/utils/specialphrases.php b/utils/specialphrases.php
index 5b3abe405120f1a913724b903d02340898d45de5..f1a0d0d078035c75d4168922f2e9612cc7516592 100755 (executable)
--- a/utils/specialphrases.php
+++ b/utils/specialphrases.php
@@ -3,50 +3,32 @@
 
         require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
         ini_set('memory_limit', '800M');
+        ini_set('display_errors', 'stderr');
 
         $aCMDOptions = array(
                 "Import and export special phrases",
                 array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
                 array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
                 array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
-
-                array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
+                array('countries', '', 0, 1, 0, 0, 'bool', 'Create import script for country codes and names'),
+                array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create import script for search phrases '),
         );
         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
-       $aLanguageIn = array(
-                       'af',
-                       'ar',
-                       'br',
-                       'ca',
-                       'cs',
-                       'de',
-                       'en',
-                       'es',
-                       'et',
-                       'eu',
-                       'fa',
-                       'fi',
-                       'fr',
-                       'gl',
-                       'hr',
-                       'hu',
-                       'ia',
-                       'is',
-                       'it',
-                       'ja',
-                       'mk',
-                       'nl',
-                       'no',
-                       'pl',
-                       'ps',
-                       'pt',
-                       'ru',
-                       'sk',
-                       'sv',
-                       'uk',
-                       'vi',
-               );
+               include(CONST_BasePath.'/settings/phrase_settings.php');
+
+
+    if ($aCMDResult['countries']) {
+        echo "select getorcreate_country(make_standard_name('uk'), 'gb');\n";
+        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 ($aCMDResult['wiki-import'])
        {
@@ -60,9 +42,28 @@
                        {
                                foreach($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('/"/', '', $sType);
+                                       # sanity check, in case somebody added garbage in the wiki
+                                       if (preg_match('/^\\w+$/', $sClass) < 1 ||
+                                               preg_match('/^\\w+$/', $sType) < 1) {
+                                               trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
+                                               exit;
+                                       }
+                                       # 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]))
@@ -85,8 +86,6 @@
 
                foreach($aPairs as $aPair)
                {
-                       if ($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";