Black/whitelisting for special phrases

[nominatim.git] / utils / specialphrases.php

diff --git a/utils/specialphrases.php b/utils/specialphrases.php
index d928d3c2cea5bbb7835b9a7b266671e5e54f8d66..c31ac36f6f91b348d7c1afe04b8f3656baf3ad25 100755 (executable)
--- a/utils/specialphrases.php
+++ b/utils/specialphrases.php
@@ -10,47 +10,27 @@
                 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('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
+                array('countries', '', 0, 1, 0, 0, 'bool', 'Create import script for coutry codes and names'),
+                array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create import script for search phrases '),

         );
         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
         );
         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 

-       $aLanguageIn = array(
-                       'af',
-                       'ar',
-                       'br',
-                       'ca',
-                       'cs',
-                       'de',
-                       'en',
-                       'es',
-                       'et',
-                       'eu',
-                       'fa',
-                       'fi',
-                       'fr',
-                       'gl',
-                       'hr',
-                       'hu',
-                       'ia',
-                       'is',
-                       'it',
-                       'ja',
-                       'mk',
-                       'nl',
-                       'no',
-                       'pl',
-                       'ps',
-                       'pt',
-                       'ru',
-                       'sk',
-                       'sv',
-                       'uk',
-                       'vi',
-               );
+
+    if ($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'])
        {

+               include(CONST_BasePath.'/settings/phrase_settings.php');

                $aPairs = array();
 
                foreach($aLanguageIn as $sLanguage)
                $aPairs = array();
 
                foreach($aLanguageIn as $sLanguage)


@@ -72,7 +52,17 @@
                                                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]))


@@ -91,7 +81,7 @@
                        }
                }
 
                        }
                }
 

-        echo "create index idx_placex_classtype on placex (class, type);\n";
+        echo "create index idx_placex_classtype on placex (class, type);";

 
                foreach($aPairs as $aPair)
                {
 
                foreach($aPairs as $aPair)
                {


@@ -103,14 +93,14 @@
                        echo "class = '".pg_escape_string($aPair[0])."' and type = '".pg_escape_string($aPair[1])."';\n";
 
                        echo "CREATE INDEX idx_place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])."_centroid ";
                        echo "class = '".pg_escape_string($aPair[0])."' and type = '".pg_escape_string($aPair[1])."';\n";
 
                        echo "CREATE INDEX idx_place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])."_centroid ";

-                       echo "ON place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." USING GIST (centroid);\n";
+                       echo "ON place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." USING GIST (centroid) tablespace ssd;\n";

 
                        echo "CREATE INDEX idx_place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])."_place_id ";
 
                        echo "CREATE INDEX idx_place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])."_place_id ";

-                       echo "ON place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." USING btree(place_id);\n";
+                       echo "ON place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." USING btree(place_id) tablespace ssd;\n";

 
             echo "GRANT SELECT ON place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." TO \"www-data\";\n";
 
                }
 
 
             echo "GRANT SELECT ON place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($aPair[1])." TO \"www-data\";\n";
 
                }
 

-        echo "drop index idx_placex_classtype;\n";
+        echo "drop index idx_placex_classtype;";

        }
        }