From: Sarah Hoffmann <lonvia@denofr.de>
Date: Tue, 20 Mar 2012 23:40:35 +0000 (+0100)
Subject: Black/whitelisting for special phrases
X-Git-Tag: deploy~693
X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/57e88c1ab367a65ebec1d2b394c17ead0a70ef5b

Black/whitelisting for special phrases

Allows to restrict the special phrases imported from the wiki.
Blacklist allows to exclude certain class/type combinations.
Whitelist allows to define an allowed subset of types for a class.
Adjust to your liking in settings/phrase_settings.php before running
the specialphrases script.
---

diff --git a/settings/phrase_settings.php b/settings/phrase_settings.php
new file mode 100644
index 00000000..4456316e
--- /dev/null
+++ b/settings/phrase_settings.php
@@ -0,0 +1,51 @@
+<?php
+# These settings control the import of special phrases from the wiki.
+
+# Languages to download the special phrases for.
+$aLanguageIn = array(
+			'af',
+			'ar',
+			'br',
+			'ca',
+			'cs',
+			'de',
+			'en',
+			'es',
+			'et',
+			'eu',
+			'fa',
+			'fi',
+			'fr',
+			'gl',
+			'hr',
+			'hu',
+			'ia',
+			'is',
+			'it',
+			'ja',
+			'mk',
+			'nl',
+			'no',
+			'pl',
+			'ps',
+			'pt',
+			'ru',
+			'sk',
+			'sv',
+			'uk',
+			'vi',
+		);
+
+# class/type combinations to exclude
+$aTagsBlacklist = array(
+  'boundary' => array('administrative'),
+  'place' => array('house', 'houses'),
+);
+# If a class is in the white list then all types will 
+# be ignored except the ones given in the list.
+# Also use this list to exclude an entire class from
+# special phrases.
+$aTagsWhitelist = array(
+  'highway' => array('bus_stop', 'rest_area', 'raceway'),
+  'building' => array(),
+);
diff --git a/utils/specialphrases.php b/utils/specialphrases.php
index 225d293c..c31ac36f 100755
--- a/utils/specialphrases.php
+++ b/utils/specialphrases.php
@@ -15,39 +15,6 @@
         );
         getCmdOpt($_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";
@@ -63,6 +30,7 @@
 
 	if ($aCMDResult['wiki-import'])
 	{
+		include(CONST_BasePath.'/settings/phrase_settings.php');
 		$aPairs = array();
 
 		foreach($aLanguageIn as $sLanguage)
@@ -84,7 +52,17 @@
 						preg_match('/^\\w+$/', $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]))