X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/01f4723b7bcd74bda121d8268d6655940dea061f..8106ca465b556c76f26f8e60ad84cc80e509bdfd:/utils/specialphrases.php

diff --git a/utils/specialphrases.php b/utils/specialphrases.php
old mode 100755
new mode 100644
index d928d3c2..9e6c9d76
--- a/utils/specialphrases.php
+++ b/utils/specialphrases.php
@@ -1,116 +1,153 @@
-#!/usr/bin/php -Cq
 <?php
 
-        require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
-        ini_set('memory_limit', '800M');
-        ini_set('display_errors', 'stderr');
+require_once(CONST_BasePath.'/lib/init-cmd.php');
+ini_set('memory_limit', '800M');
+ini_set('display_errors', 'stderr');
 
-        $aCMDOptions = array(
-                "Import and export special phrases",
-                array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
-                array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
-                array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
+$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 import script for search phrases '),
+  );
+getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
-                array('wiki-import', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
+include(CONST_Phrase_Config);
+
+if ($aCMDResult['wiki-import']) {
+    $oNormalizer = Transliterator::createFromRules(CONST_Term_Normalization_Rules);
+    $aPairs = array();
+
+    $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) {
+        $sURL = 'https://wiki.openstreetmap.org/wiki/Special:Export/Nominatim/Special_Phrases/'.strtoupper($sLanguage);
+        $sWikiPageXML = file_get_contents($sURL);
+
+        if (!preg_match_all(
+            '#\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([^|]+) \\|\\| ([\\-YN])#',
+            $sWikiPageXML,
+            $aMatches,
+            PREG_SET_ORDER
+        )) {
+            continue;
+        }
+
+        foreach ($aMatches as $aMatch) {
+            $sLabel = trim($aMatch[1]);
+            if ($oNormalizer !== null) {
+                $sTrans = pg_escape_string($oNormalizer->transliterate($sLabel));
+            } else {
+                $sTrans = null;
+            }
+            $sClass = trim($aMatch[2]);
+            $sType = trim($aMatch[3]);
+            // hack around a bug where building=yes was imported with
+            // quotes into the wiki
+            $sType = preg_replace('/(&quot;|")/', '', $sType);
+            // sanity check, in case somebody added garbage in the wiki
+            if (preg_match('/^\\w+$/', $sClass) < 1
+                || preg_match('/^\\w+$/', $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])) {
+                case 'near':
+                    printf(
+                        "SELECT getorcreate_amenityoperator(make_standard_name('%s'), '%s', '%s', '%s', 'near');\n",
+                        pg_escape_string($sLabel),
+                        $sTrans,
+                        $sClass,
+                        $sType
+                    );
+                    break;
+                case 'in':
+                    printf(
+                        "SELECT getorcreate_amenityoperator(make_standard_name('%s'), '%s', '%s', '%s', 'in');\n",
+                        pg_escape_string($sLabel),
+                        $sTrans,
+                        $sClass,
+                        $sType
+                    );
+                    break;
+                default:
+                    printf(
+                        "SELECT getorcreate_amenity(make_standard_name('%s'), '%s', '%s', '%s');\n",
+                        pg_escape_string($sLabel),
+                        $sTrans,
+                        $sClass,
+                        $sType
+                    );
+                    break;
+            }
+        }
+    }
+
+    echo 'CREATE INDEX idx_placex_classtype ON placex (class, type);';
+
+    foreach ($aPairs as $aPair) {
+        $sql_tablespace = CONST_Tablespace_Aux_Data ? ' TABLESPACE '.CONST_Tablespace_Aux_Data : '';
+
+        printf(
+            'CREATE TABLE place_classtype_%s_%s'
+            . $sql_tablespace
+            . ' AS'
+            . ' SELECT place_id AS place_id,st_centroid(geometry) AS centroid FROM placex'
+            . " WHERE class = '%s' AND type = '%s'"
+            . ";\n",
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1]),
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1])
+        );
+
+        printf(
+            'CREATE INDEX idx_place_classtype_%s_%s_centroid'
+            . ' ON place_classtype_%s_%s USING GIST (centroid)'
+            . $sql_tablespace
+            . ";\n",
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1]),
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1])
+        );
+
+        printf(
+            'CREATE INDEX idx_place_classtype_%s_%s_place_id'
+            . ' ON place_classtype_%s_%s USING btree(place_id)'
+            . $sql_tablespace
+            . ";\n",
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1]),
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1])
+        );
+
+        printf(
+            'GRANT SELECT ON place_classtype_%s_%s TO "%s"'
+            . ";\n",
+            pg_escape_string($aPair[0]),
+            pg_escape_string($aPair[1]),
+            CONST_Database_Web_User
         );
-        getCmdOpt($_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['wiki-import'])
-	{
-		$aPairs = array();
-
-		foreach($aLanguageIn 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 = $aMatch[1];
-					$sClass = $aMatch[2];
-					$sType = $aMatch[3];
-					# hack around a bug where building=yes was imported with
-					# quotes into the wiki
-					$sType = preg_replace('/&quot;/', '', $sType);
-					# sanity check, in case somebody added garbage in the wiki
-					if (preg_match('/^\\w+$/', $sClass) < 1 ||
-						preg_match('/^\\w+$/', $sType) < 1) {
-						trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
-						exit;
-					}	
-					$aPairs[$sClass.'|'.$sType] = array($sClass, $sType);
-
-					switch(trim($aMatch[4]))
-					{
-					case 'near':
-						echo "select getorcreate_amenityoperator(make_standard_name('".pg_escape_string($sLabel)."'), '$sClass', '$sType', 'near');\n";
-						break;
-					case 'in':
-						echo "select getorcreate_amenityoperator(make_standard_name('".pg_escape_string($sLabel)."'), '$sClass', '$sType', 'in');\n";
-						break;
-					default:
-						echo "select getorcreate_amenity(make_standard_name('".pg_escape_string($sLabel)."'), '$sClass', '$sType');\n";
-						break;
-					}
-				}
-			}
-		}
-
-        echo "create index idx_placex_classtype on placex (class, type);\n";
-
-		foreach($aPairs as $aPair)
-		{
-			if ($aPair[1] == 'highway') continue;
-
-			echo "drop table if exists place_classtype_".pg_escape_string($aPair[0])."_".pg_escape_string($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";
-
-			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 "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 "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;';
+}