X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/832547f192904a9ec92e173c27a91e0874fcc757..caa8210112d608609f6825dff6893b6e9c83b7c6:/wikidata/parse.php?ds=sidebyside

diff --git a/wikidata/parse.php b/wikidata/parse.php
index 8a59c3d8..1bef50ec 100755
--- a/wikidata/parse.php
+++ b/wikidata/parse.php
@@ -1,14 +1,14 @@
 #!/usr/bin/php -Cq
 <?php
 
-$hFile = @fopen("wikidatawiki-20130623-pages-articles.xml", "r");
+$hFile = @fopen('wikidatawiki-20130623-pages-articles.xml', 'r');
 
-$hFileEntity = fopen("entity.csv", "w");
-$hFileEntityLabel = fopen("entity_label.csv", "w");
-$hFileEntityDescription = fopen("entity_description.csv", "w");
-$hFileEntityAlias = fopen("entity_alias.csv", "w");
-$hFileEntityLink = fopen("entity_link.csv", "w");
-$hFileEntityProperty = fopen("entity_property.csv", "w");
+$hFileEntity = fopen('entity.csv', 'w');
+$hFileEntityLabel = fopen('entity_label.csv', 'w');
+$hFileEntityDescription = fopen('entity_description.csv', 'w');
+$hFileEntityAlias = fopen('entity_alias.csv', 'w');
+$hFileEntityLink = fopen('entity_link.csv', 'w');
+$hFileEntityProperty = fopen('entity_property.csv', 'w');
 
 $iCount = 0;
 
@@ -20,14 +20,11 @@ if ($hFile) {
     while (($sLine = fgets($hFile, 4000000)) !== false) {
         if (substr($sLine, 0, 11) == '    <title>') {
             $sTitle = substr($sLine, 11, -9);
-        }
-        else if (substr($sLine, 0, 8) == '    <ns>') {
+        } elseif (substr($sLine, 0, 8) == '    <ns>') {
             $iNS = (int)substr($sLine, 8, -6);
-        }
-        else if (substr($sLine, 0, 8) == '    <id>') {
+        } elseif (substr($sLine, 0, 8) == '    <id>') {
             $iID = (int)substr($sLine, 8, -6);
-        }
-        else if (substr($sLine, 0, 33) == '      <text xml:space="preserve">') {
+        } elseif (substr($sLine, 0, 33) == '      <text xml:space="preserve">') {
             if ($iNS == -2) continue;
             if ($iNS == -1) continue;
             if ($iNS == 1) continue;
@@ -53,7 +50,7 @@ if ($hFile) {
             $sText = html_entity_decode(substr($sLine, 33, -8), ENT_COMPAT, 'UTF-8');
             $aArticle = json_decode($sText, true);
 
-            if (array_diff(array_keys($aArticle), array("label", "description", "aliases", "links", "entity", "claims", "datatype")) != array()) {
+            if (array_diff(array_keys($aArticle), array('label', 'description', 'aliases', 'links', 'entity', 'claims', 'datatype')) != array()) {
                 // DEBUG
                 var_dump($sTitle);
                 var_dump(array_keys($aArticle));
@@ -63,118 +60,121 @@ if ($hFile) {
 
             $iPID = $iQID = null;
             if ($aArticle['entity'][0] == 'p') {
-                $iPID = (int)substr($aArticle['entity'], 1);
-            } else if ($aArticle['entity'][0] == 'q') {
-                $iQID = (int)substr($aArticle['entity'], 1);
+                $iPID = (int) substr($aArticle['entity'], 1);
+            } elseif ($aArticle['entity'][0] == 'q') {
+                $iQID = (int) substr($aArticle['entity'], 1);
             } else {
                 continue;
             }
 
-            echo ".";
+            echo '.';
 
-            fputcsv($hFileEntity, array($iID,$sTitle,$iPID,$iQID,@$aArticle['datatype']));
+            fputcsv($hFileEntity, array($iID, $sTitle, $iPID, $iQID, @$aArticle['datatype']));
 
-            foreach($aArticle['label'] as $sLang => $sLabel) {
-                fputcsv($hFileEntityLabel, array($iID,$sLang,$sLabel));
+            foreach ($aArticle['label'] as $sLang => $sLabel) {
+                fputcsv($hFileEntityLabel, array($iID, $sLang, $sLabel));
                 // echo "insert into entity_label values (".$iID.",'".pg_escape_string($sLang)."','".pg_escape_string($sLabel)."');\n";
             }
 
-            foreach($aArticle['description'] as $sLang => $sLabel) {
-                fputcsv($hFileEntityDescription, array($iID,$sLang,$sLabel));
+            foreach ($aArticle['description'] as $sLang => $sLabel) {
+                fputcsv($hFileEntityDescription, array($iID, $sLang, $sLabel));
                 // echo "insert into entity_description values (".$iID.",'".pg_escape_string($sLang)."','".pg_escape_string($sLabel)."');\n";
             }
 
-            foreach($aArticle['aliases'] as $sLang => $aLabels) {
+            foreach ($aArticle['aliases'] as $sLang => $aLabels) {
                 $aUniqueAlias = array();
-                foreach($aLabels as $sLabel) {
+                foreach ($aLabels as $sLabel) {
                     if (!isset($aUniqueAlias[$sLabel]) && $sLabel) {
-                    fputcsv($hFileEntityAlias, array($iID,$sLang,$sLabel));
-                    // echo "insert into entity_alias values (".$iID.",'".pg_escape_string($sLang)."','".pg_escape_string($sLabel)."');\n";
-                    $aUniqueAlias[$sLabel] = true;
+                        fputcsv($hFileEntityAlias, array($iID, $sLang, $sLabel));
+                        // echo "insert into entity_alias values (".$iID.",'".pg_escape_string($sLang)."','".pg_escape_string($sLabel)."');\n";
+                        $aUniqueAlias[$sLabel] = true;
+                    }
                 }
             }
-        }
-
-        foreach($aArticle['links'] as $sLang => $sLabel) {
-            fputcsv($hFileEntityLink, array($iID,$sLang,$sLabel));
-            // echo "insert into entity_link values (".$iID.",'".pg_escape_string($sLang)."','".pg_escape_string($sLabel)."');\n";
-        }
-
-
-        if (isset($aArticle['claims'])) {
-
-            foreach($aArticle['claims'] as $iClaim => $aClaim) {
-
-                $bFail = false;
-            if ($aClaim['m'][0] == 'novalue') continue;
-            if ($aClaim['m'][0] == 'somevalue') continue;
-            $iPID = (int)$aClaim['m'][1];
-            if ($aClaim['m'][0] != 'value') $bFail = true;
-            if ($aClaim['m'][2]== 'wikibase-entityid') {
-
-                if ($aClaim['m'][3]['entity-type'] != 'item') $bFail = true;
-                fputcsv($hFileEntityProperty, array($iID,$iClaim,$iPID,null,$aClaim['m'][3]['numeric-id'],null,null));
-                // echo "insert into entity_property values (nextval('seq_entity_property'),".$iID.",".$iPID.",null,".$aClaim['m'][3]['numeric-id'].",null);\n";
-
-            } elseif ($aClaim['m'][2] == 'globecoordinate') {
-
-                if ($aClaim['m'][3]['globe'] != 'http://www.wikidata.org/entity/Q2') $bFail = true;
-                fputcsv($hFileEntityProperty, array($iID,$iClaim,$iPID,null,null,"SRID=4326;POINT(".((float)$aClaim['m'][3]['longitude'])." ".((float)$aClaim['m'][3]['latitude']).")",null));
-                // echo "insert into entity_property values (nextval('seq_entity_property'),".$iID.",".$iPID.",null,null,ST_SetSRID(ST_MakePoint(".((float)$aClaim['m'][3]['longitude']).", ".((float)$aClaim['m'][3]['latitude'])."),4326));\n";
-
-            } elseif ($aClaim['m'][2] == 'time') {
-                // TODO!
-/*
-        if ($aClaim['m'][3]['calendarmodel'] == 'http://www.wikidata.org/entity/Q1985727') {
-            // Gregorian
-            if (preg_match('#(\\+|-)0*([0-9]{4})-([0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2})Z#', $aClaim['m'][3]['time'], $aMatch)) {
-                if ((int)$aMatch[2] < 4700 && ) {
-                    $sDateString = $aMatch[2].'-'.$aMatch[3].($aClaim['m'][3]['timezone']>=0?'+':'').$aClaim['m'][3]['timezone'].($aMatch[1]=='-'?' bc':'');
-                    fputcsv($hFileEntityProperty, array($iID,$iClaim,$iPID,null,null,null,$sDateString));
-                }
-            } else {
-//              $bFail = true;
-            }
-        } elseif ( $aClaim['m'][3]['calendarmodel'] != 'http://www.wikidata.org/entity/Q1985786') {
-/ *
-            // Julian
-            if (preg_match('#(\\+|-)0*([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}:[0-9]{2}:[0-9]{2})Z#', $aClaim['m'][3]['time'], $aMatch)) {
-var_dump($aMatch);
-exit;
-$iDayCount = juliantojd(2, 11, 1732);
-var_dump($iDayCount, jdtogregorian($iDayCount));
-            } else {
-                $bFail = true;
-exit;
-            }
-exit;
-* /
-        } else {
-//          $bFail = true;
-        }
-*/
-            } elseif ($aClaim['m'][2] == 'string') {
-
-                // echo "insert into entity_property values (nextval('seq_entity_property'),".$iID.",".$iPID.",'".pg_escape_string($aClaim['m'][3])."',null,null);\n";
-                fputcsv($hFileEntityProperty, array($iID,$iClaim,$iPID,$aClaim['m'][3],null,null,null));
-
-            } else {
-
-                $bFail = true;
 
+            foreach ($aArticle['links'] as $sLang => $sLabel) {
+                fputcsv($hFileEntityLink, array($iID, $sLang, $sLabel));
+                // echo "insert into entity_link values (".$iID.",'".pg_escape_string($sLang)."','".pg_escape_string($sLabel)."');\n";
             }
 
-            // Don't care about sources:    if ($aClaim['refs'] != array()) $bFail = true;
-
-            if ($bFail) {
-                var_dump($sTitle);
-                var_dump($aClaim);
-            } else {
-                // process
-            }
-
-        }
 
+            if (isset($aArticle['claims'])) {
+                //
+                foreach ($aArticle['claims'] as $iClaim => $aClaim) {
+                    //
+                    $bFail = false;
+                    if ($aClaim['m'][0] == 'novalue') continue;
+                    if ($aClaim['m'][0] == 'somevalue') continue;
+                    $iPID = (int)$aClaim['m'][1];
+                    if ($aClaim['m'][0] != 'value') $bFail = true;
+                    if ($aClaim['m'][2]== 'wikibase-entityid') {
+                        //
+                        if ($aClaim['m'][3]['entity-type'] != 'item') $bFail = true;
+                        fputcsv($hFileEntityProperty, array($iID, $iClaim, $iPID, null, $aClaim['m'][3]['numeric-id'], null, null));
+                        // echo "insert into entity_property values (nextval('seq_entity_property'),".$iID.",".$iPID.",null,".$aClaim['m'][3]['numeric-id'].",null);\n";
+                    } elseif ($aClaim['m'][2] == 'globecoordinate') {
+                        //
+                        if ($aClaim['m'][3]['globe'] != 'http://www.wikidata.org/entity/Q2') $bFail = true;
+                        fputcsv(
+                            $hFileEntityProperty,
+                            array(
+                             $iID,
+                             $iClaim,
+                             $iPID,
+                             null,
+                             null,
+                             'SRID=4326;POINT('.((float) $aClaim['m'][3]['longitude']).' '.((float)$aClaim['m'][3]['latitude']).')', null
+                            )
+                        );
+                        /* echo "insert into entity_property values (nextval('seq_entity_property'),";
+                         * echo $iID.",".$iPID.",null,null,ST_SetSRID(ST_MakePoint(".((float)$aClaim['m'][3]['longitude']);
+                         * echo ", ".((float)$aClaim['m'][3]['latitude'])."),4326));\n";
+                         */
+                    } elseif ($aClaim['m'][2] == 'time') {
+                        // TODO!
+                        /*
+                        if ($aClaim['m'][3]['calendarmodel'] == 'http://www.wikidata.org/entity/Q1985727') {
+                            // Gregorian
+                            if (preg_match('#(\\+|-)0*([0-9]{4})-([0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2})Z#', $aClaim['m'][3]['time'], $aMatch)) {
+                                if ((int)$aMatch[2] < 4700 && ) {
+                                    $sDateString = $aMatch[2].'-'.$aMatch[3].($aClaim['m'][3]['timezone']>=0?'+':'').$aClaim['m'][3]['timezone'].($aMatch[1]=='-'?' bc':'');
+                                    fputcsv($hFileEntityProperty, array($iID,$iClaim,$iPID,null,null,null,$sDateString));
+                                }
+                            } else {
+                                // $bFail = true;
+                            }
+                        } elseif ( $aClaim['m'][3]['calendarmodel'] != 'http://www.wikidata.org/entity/Q1985786') {
+                            // Julian
+                            if (preg_match('#(\\+|-)0*([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}:[0-9]{2}:[0-9]{2})Z#', $aClaim['m'][3]['time'], $aMatch)) {
+                                var_dump($aMatch);
+                                exit;
+                                $iDayCount = juliantojd(2, 11, 1732);
+                                var_dump($iDayCount, jdtogregorian($iDayCount));
+                        } else {
+                            $bFail = true;
+                            exit;
+                        }
+                        exit;
+                    } else {
+                        // $bFail = true;
+                    }
+                    */
+                    } elseif ($aClaim['m'][2] == 'string') {
+                        // echo "insert into entity_property values (nextval('seq_entity_property'),".$iID.",".$iPID.",'".pg_escape_string($aClaim['m'][3])."',null,null);\n";
+                        fputcsv($hFileEntityProperty, array($iID, $iClaim, $iPID, $aClaim['m'][3], null, null, null));
+                    } else {
+                        $bFail = true;
+                    }
+
+                    // Don't care about sources:    if ($aClaim['refs'] != array()) $bFail = true;
+
+                    if ($bFail) {
+                        var_dump($sTitle);
+                        var_dump($aClaim);
+                    } else {
+                        // process
+                    }
+                }
             }
         }
     }