Merge pull request #700 from lonvia/structured-address-info

[nominatim.git] / utils / imports.php

diff --git a/utils/imports.php b/utils/imports.php
index 7e9de8e4a33251288277f940bb33ddfd1afe3530..8dc2c8203b3309dcc677972cb64a9fea2019756b 100755 (executable)
--- a/utils/imports.php
+++ b/utils/imports.php
@@ -1,43 +1,52 @@
 #!/usr/bin/php -Cq
 <?php
 
-       require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
-       ini_set('memory_limit', '800M');
+require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
+require_once(CONST_BasePath.'/lib/init-cmd.php');
+ini_set('memory_limit', '800M');
 
-       $aCMDOptions = array(
-               "Create and setup nominatim search system",
-               array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
-               array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
-               array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
+$aCMDOptions
+ = array(
+    "Create and setup nominatim search system",
+    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('parse-tiger', '', 0, 1, 1, 1, 'realpath', 'Convert tigger edge files to nominatim sql import'),
-       );
-       getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
+    array('parse-tiger', '', 0, 1, 1, 1, 'realpath', 'Convert tiger edge files to nominatim sql import - datafiles from 2011 or later (source: edges directory of tiger data)'),
+   );
+getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
-       $bDidSomething = false;
 
-       if (isset($aCMDResult['parse-tiger']))
-       {
-               $sTempDir = tempnam('/tmp', 'tiger');
-               unlink($sTempDir);
-               mkdir($sTempDir);
+if (isset($aCMDResult['parse-tiger'])) {
+    if (!file_exists(CONST_Tiger_Data_Path)) mkdir(CONST_Tiger_Data_Path);
 
-               foreach(glob($aCMDResult['parse-tiger'].'/??_*', GLOB_ONLYDIR) as $sStateFolder)
-               {
-                       foreach(glob($sStateFolder.'/?????_*', GLOB_ONLYDIR) as $sCountyFolder)
-                       {
-                               preg_match('#([0-9]{5})_(.*)#',basename($sCountyFolder), $aMatch);
-                               $sCountyID = $aMatch[1];
-                               $sCountyName = str_replace('_', ' ', $aMatch[2]);
-                               $sImportFile = $sCountyFolder.'/tl_2009_'.$sCountyID.'_edges.zip';
-                               if (!file_exists($sImportFile))
-                               {
-                                       echo "Missing: $sImportFile\n";
-                               }
-                               $sUnzipCmd = "unzip -d $sTempDir $sImportFile";
-var_dump($sUnzipCmd);
-exit;
-//                             exec($sUnzipCmd);
-                       }
-               }
-       }
+    $sTempDir = tempnam('/tmp', 'tiger');
+    unlink($sTempDir);
+    mkdir($sTempDir);
+
+    foreach (glob($aCMDResult['parse-tiger'].'/tl_20??_?????_edges.zip', 0) as $sImportFile) {
+        set_time_limit(30);
+        preg_match('#([0-9]{5})_(.*)#', basename($sImportFile), $aMatch);
+        $sCountyID = $aMatch[1];
+        echo "Processing ".$sCountyID."...\n";
+        $sUnzipCmd = "unzip -d $sTempDir $sImportFile";
+        exec($sUnzipCmd);
+        $sShapeFile = $sTempDir.'/'.basename($sImportFile, '.zip').'.shp';
+        if (!file_exists($sShapeFile)) {
+            echo "Failed unzip ($sImportFile)\n";
+        } else {
+            $sParseCmd = CONST_BasePath.'/utils/tigerAddressImport.py '.$sShapeFile;
+            exec($sParseCmd);
+            $sOsmFile = $sTempDir.'/'.basename($sImportFile, '.zip').'.osm1.osm';
+            if (!file_exists($sOsmFile)) {
+                echo "Failed parse ($sImportFile)\n";
+            } else {
+                copy($sOsmFile, CONST_Tiger_Data_Path.'/'.$sCountyID.'.sql');
+            }
+        }
+        // Cleanup
+        foreach (glob($sTempDir.'/*') as $sTmpFile) {
+            unlink($sTmpFile);
+        }
+    }
+}