update Tiger import script to use 2014 data

[nominatim.git] / utils / imports.php

diff --git a/utils/imports.php b/utils/imports.php
index 7e9de8e4a33251288277f940bb33ddfd1afe3530..332e5b3131d04479f989e0f354cac303ec5461ee 100755 (executable)
--- a/utils/imports.php
+++ b/utils/imports.php
@@ -3,6 +3,7 @@
 
        require_once(dirname(dirname(__FILE__)).'/lib/init-cmd.php');
        ini_set('memory_limit', '800M');
+       $sDefaultDataPath = CONST_BasePath.'/data/tiger2014';
 
        $aCMDOptions = array(
                "Create and setup nominatim search system",
@@ -10,34 +11,51 @@
                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'),
+               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($aCMDResult['parse-tiger']))
        {
+               if (!file_exists($sDefaultDataPath)) mkdir($sDefaultDataPath);
+
                $sTempDir = tempnam('/tmp', 'tiger');
                unlink($sTempDir);
                mkdir($sTempDir);
 
-               foreach(glob($aCMDResult['parse-tiger'].'/??_*', GLOB_ONLYDIR) as $sStateFolder)
+               foreach(glob($aCMDResult['parse-tiger'].'/tl_20??_?????_edges.zip', 0) as $sImportFile)
                {
-                       foreach(glob($sStateFolder.'/?????_*', GLOB_ONLYDIR) as $sCountyFolder)
+                       set_time_limit(30);
+                       preg_match('#([0-9]{5})_(.*)#',basename($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
                        {
-                               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))
+                               $sParseCmd = CONST_BasePath.'/utils/tigerAddressImport.py '.$sShapeFile;
+                               exec($sParseCmd);
+                               $sOsmFile = $sTempDir.'/'.basename($sImportFile, '.zip').'.osm1.osm';
+                               if (!file_exists($sOsmFile))
+                               {
+                                       echo "Failed parse ($sImportFile)\n";
+                               }
+                               else
                                {
-                                       echo "Missing: $sImportFile\n";
+                                       copy($sOsmFile, $sDefaultDataPath.'/'.$sCountyID.'.sql');
                                }
-                               $sUnzipCmd = "unzip -d $sTempDir $sImportFile";
-var_dump($sUnzipCmd);
-exit;
-//                             exec($sUnzipCmd);
                        }
+                       // Cleanup
+                       foreach(glob($sTempDir.'/*') as $sTmpFile)
+                       {
+                               unlink($sTmpFile);
+                       }
+
                }
        }