<?php
   /*
   Shaun Osborne, smo30@cam.ac.uk, 15Dec04
   This is about the simplest List Records OAI harvester there probably is.
   It is php command line intended. Tested only on windows PHP 5 (it needs simplexml)
   Its only purpose was for me to do harvesting on my data provider to test some problems I was having
   when attempting to do complete harvests; thats what it does - it just keeps getting and storing xml responses
   until the data provider runs out or errors.
   Fill out oaibase, metadataprefix below and you're away.
   Optionally change requestdelay and datapath (default is oaidata dir where script is) and from and to dates
   
   a little warning.. the script changes php.ini value for default_socket_timeout so if you crash out of the script
   your php.ini may be left with a value for this you do not want...

   */

  #user changeable variables
  $oaibase = "http://www.fitzmuseum.cam.ac.uk/oai/wwwopac.exe?";
  $metadataprefix="oai_24hm"; # what metadata prefix do you want ListRecords to use?
  $requestdelay=45;           # delay between each request x = seconds
  $datapath="oaidata\\";     # remember double \'s for windoze and the trailing double \'s e.g. c:\\temp\\oaidata\\
                            # can be relative to where php script is e.g oaidata\\
  $fromdate="";            #OAI from (date) yyyy-mm-dd) value
  $todate="";             #OAI to (date) yyyy-mm-dd) value
  $responsetimeout=600;  # socket timeout (seconds) on http response from repository
  
  # shouldn't need to change these
  $firstresponsename = "_firstresponse.xml";
  $listrecfile = $firstresponsename;
  $restoken=$listrecfile;
  $oaierror="";
  $x=0;
  $version="1.0";
  
#this is needed to allow for possible long response times for get_file_contents from the http calls to OAI repository
$olddstvalue = ini_set("default_socket_timeout",$responsetimeout);
if($olddstvalue == FALSE){echo "WARNING: failed to set default_socket_timeout ($responsetimeout)\n";}

#loop until we no longer get a resumption token or till error
while(strcmp("",$restoken) and !strcmp("",$oaierror))
{
if(!strcmp($restoken,$firstresponsename)) #build the first request from variables above
  {$url = $oaibase."verb=ListRecords&metadataPrefix=".$metadataprefix;
   if(strcmp("",$fromdate)){$url=$url."&from=".$fromdate;}
   if(strcmp("",$todate)){$url=$url."&to=".$todate;}
   echo "[v".$version."]Starting OAI harvest using:\n".$url."\n\n";
  }
else # else just supply the resumption token
  {
    $url = $oaibase."verb=ListRecords&resumptionToken=".$restoken;
  }
$x++;    # just count 'em
echo $x." getting [".$restoken."].. ";

$results = file_get_contents($url);    #get the data
if(!$results){echo "ERROR: from file_get_contents on url:\n".$url."\n"; myexit();}

$fh = fopen($datapath.$listrecfile,"w");
if(!$fh){echo "Can't open temporary file!\n"; myexit();}
fwrite($fh, $results);                 # write it to file
fclose($fh);
echo "done.. ";

$xml = simplexml_load_string($results);
if($xml == FALSE){echo "ERROR: loading xml string";myexit();}
$restoken = $xml->ListRecords->resumptionToken; # get the resumption token
$oaierror = $xml->error; $oairesponsedate=$xml->responseDate;
$listrecfile = $restoken.".xml";
#echo "\nDEBUG[[".$restoken."]][[".$oaierror."]][[".$listrecfile."]]\n";;
if($requestdelay > 0){echo "waiting($requestdelay)";}
sleep($requestdelay); # wait x seconds if configured to do so
$fh=null;$results=null; $xml=null; #cleanup
echo "\n";
} /* end of while */

/* we might get here out of while loop due to error so check and report any */
if(strcmp("",$oaierror))
{ echo "\n** ERROR STOP\n";
  echo "\nOAI error: ".$oaierror;
  echo "\nOAI ResponseDate: ".$oairesponsedate;
  echo "\nURL was: ".$url;
}

myexit();


/********************/
function myexit()
{
global $olddstvalue;
# any global cleanup do it here

#set default_socket_timeout back to what it was
#echo "resetting default_socket_timeout to $olddstvalue\n";
if($olddstvalue != FALSE)
  {
   $retvalue = ini_set("default_socket_timeout",$olddstvalue);
   if($retvalue == FALSE){echo "WARNING: failed to reset default_socket_timeout ($olddstvalue)\n";}
  }
exit;
}

?>