HTTP XML web search interface

<?php
////////////////////////////////////////////////////////////////////////
//
// Copyright (c) 2001, SearchHippo.com 
// [email protected]
// http://www.searchhippo.com/
//
// Note: use of this code indicates agreement to our attribution terms
// Please see http://www.searchhippo.com/partner.php
//
// Rev 1.0 - 4/20/2K+1
//
// Feel free to hack this up however you like.  The only requirements are:
//   1) You must retain this notice
//   2) You must retain attribution (see bottom of file) to searchhippo.
//   3) You must retain the redirect through our servers for clickthroughs
//
// This page should be called with 'q' as a parameter passed in through
// the url that contains the query request.  For example, if you save this
// page as "searchhippo.php", you might call 
// http://[your-domain]/searchhippo.php?q=searchhippo
// again, where q are the search critera.
//
// Have fun, enjoy, and distribute freely!
//
////////////////////////////////////////////////////////////////////////


function startElementHandler ($parser, $element_name, $element_attribs)
{
  global $recctr;
  global $curstate;
  global $recs;
  global $qrystatus;
  global $data_version;

  if ($element_name == "RESULTS") {
    $qrystatus ["count"] = trim ($element_attribs ["COUNT"]);
    $qrystatus ["head"] = trim ($element_attribs ["HEAD"]);
    $qrystatus ["tail"] = trim ($element_attribs ["TAIL"]);
    $qrystatus ["nexturl"] = trim ($element_attribs ["NEXTURL"]);
    $qrystatus ["prevurl"] = trim ($element_attribs ["PREVURL"]);
  }

  if ($element_name == "RECORD") {
    $recs [$recctr] ["id"] = trim ($element_attribs ["ID"]);
    $recs [$recctr] ["timestamp"] = trim ($element_attribs ["TIMESTAMP"]);
    $recs [$recctr] ["size"] = trim ($element_attribs ["SIZE"]);
  }

  if ($element_name == "SEARCHHIPPO") {
    $data_version = trim ($element_attribs ["VERSION"]);
  }

  $curstate = $element_name;
}

function endElementHandler ($parser, $element_name)
{
  global $recctr;
  global $curstate;
  global $recs;

  $curstate = "";

  if ($element_name == "RECORD") { $recctr++; }
}

function characterDataHandler ($parser, $data)
{
  global $recctr;
  global $curstate;
  global $recs;

  if ($curstate == "") return;
  else if ($curstate == "URL") { $recs [$recctr] ["url"] = trim ($data); }
  else if ($curstate == "TITLE") { $recs [$recctr] ["title"] = trim ($data); }
  else if ($curstate == "DESCR") { $recs [$recctr] ["descr"] = trim ($data); }
}



// global vars

$recctr = 0;             // number of records retrieved
$in_item_tag = 0;        // flag for expat parsing
$curstate = '';          // for xpath state
$recs = array ();        // returned data results
$qrystatus = array ();   // returned data status
$data_version = '';      // version of results page

$xml_parser = xml_parser_create();
if (!($xml_parser)) {
  die ("can't get xml parser!");
} else {

  xml_set_element_handler ($xml_parser, 
    "startElementHandler", 
    "endElementHandler");

  xml_set_character_data_handler ($xml_parser, 
    "characterDataHandler");

  $q = stripslashes ($q);
  $qenc = urlencode ($q);   // q is query
  $ienc = urlencode ($i);   // i is 'initial count' for paging through results
  $cenc = urlencode ($c);   // c is 'count' for how many results to retrieve.

  $fp = fopen (
      "http://www.searchhippo.com/qxml?q=$qenc&c=$cenc&i=$ienc", "r");
  if ($fp) {
    $data = fread ($fp, 4096);
    while ($data) {
      if (!xml_parse ($xml_parser, $data, feof ($fp))) { break; }
      $data = fread ($fp, 4096);
    }
  }
  xml_parser_free ($xml_parser);


// 
// Now all of our data has been read from the url and parsed.  We now
// have populated the two global arrays, $recs and $qrystatus.
// additionally, $recctr will have the number of results we retrieved
// (i.e. the size of the $recs array
//


  if ($recctr > 0) {

    if ($data_version != "1.0") {
      printf ("<!-- data_version mismatch! Got %s, expected 1.0! -->", 
	  $data_version);
    }

    if ($qrystatus ["prevurl"][0]) {
// Modify slightly for the 'local' page:
      printf ("[ <A HREF="%s%s">Prev</A> ]", 
	$PHP_SELF, strchr ($qrystatus ["prevurl"], "?"));
    }

    printf (" Results %d - %d of %s ",
      $qrystatus ["head"],
      $qrystatus ["tail"],
      $qrystatus ["count"]);

    if ($qrystatus ["nexturl"][0]) {
// Modify slightly for the 'local' page:
      printf ("[ <A HREF="%s%s">Next</A> ]", 
	$PHP_SELF, strchr ($qrystatus ["nexturl"], "?"));
    }

    echo "</CENTER><P>";


// Here we are going to loop over all the items in the $recs array and
// output as necessary.
    for ($x = 0; $x < $recctr; $x++) {
      if (strlen ($recs [$x] ["title"]) > 64) {
        $ShowTitle = sprintf ("%-.61s...", $recs [$x] ["title"]);
      } else $ShowTitle = $recs [$x] ["title"];

      if (strlen ($recs [$x] ["descr"]) > 192) {
        $ShowDescr = sprintf ("%-.189s...", $recs [$x] ["descr"]);
      } else $ShowDescr = $recs [$x] ["descr"];

      $DisplayURL = ""; 
      $Tmp = strchr ($recs [$x] ["url"], "?");
      if ($Tmp) {
        // Skip first three characters 
	for ($i = 3; $i < strlen ($Tmp); $i++) { 
	  $DisplayURL .= $Tmp [$i];
	}
      }
      else $DisplayURL = $recs [$x] ["url"];


      if (!$ShowTitle) $ShowTitle = $DisplayURL;

// Optionally, one might remove the "http://r.searchhippo.com/r?u=" piece
// from the URL *displayed*  Please keep the redirect there for the actual
// hyperlink, because we use this information in determining which 
// sites are clicked on the most which one of the variables in our 
// ranking mechanism.


// Outputting some HTML below
/*******************************************************/

      printf ("<FONT SIZE="2">%d. </FONT><A HREF="%s"><FONT SIZE="2"><B>%s</B></FONT></A><BR>",
        $x + $qrystatus ["head"],
        htmlspecialchars ($recs [$x] ["url"], ENT_NOQUOTES),
        $ShowTitle);
      if (strlen ($ShowDescr) > 0) {
        printf ("<FONT SIZE="2">%s</FONT><BR>", $ShowDescr);
      }
      printf ("<FONT SIZE="2" COLOR="#117711">%s</FONT><BR>n", 
        $DisplayURL);
  

      printf("<HR>n");
/*******************************************************/
    }

    printf ("");


  } else {
    printf ("<CENTER><FONT SIZE="2" COLOR="#CC1111">Sorry, no results available for <B>"%s"</B></FONT></CENTER>n", $q);
  }
}


// You *MUST* leave SearchHippo attribution!

echo '
<CENTER>
<IMG SRC="http://www.searchhippo.com/pics/hippo_25.gif" HEIGHT="25" WIDTH="27">
Powered by:
<A HREF="http://www.searchhippo.com/">
<FONT SIZE="+1">SearchHippo.com</FONT></A>
<IMG SRC="http://www.searchhippo.com/pics/hippo_25.gif" HEIGHT="25" WIDTH="27">
</CENTER>
';

?>