scripts/google2rss.php

   1 <?
   2 error_reporting(0);
   3 // This program is public domain. Do with this what you want.
   4 //
   5 // Disclaimer. Don't expect this to be here, to work, or to get fixed.
   6 // But if you have a question or comment, email: mailto:julian_bond@voidstar.com
   7 //
   8 // If you're using Gnews2rss you presumably find it useful.
   9 // Please email Google (news-feedback@google.com) asking them to produce RSS
  10 // directly out of Google News Search.
  11 //
  12 // And why not host it yourself to save my bandwidth costs.
  13
  14 // Note: This version requires curl support in your php installation
  15
  16 $q=$_GET['q'];
  17
  18 if ($q) {
  19   parse_html($q);
  20 } else {
  21   show_form();
  22 }
  23
  24 //****************
  25 function show_form() {
  26
  27   $server = getenv("SERVER_NAME");
  28   $request = getenv("REQUEST_URI");
  29 ?>
  30
  31    <center><B>Gnews2RSS at VoidStar.com</B></center>
  32    <p>An experimental convertor that takes a Google News search and turns it
  33 into RSS</p>
  34
  35    <form action="<? print "http://" . $server . $request; ?>">
  36      <br />Number of entries to return: <select name="num">
  37        <option name="5">5</option>
  38        <option name="5">10</option>
  39        <option name="5" selected>15</option>
  40        <option name="5">20</option>
  41        <option name="5">25</option>
  42        <option name="5">50</option>
  43        <option name="5">75</option>
  44        <option name="5">100</option>
  45      </select>
  46      <br />Search Query:<input type="text" name="q" size=50>
  47      <br /><input type="submit" value="Create RSS">
  48    </form>
  49    <br /><b>Usage</b>:
  50 http://www.voidstar.com/gnews2rss.php?num=number_of_items&q=your_query
  51    <br />
  52    <br /><b>Example</b>: <a
  53 href="http://www.voidstar.com/gnews2rss.php?q=wifi+OR+WLAN+OR+80211">http://www.voidstar.com/gnews2rss.php?q=wifi+OR+WLAN+OR+80211</a>
  54    <br />
  55    <br /><b>Disclaimer</b> Don't expect this to be here, to work, or to get
  56 fixed. But if you have a question or comment, email: <a
  57 href="mailto:julian_bond@voidstar.com">julian_bond@voidstar.com</a></li>
  58    <br />
  59    <br /><b>Notes:</b>
  60    <ul>
  61    <li>num defaults to 15</li>
  62    <li>Source <a
  63 href="http://www.voidstar.com/gnews2rss.php.txt">http://www.voidstar.com/gnews2rss.php.txt</a>
  64 If you have access to PHP, why not run your own. </li>
  65    <li>Source <a
  66 href="http://www.voidstar.com/gnews2rss.php.curl.txt">http://www.voidstar.com/gnews2rss.php.curl.txt</a>
  67 A version using curl. </li>
  68    <li>If you're using Gnews2rss you presumably find it useful. Please <a
  69 href="mailto:news-feedback@google.com">email Google</a> asking them to
  70 produce RSS directly out of Google News Search. And why not <a
  71 href="http://www.voidstar.com/gnews2rss.php.txt">host it yourself</a> to
  72 save my bandwidth costs.</li>
  73    <li>Note: Some people have had trouble making it work on their servers
  74 where no results seem to be found. This seems to be related to fopen(),
  75 certain versions of php and some check that Google does.
  76    If you get this, read the source and try uncommenting the lines just
  77 after the data is collected to see what google returns.
  78    Alternatively try the cURL version of the source, for which of course you
  79 will need cURL support in PHP.</li>
  80    <li>I'm happy to talk on email about all this, but I can't get involved
  81 in debugging your installation. Go back and read the disclaimer!</li>
  82    </ul>
  83 <?
  84 }
  85
  86 //****************
  87 function parse_html($q){
  88
  89   header("Cache-Control: public");
  90
  91   $itemregexp = "%<a class=y href=\"(.+?)\">(.+?)<br><font size=-1><font color=#6f6f6f>(.+?)</font><br></table>%is";
  92   $allowable_tags = "<A><B><BR><BLOCKQUOTE><CENTER><DD><DL><DT><HR><I><IMG><LI><OL><P><PRE><U><UL>";
  93
  94   $num = ($num) ? $num+1 : 16 ;
  95
  96   $url =
  97 "http://news.google.com/news?hl=en&num=$num&scoring=d&q=".urlencode($q);
  98
  99   if ($fp = @fopen($url, "r")) {
 100     while (!feof($fp)) $data .= fgets($fp, 128);
 101     fclose($fp);
 102   }
 103
 104 // *******************
 105 // Some people seem to have problems with google not returning anything
 106 // uncomment the following lines and comment out the content-type header
 107 // to see what google is returning.
 108
 109 //  print "<html>";
 110 //  print "<pre>";
 111 //  print htmlentities($data);
 112   header("Content-Type: text/xml");
 113
 114   $data = strstr($data,"Sorted by date</b>");
 115
 116   eregi("<title>(.*)</title>", $data, $title);
 117   $channel_title = $title[1];
 118
 119   $match_count = preg_match_all($itemregexp, $data, $items);
 120   $match_count = ($match_count > 25) ? 25 : $match_count;
 121
 122   $output .= "<?xml version=\"1.0\" encoding=\"iso-8859-1\" ?>\n";
 123   $output .= "<!-- generator=\"gnews2rss/1.0\" -->\n";
 124   $output .= "<!DOCTYPE rss >\n";
 125
 126   $output .= "<rss version=\"2.0\">\n";
 127   $output .= "  <channel>\n";
 128   $output .= "    <title>Google News Search: $q</title>\n";
 129   $output .= "    <link>". htmlentities($url) ."</link>\n";
 130   $output .= "    <description>Google News Search: $q</description>\n";
 131   $output .= "    <webMaster>julian_bond@voidstar.com</webMaster>\n";
 132   $output .= "    <language>en-us</language>\n";
 133   $output .= "    <generator>&lt;a href=\"http://www.voidstar.com/gnews2rss.php\">GNews2Rss&lt;/a></generator>\n";
 134
 135   $day = date("d");
 136   if ($day == 1 || $day == 11 || $day == 21) {
 137     $output .= "    <item>\n";
 138     $output .= "      <title>". date("d-M-y"). " Do you find Gnews2RSS useful?</title>\n";
 139     $output .= "      <link>http://www.voidstar.com/gnews2rss.php</link>\n";
 140     $output .= "      <description>If you're using Gnews2rss you presumably find it useful. Please &lt;a href=\"mailto:news-feedback@google.com\">email Google&lt;/a> asking them to produce RSS directly out of Google News Search. And why not &lt;a href=\"http://www.voidstar.com/gnews2rss.php.txt\">host it yourself&lt;/a> to save my bandwidth costs.</description>\n";
 141     $output .= "    </item>\n";
 142   }
 143
 144   for ($i=0; $i< $match_count; $i++) {
 145
 146     $item_url = $items[1][$i];
 147     $title = $items[2][$i];
 148     $title = strip_tags($title);
 149     $desc = $items[3][$i];
 150
 151     $desc = eregi_replace("&nbsp;-&nbsp;.* ago</font><br>", "<br>", $desc);
 152     $desc = strip_tags($desc, $allowable_tags);
 153     $desc = htmlspecialchars($desc);
 154
 155     $output .= "    <item>\n";
 156     $output .= "      <title>". htmlspecialchars($title) ."</title>\n";
 157     $output .= "      <link>". htmlspecialchars($item_url) ."</link>\n";
 158     $output .= "      <description>". $desc ."</description>\n";
 159     $output .= "    </item>\n";
 160   }
 161
 162   $output .= "  </channel>\n";
 163   $output .= "</rss>\n";
 164
 165   print $output;
 166
 167 //****************
 168 // More debug stuff
 169 //  print "<pre>";
 170 //  print htmlentities($output);
 171 //  print "</pre>";
 172
 173 }
 174
 175
 176 ?>