| 1 |
<?
|
| 2 |
|
| 3 |
include "functions.inc";
|
| 4 |
include "theme.inc";
|
| 5 |
|
| 6 |
class rdf {
|
| 7 |
// Contains the raw rdf file:
|
| 8 |
var $data;
|
| 9 |
|
| 10 |
// Contains the parsed rdf file:
|
| 11 |
var $title; // website name
|
| 12 |
var $items; // latest headlines
|
| 13 |
|
| 14 |
function url2sql($site, $timout = 10) {
|
| 15 |
### Connect to database:
|
| 16 |
dbconnect();
|
| 17 |
|
| 18 |
### Get channel info:
|
| 19 |
$result = mysql_query("SELECT * FROM channel WHERE site = '$site'");
|
| 20 |
|
| 21 |
if ($channel = mysql_fetch_object($result)) {
|
| 22 |
### Decode URL:
|
| 23 |
$url = parse_url($channel->rdf);
|
| 24 |
$host = $url[host];
|
| 25 |
$port = $url[port] ? $url[port] : 80;
|
| 26 |
$path = $url[path];
|
| 27 |
|
| 28 |
// print "<PRE>$url - $host - $port - $path</PRE>";
|
| 29 |
|
| 30 |
### Retrieve data from website:
|
| 31 |
$fp = fsockopen($host, $port, &$errno, &$errstr, $timout);
|
| 32 |
|
| 33 |
if ($fp) {
|
| 34 |
### Get data from URL:
|
| 35 |
fputs($fp, "GET $path HTTP/1.0\n");
|
| 36 |
fputs($fp, "User-Agent: headline grabber\n");
|
| 37 |
fputs($fp, "Host: ". $host ."\n");
|
| 38 |
fputs($fp, "Accept: */*\n\n");
|
| 39 |
|
| 40 |
while(!feof($fp)) $data .= fgets($fp, 128);
|
| 41 |
|
| 42 |
// print "<PRE>$data</PRE><HR>";
|
| 43 |
|
| 44 |
if (strstr($data, "200 OK")) {
|
| 45 |
|
| 46 |
### Remove existing entries:
|
| 47 |
$result = mysql_query("DELETE FROM headlines WHERE id = $channel->id");
|
| 48 |
|
| 49 |
### Strip all 'junk':
|
| 50 |
$data = ereg_replace("<?xml.*/image>", "", $data);
|
| 51 |
$data = ereg_replace("</rdf.*", "", $data);
|
| 52 |
$data = chop($data);
|
| 53 |
|
| 54 |
### Iterating through our data processing each entry/item:
|
| 55 |
$items = explode("</item>", $data);
|
| 56 |
$number = 0;
|
| 57 |
|
| 58 |
for (reset($items); $item = current($items); next($items)) {
|
| 59 |
### Extract data:
|
| 60 |
$link = ereg_replace(".*<link>", "", $item);
|
| 61 |
$link = ereg_replace("</link>.*", "", $link);
|
| 62 |
$title = ereg_replace(".*<title>", "", $item);
|
| 63 |
$title = ereg_replace("</title>.*", "", $title);
|
| 64 |
|
| 65 |
### Clean headlines:
|
| 66 |
$title = stripslashes(fixquotes($title));
|
| 67 |
|
| 68 |
### Count the number of stories:
|
| 69 |
$number += 1;
|
| 70 |
|
| 71 |
### Insert item in database:
|
| 72 |
$result = mysql_query("INSERT INTO headlines (id, title, link, number) VALUES('$channel->id', '$title', '$link', '$number')");
|
| 73 |
}
|
| 74 |
|
| 75 |
### Mark channels as being updated:
|
| 76 |
$result = mysql_query("UPDATE channel SET timestamp = '". time() ."' WHERE id = $channel->id");
|
| 77 |
}
|
| 78 |
else print "<HR>RDF parser: 404 error?<BR><BR><PRE>$data</PRE><HR>";
|
| 79 |
}
|
| 80 |
}
|
| 81 |
}
|
| 82 |
|
| 83 |
function displayHeadlines($site, $timout = 1800) {
|
| 84 |
global $theme;
|
| 85 |
|
| 86 |
### Connect to database:
|
| 87 |
dbconnect();
|
| 88 |
|
| 89 |
### Get channel info:
|
| 90 |
$result = mysql_query("SELECT * FROM channel WHERE site = '$site'");
|
| 91 |
|
| 92 |
if ($channel = mysql_fetch_object($result)) {
|
| 93 |
|
| 94 |
### Check to see whether we have to update our headlines first:
|
| 95 |
if (time() - $channel->timestamp > $timout) $this->url2sql($site);
|
| 96 |
|
| 97 |
### Grab headlines from database:
|
| 98 |
$result = mysql_query("SELECT * FROM headlines WHERE id = $channel->id ORDER BY number");
|
| 99 |
while ($headline = mysql_fetch_object($result)) {
|
| 100 |
$content .= "<LI><A HREF=\"$headline->link\">$headline->title</A></LI>";
|
| 101 |
}
|
| 102 |
### Add timestamp:
|
| 103 |
$update = round((time() - $channel->timestamp) / 60);
|
| 104 |
$content .= "<P ALIGN=\"right\">[ <A HREF=\"rdf.php?op=reset&id=$channel->id\"><FONT COLOR=\"$theme->hlcolor2\">reset</FONT></A> | updated $update min. ago ]</P>";
|
| 105 |
|
| 106 |
### Display box:
|
| 107 |
$theme->box("$channel->site", $content);
|
| 108 |
}
|
| 109 |
else print "<P>Warning: something whiched happened: specified channel could not be found in database.</P>";
|
| 110 |
}
|
| 111 |
|
| 112 |
function addChannel($site, $url, $rdf) {
|
| 113 |
### Connect to database:
|
| 114 |
dbconnect();
|
| 115 |
|
| 116 |
### Add channel:
|
| 117 |
$query = mysql_query("INSERT INTO channel (site, url, rdf, timestamp) VALUES ('$site', '$url', '$rdf', now())");
|
| 118 |
}
|
| 119 |
|
| 120 |
function resetChannel($id) {
|
| 121 |
### Connect to database:
|
| 122 |
dbconnect();
|
| 123 |
|
| 124 |
### Delete headlines:
|
| 125 |
$result = mysql_query("DELETE FROM headlines WHERE id = $id");
|
| 126 |
|
| 127 |
### Mark channel as invalid to enforce an update:
|
| 128 |
$result = mysql_query("UPDATE channel SET timestamp = 42 WHERE id = $id");
|
| 129 |
}
|
| 130 |
}
|
| 131 |
|
| 132 |
function adminAddChannel() {
|
| 133 |
?>
|
| 134 |
<HR>
|
| 135 |
<FORM ACTION="rdf.php" METHOD="post">
|
| 136 |
<P>
|
| 137 |
<B>Site name:</B><BR>
|
| 138 |
<INPUT TYPE="text" NAME="site" SIZE="50">
|
| 139 |
</P>
|
| 140 |
|
| 141 |
<P>
|
| 142 |
<B>URL:</B><BR>
|
| 143 |
<INPUT TYPE="text" NAME="url" SIZE="50">
|
| 144 |
</P>
|
| 145 |
|
| 146 |
<P>
|
| 147 |
<B>RDF file:</B><BR>
|
| 148 |
<INPUT TYPE="text" NAME="rdf" SIZE="50">
|
| 149 |
</P>
|
| 150 |
<INPUT TYPE="submit" NAME="op" VALUE="Add RDF channel">
|
| 151 |
</FORM>
|
| 152 |
<?
|
| 153 |
}
|
| 154 |
|
| 155 |
function adminDisplayAll() {
|
| 156 |
### Connect to database:
|
| 157 |
dbconnect();
|
| 158 |
|
| 159 |
### Get channel info:
|
| 160 |
$result = mysql_query("SELECT * FROM channel ORDER BY id");
|
| 161 |
|
| 162 |
print "<TABLE BORDER=\"0\">";
|
| 163 |
while ($channel = mysql_fetch_object($result)) {
|
| 164 |
if ($state % 3 == 0) print " <TR>";
|
| 165 |
|
| 166 |
print " <TD ALIGN=\"center\" VALIGN=\"top\" WIDTH=\"33%\">";
|
| 167 |
$rdf = new rdf();
|
| 168 |
$rdf->displayHeadlines($channel->site);
|
| 169 |
print " </TD>";
|
| 170 |
|
| 171 |
if ($state % 3 == 2) print " </TR>";
|
| 172 |
|
| 173 |
$state += 1;
|
| 174 |
}
|
| 175 |
print "</TABLE>";
|
| 176 |
}
|
| 177 |
|
| 178 |
function adminDisplayInfo() {
|
| 179 |
?>
|
| 180 |
<H1>Headlines</H1>
|
| 181 |
<H3>Concept</H3>
|
| 182 |
<P>
|
| 183 |
RDF support can change a portal in a significant way: third party websites
|
| 184 |
can become <I>channels</I> in our portal without having to make 'real' deals
|
| 185 |
and with a minimum of extra work. All they need to do is to publish an RDF,
|
| 186 |
so we can include their latest updates in our portal. Yet another easy way
|
| 187 |
to add content.
|
| 188 |
</P>
|
| 189 |
<P>
|
| 190 |
That in and of itself is interesting, but it's not half so interesting as
|
| 191 |
the fact that other sites can include our headlines as well. Anyone can
|
| 192 |
grab our RDF, anyone can parse it, and anyone can put a list of our
|
| 193 |
headlines. Yet another way to generate more traffic.
|
| 194 |
</P>
|
| 195 |
|
| 196 |
<H3>Features</H3>
|
| 197 |
<P>
|
| 198 |
One of the most important features (if not the most important) is
|
| 199 |
chaching support. To avoid bogging down other portals with a continous
|
| 200 |
stream of headline grabbing, all headlines are cached and refreshed once
|
| 201 |
in a while. The 'while' can be costumized but is set to 30 minutes by
|
| 202 |
default.
|
| 203 |
</P>
|
| 204 |
<P>
|
| 205 |
You can reset a channel, that is force to update a channels headlines
|
| 206 |
and you can add new channels. If you don't know what channel to add,
|
| 207 |
check <A HREF="http://www.xmltree.com/">http://www.xmltree.com/</A>.
|
| 208 |
Make sure you don't add anything except valid RDF files!
|
| 209 |
</P>
|
| 210 |
|
| 211 |
<H3>Status</H3>
|
| 212 |
<P>
|
| 213 |
The RDF parser is still in beta and needs proper integration in the engine.
|
| 214 |
Until then this test page generates nothing more then an overview off all
|
| 215 |
subscribed channels along with their headlines: handy for news squatting. ;)
|
| 216 |
</P>
|
| 217 |
<P>
|
| 218 |
RDF files are non-proprietary and publically available. Unfortunatly,
|
| 219 |
RDF is not the only standard: another commonly used format is RSS which
|
| 220 |
would be nice to support as well.
|
| 221 |
</P>
|
| 222 |
<HR>
|
| 223 |
<?
|
| 224 |
}
|
| 225 |
|
| 226 |
$theme->header();
|
| 227 |
|
| 228 |
switch($op) {
|
| 229 |
case "reset":
|
| 230 |
$channel = new rdf();
|
| 231 |
$channel->resetChannel($id);
|
| 232 |
print "<H2>channel has been reset</H2>";
|
| 233 |
print "<A HREF=\"rdf.php\">back</A>";
|
| 234 |
break;
|
| 235 |
case "Add RDF channel":
|
| 236 |
$channel = new rdf();
|
| 237 |
$channel->addChannel($site, $url, $rdf);
|
| 238 |
// fall through:
|
| 239 |
default:
|
| 240 |
adminDisplayInfo();
|
| 241 |
adminDisplayAll();
|
| 242 |
adminAddChannel();
|
| 243 |
}
|
| 244 |
|
| 245 |
$theme->footer();
|
| 246 |
|
| 247 |
?>
|
Parent Directory
| 
Revision Log
| 
Revision Graph