| 1 |
<?php
|
| 2 |
// $Id$
|
| 3 |
/**
|
| 4 |
* All the jobs that should be done in cron-time
|
| 5 |
*
|
| 6 |
* @author Aron Novak <aaron@szentimre.hu>
|
| 7 |
* @version 0.1
|
| 8 |
* @package sna
|
| 9 |
*/
|
| 10 |
|
| 11 |
/**
|
| 12 |
* Get sql access and important functions
|
| 13 |
*/
|
| 14 |
require_once './includes/bootstrap.inc';
|
| 15 |
drupal_bootstrap(DRUPAL_BOOTSTRAP_LATE_PAGE_CACHE);
|
| 16 |
require_once 'modules/sna/common.php';
|
| 17 |
/**
|
| 18 |
* Create a graph from nodes-comments tables.
|
| 19 |
*
|
| 20 |
* @param array $edges The adjacentcy list of the graph
|
| 21 |
* @return integer The number of interactions
|
| 22 |
*/
|
| 23 |
function build_edges_from_nodes(&$edges) {
|
| 24 |
$edges = array();
|
| 25 |
$node_replies_q = "SELECT users.uid as u2, users_1.uid as u1
|
| 26 |
FROM {node} node, {users} users, {comments} comments, {users} users_1
|
| 27 |
WHERE comments.nid = node.nid AND node.uid = users.uid AND users_1.uid = comments.uid
|
| 28 |
AND users.name <> '' AND users_1.name <> ''
|
| 29 |
AND comments.pid = 0";
|
| 30 |
|
| 31 |
$comment_replies_q = "SELECT users_1.uid as 'u2', users.uid as 'u1'
|
| 32 |
FROM {users} users, {comments} comments, {comments} comments_1, {users} users_1
|
| 33 |
WHERE comments_1.cid = comments.pid
|
| 34 |
AND users.uid = comments.uid
|
| 35 |
AND users_1.uid = comments_1.uid
|
| 36 |
AND users_1.name <> ''
|
| 37 |
AND users.name <> ''";
|
| 38 |
|
| 39 |
if ((!$node_replies = db_query($node_replies_q)) || (!$comment_replies = db_query($comment_replies_q))) {
|
| 40 |
die("Database problem\n");
|
| 41 |
}
|
| 42 |
while ($line = db_fetch_array($comment_replies)) {
|
| 43 |
if ($line["u1"] != $line["u2"]) { // Do not do hitches
|
| 44 |
$edges[$line["u1"]][$line["u2"]]++;
|
| 45 |
}
|
| 46 |
}
|
| 47 |
while ($line = db_fetch_array($node_replies)) {
|
| 48 |
if ($line["u1"] != $line["u2"]) {
|
| 49 |
$edges[$line["u1"]][$line["u2"]]++;
|
| 50 |
}
|
| 51 |
}
|
| 52 |
return db_num_rows($node_replies) + db_num_rows($comment_replies);
|
| 53 |
}
|
| 54 |
|
| 55 |
/**
|
| 56 |
* Create a graph from the buddylist module data
|
| 57 |
*
|
| 58 |
* @param array $edges The adjacentcy list of the graph
|
| 59 |
* @return integer The numer of connections
|
| 60 |
*/
|
| 61 |
function build_edges_from_buddy(&$edges) {
|
| 62 |
$edges = array();
|
| 63 |
$buddy_q = "SELECT uid, buddy FROM {buddylist}";
|
| 64 |
if ((!$buddies = db_query($buddy_q))) {
|
| 65 |
die("Database problem\n");
|
| 66 |
}
|
| 67 |
while ($line = db_fetch_array($buddies)) {
|
| 68 |
$edges[$line["uid"]][$line["buddy"]]++;
|
| 69 |
}
|
| 70 |
return db_num_rows($buddies);
|
| 71 |
}
|
| 72 |
|
| 73 |
/**
|
| 74 |
* Create a graph from accesslog table. The connection is to view other's profile
|
| 75 |
*
|
| 76 |
* @param array $edges The adjacentcy list of the graph
|
| 77 |
* @return integer The number of connections
|
| 78 |
*/
|
| 79 |
function build_edges_from_stats(&$edges) {
|
| 80 |
$edges = array();
|
| 81 |
$stats_q = "SELECT uid, path FROM {accesslog} WHERE path LIKE 'user/%'";
|
| 82 |
if ((!$stats = db_query($stats_q))) {
|
| 83 |
die("Database problem\n");
|
| 84 |
}
|
| 85 |
while ($line = db_fetch_array($stats)) {
|
| 86 |
$dest = str_replace("user/", "", $line["path"]);
|
| 87 |
if (is_numeric($dest) && $dest != $line["uid"]) {
|
| 88 |
$edges[$line["uid"]][$dest]++;
|
| 89 |
}
|
| 90 |
}
|
| 91 |
return db_num_rows($stats);
|
| 92 |
}
|
| 93 |
|
| 94 |
/**
|
| 95 |
* When should we throw away edges
|
| 96 |
*
|
| 97 |
* @param integer $total Number of connections
|
| 98 |
* @return integer The thresold for throwing away edges
|
| 99 |
*/
|
| 100 |
function get_limit($total) {
|
| 101 |
if ($total < 1000) {
|
| 102 |
return FALSE;
|
| 103 |
}
|
| 104 |
else {
|
| 105 |
return 3 * log($total - 999); // Ad-hoc computation based on some experiment
|
| 106 |
}
|
| 107 |
}
|
| 108 |
|
| 109 |
/**
|
| 110 |
* Create a dot file from the graph to Graphviz
|
| 111 |
* Graphviz is a graph visualization tool
|
| 112 |
*
|
| 113 |
* @param array $edges The adjacentcy list of the graph
|
| 114 |
* @param $num_interactions Number of edges in the graph
|
| 115 |
* @return boolean The success of writing out the file
|
| 116 |
*/
|
| 117 |
function generate_graphviz_input($edges, $num_interactions) {
|
| 118 |
$dot_graph = "digraph G {\n";
|
| 119 |
$limit = intval(get_limit($num_interactions));
|
| 120 |
foreach ($edges as $u1 => $sub_arr) {
|
| 121 |
if ($u1 === 0) { // Anonymous - don't count them!
|
| 122 |
break;
|
| 123 |
}
|
| 124 |
foreach ($sub_arr as $u2 => $num) {
|
| 125 |
if ($u2 === 0) { // Anonymous - don't count them!
|
| 126 |
break;
|
| 127 |
}
|
| 128 |
if (($num > $limit) || ($limit === FALSE)) {
|
| 129 |
$dot_graph .= "\t\"". get_real_name($u1).
|
| 130 |
"\" -> \"". get_real_name($u2) .
|
| 131 |
"\" [label=". round($edges[$u1][$u2], 2) ."];\n";
|
| 132 |
}
|
| 133 |
}
|
| 134 |
}
|
| 135 |
|
| 136 |
$dot_graph .= "}";
|
| 137 |
if (!$fp = fopen(DOT_PATH, "w")) {
|
| 138 |
return FALSE;
|
| 139 |
}
|
| 140 |
// Write out the DOT file
|
| 141 |
fwrite($fp, $dot_graph);
|
| 142 |
fclose($fp);
|
| 143 |
return TRUE;
|
| 144 |
}
|
| 145 |
|
| 146 |
/**
|
| 147 |
* Create a net file from the graph to Pajek
|
| 148 |
* Pajek is a graph analizer and visualization tool
|
| 149 |
* http://vlado.fmf.uni-lj.si/pub/networks/pajek/
|
| 150 |
*
|
| 151 |
* @param array $edges The adjacentcy list of the graph
|
| 152 |
* @return boolean The success of writing out the file
|
| 153 |
*/
|
| 154 |
function generate_pajek_input($edges) {
|
| 155 |
// Count unique vertex
|
| 156 |
$edg = "*Edges\n";
|
| 157 |
$vertex = get_all_vertices();
|
| 158 |
$num_vertex = count($vertex);
|
| 159 |
for ($i = 0; $i < $num_vertex; $i++) {
|
| 160 |
$vert .= ($i + 1) ." \"". get_real_name($vertex[$i]) ."\"\n";
|
| 161 |
// in NET files we have to index points in a strict order, uid is not suitable
|
| 162 |
$real_id[$vertex[$i]] = $i + 1;
|
| 163 |
}
|
| 164 |
foreach ($edges as $vx_from => $next) {
|
| 165 |
if ($vx_from === 0) { // Anonymous - don't count them!
|
| 166 |
break;
|
| 167 |
}
|
| 168 |
foreach (array_keys($next) as $vx_to) {
|
| 169 |
if ($vx_to === 0) { // Anonymous - don't count them!
|
| 170 |
break;
|
| 171 |
}
|
| 172 |
if ($edges[$vx_from][$vx_to] > 20) {
|
| 173 |
$edg .= $real_id[$vx_from] ." ". $real_id[$vx_to] ." ". $edges[$vx_from][$vx_to] ."\n";
|
| 174 |
}
|
| 175 |
}
|
| 176 |
|
| 177 |
|
| 178 |
}
|
| 179 |
$net .= "*Vertices ". $num_vertex ."\n". $vert . $edg;
|
| 180 |
if (!$fp = fopen(NET_PATH, "w")) {
|
| 181 |
return FALSE;
|
| 182 |
}
|
| 183 |
fwrite($fp, $net);
|
| 184 |
fclose($fp);
|
| 185 |
return TRUE;
|
| 186 |
}
|
| 187 |
|
| 188 |
/**
|
| 189 |
* Write out the includable graph data
|
| 190 |
*
|
| 191 |
* @param array $edges The adjacentcy list of the graph
|
| 192 |
* @return boolean The success of writing out
|
| 193 |
*/
|
| 194 |
function put_graph($edges) {
|
| 195 |
if (!$file_s = fopen(DATA_PATH, "w")) {
|
| 196 |
return FALSE;
|
| 197 |
}
|
| 198 |
if (!lock($file_s)) {
|
| 199 |
return FALSE;
|
| 200 |
}
|
| 201 |
fwrite($file_s, '<?php $edges = ' . var_export($edges, TRUE) . ';?>');
|
| 202 |
fclose($file_s);
|
| 203 |
return TRUE;
|
| 204 |
}
|
| 205 |
|
| 206 |
/**
|
| 207 |
* Convert all the edges cost<->length
|
| 208 |
*
|
| 209 |
* @param array $edges The adjacentcy list of the graph
|
| 210 |
* @return array $edges The adjacentcy list of the graph
|
| 211 |
*/
|
| 212 |
function transform_edges($edges) {
|
| 213 |
$min_max = get_min_and_max_degree($edges);
|
| 214 |
$transformed_graph = array();
|
| 215 |
foreach (array_keys($edges) as $A) {
|
| 216 |
foreach (array_keys($edges[$A]) as $B) {
|
| 217 |
$transformed_graph[$A][$B] = get_edge_weight($edges, $A, $B, $min_max[0], $min_max[1]);
|
| 218 |
}
|
| 219 |
}
|
| 220 |
return $transformed_graph;
|
| 221 |
}
|
| 222 |
$at_start = res_start();
|
| 223 |
|
| 224 |
if (!function_exists('dba_open')) {
|
| 225 |
die('Dba support is not available.');
|
| 226 |
}
|
| 227 |
|
| 228 |
clear_cache();
|
| 229 |
$edges = array();
|
| 230 |
$graph_source = variable_get('sna_data_source', GRAPH_SOURCE);
|
| 231 |
$build_edges = 'build_edges_from_' . $graph_source;
|
| 232 |
$num_interactions = $build_edges($edges);
|
| 233 |
$edges = transform_edges($edges);
|
| 234 |
if (!put_graph($edges)) {
|
| 235 |
die('Cannot save the network into file');
|
| 236 |
}
|
| 237 |
|
| 238 |
/* Cache the most popular vertex's minimal tree */
|
| 239 |
$most_popular = sort_by_popularity($edges);
|
| 240 |
$most_popular_size = sizeof($most_popular);
|
| 241 |
$size_cache = variable_get('sna_size_cache', NUM_CACHE);
|
| 242 |
$limit = $most_popular_size < $size_cache ? $most_popular_size : $size_cache;
|
| 243 |
$at_start = res_start();
|
| 244 |
for ($i = 0; $i < $limit; $i++) {
|
| 245 |
a_to_any($edges, $most_popular[$i][1], TRUE);
|
| 246 |
print $most_popular[$i][1] . " cached\n";
|
| 247 |
}
|
| 248 |
|
| 249 |
generate_graphviz_input($edges, $num_interactions);
|
| 250 |
generate_pajek_input($edges);
|
| 251 |
print_r(res_stop($at_start));
|
| 252 |
|
| 253 |
?>
|
Parent Directory
| 
Revision Log
| 
Revision Graph