Translations of this page?:

This is an old revision of the document!

grapher.php 
  1. #!/usr/bin/php
  2. <?php
  3. if(!defined('DOKU_INC')) define('DOKU_INC', realpath(dirname(__FILE__) . '/../') . '/');
  4. require_once(DOKU_INC . 'inc/init.php');
  5.  
  6. class Grapher extends DokuCLI {
  7.  
  8.     /**
  9.      * Register options and arguments on the given $options object
  10.      *
  11.      * @param DokuCLI_Options $options
  12.      * @return void
  13.      */
  14.     protected function setup(DokuCLI_Options $options) {
  15.         $options->setHelp('Creates a graph representation of pages and media files and how they are interlinked.');
  16.         $options->registerOption(
  17.             'depth',
  18.             'Recursion depth, eg. how deep to look into the given namespaces. Use 0 for all. Default: 1',
  19.             'd', 'depth');
  20.         $options->registerOption(
  21.             'media',
  22.              "How to handle media files. 'ns' includes only media that is located in the given namespaces, ".
  23.              "'all' includes all media files and 'none' ignores the media files completely. ".
  24.              "Default: ns",
  25.             'm', 'ns|all|none');
  26.         $options->registerOption(
  27.             'format',
  28.             "The wanted output format. 'dot' is a very simple format which can be used to visualize the resulting ".
  29.             "graph with graphviz. The 'gexf' format is a more complex XML-based format which contains more info ".
  30.             "about the found nodes and can be loaded in Gephi. Default: dot",
  31.             'f', 'dot|gexf');
  32.         $options->registerOption(
  33.             'output',
  34.             "Where to store the output eg. a filename. If not given the output is written to STDOUT.",
  35.             'o', 'file');
  36.         $options->registerArgument(
  37.             'namespaces',
  38.             "Give all wiki namespaces you want to have graphed. If no namespace is given, the root ".
  39.             "namespace is assumed.",
  40.             false
  41.         );
  42.     }
  43.  
  44.     /**
  45.      * Your main program
  46.      *
  47.      * Arguments and options have been parsed when this is run
  48.      *
  49.      * @param DokuCLI_Options $options
  50.      * @return void
  51.      */
  52.     protected function main(DokuCLI_Options $options) {
  53.         $depth = $options->getOpt('depth', 1);
  54.         $media = $options->getOpt('media', 'ns');
  55.         if(!in_array($media, array('ns', 'all', 'none'))) {
  56.             $this->fatal('Bad media option: ' . $media);
  57.         }
  58.         $format = $options->getOpt('format', 'dot');
  59.         if(!in_array($format, array('dot', 'gexf'))) {
  60.             $this->fatal('Bad format option: ' . $format);
  61.         }
  62.         $output = $options->getOpt('output', '-');
  63.         if($output == '-') $output = 'php://stdout';
  64.  
  65.         $namespaces = array_map('cleanID', $options->args);
  66.         if(!count($namespaces)) $namespaces = array(''); //import from top
  67.  
  68.         $fh = @fopen($output, 'w');
  69.         if(!$fh) $this->fatal("Failed to open $output");
  70.  
  71.         $data = $this->gather_data($namespaces, $depth, $media);
  72.         if($format == 'dot') {
  73.             $this->create_dot($data, $fh);
  74.         } elseif($format == 'gexf') {
  75.             $this->create_gexf($data, $fh);
  76.         }
  77.  
  78.         fclose($fh);
  79.     }
  80.  
  81.     /**
  82.      * Find all the node and edge data for the given namespaces
  83.      * @param $namespaces
  84.      * @param int $depth
  85.      * @param string $incmedia
  86.      * @return array
  87.      */
  88.     protected function gather_data($namespaces, $depth = 0, $incmedia = 'ns') {
  89.         global $conf;
  90.         /** @var helper_plugin_translation $transplugin */
  91.         $transplugin = plugin_load('helper', 'translation');
  92.  
  93.         $pages = array();
  94.         $media = array();
  95.         foreach($namespaces as $ns) {
  96.             // find media
  97.             if($incmedia == 'ns') {
  98.                 $data = array();
  99.                 search(
  100.                     $data,
  101.                     $conf['mediadir'],
  102.                     'search_universal',
  103.                     array(
  104.                         'depth' => $depth,
  105.                         'listfiles' => true,
  106.                         'listdirs' => false,
  107.                         'pagesonly' => false,
  108.                         'skipacl' => true,
  109.                         'keeptxt' => true,
  110.                         'meta' => true,
  111.                     ),
  112.                     str_replace(':', '/', $ns)
  113.                 );
  114.  
  115.                 // go through all those media files
  116.                 while($item = array_shift($data)) {
  117.                     $media[$item['id']] = array(
  118.                         'title' => noNS($item['id']),
  119.                         'size' => $item['size'],
  120.                         'ns' => getNS($item['id']),
  121.                         'time' => $item['mtime'],
  122.                     );
  123.                 }
  124.             }
  125.  
  126.             // find pages
  127.             $data = array();
  128.             search(
  129.                 $data,
  130.                 $conf['datadir'],
  131.                 'search_universal',
  132.                 array(
  133.                     'depth' => $depth,
  134.                     'listfiles' => true,
  135.                     'listdirs' => false,
  136.                     'pagesonly' => true,
  137.                     'skipacl' => true,
  138.                     'firsthead' => true,
  139.                     'meta' => true,
  140.                 ),
  141.                 str_replace(':', '/', $ns)
  142.             );
  143.  
  144.             // ns start page
  145.             if($ns && page_exists($ns)) {
  146.                 $data[] = array(
  147.                     'id' => $ns,
  148.                     'ns' => getNS($ns),
  149.                     'title' => p_get_first_heading($ns, false),
  150.                     'size' => filesize(wikiFN($ns)),
  151.                     'mtime' => filemtime(wikiFN($ns)),
  152.                     'perm' => 16,
  153.                     'type' => 'f',
  154.                     'level' => 0,
  155.                     'open' => 1,
  156.                 );
  157.             }
  158.  
  159.             // go through all those pages
  160.             while($item = array_shift($data)) {
  161.                 $time = (int) p_get_metadata($item['id'], 'date created', false);
  162.                 if(!$time) $time = $item['mtime'];
  163.                 $lang = ($transplugin) ? $transplugin->getLangPart($item['id']) : '';
  164.  
  165.                 if($lang) $item['ns'] = preg_replace('/^' . $lang . '(:|$)/', '', $item['ns']);
  166.  
  167.                 $pages[$item['id']] = array(
  168.                     'title' => $item['title'],
  169.                     'ns' => $item['ns'],
  170.                     'size' => $item['size'],
  171.                     'time' => $time,
  172.                     'links' => array(),
  173.                     'linklabel' => array(),
  174.                     'media' => array(),
  175.                     'lang' => $lang
  176.                 );
  177.             }
  178.         }
  179.  
  180.         // now get links and media
  181.         foreach($pages as $pid => $item) {
  182.             // get instructions
  183.             $ins = p_cached_instructions(wikiFN($pid), false, $pid);
  184.             // find links and media usage
  185.             foreach($ins as $i) {
  186.                 $mid = null;
  187.  
  188.                 if($i[0] == 'internallink') {
  189.                     $id = $i[1][0];
  190.                     $label = $i[1][1];
  191.                     $exists = true;
  192.                     resolve_pageid($item['ns'], $id, $exists);
  193.                     list($id) = explode('#', $id, 2);
  194.                     if($id == $pid) continue; // skip self references
  195.                     if($exists && isset($pages[$id])) {
  196.                         $pages[$pid]['links'][] = $id;
  197.                         $pages[$pid]['linklabel'][] = $label;
  198.                     }
  199.                     if(is_array($i[1][1]) && $i[1][1]['type'] == 'internalmedia') {
  200.                         $mid = $i[1][1]['src']; // image link
  201.                     } else {
  202.                         continue; // we're done here
  203.                     }
  204.                 }
  205.  
  206.                 if($i[0] == 'internalmedia') {
  207.                     $mid = $i[1][0];
  208.                 }
  209.  
  210.                 if(is_null($mid)) continue;
  211.                 if($incmedia == 'none') continue; // no media wanted
  212.  
  213.                 $exists = true;
  214.                 resolve_mediaid($item['ns'], $mid, $exists);
  215.                 list($mid) = explode('#', $mid, 2);
  216.                 $mid = cleanID($mid);
  217.  
  218.                 if($exists) {
  219.                     if($incmedia == 'all') {
  220.                         if(!isset($media[$mid])) { //add node
  221.                             $media[$mid] = array(
  222.                                 'size' => filesize(mediaFN($mid)),
  223.                                 'time' => filemtime(mediaFN($mid)),
  224.                                 'ns' => getNS($mid),
  225.                                 'title' => noNS($mid),
  226.                             );
  227.                         }
  228.                         $pages[$pid]['media'][] = $mid;
  229.                     } elseif(isset($media[$mid])) {
  230.                         $pages[$pid]['media'][] = $mid;
  231.                     }
  232.                 }
  233.             }
  234.  
  235.             // clean up duplicates
  236.             $pages[$pid]['links'] = array_unique($pages[$pid]['links']);
  237.             $pages[$pid]['media'] = array_unique($pages[$pid]['media']);
  238.         }
  239.         //var_dump($pages);
  240.  
  241.         return array('pages' => $pages, 'media' => $media);
  242.  
  243.  
  244.     }
  245.  
  246.     /**
  247.      * Create a Graphviz dot representation
  248.      *
  249.      * @param array $data
  250.      * @param resource $fh
  251.      */
  252.     protected function create_dot(&$data, $fh) {
  253.         $pages =& $data['pages'];
  254.         $media =& $data['media'];
  255.  
  256.         fwrite($fh, "digraph G {\n");
  257.         fwrite($fh, "bgcolor=\"white\";
  258. node [shape = \"circle\", fontsize=16,	color=black, bgcolor=aquamarine];
  259. edge [arrowsize=2, color=black];
  260. \n");
  261.  
  262.         // create all nodes first
  263.         foreach($pages as $id => $page) {
  264.             fwrite($fh, "  \"page-$id\" [ 
  265.             shape=note,
  266.             label=\"{$page['title']}\",\n");
  267.             // look for links
  268.             if (!empty( $page['links'][0]))
  269. 					fwrite($fh,"URL=\"/". str_replace(':','/',$page['links'][0]) ."\"");
  270. 			else 	fwrite($fh,"URL=\"#\"");
  271. 			// look for media
  272.             if (!empty( $page['media'][0]))
  273.             {
  274. 				$path = explode (":", $page['media'][0]);
  275. 				$file = count($path );
  276. 				$folder="";
  277. 				for($i=0 ; $i < $file-1 ; $i++)
  278. 				{
  279. 					$folder = $folder . $path[$i] . "/";
  280. 				}
  281. 				fwrite($fh,"\n // " . $page['media'][0] . " ### " . $file ." ### ". $path[0] . " ### ". $path[1]);
  282. 				fwrite($fh,",\n           labelloc=\"b\",\n          shapefile=\"".DOKU_INC. "data/media/" . $folder . "300x300-" . $path[$file-1] . "\"");
  283. 			}
  284. 			fwrite($fh,";\n ];\n");
  285.         }
  286.                  //  var_dump($pages);
  287. /*
  288.        foreach($media as $id => $item) {
  289.             fwrite($fh, "    \"media-$id\" [shape=box, label=\"$id\", color=sandybrown, fontname=Helvetica];\n");
  290.         }
  291.   */      
  292.         // now create all the links
  293.         foreach($pages as $id => $page) {
  294.             foreach($page['links'] as $link) {
  295. 				foreach($page['linklabel'] as $label) {
  296. 					fwrite($fh, "    \"page-$id\" -> \"page-$link\" [color=navy;label=\"".$label."\"];\n");
  297. 				}
  298. 			}
  299.             /*
  300.             foreach($page['media'] as $link) {
  301.                 fwrite($fh, "    \"page-$id\" -> \"media-$link\" [color=firebrick];\n");
  302.             }
  303.             */
  304.         }
  305.         fwrite($fh, "}\n");
  306.     }
  307.  
  308.     /**
  309.      * Create a GEXF representation
  310.      *
  311.      * @param array $data
  312.      * @param resource $fh
  313.      */
  314.     protected function create_gexf(&$data, $fh) {
  315.         $pages =& $data['pages'];
  316.         $media =& $data['media'];
  317.  
  318.         fwrite($fh, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
  319.         fwrite(
  320.             $fh, "<gexf xmlns=\"http://www.gexf.net/1.1draft\" version=\"1.1\"
  321.                    xmlns:viz=\"http://www.gexf.net/1.1draft/viz\">\n"
  322.         );
  323.         fwrite($fh, "    <meta lastmodifieddate=\"" . date('Y-m-d H:i:s') . "\">\n");
  324.         fwrite($fh, "        <creator>DokuWiki</creator>\n");
  325.         fwrite($fh, "    </meta>\n");
  326.         fwrite($fh, "    <graph mode=\"dynamic\" defaultedgetype=\"directed\">\n");
  327.  
  328.         // define attributes
  329.         fwrite($fh, "        <attributes class=\"node\">\n");
  330.         fwrite($fh, "            <attribute id=\"title\" title=\"Title\" type=\"string\" />\n");
  331.         fwrite($fh, "            <attribute id=\"lang\" title=\"Language\" type=\"string\" />\n");
  332.         fwrite($fh, "            <attribute id=\"ns\" title=\"Namespace\" type=\"string\" />\n");
  333.         fwrite($fh, "            <attribute id=\"type\" title=\"Type\" type=\"liststring\">\n");
  334.         fwrite($fh, "                <default>page|media</default>\n");
  335.         fwrite($fh, "            </attribute>\n");
  336.         fwrite($fh, "            <attribute id=\"time\" title=\"Created\" type=\"long\" />\n");
  337.         fwrite($fh, "            <attribute id=\"size\" title=\"File Size\" type=\"long\" />\n");
  338.         fwrite($fh, "        </attributes>\n");
  339.  
  340.         // create all nodes first
  341.         fwrite($fh, "        <nodes>\n");
  342.         foreach($pages as $id => $item) {
  343.             $title = htmlspecialchars($item['title']);
  344.             $lang = htmlspecialchars($item['lang']);
  345.             fwrite($fh, "            <node id=\"page-$id\" label=\"$id\" start=\"{$item['time']}\">\n");
  346.             fwrite($fh, "               <attvalues>\n");
  347.             fwrite($fh, "                   <attvalue for=\"type\" value=\"page\" />\n");
  348.             fwrite($fh, "                   <attvalue for=\"title\" value=\"$title\" />\n");
  349.             fwrite($fh, "                   <attvalue for=\"lang\" value=\"$lang\" />\n");
  350.             fwrite($fh, "                   <attvalue for=\"ns\" value=\"{$item['ns']}\" />\n");
  351.             fwrite($fh, "                   <attvalue for=\"time\" value=\"{$item['time']}\" />\n");
  352.             fwrite($fh, "                   <attvalue for=\"size\" value=\"{$item['size']}\" />\n");
  353.             fwrite($fh, "               </attvalues>\n");
  354.             fwrite($fh, "               <viz:shape value=\"square\" />\n");
  355.             fwrite($fh, "               <viz:color r=\"173\" g=\"216\" b=\"230\" />\n");
  356.             fwrite($fh, "            </node>\n");
  357.         }
  358.         foreach($media as $id => $item) {
  359.             $title = htmlspecialchars($item['title']);
  360.             $lang = htmlspecialchars($item['lang']);
  361.             fwrite($fh, "            <node id=\"media-$id\" label=\"$id\" start=\"{$item['time']}\">\n");
  362.             fwrite($fh, "               <attvalues>\n");
  363.             fwrite($fh, "                   <attvalue for=\"type\" value=\"media\" />\n");
  364.             fwrite($fh, "                   <attvalue for=\"title\" value=\"$title\" />\n");
  365.             fwrite($fh, "                   <attvalue for=\"lang\" value=\"$lang\" />\n");
  366.             fwrite($fh, "                   <attvalue for=\"ns\" value=\"{$item['ns']}\" />\n");
  367.             fwrite($fh, "                   <attvalue for=\"time\" value=\"{$item['time']}\" />\n");
  368.             fwrite($fh, "                   <attvalue for=\"size\" value=\"{$item['size']}\" />\n");
  369.             fwrite($fh, "               </attvalues>\n");
  370.             fwrite($fh, "               <viz:shape value=\"disc\" />\n");
  371.             fwrite($fh, "               <viz:color r=\"244\" g=\"164\" b=\"96\" />\n");
  372.             fwrite($fh, "            </node>\n");
  373.         }
  374.         fwrite($fh, "        </nodes>\n");
  375.  
  376.         // now create all the edges
  377.         fwrite($fh, "        <edges>\n");
  378.         $cnt = 0;
  379.         foreach($pages as $id => $page) {
  380.             foreach($page['links'] as $link) {
  381.                 $cnt++;
  382.                 fwrite($fh, "            <edge id=\"$cnt\" source=\"page-$id\" target=\"page-$link\" />\n");
  383.             }
  384.             foreach($page['media'] as $link) {
  385.                 $cnt++;
  386.                 fwrite($fh, "            <edge id=\"$cnt\" source=\"page-$id\" target=\"media-$link\" />\n");
  387.             }
  388.         }
  389.         fwrite($fh, "        </edges>\n");
  390.  
  391.         fwrite($fh, "    </graph>\n");
  392.         fwrite($fh, "</gexf>\n");
  393.     }
  394.  
  395. }
  396.  
  397. $grapher = new Grapher();
  398. $grapher->run();
You could leave a comment if you were logged in.
  • dev/grapher-php.1601992769.txt.gz
  • Last modified: 2020/10/06 13:59
  • by bab