├── README.md
└── grapher.php


/README.md:
--------------------------------------------------------------------------------
 1 | # DokuWiki Grapher
 2 | 
 3 | This is a simple script to generate a directed graph description from DokuWiki link structures. Read the [introductional blog post](https://www.splitbrain.org/blog/2010-08/02-graphing_dokuwiki_help_needed) for some more info.
 4 | 
 5 | ## Installing
 6 | 
 7 | Download the [grapher.php](https://raw.githubusercontent.com/splitbrain/dokuwiki-grapher/master/grapher.php) file into your DokuWiki ``bin`` directory. Then run it from command line.
 8 | 
 9 | ## Usage
10 | 
11 | See ``bin/grapher.php --help``:
12 | 
13 | ```
14 | USAGE: grapher.php <OPTIONS> [<namespaces>]
15 | 
16 |   Creates a graph representation of pages and media files and how they    
17 |   are interlinked.                                                        
18 |                                                                           
19 | 
20 |   OPTIONS
21 | 
22 |   -d <depth>, --depth Recursion depth, eg. how deep to look into the      
23 |   <depth>             given namespaces. Use 0 for all. Default: 1         
24 | 
25 |   -m <ns|all|none>,   How to handle media files. 'ns' includes only media 
26 |   --media             that is located in the given namespaces, 'all'      
27 |   <ns|all|none>       includes all media files and 'none' ignores the     
28 |                       media files completely. Default: ns                 
29 | 
30 |   -f <dot|gexf>,      The wanted output format. 'dot' is a very simple    
31 |   --format <dot|gexf> format which can be used to visualize the resulting 
32 |                       graph with graphviz. The 'gexf' format is a more    
33 |                       complex XML-based format which contains more info   
34 |                       about the found nodes and can be loaded in Gephi.   
35 |                       Default: dot                                        
36 | 
37 |   -o <file>, --output Where to store the output eg. a filename. If not    
38 |   <file>              given the output is written to STDOUT.              
39 | 
40 |   --no-colors         Do not use any colors in output. Useful when piping 
41 |                       output to other tools or files.                     
42 | 
43 |   -h, --help          Display this help screen and exit immeadiately.     
44 | 
45 | 
46 |   <namespaces>        Give all wiki namespaces you want to have graphed.  
47 |                       If no namespace is given, the root namespace is     
48 |                       assumed.
49 | ```
50 | 
51 | ## Visualize
52 | 
53 | Run the created file through [GraphViz](http://www.graphviz.org/) or [Gephi](https://gephi.org/).


--------------------------------------------------------------------------------
/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 |                     'media' => array(),
174 |                     'lang' => $lang
175 |                 );
176 |             }
177 |         }
178 | 
179 |         // now get links and media
180 |         foreach($pages as $pid => $item) {
181 |             // get instructions
182 |             $ins = p_cached_instructions(wikiFN($pid), false, $pid);
183 |             // find links and media usage
184 |             foreach($ins as $i) {
185 |                 $mid = null;
186 | 
187 |                 if($i[0] == 'internallink') {
188 |                     $id = $i[1][0];
189 |                     $exists = true;
190 |                     resolve_pageid($item['ns'], $id, $exists);
191 |                     list($id) = explode('#', $id, 2);
192 |                     if($id == $pid) continue; // skip self references
193 |                     if($exists && isset($pages[$id])) {
194 |                         $pages[$pid]['links'][] = $id;
195 |                     }
196 |                     if(is_array($i[1][1]) && $i[1][1]['type'] == 'internalmedia') {
197 |                         $mid = $i[1][1]['src']; // image link
198 |                     } else {
199 |                         continue; // we're done here
200 |                     }
201 |                 }
202 | 
203 |                 if($i[0] == 'internalmedia') {
204 |                     $mid = $i[1][0];
205 |                 }
206 | 
207 |                 if(is_null($mid)) continue;
208 |                 if($incmedia == 'none') continue; // no media wanted
209 | 
210 |                 $exists = true;
211 |                 resolve_mediaid($item['ns'], $mid, $exists);
212 |                 list($mid) = explode('#', $mid, 2);
213 |                 $mid = cleanID($mid);
214 | 
215 |                 if($exists) {
216 |                     if($incmedia == 'all') {
217 |                         if(!isset($media[$mid])) { //add node
218 |                             $media[$mid] = array(
219 |                                 'size' => filesize(mediaFN($mid)),
220 |                                 'time' => filemtime(mediaFN($mid)),
221 |                                 'ns' => getNS($mid),
222 |                                 'title' => noNS($mid),
223 |                             );
224 |                         }
225 |                         $pages[$pid]['media'][] = $mid;
226 |                     } elseif(isset($media[$mid])) {
227 |                         $pages[$pid]['media'][] = $mid;
228 |                     }
229 |                 }
230 |             }
231 | 
232 |             // clean up duplicates
233 |             $pages[$pid]['links'] = array_unique($pages[$pid]['links']);
234 |             $pages[$pid]['media'] = array_unique($pages[$pid]['media']);
235 |         }
236 | 
237 |         return array('pages' => $pages, 'media' => $media);
238 |     }
239 | 
240 |     /**
241 |      * Create a Graphviz dot representation
242 |      *
243 |      * @param array $data
244 |      * @param resource $fh
245 |      */
246 |     protected function create_dot(&$data, $fh) {
247 |         $pages =& $data['pages'];
248 |         $media =& $data['media'];
249 | 
250 |         fwrite($fh, "digraph G {\n");
251 |         // create all nodes first
252 |         foreach($pages as $id => $page) {
253 |             fwrite($fh, "    \"page-$id\" [shape=note, label=\"$id\\n{$page['title']}\", color=lightblue, fontname=Helvetica];\n");
254 |         }
255 |         foreach($media as $id => $item) {
256 |             fwrite($fh, "    \"media-$id\" [shape=box, label=\"$id\", color=sandybrown, fontname=Helvetica];\n");
257 |         }
258 |         // now create all the links
259 |         foreach($pages as $id => $page) {
260 |             foreach($page['links'] as $link) {
261 |                 fwrite($fh, "    \"page-$id\" -> \"page-$link\" [color=navy];\n");
262 |             }
263 |             foreach($page['media'] as $link) {
264 |                 fwrite($fh, "    \"page-$id\" -> \"media-$link\" [color=firebrick];\n");
265 |             }
266 |         }
267 |         fwrite($fh, "}\n");
268 |     }
269 | 
270 |     /**
271 |      * Create a GEXF representation
272 |      *
273 |      * @param array $data
274 |      * @param resource $fh
275 |      */
276 |     protected function create_gexf(&$data, $fh) {
277 |         $pages =& $data['pages'];
278 |         $media =& $data['media'];
279 | 
280 |         fwrite($fh, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
281 |         fwrite(
282 |             $fh, "<gexf xmlns=\"http://www.gexf.net/1.1draft\" version=\"1.1\"
283 |                    xmlns:viz=\"http://www.gexf.net/1.1draft/viz\">\n"
284 |         );
285 |         fwrite($fh, "    <meta lastmodifieddate=\"" . date('Y-m-d H:i:s') . "\">\n");
286 |         fwrite($fh, "        <creator>DokuWiki</creator>\n");
287 |         fwrite($fh, "    </meta>\n");
288 |         fwrite($fh, "    <graph mode=\"dynamic\" defaultedgetype=\"directed\">\n");
289 | 
290 |         // define attributes
291 |         fwrite($fh, "        <attributes class=\"node\">\n");
292 |         fwrite($fh, "            <attribute id=\"title\" title=\"Title\" type=\"string\" />\n");
293 |         fwrite($fh, "            <attribute id=\"lang\" title=\"Language\" type=\"string\" />\n");
294 |         fwrite($fh, "            <attribute id=\"ns\" title=\"Namespace\" type=\"string\" />\n");
295 |         fwrite($fh, "            <attribute id=\"type\" title=\"Type\" type=\"liststring\">\n");
296 |         fwrite($fh, "                <default>page|media</default>\n");
297 |         fwrite($fh, "            </attribute>\n");
298 |         fwrite($fh, "            <attribute id=\"time\" title=\"Created\" type=\"long\" />\n");
299 |         fwrite($fh, "            <attribute id=\"size\" title=\"File Size\" type=\"long\" />\n");
300 |         fwrite($fh, "        </attributes>\n");
301 | 
302 |         // create all nodes first
303 |         fwrite($fh, "        <nodes>\n");
304 |         foreach($pages as $id => $item) {
305 |             $title = htmlspecialchars($item['title']);
306 |             $lang = htmlspecialchars($item['lang']);
307 |             fwrite($fh, "            <node id=\"page-$id\" label=\"$id\" start=\"{$item['time']}\">\n");
308 |             fwrite($fh, "               <attvalues>\n");
309 |             fwrite($fh, "                   <attvalue for=\"type\" value=\"page\" />\n");
310 |             fwrite($fh, "                   <attvalue for=\"title\" value=\"$title\" />\n");
311 |             fwrite($fh, "                   <attvalue for=\"lang\" value=\"$lang\" />\n");
312 |             fwrite($fh, "                   <attvalue for=\"ns\" value=\"{$item['ns']}\" />\n");
313 |             fwrite($fh, "                   <attvalue for=\"time\" value=\"{$item['time']}\" />\n");
314 |             fwrite($fh, "                   <attvalue for=\"size\" value=\"{$item['size']}\" />\n");
315 |             fwrite($fh, "               </attvalues>\n");
316 |             fwrite($fh, "               <viz:shape value=\"square\" />\n");
317 |             fwrite($fh, "               <viz:color r=\"173\" g=\"216\" b=\"230\" />\n");
318 |             fwrite($fh, "            </node>\n");
319 |         }
320 |         foreach($media as $id => $item) {
321 |             $title = htmlspecialchars($item['title']);
322 |             $lang = htmlspecialchars($item['lang']);
323 |             fwrite($fh, "            <node id=\"media-$id\" label=\"$id\" start=\"{$item['time']}\">\n");
324 |             fwrite($fh, "               <attvalues>\n");
325 |             fwrite($fh, "                   <attvalue for=\"type\" value=\"media\" />\n");
326 |             fwrite($fh, "                   <attvalue for=\"title\" value=\"$title\" />\n");
327 |             fwrite($fh, "                   <attvalue for=\"lang\" value=\"$lang\" />\n");
328 |             fwrite($fh, "                   <attvalue for=\"ns\" value=\"{$item['ns']}\" />\n");
329 |             fwrite($fh, "                   <attvalue for=\"time\" value=\"{$item['time']}\" />\n");
330 |             fwrite($fh, "                   <attvalue for=\"size\" value=\"{$item['size']}\" />\n");
331 |             fwrite($fh, "               </attvalues>\n");
332 |             fwrite($fh, "               <viz:shape value=\"disc\" />\n");
333 |             fwrite($fh, "               <viz:color r=\"244\" g=\"164\" b=\"96\" />\n");
334 |             fwrite($fh, "            </node>\n");
335 |         }
336 |         fwrite($fh, "        </nodes>\n");
337 | 
338 |         // now create all the edges
339 |         fwrite($fh, "        <edges>\n");
340 |         $cnt = 0;
341 |         foreach($pages as $id => $page) {
342 |             foreach($page['links'] as $link) {
343 |                 $cnt++;
344 |                 fwrite($fh, "            <edge id=\"$cnt\" source=\"page-$id\" target=\"page-$link\" />\n");
345 |             }
346 |             foreach($page['media'] as $link) {
347 |                 $cnt++;
348 |                 fwrite($fh, "            <edge id=\"$cnt\" source=\"page-$id\" target=\"media-$link\" />\n");
349 |             }
350 |         }
351 |         fwrite($fh, "        </edges>\n");
352 | 
353 |         fwrite($fh, "    </graph>\n");
354 |         fwrite($fh, "</gexf>\n");
355 |     }
356 | 
357 | }
358 | 
359 | $grapher = new Grapher();
360 | $grapher->run();
361 | 


--------------------------------------------------------------------------------