ARC2_ErdfExtractor::__construct() - Code Metrics - Enalean/tuleap - Measure and Improve Code Quality continuously with Scrutinizer

ARC2_ErdfExtractor::__construct() A
last analyzed 2016-01-22 17:23 UTC

↳ Parent: ARC2_ErdfExtractor

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Metric	Value
dl	0
loc	3
rs	10
cc	1
eloc	2
nc	1
nop	2

<?php
/*
homepage: http://arc.semsol.org/
license:  http://arc.semsol.org/license

class:    ARC2 eRDF Extractor (w/o link title generation)
author:   Benjamin Nowack
version:  2009-02-09 (Tweak: getRootNode returns 1st node if html tag is not found)
*/

ARC2::inc('RDFExtractor');

class ARC2_ErdfExtractor extends ARC2_RDFExtractor {

  function __construct($a = '', &$caller) {
    parent::__construct($a, $caller);
  }
  
  function ARC2_ErdfExtractor($a = '', &$caller) {
    $this->__construct($a, $caller);
  }

  function __init() {
    parent::__init();
  }

  /*  */
  
  function extractRDF() {
    if (!isset($this->caller->detected_formats['erdf'])) return 0;
    $root_node = $this->getRootNode();
    $base = $this->getDocBase();
    $ns = $this->getNamespaces();
    $context = array(
      'base' => $base,
      'prev_res' => $base,
      'cur_res' => $base,
      'ns' => $ns,
      'lang' => '',
    );
    $this->processNode($root_node, $context);
  }
  
  /*  */
  
  function getRootNode() {
    foreach ($this->nodes as $id => $node) {
      if ($node['tag'] == 'html') {
        return $node;
      }
    }
    return $this->nodes[0];
  }
  
  function getNamespaces() {
    $r = array(
      'rdf' => 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
      'rdfs' => 'http://www.w3.org/2000/01/rdf-schema#'
    );
    foreach ($this->nodes as $id => $node) {
      if (preg_match('/^(link|a)$/', $node['tag']) && isset($node['a']['rel']) && preg_match('/schema\.([^\s]+)/is', $node['a']['rel'], $m) && isset($node['a']['href uri'])) {
        $r[$m[1]] = $node['a']['href uri'];
      }
    }
    return $r;
  }

  /*  */
  
  function processNode($n, $ct) {
    /* context */
    //$ct['lang'] = $this->v('xml:lang', $ct['lang'], $n['a']);
    $ct['lang'] = '';
    $ct['prop_uris'] = $this->getPropertyURIs($n, $ct);
    $ct['prev_res'] = $ct['cur_res'];
    $ct['cur_res'] = $this->getCurrentResourceURI($n, $ct);
    $ct['cur_obj_id'] = $this->getCurrentObjectID($n, $ct);
    $ct['cur_obj_literal'] = $this->getCurrentObjectLiteral($n, $ct);
    /* triple production (http://research.talis.com/2005/erdf/wiki/Main/SummaryOfTripleProductionRules) */
    foreach ($ct['prop_uris'] as $type => $uris) {
      foreach ($uris as $uri) {
        $rdf_type = preg_match('/^ /', $uri) ? 1 : 0;
        /* meta + name */
        if (($type == 'name') && ($n['tag'] == 'meta')) {
          $t = array(
            's' => $ct['cur_res'],
            's_type' => 'uri',
            'p' => $uri, 
            'o' => $ct['cur_obj_literal']['value'],
            'o_type' => 'literal',
            'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
            'o_datatype' => $ct['cur_obj_literal']['datatype'],
          );
          $this->addT($t);
        }
        /* class */
        if ($type == 'class') {
          if ($rdf_type) {
            $s = $this->v('href uri', $ct['cur_res'], $n['a']);
            $s = $this->v('src uri', $s, $n['a']);
            $t = array(
              's' => $s,
              's_type' => 'uri',
              'p' => $ct['ns']['rdf'] . 'type', 
              'o' => trim($uri),
              'o_type' => 'uri',
              'o_lang' => '',
              'o_datatype' => '',
            );
          }
          elseif (isset($n['a']['id'])) {/* used as object */
            $t = array(
              's' => $ct['prev_res'],
              's_type' => 'uri',
              'p' => $uri, 
              'o' => $ct['cur_res'],
              'o_type' => 'uri',
              'o_lang' => '',
              'o_datatype' => '',
            );
          }
          else {
            $t = array(
              's' => $ct['cur_res'],
              's_type' => 'uri',
              'p' => $uri, 
              'o' => $ct['cur_obj_literal']['value'],
              'o_type' => 'literal',
              'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
              'o_datatype' => $ct['cur_obj_literal']['datatype'],
            );
            if (($o = $this->v('src uri', '', $n['a'])) || ($o = $this->v('href uri', '', $n['a']))) {
              if (!$ct['prop_uris']['rel'] && !$ct['prop_uris']['rev']) {
                $t['o'] = $o;
                $t['o_type'] = 'uri';
                $t['o_lang'] = '';
                $t['o_datatype'] = '';
              }
            }
          }
          $this->addT($t);
        }
        /* rel */
        if ($type == 'rel') {
          if (($o = $this->v('src uri', '', $n['a'])) || ($o = $this->v('href uri', '', $n['a']))) {
            $t = array(
              's' => $ct['cur_res'],
              's_type' => 'uri',
              'p' => $uri, 
              'o' => $o,
              'o_type' => 'uri',
              'o_lang' => '',
              'o_datatype' => '',
            );
            $this->addT($t);
          }
        }
        /* rev */
        if ($type == 'rev') {
          if (($s = $this->v('src uri', '', $n['a'])) || ($s = $this->v('href uri', '', $n['a']))) {
            $t = array(
              's' => $s,
              's_type' => 'uri',
              'p' => $uri, 
              'o' => $ct['cur_res'],
              'o_type' => 'uri',
              'o_lang' => '',
              'o_datatype' => '',
            );
            $this->addT($t);
          }
        }
      }
    }
    /* imgs */
    if ($n['tag'] == 'img') {
      if (($s = $this->v('src uri', '', $n['a'])) && $ct['cur_obj_literal']['value']) {
        $t = array(
          's' => $s,
          's_type' => 'uri',
          'p' => $ct['ns']['rdfs'] . 'label', 
          'o' => $ct['cur_obj_literal']['value'],
          'o_type' => 'literal',
          'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
          'o_datatype' => $ct['cur_obj_literal']['datatype'],
        );
        $this->addT($t);
      }
    }
    /* anchors */
    if ($n['tag'] == 'a') {
      if (($s = $this->v('href uri', '', $n['a'])) && $ct['cur_obj_literal']['value']) {
        $t = array(
          's' => $s,
          's_type' => 'uri',
          'p' => $ct['ns']['rdfs'] . 'label', 
          'o' => $ct['cur_obj_literal']['value'],
          'o_type' => 'literal',
          'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
          'o_datatype' => $ct['cur_obj_literal']['datatype'],
        );
        $this->addT($t);
      }
    }
    /* recurse */
    if ($n['tag'] == 'a') {
      $ct['cur_res'] = $ct['cur_obj_id'];
    }
    $sub_nodes = $this->getSubNodes($n);
    foreach ($sub_nodes as $sub_node) {
      $this->processNode($sub_node, $ct);
    }
  }

  /*  */
  
  function getPropertyURIs($n, $ct) {
    $r = array();
    foreach (array('rel', 'rev', 'class', 'name', 'src') as $type) {
      $r[$type] = array();
      $vals = $this->v($type . ' m', array(), $n['a']);
      foreach ($vals as $val) {
        if (!trim($val)) continue;
        list($uri, $sub_v) = $this->xQname(trim($val, '- '), $ct['base'], $ct['ns'], $type);
<?php

function returnThreeValues() {
    return array('a', 'b', 'c');
}

list($a, $b, $c) = returnThreeValues();

print $a . " - " . $c;
        if (!$uri) continue;
        $rdf_type = preg_match('/^-/', trim($val)) ? 1 : 0;
        $r[$type][] = $rdf_type ? ' ' . $uri : $uri;
      }
    }
    return $r;
  }

  function getCurrentResourceURI($n, $ct) {
    if (isset($n['a']['id'])) {
      list($r, $sub_v) = $this->xURI('#' . $n['a']['id'], $ct['base'], $ct['ns']);
<?php

function returnThreeValues() {
    return array('a', 'b', 'c');
}

list($a, $b, $c) = returnThreeValues();

print $a . " - " . $c;
      return $r;
    }
    return $ct['cur_res'];
  }
  
  function getCurrentObjectID($n, $ct) {
    foreach (array('href', 'src') as $a) {
      if (isset($n['a'][$a])) {
        list($r, $sub_v) = $this->xURI($n['a'][$a], $ct['base'], $ct['ns']);
<?php

function returnThreeValues() {
    return array('a', 'b', 'c');
}

list($a, $b, $c) = returnThreeValues();

print $a . " - " . $c;
        return $r;
      }
    }
    return $this->createBnodeID();
  }

  function getCurrentObjectLiteral($n, $ct) {
    $r = array('value' => '', 'lang' => $ct['lang'], 'datatype' => '');
    if (isset($n['a']['content'])) {
      $r['value'] = $n['a']['content'];
    }
    elseif (isset($n['a']['title'])) {
      $r['value'] = $n['a']['title'];
    }
    else {
      $r['value'] = $this->getPlainContent($n);
    }
    return $r;
  }
  
  /*  */
  
  function xURI($v, $base, $ns, $attr_type = '') {
    if ((list($sub_r, $sub_v) = $this->xQname($v, $base, $ns)) && $sub_r) {
      return array($sub_r, $sub_v);
    }
    if (preg_match('/^(rel|rev|class|name)$/', $attr_type) && preg_match('/^[a-z0-9]+$/', $v)) {
      return array(0, $v);
    }
    return array($this->calcURI($v, $base), '');
  }
  
  function xQname($v, $base, $ns) {
    if ($sub_r = $this->x('([a-z0-9\-\_]+)[\-\.]([a-z0-9\-\_]+)', $v)) {
      if (isset($ns[$sub_r[1]])) {
        return array($ns[$sub_r[1]] . $sub_r[2], '');
      }
    }
    return array(0, $v);
  }
  
  /*  */

}


1			<?php
2			/*
3			homepage: http://arc.semsol.org/
4			license: http://arc.semsol.org/license
5
6			class: ARC2 eRDF Extractor (w/o link title generation)
7			author: Benjamin Nowack
8			version: 2009-02-09 (Tweak: getRootNode returns 1st node if html tag is not found)
9			*/
10
11			ARC2::inc('RDFExtractor');
12
13			class ARC2_ErdfExtractor extends ARC2_RDFExtractor {
14
15			function __construct($a = '', &$caller) {
16			parent::__construct($a, $caller);
17			}
18
19			function ARC2_ErdfExtractor($a = '', &$caller) {
20			$this->__construct($a, $caller);
21			}
22
23			function __init() {
24			parent::__init();
25			}
26
27			/* */
28
29			function extractRDF() {
30			if (!isset($this->caller->detected_formats['erdf'])) return 0;
31			$root_node = $this->getRootNode();
32			$base = $this->getDocBase();
33			$ns = $this->getNamespaces();
34			$context = array(
35			'base' => $base,
36			'prev_res' => $base,
37			'cur_res' => $base,
38			'ns' => $ns,
39			'lang' => '',
40			);
41			$this->processNode($root_node, $context);
42			}
43
44			/* */
45
46			function getRootNode() {
47			foreach ($this->nodes as $id => $node) {
48			if ($node['tag'] == 'html') {
49			return $node;
50			}
51			}
52			return $this->nodes[0];
53			}
54
55			function getNamespaces() {
56			$r = array(
57			'rdf' => 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
58			'rdfs' => 'http://www.w3.org/2000/01/rdf-schema#'
59			);
60			foreach ($this->nodes as $id => $node) {
61			if (preg_match('/^(link\|a)$/', $node['tag']) && isset($node['a']['rel']) && preg_match('/schema\.([^\s]+)/is', $node['a']['rel'], $m) && isset($node['a']['href uri'])) {
62			$r[$m[1]] = $node['a']['href uri'];
63			}
64			}
65			return $r;
66			}
67
68			/* */
69
70			function processNode($n, $ct) {
71			/* context */
72			//$ct['lang'] = $this->v('xml:lang', $ct['lang'], $n['a']);
73			$ct['lang'] = '';
74			$ct['prop_uris'] = $this->getPropertyURIs($n, $ct);
75			$ct['prev_res'] = $ct['cur_res'];
76			$ct['cur_res'] = $this->getCurrentResourceURI($n, $ct);
77			$ct['cur_obj_id'] = $this->getCurrentObjectID($n, $ct);
78			$ct['cur_obj_literal'] = $this->getCurrentObjectLiteral($n, $ct);
79			/* triple production (http://research.talis.com/2005/erdf/wiki/Main/SummaryOfTripleProductionRules) */
80			foreach ($ct['prop_uris'] as $type => $uris) {
81			foreach ($uris as $uri) {
82			$rdf_type = preg_match('/^ /', $uri) ? 1 : 0;
83			/* meta + name */
84			if (($type == 'name') && ($n['tag'] == 'meta')) {
85			$t = array(
86			's' => $ct['cur_res'],
87			's_type' => 'uri',
88			'p' => $uri,
89			'o' => $ct['cur_obj_literal']['value'],
90			'o_type' => 'literal',
91			'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
92			'o_datatype' => $ct['cur_obj_literal']['datatype'],
93			);
94			$this->addT($t);
95			}
96			/* class */
97			if ($type == 'class') {
98			if ($rdf_type) {
99			$s = $this->v('href uri', $ct['cur_res'], $n['a']);
100			$s = $this->v('src uri', $s, $n['a']);
101			$t = array(
102			's' => $s,
103			's_type' => 'uri',
104			'p' => $ct['ns']['rdf'] . 'type',
105			'o' => trim($uri),
106			'o_type' => 'uri',
107			'o_lang' => '',
108			'o_datatype' => '',
109			);
110			}
111			elseif (isset($n['a']['id'])) {/* used as object */
112			$t = array(
113			's' => $ct['prev_res'],
114			's_type' => 'uri',
115			'p' => $uri,
116			'o' => $ct['cur_res'],
117			'o_type' => 'uri',
118			'o_lang' => '',
119			'o_datatype' => '',
120			);
121			}
122			else {
123			$t = array(
124			's' => $ct['cur_res'],
125			's_type' => 'uri',
126			'p' => $uri,
127			'o' => $ct['cur_obj_literal']['value'],
128			'o_type' => 'literal',
129			'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
130			'o_datatype' => $ct['cur_obj_literal']['datatype'],
131			);
132			if (($o = $this->v('src uri', '', $n['a'])) \|\| ($o = $this->v('href uri', '', $n['a']))) {
133			if (!$ct['prop_uris']['rel'] && !$ct['prop_uris']['rev']) {
134			$t['o'] = $o;
135			$t['o_type'] = 'uri';
136			$t['o_lang'] = '';
137			$t['o_datatype'] = '';
138			}
139			}
140			}
141			$this->addT($t);
142			}
143			/* rel */
144			if ($type == 'rel') {
145			if (($o = $this->v('src uri', '', $n['a'])) \|\| ($o = $this->v('href uri', '', $n['a']))) {
146			$t = array(
147			's' => $ct['cur_res'],
148			's_type' => 'uri',
149			'p' => $uri,
150			'o' => $o,
151			'o_type' => 'uri',
152			'o_lang' => '',
153			'o_datatype' => '',
154			);
155			$this->addT($t);
156			}
157			}
158			/* rev */
159			if ($type == 'rev') {
160			if (($s = $this->v('src uri', '', $n['a'])) \|\| ($s = $this->v('href uri', '', $n['a']))) {
161			$t = array(
162			's' => $s,
163			's_type' => 'uri',
164			'p' => $uri,
165			'o' => $ct['cur_res'],
166			'o_type' => 'uri',
167			'o_lang' => '',
168			'o_datatype' => '',
169			);
170			$this->addT($t);
171			}
172			}
173			}
174			}
175			/* imgs */
176			if ($n['tag'] == 'img') {
177			if (($s = $this->v('src uri', '', $n['a'])) && $ct['cur_obj_literal']['value']) {
178			$t = array(
179			's' => $s,
180			's_type' => 'uri',
181			'p' => $ct['ns']['rdfs'] . 'label',
182			'o' => $ct['cur_obj_literal']['value'],
183			'o_type' => 'literal',
184			'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
185			'o_datatype' => $ct['cur_obj_literal']['datatype'],
186			);
187			$this->addT($t);
188			}
189			}
190			/* anchors */
191			if ($n['tag'] == 'a') {
192			if (($s = $this->v('href uri', '', $n['a'])) && $ct['cur_obj_literal']['value']) {
193			$t = array(
194			's' => $s,
195			's_type' => 'uri',
196			'p' => $ct['ns']['rdfs'] . 'label',
197			'o' => $ct['cur_obj_literal']['value'],
198			'o_type' => 'literal',
199			'o_lang' => $ct['cur_obj_literal']['datatype'] ? '' : $ct['cur_obj_literal']['lang'],
200			'o_datatype' => $ct['cur_obj_literal']['datatype'],
201			);
202			$this->addT($t);
203			}
204			}
205			/* recurse */
206			if ($n['tag'] == 'a') {
207			$ct['cur_res'] = $ct['cur_obj_id'];
208			}
209			$sub_nodes = $this->getSubNodes($n);
210			foreach ($sub_nodes as $sub_node) {
211			$this->processNode($sub_node, $ct);
212			}
213			}
214
215			/* */
216
217			function getPropertyURIs($n, $ct) {
218			$r = array();
219			foreach (array('rel', 'rev', 'class', 'name', 'src') as $type) {
220			$r[$type] = array();
221			$vals = $this->v($type . ' m', array(), $n['a']);
222			foreach ($vals as $val) {
223			if (!trim($val)) continue;
224			list($uri, $sub_v) = $this->xQname(trim($val, '- '), $ct['base'], $ct['ns'], $type);
			0 ignored issues – show Unused Code introduced 2015-12-16 20:30 UTC by Report Bug Copy Issue Report The assignment to `$sub_v` is unused. Consider omitting it like so `list($first,,$third)`. This checks looks for assignemnts to variables using the `list(...)` function, where not all assigned variables are subsequently used. Consider the following code example. <?php function returnThreeValues() { return array('a', 'b', 'c'); } list($a, $b, $c) = returnThreeValues(); print $a . " - " . $c; Only the variables `$a` and `$c` are used. There was no need to assign `$b`. Instead, the list call could have been. list($a,, $c) = returnThreeValues(); Loading history...
225			if (!$uri) continue;
226			$rdf_type = preg_match('/^-/', trim($val)) ? 1 : 0;
227			$r[$type][] = $rdf_type ? ' ' . $uri : $uri;
228			}
229			}
230			return $r;
231			}
232
233			function getCurrentResourceURI($n, $ct) {
234			if (isset($n['a']['id'])) {
235			list($r, $sub_v) = $this->xURI('#' . $n['a']['id'], $ct['base'], $ct['ns']);
			0 ignored issues – show Unused Code introduced 2015-12-16 20:30 UTC by Report Bug Copy Issue Report The assignment to `$sub_v` is unused. Consider omitting it like so `list($first,,$third)`. This checks looks for assignemnts to variables using the `list(...)` function, where not all assigned variables are subsequently used. Consider the following code example. <?php function returnThreeValues() { return array('a', 'b', 'c'); } list($a, $b, $c) = returnThreeValues(); print $a . " - " . $c; Only the variables `$a` and `$c` are used. There was no need to assign `$b`. Instead, the list call could have been. list($a,, $c) = returnThreeValues(); Loading history...
236			return $r;
237			}
238			return $ct['cur_res'];
239			}
240
241			function getCurrentObjectID($n, $ct) {
242			foreach (array('href', 'src') as $a) {
243			if (isset($n['a'][$a])) {
244			list($r, $sub_v) = $this->xURI($n['a'][$a], $ct['base'], $ct['ns']);
			0 ignored issues – show Unused Code introduced 2015-12-16 20:30 UTC by Report Bug Copy Issue Report The assignment to `$sub_v` is unused. Consider omitting it like so `list($first,,$third)`. This checks looks for assignemnts to variables using the `list(...)` function, where not all assigned variables are subsequently used. Consider the following code example. <?php function returnThreeValues() { return array('a', 'b', 'c'); } list($a, $b, $c) = returnThreeValues(); print $a . " - " . $c; Only the variables `$a` and `$c` are used. There was no need to assign `$b`. Instead, the list call could have been. list($a,, $c) = returnThreeValues(); Loading history...
245			return $r;
246			}
247			}
248			return $this->createBnodeID();
249			}
250
251			function getCurrentObjectLiteral($n, $ct) {
252			$r = array('value' => '', 'lang' => $ct['lang'], 'datatype' => '');
253			if (isset($n['a']['content'])) {
254			$r['value'] = $n['a']['content'];
255			}
256			elseif (isset($n['a']['title'])) {
257			$r['value'] = $n['a']['title'];
258			}
259			else {
260			$r['value'] = $this->getPlainContent($n);
261			}
262			return $r;
263			}
264
265			/* */
266
267			function xURI($v, $base, $ns, $attr_type = '') {
268			if ((list($sub_r, $sub_v) = $this->xQname($v, $base, $ns)) && $sub_r) {
269			return array($sub_r, $sub_v);
270			}
271			if (preg_match('/^(rel\|rev\|class\|name)$/', $attr_type) && preg_match('/^[a-z0-9]+$/', $v)) {
272			return array(0, $v);
273			}
274			return array($this->calcURI($v, $base), '');
275			}
276
277			function xQname($v, $base, $ns) {
278			if ($sub_r = $this->x('([a-z0-9\-\_]+)[\-\.]([a-z0-9\-\_]+)', $v)) {
279			if (isset($ns[$sub_r[1]])) {
280			return array($ns[$sub_r[1]] . $sub_r[2], '');
281			}
282			}
283			return array(0, $v);
284			}
285
286			/* */
287
288			}
289

Enalean / tuleap

GitHub Access Token became invalid

ARC2_ErdfExtractor::__construct() A last analyzed 2016-01-22 17:23 UTC

Complexity

Size

Duplication

Duplication Side-by-Side

Filter issues like

ARC2_ErdfExtractor::__construct() A
last analyzed 2016-01-22 17:23 UTC