' . htmlspecialchars(print_r($n, 1)) . ''; + if ($n['tag'] == 'comment') { + $r = ''; + } + elseif ($n['tag'] == 'cdata') { + $r = $n['a']['value']; + } + else { + $r = ''; + if ($outer) { + $r .= '<' . $n['tag']; + asort($n['a']); + if (isset($n['a']['xmlns']) && $n['a']['xmlns']['']) { + $r .= ' xmlns="' . $n['a']['xmlns'][''] . '"'; + } + foreach ($n['a'] as $a => $val) { + if (!is_array($val) && isset($n['a'][$a . ' uri'])) $val = $n['a'][$a . ' uri']; + $r .= preg_match('/^[^\s]+$/', $a) && !is_array($val) ? ' ' . $a . '="' . addslashes($val) . '"' : ''; + } + $r .= $n['empty'] ? '/>' : '>'; + } + if (!$n['empty']) { + $r .= $this->v('cdata', '', $n); + $sub_nodes = $this->getSubNodes($n); + foreach ($sub_nodes as $sub_n) { + $r .= $this->getContent($sub_n, 1, 0); + } + if ($outer) { + $r .= ''; + } + } + } + return ($trim && !$this->keep_cdata_ws) ? trim($r) : $r; + } + + /* */ + + function getDocID($n) { + $id = $n['id']; + $k = 'doc_' . $id; + if (!isset($this->caller->cache[$k])) { + $this->caller->cache[$k] = $n['doc_url']; + } + return $this->caller->cache[$k]; + } + + function getDocOwnerID($n) { + return '_:owner_of_' . $this->normalize($this->getDocID($n)); + } + + /* */ + + function normalize($v) { + $v = preg_replace('/[\W\s]+/is', '_', strip_tags(strtolower($v))); + $v = preg_replace('/http/', '', $v); + $v = preg_replace('/[\_]+/', '_', $v); + //$v = substr($v, 0, 30); + $v = trim($v, '_'); + return $v; + } + + /* */ + +} diff --git a/arc2-2.2.4/extractors/ARC2_RdfaExtractor.php b/arc2-2.2.4/extractors/ARC2_RdfaExtractor.php new file mode 100755 index 0000000..b0a00b0 --- /dev/null +++ b/arc2-2.2.4/extractors/ARC2_RdfaExtractor.php @@ -0,0 +1,385 @@ +' . htmlspecialchars(print_r($this->nodes, 1)) . ''; + if (!isset($this->caller->detected_formats['rdfa'])) return 0; + $root_node = $this->getRootNode(); + //$base = $this->v('xml:base', $this->getDocBase(), $root_node['a']); + $base = $this->getDocBase(); + $context = array( + 'base' => $base, + 'p_s' => $base, + 'p_o' => '', + 'ns' => array(), + 'inco_ts' => array(), + 'lang' => '', + ); + $this->processNode($root_node, $context, 0); + } + + /* */ + + function getRootNode() { + foreach ($this->nodes as $id => $node) { + if ($node['tag'] == 'html') { + return $node; + } + } + return $this->nodes[0]; + } + + /* */ + + function processNode($n, $ct, $level) { + if ($n['tag']=='cdata' || $n['tag']=='comment') return null; /* patch by tobyink */ + $ts_added = 0; + /* step 1 */ + $lct = array(); + $lct['prev_s'] = $this->v('prev_s', $this->v('p_s', '', $ct), $ct); + $lct['recurse'] = 1; + $lct['skip'] = 0; + $lct['new_s'] = ''; + $lct['cur_o_res'] = ''; + $lct['inco_ts'] = array(); + $lct['base'] = $ct['base']; + //$lct['base'] = $this->v('xml:base', $ct['base'], $n['a']); + /* step 2 */ + $lct['ns'] = array_merge($ct['ns'], $this->v('xmlns', array(), $n['a'])); + /* step 3 */ + $lct['lang'] = $this->v('xml:lang', $ct['lang'], $n['a']); + /* step 4 */ + $rel_uris = $this->getAttributeURIs($n, $ct, $lct, 'rel'); + $rev_uris = $this->getAttributeURIs($n, $ct, $lct, 'rev'); + if (!$rel_uris && !$rev_uris) { + foreach (array('about', 'src', 'resource', 'href') as $attr) { + if (isset($n['a'][$attr]) && (list($uri, $sub_v) = $this->xURI($n['a'][$attr], $lct['base'], $lct['ns'], '', $lct)) && $uri) { + $lct['new_s'] = $uri; + break; + } + } + if (!$lct['new_s']) { + if (preg_match('/(head|body)/i', $n['tag'])) { + $lct['new_s'] = $lct['base']; + } + elseif ($this->getAttributeURIs($n, $ct, $lct, 'typeof')) { + $lct['new_s'] = $this->createBnodeID(); + } + elseif ($ct['p_o']) { + $lct['new_s'] = $ct['p_o']; + //$lct['skip'] = 1; + if(!isset($n['a']['property'])) $lct['skip'] = 1;/* patch by masaka */ + } + } + } + /* step 5 */ + else { + foreach (array('about', 'src') as $attr) { + if (isset($n['a'][$attr]) && (list($uri, $sub_v) = $this->xURI($n['a'][$attr], $lct['base'], $lct['ns'], '', $lct)) && $uri) { + $lct['new_s'] = $uri; + break; + } + } + if (!$lct['new_s']) { + if (preg_match('/(head|body)/i', $n['tag'])) { + $lct['new_s'] = $lct['base']; + } + elseif ($this->getAttributeURIs($n, $ct, $lct, 'typeof')) { + $lct['new_s'] = $this->createBnodeID(); + } + elseif ($ct['p_o']) { + $lct['new_s'] = $ct['p_o']; + } + } + foreach (array('resource', 'href') as $attr) { + if (isset($n['a'][$attr]) && (list($uri, $sub_v) = $this->xURI($n['a'][$attr], $lct['base'], $lct['ns'], '', $lct)) && $uri) { + $lct['cur_o_res'] = $uri; + break; + } + } + } + /* step 6 */ + if ($lct['new_s']) { + if ($uris = $this->getAttributeURIs($n, $ct, $lct, 'typeof')) { + foreach ($uris as $uri) { + $this->addT(array( + 's' => $lct['new_s'], + 's_type' => preg_match('/^\_\:/', $lct['new_s']) ? 'bnode' : 'uri', + 'p' => 'http://www.w3.org/1999/02/22-rdf-syntax-ns#type', + 'o' => $uri, + 'o_type' => 'uri', + 'o_lang' => '', + 'o_datatype' => '', + )); + $ts_added = 1; + } + } + /* step 7 */ + if ($lct['cur_o_res']) { + if ($rel_uris) { + foreach ($rel_uris as $uri) { + $this->addT(array( + 's' => $lct['new_s'], + 's_type' => preg_match('/^\_\:/', $lct['new_s']) ? 'bnode' : 'uri', + 'p' => $uri, + 'o' => $lct['cur_o_res'], + 'o_type' => preg_match('/^\_\:/', $lct['cur_o_res']) ? 'bnode' : 'uri', + 'o_lang' => '', + 'o_datatype' => '', + )); + $ts_added = 1; + } + } + if ($rev_uris) { + foreach ($rev_uris as $uri) { + $this->addT(array( + 's' => $lct['cur_o_res'], + 's_type' => preg_match('/^\_\:/', $lct['cur_o_res']) ? 'bnode' : 'uri', + 'p' => $uri, + 'o' => $lct['new_s'], + 'o_type' => preg_match('/^\_\:/', $lct['new_s']) ? 'bnode' : 'uri', + 'o_lang' => '', + 'o_datatype' => '', + )); + $ts_added = 1; + } + } + } + } + /* step 8 */ + if (!$lct['cur_o_res']) { + if ($rel_uris || $rev_uris) { + $lct['cur_o_res'] = $this->createBnodeID(); + foreach ($rel_uris as $uri) { + $lct['inco_ts'][] = array('p' => $uri, 'dir' => 'fwd'); + } + foreach ($rev_uris as $uri) { + $lct['inco_ts'][] = array('p' => $uri, 'dir' => 'rev'); + } + } + } + /* step 10 */ + if (!$lct['skip'] && ($new_s = $lct['new_s'])) { + //if ($new_s = $lct['new_s']) { + if ($uris = $this->getAttributeURIs($n, $ct, $lct, 'property')) { + foreach ($uris as $uri) { + $lct['cur_o_lit'] = $this->getCurrentObjectLiteral($n, $lct, $ct); + $this->addT(array( + 's' => $lct['new_s'], + 's_type' => preg_match('/^\_\:/', $lct['new_s']) ? 'bnode' : 'uri', + 'p' => $uri, + 'o' => $lct['cur_o_lit']['value'], + 'o_type' => 'literal', + 'o_lang' => $lct['cur_o_lit']['lang'], + 'o_datatype' => $lct['cur_o_lit']['datatype'], + )); + $ts_added = 1; + if ($lct['cur_o_lit']['datatype'] == 'http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral') { + $lct['recurse'] = 0; + } + } + } + } + /* step 11 (10) */ + $complete_triples = 0; + if ($lct['recurse']) { + if ($lct['skip']) { + $new_ct = array_merge($ct, array('base' => $lct['base'], 'lang' => $lct['lang'], 'ns' => $lct['ns'])); + } + else { + $new_ct = array( + 'base' => $lct['base'], + 'p_s' => $lct['new_s'] ? $lct['new_s'] : $ct['p_s'], + 'p_o' => $lct['cur_o_res'] ? $lct['cur_o_res'] : ($lct['new_s'] ? $lct['new_s'] : $ct['p_s']), + 'ns' => $lct['ns'], + 'inco_ts' => $lct['inco_ts'], + 'lang' => $lct['lang'] + ); + } + $sub_nodes = $this->getSubNodes($n); + foreach ($sub_nodes as $sub_node) { + if ($this->processNode($sub_node, $new_ct, $level+1)) { + $complete_triples = 1; + } + } + } + /* step 12 (11) */ + $other = 0; + if ($ts_added || $complete_triples || ($lct['new_s'] && !preg_match('/^\_\:/', $lct['new_s'])) || ($other == 1)) { + //if (!$lct['skip'] && ($complete_triples || ($lct['new_s'] && !preg_match('/^\_\:/', $lct['new_s'])))) { + foreach ($ct['inco_ts'] as $inco_t) { + if ($inco_t['dir'] == 'fwd') { + $this->addT(array( + 's' => $ct['p_s'], + 's_type' => preg_match('/^\_\:/', $ct['p_s']) ? 'bnode' : 'uri', + 'p' => $inco_t['p'], + 'o' => $lct['new_s'], + 'o_type' => preg_match('/^\_\:/', $lct['new_s']) ? 'bnode' : 'uri', + 'o_lang' => '', + 'o_datatype' => '', + )); + } + elseif ($inco_t['dir'] == 'rev') { + $this->addT(array( + 's' => $lct['new_s'], + 's_type' => preg_match('/^\_\:/', $lct['new_s']) ? 'bnode' : 'uri', + 'p' => $inco_t['p'], + 'o' => $ct['p_s'], + 'o_type' => preg_match('/^\_\:/', $ct['p_s']) ? 'bnode' : 'uri', + 'o_lang' => '', + 'o_datatype' => '', + )); + } + } + } + /* step 13 (12) (result flag) */ + if ($ts_added) return 1; + if ($lct['new_s'] && !preg_match('/^\_\:/', $lct['new_s'])) return 1; + if ($complete_triples) return 1; + return 0; + } + + /* */ + + function getAttributeURIs($n, $ct, $lct, $attr) { + $vals = ($val = $this->v($attr, '', $n['a'])) ? explode(' ', $val) : array(); + $r = array(); + foreach ($vals as $val) { + if(!trim($val)) continue; + if ((list($uri, $sub_v) = $this->xURI(trim($val), $lct['base'], $lct['ns'], $attr, $lct)) && $uri) { + $r[] = $uri; + } + } + return $r; + } + + /* */ + + function getCurrentObjectLiteral($n, $lct, $ct) { + $xml_val = $this->getContent($n); + $plain_val = $this->getPlainContent($n, 0, 0); + if (function_exists('html_entity_decode')) { + $plain_val = html_entity_decode($plain_val, ENT_QUOTES); + } + $dt = $this->v('datatype', '', $n['a']); + list($dt_uri, $sub_v) = $this->xURI($dt, $lct['base'], $lct['ns'], '', $lct); + $dt = $dt ? $dt_uri : $dt; + $r = array('value' => '', 'lang' => $lct['lang'], 'datatype' => $dt); + if (isset($n['a']['content'])) { + $r['value'] = $n['a']['content']; + if (function_exists('html_entity_decode')) { + $r['value'] = html_entity_decode($r['value'], ENT_QUOTES); + } + } + elseif ($xml_val == $plain_val) { + $r['value'] = $plain_val; + } + elseif (!preg_match('/[\<\>]/', $xml_val)) { + $r['value'] = $xml_val; + } + elseif (isset($n['a']['datatype']) && ($dt != 'http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral')) { + $r['value'] = $plain_val; + } + elseif (!isset($n['a']['datatype']) || ($dt == 'http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral')) { + $r['value'] = $this->injectXMLDeclarations($xml_val, $lct['ns'], $lct['lang']); + $r['datatype'] = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral'; + } + return $r; + } + + function injectXMLDeclarations($val, $ns, $lang) {//@@todo proper node rebuilding */ + $lang_code = $lang ? ' xml:lang="' . $lang . '"' : ''; + /* ns */ + $val = preg_replace('/<([a-z0-9]+)([\>\s])/is', '<\\1 xmlns="http://www.w3.org/1999/xhtml"' . $lang_code . '\\2', $val); + foreach ($ns as $prefix => $uri) { + if ($prefix && ($pos = strpos(' ' . $val, '<' . $prefix . ':'))) { + $val = substr($val, 0, $pos - 1) . preg_replace('/^(<' . $prefix . '\:[^\>\s]+)/', '\\1 xmlns:' . $prefix. '="' . $uri . '"' . $lang_code, substr($val, $pos - 1)); + } + } + /* remove accidentally added xml:lang and xmlns= */ + $val = preg_replace('/(\<[^\>]*)( xml\:lang[^\s\>]+)([^\>]*)(xml\:lang[^\s\>]+)/s', '\\1\\3\\4', $val); + $val = preg_replace('/(\<[^\>]*)( xmlns=[^\s\>]+)([^\>]*)(xmlns=[^\s\>]+)/s', '\\1\\3\\4', $val); + return $val; + } + + /* */ + + function xURI($v, $base, $ns, $attr_type = '', $lct = '') { + if ((list($sub_r, $sub_v) = $this->xBlankCURIE($v, $base, $ns)) && $sub_r) { + return array($sub_r, $sub_v); + } + if ((list($sub_r, $sub_v) = $this->xSafeCURIE($v, $base, $ns, $lct)) && $sub_r) { + return array($sub_r, $sub_v); + } + if ((list($sub_r, $sub_v) = $this->xCURIE($v, $base, $ns)) && $sub_r) { + return array($sub_r, $sub_v); + } + if (preg_match('/^(rel|rev)$/', $attr_type) && preg_match('/^\s*(alternate|appendix|bookmark|cite|chapter|contents|copyright|glossary|help|icon|index|last|license|meta|next|p3pv1|prev|role|section|stylesheet|subsection|start|up)(\s|$)/is', $v, $m)) { + return array('http://www.w3.org/1999/xhtml/vocab#' . strtolower($m[1]), preg_replace('/^\s*' . $m[1]. '/is', '', $v)); + } + if (preg_match('/^(rel|rev)$/', $attr_type) && preg_match('/^[a-z0-9\.]+$/i', $v)) { + return array(0, $v); + } + return array($this->calcURI($v, $base), ''); + } + + function xBlankCURIE($v, $base, $ns) { + if ($sub_r = $this->x('\[\_\:\]', $v)) { + $this->empty_bnode = isset($this->empty_bnode) ? $this->empty_bnode : $this->createBnodeID(); + return array($this->empty_bnode, ''); + } + if ($sub_r = $this->x('\[?(\_\:[a-z0-9\_\-]+)\]?', $v)) { + return array($sub_r[1], ''); + } + return array(0, $v); + } + + function xSafeCURIE($v, $base, $ns, $lct = '') { + /* empty */ + if ($sub_r = $this->x('\[\]', $v)) { + $r = $lct ? $lct['prev_s'] : $base;/* should be current subject value */ + return $sub_r[1] ? array($r, $sub_r[1]) : array($r, ''); + } + if ($sub_r = $this->x('\[([^\:]*)\:([^\]]*)\]', $v)) { + if (!$sub_r[1]) return array('http://www.w3.org/1999/xhtml/vocab#' . $sub_r[2], ''); + if (isset($ns[$sub_r[1]])) { + return array($ns[$sub_r[1]] . $sub_r[2], ''); + } + } + return array(0, $v); + } + + function xCURIE($v, $base, $ns) { + if ($sub_r = $this->x('([a-z0-9\-\_]*)\:([^\s]+)', $v)) { + if (!$sub_r[1]) return array('http://www.w3.org/1999/xhtml/vocab#' . $sub_r[2], ''); + if (isset($ns[$sub_r[1]])) { + return array($ns[$sub_r[1]] . $sub_r[2], ''); + } + } + return array(0, $v); + } + + /* */ + +} diff --git a/arc2-2.2.4/extractors/ARC2_TwitterProfilePicExtractor.php b/arc2-2.2.4/extractors/ARC2_TwitterProfilePicExtractor.php new file mode 100755 index 0000000..048cc63 --- /dev/null +++ b/arc2-2.2.4/extractors/ARC2_TwitterProfilePicExtractor.php @@ -0,0 +1,45 @@ +a['ns']['foaf'] = 'http://xmlns.com/foaf/0.1/'; + $this->a['ns']['mf'] = 'http://poshrdf.org/ns/mf#'; + } + + /* */ + + function extractRDF() { + $t_vals = array(); + $t = ''; + foreach ($this->nodes as $n) { + if (isset($n['tag']) && ($n['tag'] == 'img') && ($this->v('id', '', $n['a']) == 'profile-image')) { + $t_vals['vcard_id'] = $this->getDocID($n) . '#resource(side/1/2/1)'; + $t .= '?vcard_id mf:photo <' . $n['a']['src'] . '> . '; + break; + } + } + if ($t) { + $doc = $this->getFilledTemplate($t, $t_vals, $n['doc_base']); + $this->addTs(ARC2::getTriplesFromIndex($doc)); + } + } + + /* */ + +} diff --git a/arc2-2.2.4/parsers/ARC2_AtomParser.php b/arc2-2.2.4/parsers/ARC2_AtomParser.php new file mode 100755 index 0000000..cabef40 --- /dev/null +++ b/arc2-2.2.4/parsers/ARC2_AtomParser.php @@ -0,0 +1,245 @@ +triples = array(); + $this->target_encoding = ''; + $this->t_count = 0; + $this->added_triples = array(); + $this->skip_dupes = false; + $this->bnode_prefix = $this->v('bnode_prefix', 'arc'.substr(md5(uniqid(rand())), 0, 4).'b', $this->a); + $this->bnode_id = 0; + $this->cache = array(); + $this->allowCDataNodes = 0; + } + + /* */ + + function done() { + $this->extractRDF(); + } + + /* */ + + function setReader(&$reader) { + $this->reader = $reader; + } + + function createBnodeID(){ + $this->bnode_id++; + return '_:' . $this->bnode_prefix . $this->bnode_id; + } + + function addT($t) { + //if (!isset($t['o_datatype'])) + if ($this->skip_dupes) { + //$h = md5(print_r($t, 1)); + $h = md5(serialize($t)); + if (!isset($this->added_triples[$h])) { + $this->triples[$this->t_count] = $t; + $this->t_count++; + $this->added_triples[$h] = true; + } + } + else { + $this->triples[$this->t_count] = $t; + $this->t_count++; + } + } + + function getTriples() { + return $this->v('triples', array()); + } + + function countTriples() { + return $this->t_count; + } + + function getSimpleIndex($flatten_objects = 1, $vals = '') { + return ARC2::getSimpleIndex($this->getTriples(), $flatten_objects, $vals); + } + + /* */ + + function extractRDF() { + $index = $this->getNodeIndex(); + //print_r($index); + $this->rdf = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'; + $this->atom = 'http://www.w3.org/2005/Atom'; + $this->rss = 'http://purl.org/rss/1.0/'; + $this->dc = 'http://purl.org/dc/elements/1.1/'; + $this->sioc = 'http://rdfs.org/sioc/ns#'; + $this->dct = 'http://purl.org/dc/terms/'; + $this->content = 'http://purl.org/rss/1.0/modules/content/'; + $this->enc = 'http://purl.oclc.org/net/rss_2.0/enc#'; + $this->mappings = array( + 'feed' => $this->rss . 'channel', + 'entry' => $this->rss . 'item', + 'title' => $this->rss . 'title', + 'link' => $this->rss . 'link', + 'summary' => $this->rss . 'description', + 'content' => $this->content . 'encoded', + 'id' => $this->dc . 'identifier', + 'author' => $this->dc . 'creator', + 'category' => $this->dc . 'subject', + 'updated' => $this->dc . 'date', + 'source' => $this->dc . 'source', + ); + $this->dt_props = array( + $this->dc . 'identifier', + $this->rss . 'link' + ); + foreach ($index as $p_id => $nodes) { + foreach ($nodes as $pos => $node) { + $tag = $this->v('tag', '', $node); + if ($tag == 'feed') { + $struct = $this->extractChannel($index[$node['id']]); + $triples = ARC2::getTriplesFromIndex($struct); + foreach ($triples as $t) { + $this->addT($t); + } + } + elseif ($tag == 'entry') { + $struct = $this->extractItem($index[$node['id']]); + $triples = ARC2::getTriplesFromIndex($struct); + foreach ($triples as $t) { + $this->addT($t); + } + } + } + } + } + + function extractChannel($els) { + list($props, $sub_index) = $this->extractProps($els, 'channel'); + $uri = $props[$this->rss . 'link'][0]['value']; + return ARC2::getMergedIndex(array($uri => $props), $sub_index); + } + + function extractItem($els) { + list($props, $sub_index) = $this->extractProps($els, 'item'); + $uri = $props[$this->rss . 'link'][0]['value']; + return ARC2::getMergedIndex(array($uri => $props), $sub_index); + } + + function extractProps($els, $container) { + $r = array($this->rdf . 'type' => array(array('value' => $this->rss . $container, 'type' => 'uri'))); + $sub_index = array(); + foreach ($els as $info) { + /* key */ + $tag = $info['tag']; + if (!preg_match('/^[a-z0-9]+\:/i', $tag)) { + $k = isset($this->mappings[$tag]) ? $this->mappings[$tag] : ''; + } + elseif (isset($this->mappings[$tag])) { + $k = $this->mappings[$tag]; + } + else {/* qname */ + $k = $this->expandPName($tag); + } + //echo $k . "\n"; + if (($container == 'channel') && ($k == $this->rss . 'item')) continue; + /* val */ + $v = trim($info['cdata']); + if (!$v) $v = $this->v('href uri', '', $info['a']); + /* prop */ + if ($k) { + /* content handling */ + if (in_array($k, array($this->rss . 'description', $this->content . 'encoded'))) { + $v = $this->getNodeContent($info); + } + /* source handling */ + elseif ($k == $this->dc . 'source') { + $sub_nodes = $this->node_index[$info['id']]; + foreach ($sub_nodes as $sub_pos => $sub_info) { + if ($sub_info['tag'] == 'id') { + $v = trim($sub_info['cdata']); + } + } + } + /* link handling */ + elseif ($k == $this->rss . 'link') { + if ($link_type = $this->v('type', '', $info['a'])) { + $k2 = $this->dc . 'format'; + if (!isset($sub_index[$v])) $sub_index[$v] = array(); + if (!isset($sub_index[$v][$k2])) $sub_index[$v][$k2] = array(); + $sub_index[$v][$k2][] = array('value' => $link_type, 'type' => 'literal'); + } + } + /* author handling */ + elseif ($k == $this->dc . 'creator') { + $sub_nodes = $this->node_index[$info['id']]; + foreach ($sub_nodes as $sub_pos => $sub_info) { + if ($sub_info['tag'] == 'name') { + $v = trim($sub_info['cdata']); + } + if ($sub_info['tag'] == 'uri') { + $k2 = $this->sioc . 'has_creator'; + $v2 = trim($sub_info['cdata']); + if (!isset($r[$k2])) $r[$k2] = array(); + $r[$k2][] = array('value' => $v2, 'type' => 'uri'); + } + } + } + /* date handling */ + elseif (in_array($k, array($this->dc . 'date', $this->dct . 'modified'))) { + if (!preg_match('/^[0-9]{4}/', $v) && ($sub_v = strtotime($v)) && ($sub_v != -1)) { + $tz = date('Z', $sub_v); /* timezone offset */ + $sub_v -= $tz; /* utc */ + $v = date('Y-m-d\TH:i:s\Z', $sub_v); + } + } + /* tag handling */ + elseif ($k == $this->dc . 'subject') { + $v = $this->v('term', '', $info['a']); + } + /* other attributes in closed tags */ + elseif (!$v && ($info['state'] == 'closed') && $info['a']) { + foreach ($info['a'] as $sub_k => $sub_v) { + if (!preg_match('/(xmlns|\:|type)/', $sub_k)) { + $v = $sub_v; + break; + } + } + } + if (!isset($r[$k])) $r[$k] = array(); + $r[$k][] = array('value' => $v, 'type' => in_array($k, $this->dt_props) || !preg_match('/^[a-z0-9]+\:[^\s]+$/is', $v) ? 'literal' : 'uri'); + } + } + return array($r, $sub_index); + } + + function initXMLParser() { + if (!isset($this->xml_parser)) { + $enc = preg_match('/^(utf\-8|iso\-8859\-1|us\-ascii)$/i', $this->getEncoding(), $m) ? $m[1] : 'UTF-8'; + $parser = xml_parser_create($enc); + xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 0); + xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0); + xml_set_element_handler($parser, 'open', 'close'); + xml_set_character_data_handler($parser, 'cData'); + xml_set_start_namespace_decl_handler($parser, 'nsDecl'); + xml_set_object($parser, $this); + $this->xml_parser = $parser; + } + } + + /* */ + + +} diff --git a/arc2-2.2.4/parsers/ARC2_CBJSONParser.php b/arc2-2.2.4/parsers/ARC2_CBJSONParser.php new file mode 100755 index 0000000..d59f77e --- /dev/null +++ b/arc2-2.2.4/parsers/ARC2_CBJSONParser.php @@ -0,0 +1,267 @@ + + * @license http://arc.semsol.org/license + * @homepage
' . htmlspecialchars(print_r($n, 1)) . ''; + if ($n['tag'] == 'cdata') { + $r = $n['a']['value']; + } + else { + $r = ''; + if ($outer) { + $r .= '<' . $n['tag']; + asort($n['a']); + if (isset($n['a']['xmlns']) && $n['a']['xmlns']['']) { + $r .= ' xmlns="' . $n['a']['xmlns'][''] . '"'; + } + foreach ($n['a'] as $a => $val) { + $r .= preg_match('/^[^\s]+$/', $a) && !is_array($val) ? ' ' . $a . '="' . addslashes($val) . '"' : ''; + } + $r .= $n['empty'] ? '/>' : '>'; + } + if (!$n['empty']) { + $r .= $this->v('cdata', '', $n); + $sub_nodes = $this->getSubNodes($n); + foreach ($sub_nodes as $sub_n) { + $r .= $this->getNodeContent($sub_n, 1, 0); + } + if ($outer) { + $r .= ''; + } + } + } + return ($trim && !$this->keep_cdata_ws) ? trim($r) : $r; + } + + /* */ + + function pushNode($n) { + $n['id'] = $this->node_count; + $this->nodes[$this->node_count] = $n; + $this->node_count++; + } + + function getCurNode($t = '') { + $i = 1; + do { + $r = $this->node_count ? $this->nodes[$this->node_count - $i] : 0; + $found = (!$t || ($r['tag'] == $t)) ? 1 : 0; + $i++; + } while (!$found && isset($this->nodes[$this->node_count - $i])); + return $r; + } + + function updateNode($node) {/* php4-save */ + $this->nodes[$node['id']] = $node; + } + + /* */ + + function initXMLParser() { + if (!isset($this->xml_parser)) { + $enc = preg_match('/^(utf\-8|iso\-8859\-1|us\-ascii)$/i', $this->getEncoding(), $m) ? $m[1] : 'UTF-8'; + $parser = xml_parser_create_ns($enc, ''); + xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 0); + xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0); + xml_set_element_handler($parser, 'open', 'close'); + xml_set_character_data_handler($parser, 'cData'); + xml_set_start_namespace_decl_handler($parser, 'nsDecl'); + xml_set_object($parser, $this); + $this->xml_parser = $parser; + } + } + + /* */ + + function open($p, $t, $a) { + $t_exact = $t; + //echo "
\n".'opening '.$t . ' ' . print_r($a, 1); flush(); + //echo "
\n".'opening '.$t; flush(); + $t = strpos($t, ':') ? $t : strtolower($t); + /* base check */ + $base = ''; + if (($t == 'base') && isset($a['href'])) { + $this->base = $a['href']; + $base = $a['href']; + } + /* URIs */ + foreach (array('href', 'src', 'id') as $uri_a) { + if (isset($a[$uri_a])) { + $a[$uri_a . ' uri'] = ($uri_a == 'id') ? $this->calcURI('#'.$a[$uri_a]) : $this->calcURI($a[$uri_a]); + } + } + /* ns */ + if ($a) { + foreach ($a as $k => $v) { + if (strpos($k, 'xmlns') === 0) { + $this->nsDecl($p, trim(substr($k, 5), ':'), $v); + } + } + } + /* node */ + $node = array( + 'tag' => $t, + 'tag_exact' => $t_exact, + 'a' => $a, + 'level' => $this->level, + 'pos' => 0, + 'p_id' => $this->node_count-1, + 'state' => 'open', + 'empty' => 0, + 'cdata' =>'' + ); + if ($base) { + $node['base'] = $base; + } + /* parent/sibling */ + if ($this->node_count) { + $l = $this->level; + $prev_node = $this->getCurNode(); + if ($prev_node['level'] == $l) { + $node['p_id'] = $prev_node['p_id']; + $node['pos'] = $prev_node['pos']+1; + } + elseif($prev_node['level'] > $l) { + while($prev_node['level'] > $l) { + if (!isset($this->nodes[$prev_node['p_id']])) { + //$this->addError('nesting mismatch: tag is ' . $t . ', level is ' . $l . ', prev_level is ' . $prev_node['level'] . ', prev_node p_id is ' . $prev_node['p_id']); + break; + } + $prev_node = $this->nodes[$prev_node['p_id']]; + } + $node['p_id'] = $prev_node['p_id']; + $node['pos'] = $prev_node['pos']+1; + } + } + $this->pushNode($node); + $this->level++; + /* cdata */ + $this->cur_cdata=""; + } + + function close($p, $t, $empty = 0) { + //echo "
\n".'closing '.$t; flush(); + $node = $this->getCurNode($t); + $node['state'] = 'closed'; + $node['empty'] = $empty; + $this->updateNode($node); + $this->level--; + } + + function cData($p, $d) { + //echo trim($d) ? "
\n".'cdata: ' . $d : ''; flush(); + $node = $this->getCurNode(); + if($node['state'] == 'open') { + $node['cdata'] .= $d; + $this->updateNode($node); + } + else {/* cdata is sibling of node */ + if ($this->allowCDataNodes) { + $this->open($p, 'cdata', array('value' => $d)); + $this->close($p, 'cdata'); + } + } + } + + function nsDecl($p, $prf, $uri) { + if (is_array($uri)) return 1; + $this->ns[$prf] = $uri; + $this->nsp[$uri] = isset($this->nsp[$uri]) ? $this->nsp[$uri] : $prf; + } + + /* */ + +} \ No newline at end of file diff --git a/arc2-2.2.4/parsers/ARC2_RDFParser.php b/arc2-2.2.4/parsers/ARC2_RDFParser.php new file mode 100755 index 0000000..44dd778 --- /dev/null +++ b/arc2-2.2.4/parsers/ARC2_RDFParser.php @@ -0,0 +1,139 @@ + + * @license http://arc.semsol.org/license + * @homepage
- ' . $r . $n . $ind . '
mdAttrs($s, $main_type) . '>
+
+ ';
+ }
+ return $r;
+ }
+
+ function getObjectValue($o, $p) {
+ if ($o['type'] == 'uri') {
+ if (preg_match('/(jpe?g|gif|png)$/i', $o['value'])) {
+ return $this->getImageObjectValue($o, $p);
+ }
+ return $this->getURIObjectValue($o, $p);
+ }
+ if ($o['type'] == "bnode") {
+ return $this->getBNodeObjectValue($o, $p);
+ }
+ return $this->getLiteralObjectValue($o, $p);
+ }
+
+ function getImageObjectValue($o, $p) {
+ return '' . ucfirst($this->getLabel($s, $ps)) . '
+ '; + /* arcs */ + foreach ($ps as $p => $os) { + $p_cls = strtolower($this->getPName($p)); + $p_cls = str_replace(':', '-', $p_cls); + $r .= ' +
+ ' . ucfirst($this->getLabel($p)) . ':
+
+ ';
+ }
+ /* /node */
+ $r .= '
+
+ -
+ ';
+ $oc = count($os);
+ foreach ($os as $i => $o) {
+ $val = $this->getObjectValue($o, $p);
+ $cls = '';
+ if ($i == 0) $cls .= ($cls ? ' ' : '') . 'first';
+ if ($i == $oc - 1) $cls .= ($cls ? ' ' : '') . 'last';
+ $r .= $n . '
' . $o['value'] . '
';
+ return 'An unnamed resource
';
+ }
+
+ function getLiteralObjectValue($o, $p) {
+ return '' . $o['value'] . '
';
+ }
+
+ /* */
+
+ function getObjectLabel($id) {
+ $r = $this->extractTermLabel($id);
+ if (!$this->label_store) return $r;
+ $q = '
+ SELECT ?val WHERE {
+ <' . $id . '> ?p ?val .
+ FILTER(REGEX(str(?p), "(label|title|name|summary)$"))
+ } LIMIT 1
+ ';
+ $row = $this->label_store->query($q, 'row');
+ return $row ? $row['val'] : $r;
+ }
+
+ /* */
+
+}
+
diff --git a/arc2-2.2.4/serializers/ARC2_NTriplesSerializer.php b/arc2-2.2.4/serializers/ARC2_NTriplesSerializer.php
new file mode 100755
index 0000000..6e95352
--- /dev/null
+++ b/arc2-2.2.4/serializers/ARC2_NTriplesSerializer.php
@@ -0,0 +1,192 @@
+
+ * @homepage
+
+ ';
+ }
+ return $r;
+ }
+
+ function getObjectValue($o) {
+ if ($o['type'] == 'uri') {
+ if (preg_match('/(jpe?g|gif|png)$/i', $o['value'])) {
+ return $this->getImageObjectValue($o);
+ }
+ return $this->getURIObjectValue($o);
+ }
+ if ($o['type'] == "bnode") {
+ return $this->getBNodeObjectValue($o);
+ }
+ return $this->getLiteralObjectValue($o);
+ }
+
+ function getImageObjectValue($o) {
+ return '' . $this->getLabel($s, $ps) . '
+ '; + /* arcs */ + foreach ($ps as $p => $os) { + $r .= ' +
+ ' . ucfirst($this->getLabel($p)) . '
+ ';
+ foreach ($os as $o) {
+ $r .= $n . $this->getObjectValue($o);
+ }
+ $r .= '
+
+ ';
+ }
+ /* node */
+ $r .= '
+
+ An unnamed resource
';
+ }
+
+ function getLiteralObjectValue($o) {
+ return '' . $o['value'] . '
';
+ }
+
+ /* */
+
+}
+
diff --git a/arc2-2.2.4/serializers/ARC2_RDFJSONSerializer.php b/arc2-2.2.4/serializers/ARC2_RDFJSONSerializer.php
new file mode 100755
index 0000000..8cc90a1
--- /dev/null
+++ b/arc2-2.2.4/serializers/ARC2_RDFJSONSerializer.php
@@ -0,0 +1,89 @@
+
+ * @license http://arc.semsol.org/license
+ * @homepage
+ ' . ($this->p('output') != 'htmltab' ? '
+ ';
+ $this->handleEmptyRequest(1);
+ }
+ }
+ else {
+ $this->handleEmptyRequest();
+ }
+ }
+
+ function go($auto_setup = 0) {
+ $this->handleRequest($auto_setup);
+ $this->sendHeaders();
+ echo $this->getResult();
+ }
+
+ /* */
+
+ function handleImgRequest($img) {
+ $this->setHeader('content-type', 'Content-type: image/gif');
+ $imgs = array(
+ 'bg_body' => base64_decode('R0lGODlhAQBkAMQAAPf39/Hx8erq6vPz8/Ly8u/v7+np6fT09Ovr6/b29u3t7ejo6Pz8/Pv7+/39/fr6+vj4+P7+/vn5+f///wAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACH5BAAAAAAALAAAAAABAGQAAAUp4GIIiFIExHAkAAC9cAxJdG3TT67vTe//jKBQ6Cgaj5GkcpmcOJ/QZwgAOw=='),
+ );
+ $this->result = isset($imgs[$img]) ? $imgs[$img] : '';
+ $this->sendHeaders();
+ echo $this->getResult();
+ exit;
+ }
+
+ /* */
+
+ function handleEmptyRequest($force = 0) {
+ /* service description */
+ $formats = array(
+ 'rdfxml' => 'RDFXML', 'rdf+xml' => 'RDFXML', 'html' => 'HTML'
+ );
+ if (!$force && $this->getResultFormat($formats, 'html') != 'HTML') {
+ $this->handleServiceDescriptionRequest();
+ }
+ else {
+ $this->setHeader('content-type', 'Content-type: text/html; charset=utf-8');
+ $this->result = $this->getHTMLFormDoc();
+ }
+ }
+
+ /* */
+
+ function handleServiceDescriptionRequest() {
+ $q = '
+ PREFIX void: ' . htmlspecialchars($this->getResult()) . '' : $this->getResult()) . ' +
' . $this->getHTMLDocHeading() . '
+
+
+ ' . $this->getHTMLDocForm() .'
+ ' . ($this->p('show_inline') ? $this->query_result : '') . '
+
+ ';
+ }
+
+ function getHTMLDocForm() {
+ $q = $this->p('query') ? htmlspecialchars($this->p('query')) : "SELECT * WHERE {\n GRAPH ?g { ?s ?p ?o . }\n}\nLIMIT 10";
+ return '
+
+ ';
+ }
+
+ function getHTMLDocOptions() {
+ $sel = $this->p('output');
+ $sel_code = ' selected="selected"';
+ return '
+
+
+ ';
+ }
+
+ /* */
+
+}
diff --git a/arc2-2.2.4/store/ARC2_StoreHelper.php b/arc2-2.2.4/store/ARC2_StoreHelper.php
new file mode 100755
index 0000000..9ce9c58
--- /dev/null
+++ b/arc2-2.2.4/store/ARC2_StoreHelper.php
@@ -0,0 +1,66 @@
+store = $this->caller;
+ }
+
+ /* */
+
+ function changeNamespaceURI($old_uri, $new_uri) {
+ $id_changes = 0;
+ $t_changes = 0;
+ /* table lock */
+ if ($this->store->getLock()) {
+ $con = $this->store->getDBCon();
+ foreach (array('id', 's', 'o') as $id_col) {
+ $tbl = $this->store->getTablePrefix() . $id_col . '2val';
+ $sql = 'SELECT id, val FROM ' . $tbl . ' WHERE val LIKE "' . mysql_real_escape_string($old_uri, $con). '%"';
+ $rs = mysql_query($sql, $con);
+ if (!$rs) continue;
+ while ($row = mysql_fetch_array($rs)) {
+ $new_val = str_replace($old_uri, $new_uri, $row['val']);
+ $new_id = $this->store->getTermID($new_val, $id_col);
+ if (!$new_id) {/* unknown ns uri, overwrite current id value */
+ $sub_sql = "UPDATE " . $tbl . " SET val = '" . mysql_real_escape_string($new_val, $con) . "' WHERE id = " . $row['id'];
+ $sub_r = mysql_query($sub_sql, $con);
+ $id_changes++;
+ }
+ else {/* replace ids */
+ $t_tbls = $this->store->getTables();
+ foreach ($t_tbls as $t_tbl) {
+ if (preg_match('/^triple/', $t_tbl)) {
+ foreach (array('s', 'p', 'o', 'o_lang_dt') as $t_col) {
+ $sub_sql = "UPDATE " . $this->store->getTablePrefix() . $t_tbl . " SET " . $t_col . " = " . $new_id . " WHERE " . $t_col . " = " . $row['id'];
+ $sub_r = mysql_query($sub_sql, $con);
+ $t_changes += mysql_affected_rows($con);
+ }
+ }
+ }
+ }
+ }
+ }
+ $this->store->releaseLock();
+ }
+ return array('id_replacements' => $id_changes, 'triple_updates' => $t_changes);
+ }
+
+ /* */
+
+}
diff --git a/arc2-2.2.4/store/ARC2_StoreInsertQueryHandler.php b/arc2-2.2.4/store/ARC2_StoreInsertQueryHandler.php
new file mode 100755
index 0000000..bffdab5
--- /dev/null
+++ b/arc2-2.2.4/store/ARC2_StoreInsertQueryHandler.php
@@ -0,0 +1,54 @@
+
+ * @license http://arc.semsol.org/license
+ * @homepage + This interface implements + SPARQL and + SPARQL+ via HTTP Bindings. +
++ Enabled operations: ' . join(', ', $this->getFeatures()) . ' +
++ Max. number of results : ' . $this->v('endpoint_max_limit', 'unrestricted', $this->a) . ' +
+' . htmlspecialchars($q) . '
' . $nl ;
+ if ($errors || $rest) {
+ $pass = 0;
+ $r .= htmlspecialchars($nl . $nl . print_r($errors, 1) . $nl . print_r($rest, 1));
+ }
+ else {
+ $pass = 1;
+ $r .= htmlspecialchars($nl . $nl . print_r($infos, 1));
+ }
+ return array('pass' => $pass, 'info' => $r);
+ }
+
+ /* */
+
+ function runNegativeSyntaxTest($id) {
+ $nl = "\n";
+ $r = '';
+ /* get action */
+ $q = '
+ PREFIX mf: ' . htmlspecialchars($q) . '
' . $nl ;
+ if ($errors || $rest) {
+ $pass = 1;
+ $r .= htmlspecialchars($nl . $nl . print_r($errors, 1) . $nl . print_r($rest, 1));
+ }
+ else {
+ $pass = 0;
+ $r .= htmlspecialchars($nl . $nl . print_r($infos, 1));
+ }
+ return array('pass' => $pass, 'info' => $r);
+ }
+
+ /* */
+
+ function runQueryEvaluationTest($id) {
+ $nl = "\n";
+ $r = '';
+ /* get action */
+ $q = '
+ PREFIX mf: ' . $$var . $nl . $nl . htmlspecialchars(${$var . '_value'}) . ''; + } + } + /* query infos */ + ARC2::inc('SPARQLPlusParser'); + $parser = new ARC2_SPARQLPlusParser($this->a, $this); + $parser->parse($query_value, $query); + $infos = $parser->getQueryInfos(); + $rest = $parser->getUnparsedCode(); + $errors = $parser->getErrors(); + $q_type = !$errors ? $infos['query']['type'] : ''; + /* add data */ + $dsets = array(); + $gdsets = array(); + if ($data) { + $dsets = is_array($data) ? array_merge($dsets, $data) : array_merge($dsets, array($data)); + } + if ($graph_data) { + $gdsets = is_array($graph_data) ? array_merge($gdsets, $graph_data) : array_merge($gdsets, array($graph_data)); + } + if (!$dsets && !$gdsets) { + foreach ($infos['query']['dataset'] as $set) { + if ($set['named']) { + $gdsets[] = $set['graph']; + } + else { + $dsets[] = $set['graph']; + } + } + } + $store = $this->data_store; + $store->reset(); + foreach ($dsets as $graph) { + $qr = $store->query('LOAD <' .$graph. '>'); + } + foreach ($gdsets as $graph) { + $qr = $store->query('LOAD <' .$graph. '> INTO <' .$graph. '>'); + } + /* run query */ + if ($query) { + $sql = $store->query($query_value, 'sql', $query); + $qr = $store->query($query_value, '', $query); + $qr_result = $qr['result']; + if ($q_type == 'select') { + $qr_result = $this->adjustBnodes($qr['result'], $id); + } + elseif ($q_type == 'construct') { + $ser = ARC2::getTurtleSerializer($this->a); + $qr_result = $ser->getSerializedIndex($qr_result); + } + } + //echo '
query result: ' . $nl . htmlspecialchars(print_r($qr_result, 1)) . ''; + if (!$query || $errors || $rest) { + return array('pass' => 0, 'info' => 'query could not be parsed' . htmlspecialchars($query_value)); + } + $m = 'isSame' . $q_type . 'Result'; + $sub_r = $this->$m($qr_result, $result_value, $result, $id); + $pass = $sub_r['pass']; + if (in_array($id, array( + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/sort/manifest#dawg-sort-6', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/sort/manifest#dawg-sort-8', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/sort/manifest#dawg-sort-builtin', + ))) { + $pass = 0; /* manually checked 2007-09-18 */ + } + if (in_array($id, array( + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/sort/manifest#dawg-sort-function', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/reduced/manifest#reduced-1', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/reduced/manifest#reduced-2', + ))) { + $pass = 1; /* manually checked 2007-11-28 */ + } + $pass_info = $sub_r['info']; + $info = print_r($pass_info, 1) . $nl; + $info .= '
sql: ' . $nl . htmlspecialchars($sql['result']) . '
'; + $info .= $pass ? '' : print_r($graph_data, 1) . $nl . htmlspecialchars(print_r($graph_data_value, 1)) . '
'; + $info .= $pass ? '' : print_r($data, 1) . $nl . htmlspecialchars(print_r($data_value, 1)) . '
'; + $info .= $pass ? '' : $query . $nl . htmlspecialchars($query_value) . '
'; + $info .= $pass ? '' : '
query result: ' . $nl . htmlspecialchars(print_r($qr_result, 1)) . '' . '
'; + $info .= $pass ? '' : print_r($infos, 1); + return array('pass' => $pass, 'info' => $info); + } + + /* */ + + function isSameSelectResult($qr, $result, $result_base) { + if (strpos($result, 'http://www.w3.org/2001/sw/DataAccess/tests/result-set#')) { + $parser = ARC2::getRDFParser($this->a); + $parser->parse($result_base, $result); + $index = $parser->getSimpleIndex(0); + //echo '
' . print_r($index, 1) .''; + $valid_qr = $this->buildTurtleSelectQueryResult($index); + } + else { + $parser = ARC2::getSPARQLXMLResultParser($this->a); + $parser->parse('', $result); + $valid_qr = $parser->getStructure(); + } + if (isset($valid_qr['boolean'])) { + $pass = $valid_qr['boolean'] == $this->v('boolean', '', $qr); + } + else { + $pass = 1; + if (count($valid_qr['variables']) != count($qr['variables'])) { + $pass = 0; + } + if (count($valid_qr['rows']) != count($qr['rows'])) { + $pass = 0; + } + if ($pass) { + foreach ($valid_qr['variables'] as $var) { + if (!in_array($var, $qr['variables'])) { + $pass = 0; + break; + } + } + } + if ($pass) { + $index = $this->buildArrayHashIndex($qr['rows']); + $valid_index = $this->buildArrayHashIndex($valid_qr['rows']); + if (($diff = array_diff($index, $valid_index)) || ($diff = array_diff($valid_index, $index))) { + $pass = 0; + //echo '
' . print_r($diff, 1) . ''; + } + } + } + return array('pass' => $pass, 'info' => $valid_qr); + } + + /* */ + + function isSameConstructResult($qr, $result, $result_base, $test) { + $parser = ARC2::getRDFParser($this->a); + $parser->parse('', $result); + $valid_triples = $parser->getTriples(); + $parser = ARC2::getRDFParser($this->a); + $parser->parse('', $qr); + $triples = $parser->getTriples(); + $info = '
' . print_r($valid_triples, 1) .''; + $info = ''; + + //echo '
' . print_r($index, 1) .''; + $pass = 0; + if (in_array($test, array(/* manually checked 2007-09-21 */ + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/construct/manifest#construct-1', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/construct/manifest#construct-2', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/construct/manifest#construct-3', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/construct/manifest#construct-4', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/construct/manifest#construct-5', + ))) { + $pass = 1; + } + return array('pass' => $pass, 'info' => $valid_triples); + } + + /* */ + + function isSameAskResult($qr, $result, $result_base) { + if (preg_match('/(true|false)\.(ttl|n3)$/', $result_base, $m)) { + $valid_r = $m[1]; + } + else { + $valid_r = preg_match('/boolean\>([^\<]+)/s', $result, $m) ? trim($m[1]) : '-'; + } + $r = ($qr === true) ? 'true' : 'false'; + $pass = ($r == $valid_r) ? 1 : 0; + return array('pass' => $pass, 'info' => $valid_r); + } + + /* */ + + function buildTurtleSelectQueryResult($index) { + $rs = 'http://www.w3.org/2001/sw/DataAccess/tests/result-set#'; + $rdf = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'; + $r = array('variables' => array(), 'rows' => array()); + foreach ($index as $node => $props) { + $types = $this->v($rdf . 'type', array(), $props); + foreach ($types as $type) { + if ($type['value'] == $rs . 'ResultSet') { + $vars = $this->v($rs . 'resultVariable', array(), $props); + foreach ($vars as $var) { + $r['variables'][] = $var['value']; + } + } + } + $bindings = $this->v($rs . 'binding', array(), $props); + if ($bindings) { + $row = array(); + foreach ($bindings as $binding) { + $binding_id = $binding['value']; + $var = $index[$binding_id][$rs . 'variable'][0]['value']; + $val = $index[$binding_id][$rs . 'value'][0]['value']; + $val_type = $index[$binding_id][$rs . 'value'][0]['type']; + //$val_type = preg_match('/literal/', $val_type) ? 'literal' : $val_type; + $row[$var] = $val; + $row[$var . ' type'] = $val_type; + if ($dt = $this->v('datatype', 0, $index[$binding_id][$rs . 'value'][0])) { + $row[$var . ' datatype'] = $dt; + } + if ($lang = $this->v('lang', 0, $index[$binding_id][$rs . 'value'][0])) { + $row[$var . ' lang'] = $lang; + } + } + $r['rows'][] = $row; + } + } + return $r; + } + + /* */ + + function buildArrayHashIndex($rows) { + $r = array(); + foreach ($rows as $row) { + $hash = ''; + ksort($row); + foreach ($row as $k => $v) { + $hash .= is_numeric($k) ? '' : ' ' . md5($k) . ' ' . md5($v); + } + $r[] = $hash; + } + return $r; + } + + /* */ + + function adjustBnodes($result, $data) { + $mappings = array( + '_:b1371233574_bob' => '_:b10', + '_:b1114277307_alice' => '_:b1f', + '_:b1368422168_eve' => '_:b20', + '_:b1638119969_fred' => '_:b21', + + '_:b288335586_a' => array( + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#no-distinct-3' => '_:b0', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#distinct-3' => '_:b0', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#distinct-9' => '_:b0', + 'http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#no-distinct-9' => '_:b0', + 'default' => '_:bn5', + ), + ); + if (isset($result['rows'])) { + foreach ($result['rows'] as $i => $row) { + foreach ($result['variables'] as $var) { + if (isset($row[$var]) && isset($mappings[$row[$var]])) { + if (is_array($mappings[$row[$var]])) { + $result['rows'][$i][$var] = isset($mappings[$row[$var]][$data]) ? $mappings[$row[$var]][$data] : $mappings[$row[$var]]['default']; + } + else { + $result['rows'][$i][$var] = $mappings[$row[$var]]; + } + } + } + } + } + return $result; + } + +} diff --git a/arc2-2.2.4/tests/data/atom/feed.atom b/arc2-2.2.4/tests/data/atom/feed.atom new file mode 100755 index 0000000..446c131 --- /dev/null +++ b/arc2-2.2.4/tests/data/atom/feed.atom @@ -0,0 +1,20 @@ + +
My name is alice
" + }, + + "friend" : { + "type": "bnode", + "value": "r2" + } + },{ + "x" : { + "type": "bnode", + "value": "r2" + }, + + "hpage" : { + "type": "uri", + "value": "http://work.example.org/bob/" + }, + + "name" : { + "type": "literal", + "value": "Bob", + "xml:lang": "en" + }, + + "mbox" : { + "type": "uri", + "value": "mailto:bob@work.example.org" + }, + + "friend" : { + "type": "bnode", + "value": "r1" + } + } + ] + } + } diff --git a/arc2-2.2.4/tests/data/nt/test.nt b/arc2-2.2.4/tests/data/nt/test.nt new file mode 100755 index 0000000..3e52881 --- /dev/null +++ b/arc2-2.2.4/tests/data/nt/test.nt @@ -0,0 +1,78 @@ +# +# Copyright World Wide Web Consortium, (Massachusetts Institute of +# Technology, Institut National de Recherche en Informatique et en +# Automatique, Keio University). +# +# All Rights Reserved. +# +# Please see the full Copyright clause at +# [ … ] ." ;
+ mf:action ;