From 4a6c9e445149e42a4f81d5140296e7770c60bc6c Mon Sep 17 00:00:00 2001 From: Brion Vibber Date: Thu, 18 Mar 2010 17:55:21 -0700 Subject: Work around weird bug with HTML normalization via PHP DOM module; if source had xmlns and xml:lang I ended up with double output, breaking the subsequent parsing. Will have to track this down later and report upstream if not already resolved. --- plugins/OStatus/extlib/hkit/hkit.class.php | 2 +- plugins/OStatus/lib/discoveryhints.php | 21 ++++++++++++++++++++- 2 files changed, 21 insertions(+), 2 deletions(-) (limited to 'plugins') diff --git a/plugins/OStatus/extlib/hkit/hkit.class.php b/plugins/OStatus/extlib/hkit/hkit.class.php index c3a54cff6..fec6f4d8f 100644 --- a/plugins/OStatus/extlib/hkit/hkit.class.php +++ b/plugins/OStatus/extlib/hkit/hkit.class.php @@ -472,4 +472,4 @@ } -?> \ No newline at end of file +?> diff --git a/plugins/OStatus/lib/discoveryhints.php b/plugins/OStatus/lib/discoveryhints.php index 4da2ec0f1..0273b5a92 100644 --- a/plugins/OStatus/lib/discoveryhints.php +++ b/plugins/OStatus/lib/discoveryhints.php @@ -174,6 +174,26 @@ class DiscoveryHints { error_reporting($old); if ($ok) { + // If the original had xmlns or xml:lang attributes on the + // , we seen to end up with duplicates, which causes + // parse errors. Remove em! + // + // For some reason we have to iterate and remove them twice, + // *plus* they don't show up on hasAttribute() or removeAttribute(). + // This might be some weird bug in PHP or libxml2, uncertain if + // it affects other folks consistently. + $root = $dom->documentElement; + foreach ($root->attributes as $i => $x) { + if ($i == 'xmlns' || $i == 'xml:lang') { + $root->removeAttributeNode($x); + } + } + foreach ($root->attributes as $i => $x) { + if ($i == 'xmlns' || $i == 'xml:lang') { + $root->removeAttributeNode($x); + } + } + // hKit doesn't give us a chance to pass the source URL for // resolving relative links, such as the avatar photo on a // Google profile. We'll slip it into a tag if there's @@ -192,7 +212,6 @@ class DiscoveryHints { $head = $heads->item(0); } else { $head = $dom->createElement('head'); - $root = $dom->documentRoot; if ($root->firstChild) { $root->insertBefore($head, $root->firstChild); } else { -- cgit v1.2.3-54-g00ecf From 17c50f338ceb574780476f6b788f48e2d7d06017 Mon Sep 17 00:00:00 2001 From: Evan Prodromou Date: Thu, 18 Mar 2010 20:52:00 -0500 Subject: Remove hkit and do our own hcard parsing Parsing hcards for the data we need wasn't hard enough to justify using hkit. It was dependent on a number of external systems (something to run tidy), and only could handle XHTML. We now parse HTML with the PHP dom libraries used elsewhere, and scrape out our own hcards. Seems to work nicer and faster and most of all works with Google Buzz profile URLs. --- plugins/OStatus/extlib/hkit/hcard.profile.php | 105 ------ plugins/OStatus/extlib/hkit/hkit.class.php | 475 -------------------------- plugins/OStatus/lib/discoveryhints.php | 194 +++++++---- 3 files changed, 130 insertions(+), 644 deletions(-) delete mode 100644 plugins/OStatus/extlib/hkit/hcard.profile.php delete mode 100644 plugins/OStatus/extlib/hkit/hkit.class.php (limited to 'plugins') diff --git a/plugins/OStatus/extlib/hkit/hcard.profile.php b/plugins/OStatus/extlib/hkit/hcard.profile.php deleted file mode 100644 index 6ec0dc890..000000000 --- a/plugins/OStatus/extlib/hkit/hcard.profile.php +++ /dev/null @@ -1,105 +0,0 @@ -root_class = 'vcard'; - - $this->classes = array( - 'fn', array('honorific-prefix', 'given-name', 'additional-name', 'family-name', 'honorific-suffix'), - 'n', array('honorific-prefix', 'given-name', 'additional-name', 'family-name', 'honorific-suffix'), - 'adr', array('post-office-box', 'extended-address', 'street-address', 'postal-code', 'country-name', 'type', 'region', 'locality'), - 'label', 'bday', 'agent', 'nickname', 'photo', 'class', - 'email', array('type', 'value'), - 'category', 'key', 'logo', 'mailer', 'note', - 'org', array('organization-name', 'organization-unit'), - 'tel', array('type', 'value'), - 'geo', array('latitude', 'longitude'), - 'tz', 'uid', 'url', 'rev', 'role', 'sort-string', 'sound', 'title' - ); - - // classes that must only appear once per card - $this->singles = array( - 'fn' - ); - - // classes that are required (not strictly enforced - give at least one!) - $this->required = array( - 'fn' - ); - - $this->att_map = array( - 'fn' => array('IMG|alt'), - 'url' => array('A|href', 'IMG|src', 'AREA|href'), - 'photo' => array('IMG|src'), - 'bday' => array('ABBR|title'), - 'logo' => array('IMG|src'), - 'email' => array('A|href'), - 'geo' => array('ABBR|title') - ); - - - $this->callbacks = array( - 'url' => array($this, 'resolvePath'), - 'photo' => array($this, 'resolvePath'), - 'logo' => array($this, 'resolvePath'), - 'email' => array($this, 'resolveEmail') - ); - - - - function hKit_hcard_post($a) - { - - foreach ($a as &$vcard){ - - hKit_implied_n_optimization($vcard); - hKit_implied_n_from_fn($vcard); - - } - - return $a; - - } - - - function hKit_implied_n_optimization(&$vcard) - { - if (array_key_exists('fn', $vcard) && !is_array($vcard['fn']) && - !array_key_exists('n', $vcard) && (!array_key_exists('org', $vcard) || $vcard['fn'] != $vcard['org'])){ - - if (sizeof(explode(' ', $vcard['fn'])) == 2){ - $patterns = array(); - $patterns[] = array('/^(\S+),\s*(\S{1})$/', 2, 1); // Lastname, Initial - $patterns[] = array('/^(\S+)\s*(\S{1})\.*$/', 2, 1); // Lastname Initial(.) - $patterns[] = array('/^(\S+),\s*(\S+)$/', 2, 1); // Lastname, Firstname - $patterns[] = array('/^(\S+)\s*(\S+)$/', 1, 2); // Firstname Lastname - - foreach ($patterns as $pattern){ - if (preg_match($pattern[0], $vcard['fn'], $matches) === 1){ - $n = array(); - $n['given-name'] = $matches[$pattern[1]]; - $n['family-name'] = $matches[$pattern[2]]; - $vcard['n'] = $n; - - - break; - } - } - } - } - } - - - function hKit_implied_n_from_fn(&$vcard) - { - if (array_key_exists('fn', $vcard) && is_array($vcard['fn']) - && !array_key_exists('n', $vcard) && (!array_key_exists('org', $vcard) || $vcard['fn'] != $vcard['org'])){ - - $vcard['n'] = $vcard['fn']; - } - - if (array_key_exists('fn', $vcard) && is_array($vcard['fn'])){ - $vcard['fn'] = $vcard['fn']['text']; - } - } - -?> \ No newline at end of file diff --git a/plugins/OStatus/extlib/hkit/hkit.class.php b/plugins/OStatus/extlib/hkit/hkit.class.php deleted file mode 100644 index c3a54cff6..000000000 --- a/plugins/OStatus/extlib/hkit/hkit.class.php +++ /dev/null @@ -1,475 +0,0 @@ -' . implode(', ', $missing) . ''); - - } - - - public function getByURL($profile='', $url='') - { - - if ($profile=='' || $url == '') return false; - - $this->loadProfile($profile); - - $source = $this->loadURL($url); - - if ($source){ - $tidy_xhtml = $this->tidyThis($source); - - $fragment = false; - - if (strrchr($url, '#')) - $fragment = array_pop(explode('#', $url)); - - $doc = $this->loadDoc($tidy_xhtml, $fragment); - $s = $this->processNodes($doc, $this->classes); - $s = $this->postProcess($profile, $s); - - return $s; - }else{ - return false; - } - } - - public function getByString($profile='', $input_xml='') - { - if ($profile=='' || $input_xml == '') return false; - - $this->loadProfile($profile); - - $doc = $this->loadDoc($input_xml); - $s = $this->processNodes($doc, $this->classes); - $s = $this->postProcess($profile, $s); - - return $s; - - } - - private function processNodes($items, $classes, $allow_includes=true){ - - $out = array(); - - foreach($items as $item){ - $data = array(); - - for ($i=0; $ixpath($xpath); - - if ($results){ - foreach ($results as $result){ - if (isset($classes[$i+1]) && is_array($classes[$i+1])){ - $nodes = $this->processNodes($results, $classes[$i+1]); - if (sizeof($nodes) > 0){ - $nodes = array_merge(array('text'=>$this->getNodeValue($result, $classes[$i])), $nodes); - $data[$classes[$i]] = $nodes; - }else{ - $data[$classes[$i]] = $this->getNodeValue($result, $classes[$i]); - } - - }else{ - if (isset($data[$classes[$i]])){ - if (is_array($data[$classes[$i]])){ - // is already an array - append - $data[$classes[$i]][] = $this->getNodeValue($result, $classes[$i]); - - }else{ - // make it an array - if ($classes[$i] == 'value'){ // unless it's the 'value' of a type/value pattern - $data[$classes[$i]] .= $this->getNodeValue($result, $classes[$i]); - }else{ - $old_val = $data[$classes[$i]]; - $data[$classes[$i]] = array($old_val, $this->getNodeValue($result, $classes[$i])); - $old_val = false; - } - } - }else{ - // set as normal value - $data[$classes[$i]] = $this->getNodeValue($result, $classes[$i]); - - } - } - - // td@headers pattern - if (strtoupper(dom_import_simplexml($result)->tagName)== "TD" && $result['headers']){ - $include_ids = explode(' ', $result['headers']); - $doc = $this->doc; - foreach ($include_ids as $id){ - $xpath = "//*[@id='$id']/.."; - $includes = $doc->xpath($xpath); - foreach ($includes as $include){ - $tmp = $this->processNodes($include, $this->classes); - if (is_array($tmp)) $data = array_merge($data, $tmp); - } - } - } - } - } - } - $result = false; - } - - // include-pattern - if ($allow_includes){ - $xpath = ".//*[contains(concat(' ',normalize-space(@class),' '),' include ')]"; - $results = $item->xpath($xpath); - - if ($results){ - foreach ($results as $result){ - $tagName = strtoupper(dom_import_simplexml($result)->tagName); - if ((($tagName == "OBJECT" && $result['data']) || ($tagName == "A" && $result['href'])) - && preg_match('/\binclude\b/', $result['class'])){ - $att = ($tagName == "OBJECT" ? 'data' : 'href'); - $id = str_replace('#', '', $result[$att]); - $doc = $this->doc; - $xpath = "//*[@id='$id']"; - $includes = $doc->xpath($xpath); - foreach ($includes as $include){ - $include = simplexml_load_string(''.$include->asXML().''); // don't ask. - $tmp = $this->processNodes($include, $this->classes, false); - if (is_array($tmp)) $data = array_merge($data, $tmp); - } - } - } - } - } - $out[] = $data; - } - - if (sizeof($out) > 1){ - return $out; - }else if (isset($data)){ - return $data; - }else{ - return array(); - } - } - - - private function getNodeValue($node, $className) - { - - $tag_name = strtoupper(dom_import_simplexml($node)->tagName); - $s = false; - - // ignore DEL tags - if ($tag_name == 'DEL') return $s; - - // look up att map values - if (array_key_exists($className, $this->att_map)){ - - foreach ($this->att_map[$className] as $map){ - if (preg_match("/$tag_name\|/", $map)){ - $s = ''.$node[array_pop($foo = explode('|', $map))]; - } - } - } - - // if nothing and OBJ, try data. - if (!$s && $tag_name=='OBJECT' && $node['data']) $s = ''.$node['data']; - - // if nothing and IMG, try alt. - if (!$s && $tag_name=='IMG' && $node['alt']) $s = ''.$node['alt']; - - // if nothing and AREA, try alt. - if (!$s && $tag_name=='AREA' && $node['alt']) $s = ''.$node['alt']; - - //if nothing and not A, try title. - if (!$s && $tag_name!='A' && $node['title']) $s = ''.$node['title']; - - - // if nothing found, go with node text - $s = ($s ? $s : implode(array_filter($node->xpath('child::node()'), array(&$this, "filterBlankValues")), ' ')); - - // callbacks - if (array_key_exists($className, $this->callbacks)){ - $s = preg_replace_callback('/.*/', $this->callbacks[$className], $s, 1); - } - - // trim and remove line breaks - if ($tag_name != 'PRE'){ - $s = trim(preg_replace('/[\r\n\t]+/', '', $s)); - $s = trim(preg_replace('/(\s{2})+/', ' ', $s)); - } - - return $s; - } - - private function filterBlankValues($s){ - return preg_match("/\w+/", $s); - } - - - private function tidyThis($source) - { - switch ( $this->tidy_mode ) - { - case 'exec': - $tmp_file = $this->tmp_dir.md5($source).'.txt'; - file_put_contents($tmp_file, $source); - exec("tidy -utf8 -indent -asxhtml -numeric -bare -quiet $tmp_file", $tidy); - unlink($tmp_file); - return implode("\n", $tidy); - break; - - case 'php': - $tidy = tidy_parse_string($source); - return tidy_clean_repair($tidy); - break; - - default: - return $source; - break; - } - - } - - - private function loadProfile($profile) - { - require_once("$profile.profile.php"); - } - - - private function loadDoc($input_xml, $fragment=false) - { - $xml = simplexml_load_string($input_xml); - - $this->doc = $xml; - - if ($fragment){ - $doc = $xml->xpath("//*[@id='$fragment']"); - $xml = simplexml_load_string($doc[0]->asXML()); - $doc = null; - } - - // base tag - if ($xml->head->base['href']) $this->base = $xml->head->base['href']; - - // xml:base attribute - PITA with SimpleXML - preg_match('/xml:base="(.*)"/', $xml->asXML(), $matches); - if (is_array($matches) && sizeof($matches)>1) $this->base = $matches[1]; - - return $xml->xpath("//*[contains(concat(' ',normalize-space(@class),' '),' $this->root_class ')]"); - - } - - - private function loadURL($url) - { - $this->url = $url; - - if ($this->tidy_mode == 'proxy' && $this->tidy_proxy != ''){ - $url = $this->tidy_proxy . $url; - } - - return @file_get_contents($url); - - } - - - private function postProcess($profile, $s) - { - $required = $this->required; - - if (is_array($s) && array_key_exists($required[0], $s)){ - $s = array($s); - } - - $s = $this->dedupeSingles($s); - - if (function_exists('hKit_'.$profile.'_post')){ - $s = call_user_func('hKit_'.$profile.'_post', $s); - } - - $s = $this->removeTextVals($s); - - return $s; - } - - - private function resolvePath($filepath) - { // ugly code ahoy: needs a serious tidy up - - $filepath = $filepath[0]; - - $base = $this->base; - $url = $this->url; - - if ($base != '' && strpos($base, '://') !== false) - $url = $base; - - $r = parse_url($url); - $domain = $r['scheme'] . '://' . $r['host']; - - if (!isset($r['path'])) $r['path'] = '/'; - $path = explode('/', $r['path']); - $file = explode('/', $filepath); - $new = array(''); - - if (strpos($filepath, '://') !== false || strpos($filepath, 'data:') !== false){ - return $filepath; - } - - if ($file[0] == ''){ - // absolute path - return ''.$domain . implode('/', $file); - }else{ - // relative path - if ($path[sizeof($path)-1] == '') array_pop($path); - if (strpos($path[sizeof($path)-1], '.') !== false) array_pop($path); - - foreach ($file as $segment){ - if ($segment == '..'){ - array_pop($path); - }else{ - $new[] = $segment; - } - } - return ''.$domain . implode('/', $path) . implode('/', $new); - } - } - - private function resolveEmail($v) - { - $parts = parse_url($v[0]); - return ($parts['path']); - } - - - private function dedupeSingles($s) - { - $singles = $this->singles; - - foreach ($s as &$item){ - foreach ($singles as $classname){ - if (array_key_exists($classname, $item) && is_array($item[$classname])){ - if (isset($item[$classname][0])) $item[$classname] = $item[$classname][0]; - } - } - } - - return $s; - } - - private function removeTextVals($s) - { - foreach ($s as $key => &$val){ - if ($key){ - $k = $key; - }else{ - $k = ''; - } - - if (is_array($val)){ - $val = $this->removeTextVals($val); - }else{ - if ($k == 'text'){ - $val = ''; - } - } - } - - return array_filter($s); - } - - } - - -?> \ No newline at end of file diff --git a/plugins/OStatus/lib/discoveryhints.php b/plugins/OStatus/lib/discoveryhints.php index db13793dd..1bb0ad2ae 100644 --- a/plugins/OStatus/lib/discoveryhints.php +++ b/plugins/OStatus/lib/discoveryhints.php @@ -63,49 +63,12 @@ class DiscoveryHints { static function hcardHints($body, $url) { - common_debug("starting tidy"); - - $body = self::_tidy($body); - - common_debug("done with tidy"); - - set_include_path(get_include_path() . PATH_SEPARATOR . INSTALLDIR . '/plugins/OStatus/extlib/hkit/'); - require_once('hkit.class.php'); - - $h = new hKit; - - $hcards = $h->getByString('hcard', $body); - - if (empty($hcards)) { - return array(); - } - - if (count($hcards) == 1) { - $hcard = $hcards[0]; - } else { - foreach ($hcards as $try) { - if (array_key_exists('url', $try)) { - if (is_string($try['url']) && $try['url'] == $url) { - $hcard = $try; - break; - } else if (is_array($try['url'])) { - foreach ($try['url'] as $tryurl) { - if ($tryurl == $url) { - $hcard = $try; - break 2; - } - } - } - } - } - // last chance; grab the first one - if (empty($hcard)) { - $hcard = $hcards[0]; - } - } + $hcard = self::_hcard($body, $url); $hints = array(); + // XXX: don't copy stuff into an array and then copy it again + if (array_key_exists('nickname', $hcard)) { $hints['nickname'] = $hcard['nickname']; } @@ -117,7 +80,7 @@ class DiscoveryHints { } if (array_key_exists('photo', $hcard)) { - $hints['avatar'] = $hcard['photo']; + $hints['avatar'] = $hcard['photo'][0]; } if (array_key_exists('note', $hcard)) { @@ -144,39 +107,142 @@ class DiscoveryHints { return $hints; } - private static function _tidy($body) + static function _hcard($body, $url) { - if (function_exists('tidy_parse_string')) { - common_debug("Tidying with extension"); - $text = tidy_parse_string($body); - $text = tidy_clean_repair($text); - return $body; - } else if ($fullpath = self::_findProgram('tidy')) { - common_debug("Tidying with program $fullpath"); - $tempfile = tempnam('/tmp', 'snht'); // statusnet hcard tidy - file_put_contents($tempfile, $source); - exec("$fullpath -utf8 -indent -asxhtml -numeric -bare -quiet $tempfile", $tidy); - unlink($tempfile); - return implode("\n", $tidy); + // DOMDocument::loadHTML may throw warnings on unrecognized elements. + + $old = error_reporting(error_reporting() & ~E_WARNING); + + $doc = new DOMDocument(); + $doc->loadHTML($body); + + error_reporting($old); + + $xp = new DOMXPath($doc); + + $hcardNodes = self::_getChildrenByClass($doc->documentElement, 'vcard', $xp); + + $hcards = array(); + + for ($i = 0; $i < $hcardNodes->length; $i++) { + + $hcardNode = $hcardNodes->item($i); + + $hcard = self::_hcardFromNode($hcardNode, $xp, $url); + + $hcards[] = $hcard; + } + + $repr = null; + + foreach ($hcards as $hcard) { + if (in_array($url, $hcard['url'])) { + $repr = $hcard; + break; + } + } + + if (!is_null($repr)) { + return $repr; + } else if (count($hcards) > 0) { + return $hcards[0]; } else { - common_debug("Not tidying."); - return $body; + return null; } } - private static function _findProgram($name) + function _getChildrenByClass($el, $cls, $xp) + { + // borrowed from hkit. Thanks dudes! + + $qry = ".//*[contains(concat(' ',normalize-space(@class),' '),' $cls ')]"; + + $nodes = $xp->query($qry, $el); + + return $nodes; + } + + function _hcardFromNode($hcardNode, $xp, $base) { - $path = $_ENV['PATH']; + $hcard = array(); + + $hcard['url'] = array(); + + $urlNodes = self::_getChildrenByClass($hcardNode, 'url', $xp); + + for ($j = 0; $j < $urlNodes->length; $j++) { + + $urlNode = $urlNodes->item($j); + + if ($urlNode->hasAttribute('href')) { + $url = $urlNode->getAttribute('href'); + } else { + $url = $urlNode->textContent; + } + + $hcard['url'][] = self::_rel2abs($url, $base); + } + + $hcard['photo'] = array(); + + $photoNodes = self::_getChildrenByClass($hcardNode, 'photo', $xp); + + for ($j = 0; $j < $photoNodes->length; $j++) { + $photoNode = $photoNodes->item($j); + if ($photoNode->hasAttribute('src')) { + $url = $photoNode->getAttribute('src'); + } else if ($photoNode->hasAttribute('href')) { + $url = $photoNode->getAttribute('href'); + } else { + $url = $photoNode->textContent; + } + $hcard['photo'][] = self::_rel2abs($url, $base); + } + + $singles = array('nickname', 'note', 'fn', 'n', 'adr'); - $parts = explode(':', $path); + foreach ($singles as $single) { - foreach ($parts as $part) { - $fullpath = $part . '/' . $name; - if (is_executable($fullpath)) { - return $fullpath; + $nodes = self::_getChildrenByClass($hcardNode, $single, $xp); + + if ($nodes->length > 0) { + $node = $nodes->item(0); + $hcard[$single] = $node->textContent; } } - return null; + return $hcard; + } + + // XXX: this is a first pass; we probably need + // to handle things like ../ and ./ and so on + + static function _rel2abs($rel, $wrt) + { + $parts = parse_url($rel); + + if ($parts === false) { + return false; + } + + // If it's got a scheme, use it + + if ($parts['scheme'] != '') { + return $rel; + } + + $w = parse_url($wrt); + + $base = $w['scheme'].'://'.$w['host']; + + if ($rel[0] == '/') { + return $base.$rel; + } + + $wp = explode('/', $w['path']); + + array_pop($wp); + + return $base.implode('/', $wp).'/'.$rel; } } -- cgit v1.2.3-54-g00ecf