Home » Keires_Feed » Timeline » 395
Diffs
Keires_Feed/trunk/src/Keires/ExceptionWrapper.php
+<?php +/** + * Keires_Feed + * + * Feed Parser like XML_Feed_Parser + * + * PHP version 5 + * + * LICENSE: This source file is subject to version 3.0 of the PHP license + * that is available through the world-wide-web at the following URI: + * http://www.php.net/license/3_0.txt. If you did not receive a copy of + * the PHP License and are unable to obtain it through the web, please + * send a note to license@php.net so we can mail you a copy immediately. + * + * @category XML + * @package Keires_Feed + * @author KOYAMA Tetsuji <koyama@hoge.hoge.org> + * @copyright 2008-2009 KOYAMA Tetsuji + * @license http://www.php.net/license/3_0.txt PHP License 3.0 + * @version svn: $Id$ + * @link http://openpear.org/package/Keires_Feed + * @since File available since Release 0.1 + */ + +require_once 'PEAR.php'; +require_once 'PEAR/Exception.php'; + +class Keires_ExceptionWrapper { + + static public function init() { + static $initialized = false; + if (!$initialized) { + PEAR::setErrorHandling(PEAR_ERROR_CALLBACK, + array('Keires_ExceptionWrapper', + 'handleError')); + $initialized = true; + } + } + + static public function wrap($obj) { + if (is_a($obj, 'PEAR')) { + $obj->setErrorHandling(PEAR_ERROR_CALLBACK, + array('Keires_ExceptionWrapper', + 'handleError')); + } + } + + static public function handleError($err) { + throw self::factory($err); + } + + static public function factory($err) { + // default exception class + $class = 'PEAR_Exception'; + + return new $class($err->getMessage(), $err->getCode()); + } +} + +?> \ No newline at end of file 属性に変更があったパス: Keires_Feed/trunk/src/Keires/ExceptionWrapper.php ___________________________________________________________________ 追加: svn:keywords + Id 追加: svn:eol-style + native
Keires_Feed/trunk/src/Keires/FeedParser.php
+<?php +/** + * Keires_Feed + * + * Feed Parser like XML_Feed_Parser + * + * PHP version 5 + * + * LICENSE: This source file is subject to version 3.0 of the PHP license + * that is available through the world-wide-web at the following URI: + * http://www.php.net/license/3_0.txt. If you did not receive a copy of + * the PHP License and are unable to obtain it through the web, please + * send a note to license@php.net so we can mail you a copy immediately. + * + * @category XML + * @package Keires_Feed + * @author KOYAMA Tetsuji <koyama@hoge.hoge.org> + * @copyright 2008-2009 KOYAMA Tetsuji + * @license http://www.php.net/license/3_0.txt PHP License 3.0 + * @version svn: $Id$ + * @link http://openpear.org/package/Keires_Feed + * @since File available since Release 0.1 + * + * based on Rasmus Lerdorf's simple_rss.php + * ( http://lerdorf.com/php/simple_rss.phps + */ + +require_once 'Keires/Feed.php'; + +class Keires_FeedParser { + + protected $xmlversion = '1.0'; + protected $encoding = null; + protected $xml = null; + protected $feed = null; + + public function __construct($xml, $options = null) { + $this->xml = $xml; + $this->analyze(); + } + + public function getFeed() { + return $this->feed; + } + + static protected function firstLine(&$text) { + $p1 = strpos($text, "\r"); + $p2 = strpos($text, "\n"); + $p = FALSE; + + if ($p1 === FALSE) { + $p = $p2; + } else { + if ($p2 === FALSE) { + $p = $p1; + } else { + $p = min($p1, $p2); + } + } + if ($p === FALSE) { + // no or long length line + return substr($text, 0, 256); + } + return substr($text, 0, $p); + } + + protected function analyze() { + $xml = $this->xml; + + $first_line = $this->firstLine($xml); + $nmatch = preg_match('!<?xml version=[\'"](.+)[\'"] encoding=[\'"](.+)[\'"]\s*\?>!i', + $first_line, $matches); + if ($nmatch != 1) { + throw new Keires_Exception('Invalid XML document', + KEIRES_ERROR_INVALID_XML); + } + + $this->xmlversion = $matches[1]; + $this->encoding = $matches[2]; + } + + public function parse() { + $options = array( + 'version' => $this->xmlversion, + 'encoding' => $this->encoding, + ); + $feed = self::parseXML($this->xml, $options); + $this->feed = $feed; + } + + public function type() { + if (empty($this->feed)) { + throw new Keires_Exception('feed is not parsed yet', + KEIRES_ERROR_UNPARSED); + } + return $this->feed['_type']; + } + + public function items() { + if (empty($this->feed)) { + throw new Keires_Exception('feed not parsed yet', + KEIRES_ERROR_UNPARSED); + } + return $this->feed['item']; + } + + static public function feedType($dom) { + $type = null; + foreach ($dom->childNodes as $child) { + $node = $child->nodeName; + switch ($node) { + case 'rss': + $type = 'rss'; + break; + case 'feed': + $type = 'atom'; + break; + default: + if (strncmp($node, 'rdf', 3) === 0) { + $type = 'rss'; + } + } + if (!empty($type)) { + break; + } + } + return $type; + } + + static public function parseXML($input_xml, $options) { + $version = array_val($options, 'version', '1.0'); + $encoding = array_val($options, 'encoding', 'UTF-8'); + + $dom = new DOMDocument($version, $encoding); + + $result = $dom->loadXML($input_xml); + if ($result === FALSE) { + throw new Keires_Exception('DOMDocument load error', + KEIRES_ERROR_DOM_LOADING); + } + $feedtype = self::feedType($dom); + + // Pick out the namespaces that apply to this doc. + // We need to do this from DOM because simplexml does't see the + // special xmlns attributes because of the way libxml2 handles them. + $xpath = new DOMXPath($dom); + $ns = array(''=>NULL); + foreach($xpath->query("namespace::*") as $v) { + if($v->localName!='xml') $ns[$v->localName] = $v->nodeValue; + } + $atom10 = $atom03 = false; + if ($feedtype != 'rss') { + if (in_array('http://www.w3.org/2005/Atom', $ns)) { + $atom10 = true; + unset($ns['xmlns']); + } else if (in_array('http://purl.org/atom/ns#', $ns)) { + $atom03 = true; + unset($ns['xmlns']); + } + } + + // Ok, now we can switch to simplexml + $xml = simplexml_import_dom($dom); + if ($xml === null) { + throw new Keires_Exception('XML format error', + KEIRES_ERROR_INVALID_XML); + } + $feed = array(); + + // Pull out the root attributes - usually just version + foreach($xml->attributes() as $k=>$v) $feed[$k] = (string)$v; + + // We will deal with the items separately, so start by only looking + // at the stuff leading up to the items checking each namespace. + $rss1 = false; + if ($atom10 || $atom03) { + $top = $xml; + } else { + $top = $xml->channel; + } + foreach($ns as $alias=>$uri) foreach($top->children($uri) as $key=>$val) { + if($key=="item" || $key=="entry") continue; + if($key=="items") { + $rss1 = true; continue; + } + if(!$val->children()) { + $feed[$key][0] = (string)$val; + foreach($ns as $a=>$u) foreach($val->attributes($u) as $at=>$atv) { + $feed[$key][$at] = (string)$atv; + } + } else { + foreach($val->children() as $k=>$v) { + $feed[$key][$k] = (string)$v; + foreach($v->attributes() as $at=>$atv) { + $feed[$k][$at] = (string)$atv; + } + } + } + } + + // Now we deal with the items + // Atom and RSS1 have the feed items a level higher than RSS2 + $i = 0; + if($rss1) { $feed['_type']='rss1.0'; $items = $xml->item; } + else if($atom10) { $feed['_type']='atom1.0'; $items = $xml->entry; } + else if($atom03) { $feed['_type']='atom0.3'; $items = $xml->entry; } + else { + if($feed['version']=='2.0') $feed['_type']='rss2.0'; + else if($feed['version']=='0.91') $feed['_type']='rss0.91'; + $items = $xml->channel->item; + } + foreach($items as $key=>$val) { + foreach($ns as $a=>$u) foreach($val->attributes($u) as $at=>$atv) { + $feed['item'][$i][$at] = (string)$atv; + } + foreach($ns as $alias=>$uri) { + foreach($val->children($uri) as $k=>$v) { + $feed['item'][$i][$k][0] = (string)$v; + foreach($v->attributes() as $at=>$atv) { + $at_val = (string)$atv; + if ($atom10 || $atom03) { + $feed['item'][$i][$k][$at][] = $at_val; + // Don't even try parsing this stuff, just pass it through. + if($at_val=='xhtml' || $at_val=='html' || $at_val=='text') { + $tags = $v->children(); + $feed['item'][$i][$k]['text'] = $tags->asXML(); + } + } else $feed['item'][$i][$k][$at] = $at_val; + + } + } + } + $i++; + } + return $feed; + } +} + +?> 属性に変更があったパス: Keires_Feed/trunk/src/Keires/FeedParser.php ___________________________________________________________________ 追加: svn:keywords + Id 追加: svn:eol-style + native
Keires_Feed/trunk/src/Keires/Feed.php
+<?php +/** + * Keires_Feed + * + * Feed Parser like XML_Feed_Parser + * + * PHP version 5 + * + * LICENSE: This source file is subject to version 3.0 of the PHP license + * that is available through the world-wide-web at the following URI: + * http://www.php.net/license/3_0.txt. If you did not receive a copy of + * the PHP License and are unable to obtain it through the web, please + * send a note to license@php.net so we can mail you a copy immediately. + * + * @category XML + * @package Keires_Feed + * @author KOYAMA Tetsuji <koyama@hoge.hoge.org> + * @copyright 2008-2009 KOYAMA Tetsuji + * @license http://www.php.net/license/3_0.txt PHP License 3.0 + * @version svn: $Id$ + * @link http://openpear.org/package/Keires_Feed + * @since File available since Release 0.1 + */ + +require_once 'Keires/FeedParser.php'; +require_once 'Keires/ExceptionWrapper.php'; +require_once 'HTTP/Request.php'; // PEAR::HTTP_Request + +/* + * Error codes + */ +define('KEIRES_ERROR_INVALID_PARAM', 1); +define('KEIRES_ERROR_OPEN_FILE', 2); +define('KEIRES_ERROR_EMPTY_CONTENTS', 3); +define('KEIRES_ERROR_UNPARSED', 4); +define('KEIRES_ERROR_INVALID_XML', 5); +define('KEIRES_ERROR_DOM_LOADING', 6); + +/* + * Utility function + */ +if (!function_exists('array_val')) { + function array_val(&$data, $key, $default = null) { + if (!is_array($data)) { + return $default; + } + return isset($data[$key])? $data[$key]: $default; + } +} + +/* + * Excepiton + */ +class Keires_Exception extends Exception { + // empty +} + +class Keires_Feed { + protected $contents = null; + protected $parser = null; + + public function __construct($url, $options = null) { + $noreq = array_val($options, 'noreq', false); + $ua = array_val($options, 'user_agent'); + + if ($noreq) { + return; + } + + if (empty($url)) { + throw new Keires_Exception('url is required', + KEIRES_ERROR_INVALID_PARAM); + } + + $http_opts = array( + 'allowRedirects' => true, + ); + $req = new HTTP_Request($url, $http_opts); + if (!empty($ua)) { + $req->addHeader('User-Agent', $ua); + } + $result = $req->sendRequest(); + if (PEAR::isError($result)) { + Keires_ExceptionWrapper::handleError($result); + } + + $code = $req->getResponseCode(); + if ($code != 200) { + throw new Keires_Exception('path open failed: '. $url, + KEIRES_ERROR_OPEN_FILE); + } + + $cont = $req->getResponseBody(); + if ($cont === false) { + throw new Keires_Exception('path open failed: '. $url, + KEIRES_ERROR_OPEN_FILE); + } + + $this->contents = $cont; + } + + public function setContents($contents) { + $this->contents = $contents; + } + + public function getContents() { + return $this->contents; + } + + public function parse($options = null) { + if (empty($this->contents)) { + throw new Keires_Exception('empty contents', + KEIRES_ERROR_EMPTY_CONTENTS); + } + $parser = new Keires_FeedParser($this->contents, $options); + $parser->parse(); + + $this->parser = $parser; + } + + public function getItems() { + if (empty($this->parser)) { + throw new Keires_Exception('Not parsed yet', + KEIRES_ERROR_UNPARSED); + } + $type = $this->parser->type(); + $item = $this->parser->items(); + return new Keires_FeedItem($item, $type); + } +} + +class Keires_FeedItem implements Iterator { + protected $pos = 0; + protected $type = null; + protected $items = null; + + public function __construct($items, $type) { + if (empty($items) || empty($type)) { + throw new Keires_Exception('Empty params', + KEIRES_ERROR_INVALID_PARAM); + } + $this->type = $type; + $this->items = $items; + } + + static protected function getVal($item, $key) { + $v = array_val($item, $key); + for (;;) { + if (is_array($v)) { + $v = $v[0]; + } else { + break; + } + } + return $v; + } + + static public function getValKeys($item, $keys) { + if (!is_array($keys)) { + return null; + } + foreach ($keys as $key) { + $data = self::getVal($item, $key); + if (!empty($data)) { + return $data; + } + } + return null; + } + + static public function parseEntryRSS($item) { + $datekeys = array('date', 'pubDate'); + $pubdate = self::getValKeys($item, $datekeys); + $entry = array( + 'url' => self::getVal($item, 'link'), + 'title' => self::getVal($item, 'title'), + 'description' => self::getVal($item, 'description'), + 'content' => self::getVal($item, 'encoded'), + 'creator' => self::getVal($item, 'creator'), + 'pubdate' => $pubdate, + ); + return $entry; + } + + static public function parseEntryATOM($item) { + // get url + $link = $item['link']; + $url = null; + + // for feedburner + $url = self::getVal($item, 'origLink'); + + if (empty($url)) { + // first: find <link rel="alternate"> + foreach ($link['rel'] as $idx => $rel) { + if ($rel === 'alternate') { + $url = $link['href'][$idx]; + break; + } + } + } + if (empty($url)) { + // second: <link type="text/html"> + foreach ($link['type'] as $idx => $type) { + if (($type === 'text/html') || + ($type === 'application/xhtml+xml')){ + $url = $link['href'][$idx]; + break; + } + } + } + if (empty($url)) { + // for lifehacking.jp + $url = self::getVal($item, 'id'); + } + // get published date + $datekeys = array( + 'created', + 'published', + 'modified', + 'issued', + ); + $pubdate = self::getValKeys($item, $datekeys); + $entry = array( + 'url' => $url, + 'title' => self::getVal($item, 'title'), + 'description' => self::getVal($item, 'summary'), + 'creator' => self::getVal($item, 'author'), + 'pubdate' => $pubdate, + ); + return $entry; + } + + public function rewind() { + $this->pos = 0; + } + + public function current() { + $item = $this->items[$this->pos]; + $entry = null; + if (strncmp($this->type, 'rss', 3) === 0) { + $entry = self::parseEntryRSS($item); + } else if (strncmp($this->type, 'atom', 4) === 0) { + $entry = self::parseEntryATOM($item); + } + return $entry; + } + + public function key() { + return $this->pos; + } + + public function next() { + ++$this->pos; + } + + public function valid() { + return ($this->pos < count($this->items)); + } +} + +?> 属性に変更があったパス: Keires_Feed/trunk/src/Keires/Feed.php ___________________________________________________________________ 追加: svn:keywords + Id 追加: svn:eol-style + native
Keires_Feed/trunk/examples/feedparser.php
+<?php +// $Id$ + +require_once 'Keires/FeedParser.php'; + +function usage() { + echo "Usage: program feed_url\n"; +} + +try { + if ($argc != 2) { + usage(); + exit; + } + + $url = $argv[1]; + $xml = file_get_contents($url); + + $parser = new Keires_FeedParser($xml); + $parser->parse(); + var_dump($parser->getFeed()); + +} catch (Exception $e) { + die($e->getMessage()); +} + +?> \ No newline at end of file 属性に変更があったパス: Keires_Feed/trunk/examples/feedparser.php ___________________________________________________________________ 追加: svn:keywords + Id 追加: svn:eol-style + native
Keires_Feed/trunk/examples/feed.php
+<?php +// $Id$ + +require_once 'Keires/Feed.php'; + +function usage() { + echo "Usage: program feed_url\n"; +} + +try { + if ($argc != 2) { + usage(); + exit; + } + + $url = $argv[1]; + $xml = file_get_contents($url); + + $opt = array( + 'noreq' => true, + ); + $feed = new Keires_Feed(null, $opt); + $feed->setContents($xml); + + $feed->parse(); + + $items = $feed->getItems(); + + foreach ($items as $item) { + var_dump($item); + } + +} catch (Exception $e) { + die($e->getMessage()); + } + + + +?> \ No newline at end of file 属性に変更があったパス: Keires_Feed/trunk/examples/feed.php ___________________________________________________________________ 追加: svn:keywords + Id 追加: svn:eol-style + native