PukiWiki plugin for J-STAGE API

jstage.inc.php

<?php
/*
 * J-STAGE API plugin
 */

define('PLUGIN_JSTAGE_USAGE', '#jstage(issn=xxxx-xxxx)');
define('JSTAGE_API_BASEURL', 'http://api.jstage.jst.go.jp/searchapi/do');

define('DEFAULT_HEADING_LEVEL', 2);

function plugin_jstage_convert()
{
	static $_xml;
	if (! isset($_xml)) $_xml = extension_loaded('xml');
	if (! $_xml) return '#jstage: xml extension is not found<br />' . "\n";
	$num = func_num_args();
	if ($num == 0) return PLUGIN_JSTAGE_USAGE .'<br />' . "\n";
	$argv = func_get_args();
	$params = trim($argv[0]);
	if (strlen($params) == 0) return ''. PLUGIN_JSTAGE_USAGE . '<br />' . "\n";
	//return encode($params);
	$timestamp = FALSE;
	$cachehour = 24; // cache expires after a day (default)
	$level = DEFAULT_HEADING_LEVEL;
	if ($num > 1 && intval($argv[1]) > 0) {
		$level = intval($argv[1]);
	}

	list($xml, $time) = plugin_jstage_get_xml($params, $cachehour);
	$time = '';
	if ($timestamp > 0) {
		$time = '<p style="font-size:10px; font-weight:bold">Last-Modified:' .
			get_date('Y/m/d H:i:s', $time) .  '</p>';
	}

	$obj = new JSTAGE_TOC_html($xml, $level);
	return $obj->toString($time);
}

class JSTAGE_TOC_html
{
	var $items = array();
	var $class = '';
	var $level;

	function JSTAGE_TOC_html($entry, $level)
	{
		$this->level = $level;
		foreach ($entry as $item) {
		    	$link  = $item['ID'];
			$title = $item['TITLE'];
			$volume = $item['PRISM:VOLUME'];
			$number = $item['PRISM:NUMBER'];
			$spage = $item['PRISM:STARTINGPAGE'];
			$link = '<a href="'. $link .'" title="'. $title .'" rel="nofollow">'. "$title</a>  - $spage<br />\n";
			if (! is_array($this->items[$volume]))
			    $this->items[$volume] = array();
			if (! is_array($this->items[$volume][$number]))
			    $this->items[$volume][$number] = array();
			$this->items[$volume][$number][$spage] = $link;
		}
	}

	function toString($timestamp)
	{
		$retval = '';
		$volumes = array_keys($this->items);
		rsort($volumes);
		foreach ($volumes as $volume) {
		    $retval .= "<h". $this->level .">Vol.$volume</h". $this->level .">\n";
		    $numbers = array_keys($this->items[$volume]);
		    sort($numbers);
		    $next_level = $this->level + 1;
		    foreach ($numbers as $number) {
			$retval .= "<h$next_level>";
			$retval .= "No.$number";
			$retval .= "</h$next_level>";
			$spages = array_keys($this->items[$volume][$number]);
			sort( $spages );
		        foreach ($spages as $spage) {
			    $retval .= $this->items[$volume][$number][$spage];
			}
		    }
		}
		return <<<EOD
<div{$this->class}>
$retval$timestamp
</div>
EOD;
	}
}

// Get and save J-STAGE XML
function plugin_jstage_get_xml($params, $cachehour)
{
	$buf  = '';
	$time = NULL;
	$filename = CACHE_DIR . "jstage" . encode($params) . '.tmp';
	if ($cachehour) {
		// Remove expired cache
		plugin_jstage_cache_expire($cachehour);

		// Get the cache not expired
		if (is_readable($filename)) {
			$buf  = join('', file($filename));
			$time = filemtime($filename) - LOCALZONE;
		}
	}

	if ($time === NULL) {
		// Newly get XML
	    	$url = PLUGIN_JSTAGE_BASEURL .'?service=3&'. $params;
		$url = 'http://api.jstage.jst.go.jp/searchapi/do?service=3&'. $params;
		//return array($cachehour,0);
		$data = http_request($url);
		if ($data['rc'] !== 200)
			return array(FALSE, 0);

		$buf = $data['data'];
		//return array($buf,0);
		$time = UTIME;

		// Save XML into cache
		if ($cachehour) {
			$fp = fopen($filename, 'w');
			fwrite($fp, $buf);
			fclose($fp);
		}
	}

	// Parse
	$obj = new JSTAGE_TOC_XML();
	return array($obj->parse($buf),$time);
}

// Remove cache if expired limit exeed
function plugin_jstage_cache_expire($cachehour)
{
	$expire = $cachehour * 60 * 60; // Hour
	$dh = dir(CACHE_DIR);
	while (($file = $dh->read()) !== FALSE) {
		if (substr($file, -4) != '.tmp') continue;
		$file = CACHE_DIR . $file;
		$last = time() - filemtime($file);
		if ($last > $expire) unlink($file);
	}
	$dh->close();
}

// Get RSS and array() them
class JSTAGE_TOC_XML
{
	var $items;
	var $item;
	var $is_item;
	var $tag;
	var $encoding;

	function parse($buf)
	{
		$this->items   = array();
		$this->item    = array();
		$this->is_item = FALSE;
		$this->tag     = '';

		// Detect encoding
		$matches = array();
		if(preg_match('/<\?xml [^>]*\bencoding="([a-z0-9-_]+)"/i', $buf, $matches)) {
			$this->encoding = $matches[1];
		} else {
			$this->encoding = mb_detect_encoding($buf);
		}

		// Normalize to UTF-8 / ASCII
		if (! in_array(strtolower($this->encoding), array('us-ascii', 'iso-8859-1', 'utf-8'))) {
			$buf = mb_convert_encoding($buf, 'utf-8', $this->encoding);
			$this->encoding = 'utf-8';
		}

		// Parsing
		$xml_parser = xml_parser_create($this->encoding);
		xml_set_element_handler($xml_parser, array(& $this, 'start_element'), array(& $this, 'end_element'));
		xml_set_character_data_handler($xml_parser, array(& $this, 'character_data'));
		if (! xml_parse($xml_parser, $buf, 1)) {
			return(sprintf('XML error: %s at line %d in %s',
				xml_error_string(xml_get_error_code($xml_parser)),
				xml_get_current_line_number($xml_parser), $buf));
		}
		xml_parser_free($xml_parser);

		return $this->items;
	}

	function escape($str)
	{
		// Unescape already-escaped chars (&lt;, &gt;, &amp;, ...) in RSS body before htmlspecialchars()
		$str = strtr($str, array_flip(get_html_translation_table(ENT_COMPAT)));
		// Escape
		$str = htmlspecialchars($str);
		// Encoding conversion
		$str = mb_convert_encoding($str, SOURCE_ENCODING, $this->encoding);
		return trim($str);
	}

	// Tag start
	function start_element($parser, $name, $attrs)
	{
		if ($this->is_item) {
			$this->tag     = $name;
		} else if ($name == 'ENTRY') {
			$this->is_item = TRUE;
		}
	}

	// Tag end
	function end_element($parser, $name)
	{
		if (! $this->is_item || $name != 'ENTRY') return;

		$item = array_map(array(& $this, 'escape'), $this->item);
		$this->item = array();

		if (isset($item['UPDATED'])) {
			$time = plugin_jstage_get_timestamp($item['UPDATED']);
		} else if (isset($item['PUBDATE'])) {
			$time = plugin_jstage_get_timestamp($item['PUBDATE']);
			
		} else if (isset($item['DESCRIPTION']) &&
			($description = trim($item['DESCRIPTION'])) != '' &&
			($time = strtotime($description)) != -1) {
				$time -= LOCALZONE;
		} else {
			$time = time() - LOCALZONE;
		}
		$item['_TIMESTAMP'] = $time;
		$date = get_date('Y-m-d', $item['_TIMESTAMP']);

		$this->items[] = $item;
		$this->is_item        = FALSE;
	}

	function character_data($parser, $data)
	{
		if (! $this->is_item) return;
		if (! isset($this->item[$this->tag])) $this->item[$this->tag] = '';
		$this->item[$this->tag] .= $data;
	}
}

function plugin_jstage_get_timestamp($str)
{
	$str = trim($str);
	if ($str == '') return UTIME;

	$matches = array();
	if (preg_match('/(\d{4}-\d{2}-\d{2})T(\d{2}:\d{2}:\d{2})(([+-])(\d{2}):(\d{2}))?/', $str, $matches)) {
		$time = strtotime($matches[1] . ' ' . $matches[2]);
		if ($time == -1) {
			$time = UTIME;
		} else if ($matches[3]) {
			$diff = ($matches[5] * 60 + $matches[6]) * 60;
			$time += ($matches[4] == '-' ? $diff : -$diff);
		}
		return $time;
	} else {
		$time = strtotime($str);
		return ($time == -1) ? UTIME : $time - LOCALZONE;
	}
}
?>