[PATCH] News cleanup & hAtom support

Index: master/scripts/pregen_news
===================================================================
RCS file: /repository/php-master-web/scripts/pregen_news,v
retrieving revision 1.2
diff -u -p -r1.2 pregen_news
--- master/scripts/pregen_news	29 Dec 2007 14:44:08 -0000	1.2
+++ master/scripts/pregen_news	29 Feb 2008 17:17:08 -0000
@@ -1,102 +1,151 @@
 <?php /* vim: set noet ts=4 sw=4 ft=php: : */
 
+define("XMLNS_ATOM",  "http://www.w3.org/2005/Atom");
+define("XMLNS_XHTML", "http://www.w3.org/1999/xhtml");
+define("XMLNS_PHP",   "http://php.net/ns/news");
+
+function date_sort($a, $b) {
+	return $a["updated"] == $b["updated"] ? 0 : (strtotime($a["updated"]) > strtotime($b["updated"]) ? -1 : 1);
+}
+
 $XML_OPTIONS =  LIBXML_COMPACT | LIBXML_NOBLANKS | LIBXML_NOCDATA | LIBXML_NSCLEAN | LIBXML_XINCLUDE;
+
 function pregen_atom($feed, $feedDest, $newsDest) {
-	$dom = new DOMDocument("1.0", "UTF-8");
-	$dom->formatOutput = true;
-	$dom->preserveWhiteSpace = false;
-	$options = $GLOBALS["XML_OPTIONS"];
-
-	$y = date("Y");
-	if (file_exists($feed .$y. ".xml")) {
-		$dom->load($feed .$y. ".xml", $options);
-	} elseif (file_exists($feed .(--$y). ".xml")) {
-		$dom->load($feed .$y. ".xml");
-	} else {
-		trigger_error("Can't find news feed", E_USER_WARNING);
-		return;
+	$entries = format_atom_feed($feed, $y = date("Y"));
+	uasort($entries, "date_sort");
+
+	$write = '<?php $NEWS_ENTRIES = ' . var_export($entries, 1) . ';';
+	file_put_contents($newsDest, $write);
+
+	do {
+		$filename = $feed . $y .".xml";
+	} while (!file_exists($filename));
+	copy($filename, $feedDest);
+}
+
+function format_atom_feed($feed, $year) {
+	static $frontpage = 0;
+	static $conf = 0;
+
+	// Lets only go back one year
+	if ($year < (date("Y")-1)) {
+		$frontpage = $conf = 100;
+		return array();
 	}
 
+	$filename = $feed . $year .".xml";
 
-	$entrys = $dom->getElementsByTagNameNS("http://www.w3.org/2005/Atom", "entry");
-	if(!$entrys || $entrys->length < 10) {
-		$fragment = $dom->createDocumentFragment();
-		$tmp = new DOMDocument("1.0", "UTF-8");
-		$tmp->formatOutput = true;
-		$tmp->preserveWhiteSpace = false;
-
-		$tmp->load($feed .($y-1). ".xml", $options);
-		$e = $tmp->getElementsByTagNameNS("http://www.w3.org/2005/Atom", "entry");
-		foreach($e as $n) {
-			$n->setAttribute("xmlns", "http://www.w3.org/2005/Atom");
-			$n->setAttribute("xmlns:php", "http://php.net/ns/news");
-			$fragment->appendXML($tmp->saveXML($n));
+	// January 1st...
+	if (!file_exists($filename)) {
+		return format_atom_feed($feed, --$year);
+	}
+
+	$r = new XMLReader;
+	$r->open($filename, "UTF-8");
+
+	$entries = array();
+	while($r->read()) {
+		if ($r->nodeType === XMLReader::ELEMENT && $r->name === "entry") {
+			$entries[] = $current = format_atom_entry($r);
+
+			// Count the frontpage & conference entries
+			foreach($current["category"] as $dog) {
+				if ($dog["term"] == "frontpage") {
+					++$frontpage;
+					break;
+				} else {
+					++$conf;
+				}
+			}
 		}
-		$child = $dom->documentElement->appendChild($fragment);
-		$child->removeAttributeNS("http://www.w3.org/1999/xhtml", "default");
 	}
 
-	$dom->save($feedDest, $options);
-	//$dom->normalizeDocument(); # For whatever reason this doesn't seem to work :(
-	$dom->load($feedDest);
-	$NEWS = array();
-
-	foreach($entrys as $entry) {
-		$fragment = "";
-		$c = array();
-		$contentNode 	  = $entry->getElementsByTagName("content")->item(0)->firstChild;
-
-		foreach($entry->getElementsByTagName("link") as $link) {
-			if($link->getAttribute("rel") == "alternate") {
-				$self     = $link->getAttribute("href");
-				$fragment = parse_url("http://php.net" . $self, PHP_URL_FRAGMENT);
+	// Make sure we have enough entries to display (shortly after new-year..)
+	if (min($frontpage, $conf) < 5) {
+		$entries = array_merge($entries, format_atom_feed($feed, --$year));
+	}
+
+	return $entries;
+}
+
+// {{{ Parse the entry into array(element => value)
+function format_atom_entry($r) {
+	$retval = array();
+
+	while($r->read()) {
+		if ($r->nodeType !== XMLReader::ELEMENT) {
+			if ($r->nodeType === XMLReader::END_ELEMENT && $r->name === "entry") {
+				return $retval;
+			}
+			continue;
+		}
+
+		$name = $r->localName;
+		if ($r->namespaceURI === XMLNS_ATOM) {
+			switch($name) {
+			case "title":
+			case "id":
+			case "published":
+			case "updated":
+				$retval[$name] = $r->readString();
+				$retval[$name] = $r->readString();
+				break;
+
+			case "link":
+			case "category":
+				$retval[$name][] = format_attributes($r);
+				break;
+
+			case "content":
+				if ($r->hasAttributes) {
+					switch($r->getAttribute("type")) {
+					case "html":
+					case "xhtml":
+						$retval[$name] = $r->readInnerXML();
+						break;
+
+					case "text":
+						$retval[$name] = $r->readString();
+						break;
+					}
+				} else {
+					$retval[$name] = $r->readString();
+				}
 				break;
 			}
-	 	}
+		} elseif ($r->namespaceURI === XMLNS_PHP) {
+			switch($name) {
+			case "newsImage":
+				$retval[$name] = format_attributes($r);
+				$retval[$name]["content"] = $r->readString();
+				break;
 
-		$dateNode         = $entry->getElementsByTagName("published")->item(0)->firstChild;
-		$date             = date("d-M-Y", strtotime($dom->saveXML($dateNode, $options)));
-		$dateSpanNode     = $dom->createElement("span", "[$date]");
-		$dateSpanNode->setAttribute("class", "newsdate");
-
-		$contentNode->insertBefore($dateSpanNode, $contentNode->firstChild);
-		$contentNode->removeAttributeNS("http://www.w3.org/1999/xhtml", "");
-
-		$content          = $dom->saveXML($contentNode, $options);
-		$title            = $dom->saveXML($entry->getElementsByTagName("title")->item(0)->firstChild, $options);
-		$image            = $entry->getElementsByTagNameNS("http://php.net/ns/news", "newsImage");
-		if($image->length > 0) {
-			$image = array(
-				"link"    => $image->item(0)->getAttribute("link"),
-				"title"   => $image->item(0)->getAttribute("title"),
-				"image"   => $image->item(0)->nodeValue,
-			);
-		} else {
-			$image = false;
+			case "finalTeaserDate":
+				$retval[$name] = $r->readString();
+				break;
+			}
 		}
+	}
 
-		$newsEntry        = array(
-			"fragment"    => $fragment,
-			"title"       => $title,
-			"content"     => $content,
-			"image"       => $image,
-			"date"        => $date,
-		);
-		$categories       = $entry->getElementsByTagName("category");
-		for ($i=0; $i<$categories->length; $i++) {
-			$c[] = $categories->item($i)->getAttribute("term");
-		}
-		if (in_array("cfp", $c) || in_array("conferences", $c)) {
-			$NEWS["conferences"][] = $newsEntry;
-		}
-		if (in_array("frontpage", $c)) {
-			$NEWS["frontpage"][] = $newsEntry;
-		}
+	return $retval;
+} // }}}
+
+// {{{ Return all attrs for current element as an array(attr-name => attr-value)
+function format_attributes($r) {
+	$retval = array();
+
+	if (!$r->hasAttributes) {
+		return $retval;
 	}
 
-	$write = '<?php $NEWS_ENTRIES = ' . var_export($NEWS, 1) . ';';
-	file_put_contents($newsDest, $write);
-}
+	$r->moveToFirstAttribute();
+	do {
+		$retval[$r->localName] = $r->value;
+	} while($r->moveToNextAttribute());
+	$r->moveToElement();
+
+	return $retval;
+} // }}}
 
 function legacy_rss($atom, $newsDest, $confDest) {
 	$sxe = new SimpleXMLElement($atom, $GLOBALS["XML_OPTIONS"], true);
Index: web/index.php
===================================================================
RCS file: /repository/phpweb/index.php,v
retrieving revision 1.905
diff -u -p -r1.905 index.php
--- web/index.php	25 Feb 2008 13:05:43 -0000	1.905
+++ web/index.php	29 Feb 2008 17:17:09 -0000
@@ -245,7 +245,7 @@ if (is_array($CONF_TEASER) && count($CON
 /* Where the h*ll did all the news go?
  * See archives/2007.xml
  */
-print_news($NEWS_ENTRIES["frontpage"]);
+print_news($NEWS_ENTRIES, "frontpage");
 ?>
 
 <p class="center"><a href="/archive/index.php">News Archive</a></p>
Index: web/archive/2008.xml
===================================================================
RCS file: /repository/phpweb/archive/2008.xml,v
retrieving revision 1.11
diff -u -p -r1.11 2008.xml
--- web/archive/2008.xml	28 Feb 2008 23:02:42 -0000	1.11
+++ web/archive/2008.xml	29 Feb 2008 17:17:09 -0000
@@ -21,7 +21,7 @@
     <php:finalTeaserDate>2008-05-24</php:finalTeaserDate>
     <php:newsImage link="http://tek.phparch.com/" title="php|tek 2008: Chicago">phptek_2008.png</php:newsImage>
     <content type="xhtml">
-      <div xmlns="http://www.w3.org/1999/xhtml" class="description">
+      <div xmlns="http://www.w3.org/1999/xhtml">
 <a href="http://tek.phparch.com/" class="url">php|tek 2008: Chicago</a>
 <p>
 The publishers of <a href="http://www.phparch.com/">php|architect Magazine</a> are proud to announce the php|tek 2008 conference in Chicago, Illinois, USA.
@@ -46,7 +46,7 @@ For the past two years, php|architect's 
     <php:finalTeaserDate>2008-03-16</php:finalTeaserDate>
     <php:newsImage link="http://conf.phpquebec.com/en/conf2008/" title="PHP Québec conference 2008">conference_php_quebec.gif</php:newsImage>
     <content type="xhtml">
-      <div xmlns="http://www.w3.org/1999/xhtml" class="description">
+      <div xmlns="http://www.w3.org/1999/xhtml">
 <a href="http://conf.phpquebec.com/">2008 PHP Quebec Conference &amp; Job Fair</a>
 <p>
 The PHP Quebec team is pleased to present the sixth edition of the <a 
@@ -78,7 +78,7 @@ visit the website: <a href="http://conf.
     <php:newsImage xmlns="http://php.net/ns/news" link="http://www.phpconference.co.uk/" title="PHP London Conference 08">phplondon2008.png</php:newsImage>
     <link href="http://www.phpconference.co.uk/" rel="via" type="text/html"/>
     <content type="xhtml">
-      <div xmlns="http://www.w3.org/1999/xhtml" class="description">
+      <div xmlns="http://www.w3.org/1999/xhtml">
 		  <p><abbr class="dtstart" title="2008-02-29">February 29th</abbr> (Leap Year Day). phplondon.org  announce their third
             annual <a href="http://www.phpconference.co.uk" alt="phplondon.org community conference" class="url">community conference</a>
             to be held at Inmarsat, Old Street, London.</p>
Index: web/archive/2008.php
===================================================================
RCS file: /repository/phpweb/archive/2008.php,v
retrieving revision 1.1
diff -u -p -r1.1 2008.php
--- web/archive/2008.php	29 Dec 2007 15:12:04 -0000	1.1
+++ web/archive/2008.php	29 Feb 2008 17:17:09 -0000
@@ -16,13 +16,8 @@ site_header("News Archive - 2008");
 <hr />
 
 <?php
-function date_sort($a, $b) {
-	return $a["date"] == $b["date"] ? 0 : (strtotime($a["date"]) > strtotime($b["date"]) ? -1 : 1);
-}
 
-$news = array_merge($NEWS_ENTRIES["frontpage"], $NEWS_ENTRIES["conferences"]);
-uasort($news, "date_sort");
-print_news($news, 2008);
+print_news($NEWS_ENTRIES, array("conferences", "cfp", "frontpage"), 50);
 /* %s/<a href="\(.*\)"><img src="\/images\/news\/\(.*\)" alt="\(.*\)" width.*><\/a>/<?php news_image("\1", "\2", "\3"); ?>/g */
 site_footer();
 
Index: web/conferences/index.php
===================================================================
RCS file: /repository/phpweb/conferences/index.php,v
retrieving revision 1.23
diff -u -p -r1.23 index.php
--- web/conferences/index.php	25 Feb 2008 15:49:21 -0000	1.23
+++ web/conferences/index.php	29 Feb 2008 17:17:09 -0000
@@ -22,7 +22,7 @@ unset($RSIDEBAR_DATA);
 
 site_header("PHP Conferences around the world", array("layout_workaround" => $layout_workaround, 'headtags' => '<link rel="alternate" type="application/atom+xml" title="PHP: Conference announcements" href="' . $MYSITE . 'feed.atom" />'));
 
-print_news($NEWS_ENTRIES["conferences"], false, true);
+print_news($NEWS_ENTRIES, array("conferences", "cfp"), 15);
 
 site_footer(
     array("atom" => "/feed.atom") // Add a link to the feed
Index: web/include/layout.inc
===================================================================
RCS file: /repository/phpweb/include/layout.inc,v
retrieving revision 1.277
diff -u -p -r1.277 layout.inc
--- web/include/layout.inc	25 Feb 2008 15:49:21 -0000	1.277
+++ web/include/layout.inc	29 Feb 2008 17:17:09 -0000
@@ -623,24 +623,60 @@ function news_archive_sidebar()
 }
 
 // Print news
-function print_news($news, $year = false, $event = false) {
+function print_news($news, $dog, $max = 5) {
+    $count = 0;
     foreach($news as $item) {
-        if ($year && $year != date("Y", strtotime($item["date"]))) {
+        $vevent = "";
+        $ok = false;
+
+        // Only print entries in the provided s/dog/cat/ egory
+        // If its a conference, use the hCalendar container
+        foreach($item["category"] as $category) {
+            if (in_array($category["term"], (array)$dog)) {
+                $ok = true;
+                ++$count;
+            }
+            if ($category["term"] === "conferences") {
+                $vevent = " vevent";
+            }
+        }
+        if ($count > $max) {
+            break;
+        }
+        if ($ok === false) {
             continue;
         }
+
         $image = "";
-        if($item["image"]) {
-            $image = news_image($item["image"]["link"], $item["image"]["image"], $item["image"]["title"]);
+        if(isset($item["newsImage"])) {
+            $image = news_image($item["newsImage"]["link"], $item["newsImage"]["content"], $item["newsImage"]["title"]);
         }
-        $vevent = "";
-        if ($event) {
-            $vevent = " vevent";
+
+        //$id = parse_url($item["id"], PHP_URL_FRAGMENT); 5.1.2
+        $id = parse_url($item["id"]);
+        $id = $id["fragment"];
+
+        // Find the permlink
+        foreach($item["link"] as $link) {
+            if ($link["rel"] === "via") {
+                $permlink = $link["href"];
+                break;
+            }
+        }
+        if (!isset($permlink)) {
+            $permlink = "#" .$id;
         }
+
+        $newsdate = date("d-M-Y", strtotime($item["updated"]));
+
         echo <<< EOT
-<div class="newsItem$vevent">
-    $image
-    <a name="{$item["fragment"]}" id="{$item["fragment"]}"><h1 class="summary">{$item["title"]}</h1></a>
-    {$item["content"]}
+<div class="newsItem hentry$vevent">
+    <div class="newsImage">$image</div>
+    <h1 class="summary entry-title"><a name="{$id}" id="{$id}" href="{$permlink}" rel="bookmark" class="bookmark">{$item["title"]}</a></h1>
+    [<abbr class="updated newsdate" title="{$item["updated"]}">{$newsdate}</abbr>]
+    <div class="entry-content description">
+        {$item["content"]}
+    </div>
 </div>
 
 <hr />
@@ -649,3 +685,4 @@ EOT;
     }
 }
 /* vim: set et ts=4 sw=4 ft=php: : */
+
Index: web/styles/site.css
===================================================================
RCS file: /repository/phpweb/styles/site.css,v
retrieving revision 1.63
diff -u -p -r1.63 site.css
--- web/styles/site.css	5 Feb 2008 08:12:26 -0000	1.63
+++ web/styles/site.css	29 Feb 2008 17:17:09 -0000
@@ -160,10 +160,6 @@ hr {
 	color: #6666cc;
 	background-color: transparent;
 }
-div.newsItem .newsdate {
-	float:left;
-	padding-right: 10px;
-}
 
 div.indent {
 	margin-left: 50px;