diff options
author | Andrew Dolgov <[email protected]> | 2013-05-01 17:04:57 +0400 |
---|---|---|
committer | Andrew Dolgov <[email protected]> | 2013-05-01 17:04:57 +0400 |
commit | cd07592c29391ca374f78a75a10465f894d50042 (patch) | |
tree | c8bd2d30267ff5b1f6334417a304dddb5f0125ae /classes | |
parent | 7fc041c6ef2241d0611afb60fb721568376cd723 (diff) |
add basic tinyparser/atom
Diffstat (limited to 'classes')
-rw-r--r-- | classes/feedenclosure.php | 7 | ||||
-rw-r--r-- | classes/feeditem.php | 15 | ||||
-rw-r--r-- | classes/feeditem/atom.php | 118 | ||||
-rw-r--r-- | classes/feedparser.php | 101 |
4 files changed, 241 insertions, 0 deletions
diff --git a/classes/feedenclosure.php b/classes/feedenclosure.php new file mode 100644 index 000000000..d610dd7c8 --- /dev/null +++ b/classes/feedenclosure.php @@ -0,0 +1,7 @@ +<?php +class FeedEnclosure { + public $link; + public $type; + public $length; +} +?> diff --git a/classes/feeditem.php b/classes/feeditem.php new file mode 100644 index 000000000..cc39fcee7 --- /dev/null +++ b/classes/feeditem.php @@ -0,0 +1,15 @@ +<?php +class FeedItem { + abstract function get_id(); + abstract function get_date(); + abstract function get_link(); + abstract function get_title(); + abstract function get_description(); + abstract function get_content(); + abstract function get_comments_url(); + abstract function get_comments_count(); + abstract function get_categories(); + abstract function get_enclosures(); + abstract function get_author(); +} +?> diff --git a/classes/feeditem/atom.php b/classes/feeditem/atom.php new file mode 100644 index 000000000..560484eda --- /dev/null +++ b/classes/feeditem/atom.php @@ -0,0 +1,118 @@ +<?php +class FeedItem_Atom { + private $elem; + + function __construct($elem) { + $this->elem = $elem; + } + + function get_id() { + $id = $this->elem->getElementsByTagName("id")->item(0); + + if ($id) { + return $id->nodeValue; + } else { + return $this->get_link(); + } + } + + function get_date() { + + + } + + function get_link() { + $links = $this->elem->getElementsByTagName("link"); + + foreach ($links as $link) { + if ($link && $link->hasAttribute("href") && !$link->hasAttribute("rel")) { + return $link->getAttribute("href"); + } + } + } + + function get_title() { + $title = $this->elem->getElementsByTagName("title")->item(0); + + if ($title) { + return $title->nodeValue; + } + } + + function get_content() { + $content = $this->elem->getElementsByTagName("content")->item(0); + + if ($content) { + return $content->nodeValue; + } + } + + function get_description() { + $summary = $this->elem->getElementsByTagName("summary")->item(0); + + if ($summary) { + return $summary->nodeValue; + } + } + + // todo + function get_comments_url() { + + } + + // todo + function get_comments_count() { + + } + + function get_categories() { + $categories = $this->elem->getElementsByTagName("category"); + $cats = array(); + + foreach ($categories as $cat) { + if ($cat->hasAttribute("term")) + array_push($cats, $cat->getAttribute("term")); + } + + + return $cats; + } + + function get_enclosures() { + $links = $this->elem->getElementsByTagName("link"); + + $encs = array(); + + foreach ($links as $link) { + if ($link && $link->hasAttribute("href") && $link->hasAttribute("rel")) { + if ($link->getAttribute("rel") == "enclosure") { + $enc = new FeedEnclosure(); + + $enc->type = $link->getAttribute("type"); + $enc->link = $link->getAttribute("href"); + $enc->length = $link->getAttribute("length"); + + array_push($encs, $enc); + } + } + } + + return $encs; + } + + function get_author() { + $author = $this->elem->getElementsByTagName("author")->item(0); + + if ($author) { + $name = $author->getElementsByTagName("name")->item(0); + + if ($name) return $name->nodeValue; + + $email = $author->getElementsByTagName("email")->item(0); + + if ($email) return $email->nodeValue; + + } + } +} +?> diff --git a/classes/feedparser.php b/classes/feedparser.php new file mode 100644 index 000000000..274e871b9 --- /dev/null +++ b/classes/feedparser.php @@ -0,0 +1,101 @@ +<?php +class FeedParser { + private $doc; + private $error; + private $items; + private $link; + private $title; + private $type; + + const FEED_RDF = 0; + const FEED_RSS = 1; + const FEED_ATOM = 2; + + function __construct($data) { + libxml_use_internal_errors(true); + libxml_clear_errors(); + $this->doc = new DOMDocument(); + $this->doc->loadXML($data); + $this->error = $this->format_error(libxml_get_last_error()); + libxml_clear_errors(); + + $this->items = array(); + } + + function init() { + $root = $this->doc->firstChild; + + if ($root) { + switch ($root->tagName) { + case "rss": + $this->type = $this::FEED_RSS; + break; + case "feed": + $this->type = $this::FEED_ATOM; + break; + default: + $this->error = "Unknown/unsupported feed type"; + return; + } + + $xpath = new DOMXPath($this->doc); + + switch ($this->type) { + case $this::FEED_ATOM: + $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom'); + + $title = $xpath->query("//atom:feed/atom:title")->item(0); + + if ($title) { + $this->title = $title->nodeValue; + } + + $link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0); + + if ($link && $link->hasAttributes()) { + $this->link = $link->getAttribute("href"); + } + + $articles = $xpath->query("//atom:entry"); + + foreach ($articles as $article) { + array_push($this->items, new FeedItem_Atom($article)); + } + + break; + case $this::FEED_RDF: + + break; + case $this::FEED_RSS: + break; + } + } + } + + function format_error($error) { + if ($error) { + return sprintf("LibXML error %s at line %d (column %d): %s", + $error->code, $error->line, $error->column, + $error->message); + } else { + return ""; + } + } + + function error() { + return $this->error; + } + + function get_link() { + return $this->link; + } + + function get_title() { + return $this->title; + } + + function get_items() { + return $this->items; + } + +} ?> |