lib/AtomParser.php

   1 <?php
   2 /*
   3  * Copyright 2010 Sébastien Le Callonnec
   4  *
   5  * This file is part of PhpWiki.
   6  *
   7  * PhpWiki is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * PhpWiki is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
  19  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  20  */
  21 /**
  22  * This class is a poor-man Atom parser, it does no validation of the feed.
  23  * The content of an entry ("payload") is not parsed but rather returned "as-is",
  24  * as its format can be text, html or xhtml.
  25  *
  26  * @author: Sébastien Le Callonnec
  27  */
  28 require_once 'lib/XmlParser.php';
  29
  30 class AtomParser
  31     extends XmlParser
  32 {
  33     // Feed
  34     public $feed = array();
  35     public $feed_title = '';
  36     public $feed_links = array();
  37     public $feed_subtitle = '';
  38     public $feed_id = '';
  39     public $feed_updated = '';
  40     public $feed_authors = array();
  41     public $feed_contributors = array();
  42     public $generator = '';
  43     public $icon = '';
  44     public $rights = '';
  45     public $logo = '';
  46
  47     public $categories = array();
  48
  49     public $authors = array();
  50     public $contributors = array();
  51
  52     // Author, Contributor
  53     public $name = '';
  54     public $email = '';
  55     public $uri = '';
  56
  57     // Entries
  58     public $entries = array();
  59     public $inside_entry = false;
  60     public $title = '';
  61     public $updated = '';
  62     public $published = '';
  63     public $id = '';
  64     public $links = array();
  65     public $summary = '';
  66
  67     public $inside_content = false;
  68     public $content = '';
  69
  70     function tag_open($parser, $name, $attrs = '')
  71     {
  72         global $current_tag, $current_attrs;
  73
  74         $current_tag = $name;
  75         $current_attrs = $attrs;
  76
  77         if ($name == "ENTRY") {
  78             $this->inside_entry = true;
  79         } elseif ($this->inside_content) {
  80             $this->content .= $this->serialize_tag(strtolower($name), $attrs);
  81         } elseif ($name == "CONTENT") {
  82             $this->inside_content = true;
  83         }
  84     }
  85
  86     function tag_close($parser, $name, $attrs = '')
  87     {
  88         if ($name == "AUTHOR") {
  89             $an_author = $this->trim_data(array(
  90                 "name" => $this->name,
  91                 "email" => $this->email,
  92                 "uri" => $this->uri
  93             ));
  94             if ($this->inside_entry) {
  95                 $this->authors[] = $an_author;
  96             } else {
  97                 $this->feed_authors[] = $an_author;
  98             }
  99             $this->name = '';
 100             $this->email = '';
 101             $this->uri = '';
 102         } elseif ($name == "FEED") {
 103             $this->feed[] = $this->trim_data(array(
 104                 "id" => $this->feed_id,
 105                 "title" => $this->feed_title,
 106                 "links" => $this->feed_links,
 107                 "subtitle" => $this->feed_subtitle,
 108                 "updated" => $this->feed_updated,
 109                 "generator" => $this->generator,
 110                 "icon" => $this->icon,
 111                 "rights" => $this->rights,
 112                 "logo" => $this->logo,
 113                 "authors" => $this->feed_authors,
 114                 "contributors" => $this->feed_contributors
 115             ));
 116             $this->feed_title = '';
 117             $this->feed_id = '';
 118             $this->feed_links = array();
 119             $this->feed_subtitle = '';
 120             $this->feed_updated = '';
 121             $this->feed_authors = array();
 122             $this->feed_contributors = array();
 123             $this->generator = '';
 124             $this->icon = '';
 125             $this->rights = '';
 126             $this->logo = '';
 127         } elseif ($name == "ENTRY") {
 128             $this->entries[] = $this->trim_data(array(
 129                 "id" => $this->id,
 130                 "title" => $this->title,
 131                 "updated" => $this->updated,
 132                 "links" => $this->links,
 133                 "published" => $this->published,
 134                 "content" => $this->content,
 135                 "summary" => $this->summary,
 136                 "authors" => $this->authors,
 137                 "contributors" => $this->contributors
 138             ));
 139             $this->id = '';
 140             $this->title = '';
 141             $this->updated = '';
 142             $this->links = '';
 143             $this->published = '';
 144             $this->content = '';
 145             $this->authors = array();
 146             $this->contributors = array();
 147             $this->inside_entry = false;
 148         } elseif ($name == "CONTENT") {
 149             $this->inside_content = false;
 150         } elseif ($name == "CONTRIBUTOR") {
 151             $a_contributor = $this->trim_data(array(
 152                 "name" => $this->name,
 153                 "email" => $this->email
 154             ));
 155             if ($this->inside_entry) {
 156                 $this->contributors[] = $a_contributor;
 157             } else {
 158                 $this->feed_contributors[] = $a_contributor;
 159             }
 160             $this->name = '';
 161             $this->email = '';
 162         } elseif ($this->inside_content) {
 163             $this->content .= "</" . strtolower($name) . ">";
 164         }
 165     }
 166
 167     function cdata($parser, $data)
 168     {
 169         global $current_tag, $current_attrs;
 170
 171         if ($this->inside_content) {
 172             $this->content .= $data;
 173         } else {
 174             switch ($current_tag) {
 175                 case "ID":
 176                     if ($this->inside_entry)
 177                         $this->id .= $data;
 178                     else
 179                         $this->feed_id .= $data;
 180                     break;
 181                 case "LINK":
 182                     $a_link = array();
 183                     foreach ($current_attrs as $k => $v) {
 184                         $a_link[strtolower($k)] = $v;
 185                     }
 186                     if ($this->inside_entry) {
 187                         $this->links[] = $a_link;
 188                     } else {
 189                         $this->feed_links[] = $a_link;
 190                     }
 191                     break;
 192                 case "NAME":
 193                     $this->name .= $data;
 194                     break;
 195                 case "EMAIL":
 196                     $this->email .= $data;
 197                     break;
 198                 case "TITLE" :
 199                     if ($this->inside_entry)
 200                         $this->title .= $data;
 201                     else
 202                         $this->feed_title .= $data;
 203                     break;
 204                 case "UPDATED":
 205                     if ($this->inside_entry)
 206                         $this->updated .= $data;
 207                     else
 208                         $this->feed_updated .= $data;
 209                     break;
 210                 case "SUBTITLE":
 211                     $this->feed_subtitle .= $data;
 212                     break;
 213                 case "PUBLISHED":
 214                     $this->published .= $data;
 215                     break;
 216                 case "SUMMARY":
 217                     $this->summary .= $data;
 218                     break;
 219                 case "URI":
 220                     $this->uri .= $data;
 221                     break;
 222                 case "GENERATOR":
 223                     $this->generator .= $data;
 224                     break;
 225                 case "ICON":
 226                     $this->icon .= $data;
 227                     break;
 228                 case "LOGO":
 229                     $this->logo .= $data;
 230                     break;
 231                 case "RIGHTS":
 232                     $this->rights .= $data;
 233                     break;
 234             }
 235         }
 236     }
 237
 238     function trim_data($array)
 239     {
 240         return array_map(array("self", "trim_element"), $array);
 241     }
 242
 243     function trim_element($element)
 244     {
 245         if (is_array($element)) {
 246             return $this->trim_data($element);
 247         } elseif (is_string($element)) {
 248             return trim($element);
 249         }
 250     }
 251
 252     function serialize_tag($tag_name, $attributes)
 253     {
 254         $tag = "<" . $tag_name;
 255         foreach ($attributes as $k => $v) {
 256             $tag .= " " . strtolower($k) . "=\"$v\"";
 257         }
 258         $tag .= ">";
 259         return $tag;
 260     }
 261 }