rss_converter_howtoons.com.xsl

   1 <!--
   2   Stylesheet to convert Howtoons.com to RSS.
   3
   4   Copyright (C) 2014  Antonio Ospite <ao2@ao2.it>
   5
   6   This file is part of tweeper.
   7
   8   This program is free software: you can redistribute it and/or modify
   9   it under the terms of the GNU General Public License as published by
  10   the Free Software Foundation, either version 3 of the License, or
  11   (at your option) any later version.
  12
  13   This program is distributed in the hope that it will be useful,
  14   but WITHOUT ANY WARRANTY; without even the implied warranty of
  15   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16   GNU General Public License for more details.
  17
  18   You should have received a copy of the GNU General Public License
  19   along with this program.  If not, see <http://www.gnu.org/licenses/>.
  20 -->
  21
  22 <!--
  23   The RSS feed link is broken on http://howtoons.com so just work around it.
  24
  25   Howtoons uses Wordpress, so maybe this style sheet can be used as a base for
  26   scraping other Wordpress sites.
  27 -->
  28
  29 <xsl:stylesheet version="1.0"
  30     xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
  31     xmlns:php="http://php.net/xsl"
  32     xsl:extension-element-prefixes="php">
  33
  34     <xsl:output method="xml" indent="yes"/>
  35
  36     <xsl:variable name="BaseURL">
  37         <xsl:text>http://howtoons.com</xsl:text>
  38     </xsl:variable>
  39
  40     <xsl:template match="//div[contains(@id, 'post-')]">
  41         <item>
  42             <title>
  43                 <xsl:value-of select="normalize-space(.//div[@class='post-headline']//a)"/>
  44             </title>
  45             <link>
  46                 <xsl:value-of select=".//div[@class='post-headline']//a/@href"/>
  47             </link>
  48             <pubDate>
  49                 <xsl:variable name="date" select="substring-before(.//div[@class='post-byline'], ',')"/>
  50                 <!-- date format is MM.DD.YY -->
  51                 <xsl:variable name="month" select="substring($date, 1, 2)"/>
  52                 <xsl:variable name="day" select="substring($date, 4, 2)"/>
  53                 <xsl:variable name="year" select="substring($date, 7, 2)"/>
  54                 <xsl:variable name="iso-date" select="concat('20', $year, '-', $month, '-', $day)"/>
  55                 <xsl:value-of select="php:functionString('Tweeper::str_to_gmdate', $iso-date)"/>
  56             </pubDate>
  57             <description>
  58                 <xsl:text disable-output-escaping="yes">&lt;![CDATA[</xsl:text>
  59                 <xsl:copy-of select=".//div[contains(@class, 'post-bodycopy')]/p"/>
  60                 <xsl:text disable-output-escaping="yes">]]&gt;</xsl:text>
  61             </description>
  62         </item>
  63     </xsl:template>
  64
  65     <xsl:template match="/">
  66
  67         <rss version="2.0">
  68             <xsl:attribute name="xml:base"><xsl:value-of select="$BaseURL" /></xsl:attribute>
  69             <channel>
  70                 <generator>Tweeper</generator>
  71                 <title>
  72                     <xsl:value-of select="//title"/>
  73                 </title>
  74                 <link>
  75                     <xsl:value-of select="$BaseURL"/>
  76                 </link>
  77                 <description>
  78                     <xsl:text>The world's greatest D.I.Y. comic website! Tools of mass construction!</xsl:text>
  79                 </description>
  80                 <image>
  81                     <url>
  82                         <xsl:text>http://www.howtoons.com/wp-content/themes/atahualpa/images/header/tuck1000.png</xsl:text>
  83                     </url>
  84                 </image>
  85                 <xsl:apply-templates select="//div[contains(@id, 'post-')]"/>
  86             </channel>
  87         </rss>
  88     </xsl:template>
  89 </xsl:stylesheet>