tweeper.php: add infrastructure for sites using json data
[tweeper.git] / 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     exclude-result-prefixes="php">
34
35     <xsl:output method="xml" indent="yes"/>
36
37     <xsl:variable name="BaseURL">
38         <xsl:text>http://howtoons.com</xsl:text>
39     </xsl:variable>
40
41     <xsl:template match="//div[contains(@id, 'post-')]">
42         <xsl:variable name="item-permalink" select=".//div[@class='post-headline']//a/@href"/>
43         <item>
44             <title>
45                 <xsl:value-of select="normalize-space(.//div[@class='post-headline']//a)"/>
46             </title>
47             <link>
48                 <xsl:value-of select="$item-permalink"/>
49             </link>
50             <guid>
51                 <xsl:value-of select="$item-permalink"/>
52             </guid>
53             <pubDate>
54                 <xsl:variable name="date" select="substring-before(.//div[@class='post-byline'], ',')"/>
55                 <!-- date format is MM.DD.YY -->
56                 <xsl:variable name="month" select="substring($date, 1, 2)"/>
57                 <xsl:variable name="day" select="substring($date, 4, 2)"/>
58                 <xsl:variable name="year" select="substring($date, 7, 2)"/>
59                 <xsl:variable name="iso-date" select="concat('20', $year, '-', $month, '-', $day)"/>
60                 <xsl:value-of select="php:functionString('Tweeper::str_to_gmdate', $iso-date)"/>
61             </pubDate>
62             <description>
63                 <xsl:text disable-output-escaping="yes">&lt;![CDATA[</xsl:text>
64                 <xsl:copy-of select=".//div[contains(@class, 'post-bodycopy')]/p"/>
65                 <xsl:text disable-output-escaping="yes">]]&gt;</xsl:text>
66             </description>
67         </item>
68     </xsl:template>
69
70     <xsl:template match="/">
71         <xsl:variable name="channel-title" select="//title"/>
72         <xsl:variable name="channel-link" select="$BaseURL"/>
73
74         <rss version="2.0">
75             <xsl:attribute name="xml:base"><xsl:value-of select="$BaseURL" /></xsl:attribute>
76             <channel>
77                 <generator>Tweeper</generator>
78                 <title>
79                     <xsl:value-of select="$channel-title"/>
80                 </title>
81                 <link>
82                     <xsl:value-of select="$channel-link"/>
83                 </link>
84                 <description>
85                     <xsl:text>The world's greatest D.I.Y. comic website! Tools of mass construction!</xsl:text>
86                 </description>
87                 <image>
88                     <title>
89                         <xsl:value-of select="$channel-title"/>
90                     </title>
91                     <link>
92                         <xsl:value-of select="$channel-link"/>
93                     </link>
94                     <url>
95                         <xsl:text>http://www.howtoons.com/wp-content/themes/atahualpa/images/header/tuck1000.png</xsl:text>
96                     </url>
97                 </image>
98                 <xsl:apply-templates select="//div[contains(@id, 'post-')]"/>
99             </channel>
100         </rss>
101     </xsl:template>
102 </xsl:stylesheet>