5c48e19e150a8d881fb6e91e47baeaedd8160d86
[tweeper.git] / rss_converter_twitter.com.xsl
1 <!--
2   Stylesheet to convert Twitter user timelines to RSS.
3
4   Copyright (C) 2013-2014  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 <xsl:stylesheet version="1.0"
22     xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
23     xmlns:php="http://php.net/xsl"
24     xsl:extension-element-prefixes="php"
25     exclude-result-prefixes="php">
26
27     <xsl:param name="generate-enclosure"/>
28
29     <xsl:output method="xml" indent="yes"/>
30
31     <xsl:variable name="BaseURL">
32         <xsl:text>https://twitter.com</xsl:text>
33     </xsl:variable>
34
35     <!-- Identity transform -->
36     <xsl:template match="@*|node()">
37         <xsl:copy>
38             <xsl:apply-templates select="@*|node()"/>
39         </xsl:copy>
40     </xsl:template>
41
42     <!--
43          Anchors to external links provide the direct URL in the
44          data-expanded-url attribute, so use this in the href attribute too
45          instead of the default short URL which uses the t.co redirection
46          service.
47
48          NOTE: when creating an element, attributes must be processed _before_
49          adding the contents (either children or a value):
50          http://stackoverflow.com/questions/21984867/
51     -->
52     <xsl:template match="a[@data-expanded-url]">
53         <!-- Prepend and append a white space for aestethic reasons -->
54         <xsl:text> </xsl:text>
55         <a>
56             <xsl:attribute name="href">
57                 <xsl:value-of select="@data-expanded-url"/>
58             </xsl:attribute>
59             <!-- Also strip &nbsp; and &hellip; -->
60             <xsl:value-of select="translate(., '&#xA0;&#x2026;', '')"/>
61         </a>
62         <xsl:text> </xsl:text>
63     </xsl:template>
64
65     <!--
66          These are links to pic.twitter.com, use the direct link for those
67          too instead of the t.co redirections.
68     -->
69     <xsl:template match="a[@data-pre-embedded='true']">
70         <!-- Prepend and append a white space for aestethic reasons -->
71         <xsl:text> </xsl:text>
72         <a>
73             <xsl:attribute name="href">
74                 <xsl:value-of select="concat('https://', .)"/>
75             </xsl:attribute>
76             <xsl:value-of select="concat('https://', .)"/>
77         </a>
78         <xsl:text> </xsl:text>
79     </xsl:template>
80
81     <!-- Present images in a more convenient way -->
82     <xsl:template match="div[@data-image-url]">
83         <a>
84             <xsl:attribute name="href">
85                 <xsl:value-of select="concat(@data-image-url, ':orig')"/>
86             </xsl:attribute>
87             <img>
88                 <xsl:attribute name="src">
89                     <xsl:value-of select="@data-image-url"/>
90                 </xsl:attribute>
91             </img>
92         </a>
93     </xsl:template>
94
95     <xsl:template match="a[@data-expanded-url]" mode="enclosure">
96         <xsl:copy-of select="php:function('Tweeper::generateEnclosure', string(./@data-expanded-url))"/>
97     </xsl:template>
98
99     <xsl:variable name="screen-name" select="//div[@class='user-actions btn-group not-following ']/@data-screen-name"/>
100
101     <xsl:template match="//li[@data-item-id and @data-item-type='tweet']">
102         <xsl:variable name="user-name" select=".//div[contains(@class, 'js-stream-tweet')]/@data-screen-name"/>
103         <xsl:variable name="item-content" select=".//p[contains(@class, 'js-tweet-text')]"/>
104         <xsl:variable name="item-media" select=".//div[contains(@class, 'AdaptiveMedia-container')]"/>
105         <xsl:variable name="item-permalink" select="concat($BaseURL, .//div[@data-permalink-path]/@data-permalink-path)"/>
106         <item>
107             <title>
108                 <xsl:value-of select="concat($user-name, ': ')"/>
109                 <!--
110                      Prepend a space in front of the URLs which are not
111                      preceded by an open parenthesis, for aestethic reasons.
112                      Also, regex, I know: http://xkcd.com/1171/
113                 -->
114                 <xsl:variable
115                     name="processed-title"
116                     select="php:function('preg_replace', '@((?&lt;!\()(?:http[s]?://|pic.twitter.com))@', ' \1', string($item-content))"/>
117                 <!-- Also strip &nbsp; and &hellip; -->
118                 <xsl:value-of select="normalize-space(translate($processed-title, '&#xA0;&#x2026;', ''))"/>
119             </title>
120             <link>
121                 <xsl:value-of select="$item-permalink"/>
122             </link>
123             <guid>
124                 <xsl:value-of select="$item-permalink"/>
125             </guid>
126             <pubDate>
127                 <xsl:variable name="timestamp" select=".//span[contains(@class, 'js-short-timestamp')]/@data-time"/>
128                 <xsl:value-of select="php:functionString('Tweeper::epochToRssDate', number($timestamp))"/>
129             </pubDate>
130             <description>
131                 <xsl:value-of select="concat($user-name, ': ')"/>
132                 <xsl:text disable-output-escaping="yes">&lt;![CDATA[</xsl:text>
133                 <xsl:apply-templates select="$item-content/node()"/>
134                 <xsl:apply-templates select="$item-media/node()"/>
135                 <xsl:text disable-output-escaping="yes">]]&gt;</xsl:text>
136             </description>
137             <xsl:if test="$generate-enclosure = 1">
138                 <xsl:apply-templates select="$item-content//a[@data-expanded-url]" mode="enclosure"/>
139             </xsl:if>
140         </item>
141     </xsl:template>
142
143     <xsl:template match="/">
144         <xsl:variable name="channel-title">
145             <xsl:choose>
146                 <xsl:when test="$screen-name != ''">
147                     <xsl:value-of select="concat('Twitter / ', $screen-name)"/>
148                 </xsl:when>
149                 <xsl:otherwise>
150                     <xsl:value-of select="concat('Twitter / ', normalize-space(//h1[1]))"/>
151                 </xsl:otherwise>
152             </xsl:choose>
153         </xsl:variable>
154         <xsl:variable name="channel-link" select="//link[@rel='canonical']/@href"/>
155
156         <rss version="2.0">
157             <xsl:attribute name="xml:base"><xsl:value-of select="$BaseURL" /></xsl:attribute>
158             <channel>
159                 <generator>Tweeper</generator>
160                 <title>
161                     <xsl:value-of select="$channel-title"/>
162                 </title>
163                 <link>
164                     <xsl:value-of select="$channel-link"/>
165                 </link>
166                 <description>
167                     <xsl:value-of select="normalize-space(//div[@class='ProfileHeaderCard'])"/>
168                 </description>
169                 <image>
170                     <title>
171                         <xsl:value-of select="$channel-title"/>
172                     </title>
173                     <link>
174                         <xsl:value-of select="$channel-link"/>
175                     </link>
176                     <url>
177                         <xsl:value-of select="//a[contains(@class, 'profile-picture')]/@href"/>
178                     </url>
179                 </image>
180                 <xsl:apply-templates select="//ol[@id='stream-items-id']/li[@data-item-id and @data-item-type='tweet']"/>
181             </channel>
182         </rss>
183     </xsl:template>
184 </xsl:stylesheet>