src/rss_converter_twitter.com.xsl

   1 <!--
   2   Stylesheet to convert Twitter user timelines to RSS.
   3
   4   Copyright (C) 2013-2018  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     exclude-result-prefixes="php">
  25
  26     <xsl:param name="generate-enclosure"/>
  27     <xsl:param name="show-usernames"/>
  28     <xsl:param name="show-multimedia"/>
  29
  30     <xsl:output method="xml" indent="yes"/>
  31
  32     <xsl:variable name="BaseURL">
  33         <xsl:text>https://twitter.com</xsl:text>
  34     </xsl:variable>
  35
  36     <!-- Identity transform -->
  37     <xsl:template match="@*|node()">
  38         <xsl:copy>
  39             <!--
  40                 Strip the style attribute while copying elements because it may be
  41                 dangerous, see:
  42                 https://validator.w3.org/feed/docs/warning/DangerousStyleAttr.html
  43             -->
  44             <xsl:apply-templates select="@*[not(name() = 'style')]|node()"/>
  45         </xsl:copy>
  46     </xsl:template>
  47
  48     <!-- Strip leading spaces in first text node of the tweet-text. -->
  49     <xsl:template match="div[@class='tweet-text']/div/text()[1]">
  50         <xsl:value-of select="substring-after(substring-after(., ' '), ' ')"/>
  51     </xsl:template>
  52
  53     <!--
  54          Anchors to external links provide the direct URL in the
  55          data-expanded-url attribute, so use this in the href attribute too
  56          instead of the default short URL which uses the t.co redirection
  57          service.
  58
  59          NOTE: when creating an element, attributes must be processed _before_
  60          adding the contents (either children or a value):
  61          http://stackoverflow.com/questions/21984867/
  62     -->
  63     <xsl:template match="a[@data-expanded-url]">
  64         <a>
  65             <xsl:attribute name="href">
  66                 <xsl:value-of select="@data-expanded-url"/>
  67             </xsl:attribute>
  68             <xsl:value-of select="@data-expanded-url"/>
  69         </a>
  70     </xsl:template>
  71
  72     <!--
  73          These are links to pic.twitter.com, use the direct link for those
  74          too instead of the t.co redirections.
  75     -->
  76     <xsl:template match="a[@data-pre-embedded='true']">
  77         <xsl:if test="$show-multimedia = 1">
  78             <a>
  79                 <xsl:attribute name="href">
  80                     <xsl:value-of select="@data-url"/>
  81                 </xsl:attribute>
  82                 <xsl:value-of select="concat('https://', .)"/>
  83             </a>
  84         </xsl:if>
  85     </xsl:template>
  86
  87     <!-- Present images in a more convenient way -->
  88     <!-- TODO: not supported in mobile UI
  89     <xsl:template match="a[@data-pre-embedded='true' and contains(@data-url, '/photo/')]">
  90         <xsl:variable name="embedded-photo-url" select="concat('https://pbs.twimg.com/media/', @data-tco-id, '?format=jpg')"/>
  91         <a>
  92             <xsl:attribute name="href">
  93                 <xsl:value-of select="$embedded-photo-url"/>
  94             </xsl:attribute>
  95             <img style="max-width: 100%">
  96                 <xsl:attribute name="src">
  97                     <xsl:value-of select="$embedded-photo-url"/>
  98                 </xsl:attribute>
  99             </img>
 100         </a>
 101     </xsl:template>
 102     -->
 103
 104     <!-- Don't repeat background in embedded media content -->
 105     <!-- TODO: not supported in mobile UI
 106     <xsl:template match="div[contains(@class, 'PlayableMedia-player')]">
 107         <xsl:copy>
 108             <xsl:apply-templates select="@*"/>
 109             <xsl:attribute name="style">
 110                 <xsl:value-of select="concat(@style, '; background-repeat: no-repeat; background-size: 100% auto')"/>
 111             </xsl:attribute>
 112             <xsl:apply-templates select="node()"/>
 113         </xsl:copy>
 114     </xsl:template>
 115     -->
 116
 117     <xsl:template match="a[@data-expanded-url]" mode="enclosure">
 118         <xsl:copy-of select="php:functionString('Tweeper\Tweeper::generateEnclosure', ./@data-expanded-url)"/>
 119     </xsl:template>
 120
 121     <xsl:template match="a[@data-pre-embedded='true']" mode="enclosure">
 122         <xsl:copy-of select="php:functionString('Tweeper\Tweeper::generateEnclosure', @data-url)"/>
 123     </xsl:template>
 124
 125     <xsl:variable name="screen-name" select="normalize-space(substring-after(//table[@class='profile-details' or @class='main-tweet']//*[@class='username'], '@'))"/>
 126
 127     <xsl:template match="//div[contains(@class, 'timeline')]/table[@class='tweet  ']|//div[@class='main-tweet-container']/table[@class='main-tweet']">
 128         <xsl:variable name="user-name" select="normalize-space(.//*[@class='username']/text()[2])"/>
 129         <xsl:variable name="item-content" select=".//div[@class='tweet-text']/div"/>
 130         <xsl:variable name="item-media" select=".//a[@data-pre-embedded='true']"/>
 131         <xsl:variable name="item-permalink">
 132             <xsl:choose>
 133                 <xsl:when test="@href">
 134                     <xsl:value-of select="concat($BaseURL, substring-before(@href, '?'))"/>
 135                 </xsl:when>
 136                 <xsl:otherwise>
 137                     <!--
 138                         The main tweet in permalink pages do not have a timestamp tag,
 139                         just use the canonical URL as permalink.
 140                     -->
 141                     <xsl:value-of select="//link[@rel='canonical']/@href"/>
 142                 </xsl:otherwise>
 143             </xsl:choose>
 144         </xsl:variable>
 145
 146         <!-- TODO twitter mobile UI does not have a way to detect this
 147         <xsl:variable name="item-has-video" select="$item-media//*[contains(@class, 'PlayableMedia- -video')]"/>
 148         <xsl:variable name="item-has-gif" select="$item-media//*[contains(@class, 'PlayableMedia- -gif')]"/>
 149         -->
 150
 151         <item>
 152             <title>
 153                 <xsl:if test="($show-usernames = 1) or ($screen-name != $user-name)">
 154                     <xsl:value-of select="concat($user-name, ': ')"/>
 155                 </xsl:if>
 156                 <!-- TODO twitter mobile UI does not have a way to detect this
 157                 <xsl:if test="$item-has-video">
 158                     <xsl:text>(Video) </xsl:text>
 159                 </xsl:if>
 160                 -->
 161                 <!--
 162                      Prepend a space in front of the URLs which are not
 163                      preceded by an open parenthesis, for aestethic reasons.
 164                      Also, regex, I know: http://xkcd.com/1171/
 165                 -->
 166                 <xsl:variable
 167                     name="processed-title"
 168                     select="php:functionString('preg_replace', '@((?&lt;!\()(?:http[s]?://|pic.twitter.com))@', ' \1', $item-content)"/>
 169                 <!-- Also strip &nbsp; and &hellip; -->
 170                 <xsl:value-of select="normalize-space(translate($processed-title, '&#xA0;&#x2026;', ''))"/>
 171             </title>
 172             <link>
 173                 <xsl:value-of select="$item-permalink"/>
 174             </link>
 175             <guid>
 176                 <xsl:value-of select="$item-permalink"/>
 177             </guid>
 178             <pubDate>
 179                 <xsl:variable name="timestamp" select=".//td[@class='timestamp']/a|.//div[@class='metadata']/a"/>
 180                 <xsl:value-of select="php:functionString('Tweeper\Tweeper::twitterToRssDate', $timestamp)"/>
 181             </pubDate>
 182             <description>
 183                 <xsl:text disable-output-escaping="yes">&lt;![CDATA[</xsl:text>
 184                 <xsl:if test="($show-usernames = 1) or ($screen-name != $user-name)">
 185                     <xsl:value-of select="concat($user-name, ':')"/>
 186                     <xsl:element name="br"/>
 187                 </xsl:if>
 188                 <!-- TODO twitter mobile UI does not support embedded media
 189                 <xsl:if test="$item-has-video">
 190                     <xsl:text> (Video)</xsl:text>
 191                     <xsl:element name="br"/>
 192                 </xsl:if>
 193                 <xsl:if test="$item-has-gif">
 194                     <xsl:text> (GIF)</xsl:text>
 195                     <xsl:element name="br"/>
 196                 </xsl:if>
 197                 -->
 198                 <xsl:element name="span">
 199                     <xsl:attribute name="style">white-space: pre-wrap;</xsl:attribute>
 200                     <xsl:apply-templates select="$item-content/node()"/>
 201                 </xsl:element>
 202
 203                 <!-- TODO twitter mobile UI does not support embedded media
 204                 <xsl:if test="$show-multimedia = 1">
 205                     <xsl:apply-templates select="$item-media"/>
 206                 </xsl:if>
 207                 -->
 208                 <xsl:text disable-output-escaping="yes">]]&gt;</xsl:text>
 209             </description>
 210             <xsl:if test="$generate-enclosure = 1">
 211                 <xsl:apply-templates select="$item-content//a[@data-expanded-url]" mode="enclosure"/>
 212                 <xsl:apply-templates select="$item-media" mode="enclosure"/>
 213             </xsl:if>
 214         </item>
 215     </xsl:template>
 216
 217     <xsl:template match="/">
 218         <xsl:variable name="channel-title">
 219             <xsl:choose>
 220                 <xsl:when test="$screen-name != ''">
 221                     <xsl:value-of select="concat('Twitter / ', $screen-name)"/>
 222                 </xsl:when>
 223                 <xsl:otherwise>
 224                     <xsl:value-of select="concat('Twitter / ', normalize-space(//td[@id='search']//input/@value))"/>
 225                 </xsl:otherwise>
 226             </xsl:choose>
 227         </xsl:variable>
 228         <xsl:variable name="channel-link" select="//link[@rel='canonical']/@href"/>
 229         <xsl:variable name="channel-image" select="//table[@class='profile-details' or @class='main-tweet']//td[@class='avatar']//img/@src"/>
 230
 231         <rss version="2.0">
 232             <xsl:attribute name="xml:base"><xsl:value-of select="$BaseURL" /></xsl:attribute>
 233             <channel>
 234                 <generator>Tweeper</generator>
 235                 <title>
 236                     <xsl:value-of select="$channel-title"/>
 237                 </title>
 238                 <link>
 239                     <xsl:value-of select="$channel-link"/>
 240                 </link>
 241                 <description>
 242                     <xsl:value-of select="normalize-space(//table[@class='profile-details' or @class='main-tweet']//td[@class='details'])"/>
 243                 </description>
 244                 <xsl:if test="$channel-image != ''">
 245                     <image>
 246                         <title>
 247                             <xsl:value-of select="$channel-title"/>
 248                         </title>
 249                         <link>
 250                             <xsl:value-of select="$channel-link"/>
 251                         </link>
 252                         <url>
 253                             <xsl:value-of select="$channel-image"/>
 254                         </url>
 255                     </image>
 256                 </xsl:if>
 257                 <xsl:apply-templates select="//div[contains(@class, 'timeline')]/table[@class='tweet  ']|//div[@class='main-tweet-container']/table[@class='main-tweet']"/>
 258             </channel>
 259         </rss>
 260     </xsl:template>
 261 </xsl:stylesheet>