wiki_to_hdoc.xsl

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
    xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:functx="http://www.functx.com"
    exclude-result-prefixes="xs" version="2.0" xmlns="http://www.utc.fr/ics/hdoc/xhtml">

    <xsl:output method="xhtml" indent="yes"/>
    <xsl:preserve-space elements="pre"/>

    <xsl:template match="*"/>
    <xsl:template match="text()">
        <xsl:value-of select="."/>
    </xsl:template>

    <xsl:template match="html">
        <!-- Schema link -->
        <xsl:processing-instruction name="oxygen">RNGSchema="http://scenari.utc.fr/hdoc/schemas/xhtml/hdoc1-xhtml.rng" type="xml"</xsl:processing-instruction>
        <xsl:text>&#10;</xsl:text>

        <!-- html content -->
        <html>
            <head>
                <xsl:apply-templates select="head"/>
            </head>
            <body>
                <xsl:apply-templates select="body"/>
            </body>
        </html>
    </xsl:template>

    <!-- Head template -->
    <xsl:template match="head">
        <xsl:apply-templates select="title"/>
        <meta charset="utf-8"/>
        <meta name="generator" content="HdocConverter/wikipedia"/>
        <meta name="author" content="Wikipedia"/>
    </xsl:template>

    <!-- Body template -->
    <xsl:template match="body">
        <!-- If there is h1 title with firstHeading class, this has to be the first section title -->
        <xsl:apply-templates select="//h1[1]" mode="introduction"/>

        <!-- Next are all parts of the document: h2 -->
        <xsl:apply-templates select="//h2"/>
    </xsl:template>

    <xsl:template match="title">
        <title>
            <xsl:value-of select="."/>
        </title>
    </xsl:template>

    <!-- Beginning of the wikipedia page: introduction paragraph -->
    <xsl:template match="h1[1]" mode="introduction">
        <section data-hdoc-type="introduction">
            <header>
                <h1>Introduction</h1>
            </header>

            <!-- Introduction text before first h2 (first section) -->
            <xsl:if test="//p[count(preceding::h2)=0 and not(ancestor::td)]">
                <div>
                    <xsl:apply-templates select="//p[count(preceding::h2)=0 and not(ancestor::td)]"
                        mode="textOnly"/>
                </div>
            </xsl:if>
        </section>
    </xsl:template>

    <!-- Wikipedia sections and subsections -->
    <xsl:template match="h2|h3|h4|h5">
        <section>
            <!-- h3 sections are opale "grains" -->
            <xsl:if test="self::h3">
                <xsl:attribute name="data-hdoc-type">unit-of-content</xsl:attribute>
            </xsl:if>
            <header>
                <h1>
                    <xsl:apply-templates select="node()" mode="title"/>
                </h1>
            </header>

            <!-- Storing current section to know when apply template has to be called in the next parts of the template -->
            <xsl:variable name="currentSectionTitle" select="."/>


            <!-- Apply this template for all elements between the actual title h{2,3,4,5} and the next one  -->
            <xsl:variable name="nextTitle"
                select="following-sibling::*[starts-with(name(), 'h')][1]"/>

            <xsl:apply-templates
                select="following-sibling::div[contains(@class,'thumb tleft') or contains(@class,'thumb tright')] intersect $nextTitle/preceding-sibling::*"/>

            <!--             <xsl:apply-templates select="following-sibling::p/img"/>-->
            <!-- If there is text right below the section name, copy it -->
            <xsl:if
                test="not(following-sibling::*[1] intersect following-sibling::h3) and not(following-sibling::*[1] intersect following-sibling::h4) and not(following-sibling::*[1] intersect following-sibling::h5) and not(following-sibling::*[1] intersect following-sibling::h6)">
                <div>
                    <!--IIICCIIII -->
                    <xsl:for-each
                        select="following-sibling::* intersect following-sibling::h2[1]/preceding-sibling::*">
                        <xsl:if
                            test="not(preceding-sibling::h3 intersect $currentSectionTitle/following-sibling::h3) and not($currentSectionTitle/following-sibling::h4 intersect preceding-sibling::h4) and not($currentSectionTitle/following-sibling::h5 intersect preceding-sibling::h5) and not($currentSectionTitle/following-sibling::h6 intersect preceding-sibling::h6) and not(self::h3) and not(self::h4)  and not(self::h5)  and not(self::h6)">
                            <xsl:apply-templates select="." mode="textOnly"/>
                            <!--To add images it took me almost 1h and a half I believe strongly that the structure of this page can be changed  and made easy !!  -->
                        </xsl:if>
                    </xsl:for-each>
                </div>
            </xsl:if>

            <!-- Applying template of subsections if any -->
            <xsl:choose>
                <xsl:when test="self::h2">
                    <!-- h2 can have h3 subsections -->
                    <xsl:apply-templates
                        select="following-sibling::h3 intersect following-sibling::h2[1]/preceding-sibling::h3"
                    />
                </xsl:when>
                <xsl:when test="self::h3">
                    <!-- Apply template to h4 subsections of h3. These h4 are below the current h3: previous h3 of these h4 is current h3. -->
                    <xsl:for-each
                        select="following-sibling::h4 intersect following-sibling::h2[1]/preceding-sibling::h4">
                        <xsl:if test="(preceding-sibling::h3[1] intersect $currentSectionTitle)">
                            <xsl:apply-templates select="."/>
                        </xsl:if>
                    </xsl:for-each>
                </xsl:when>
                <xsl:when test="self::h4">
                    <!-- Apply template aux h5 devant qui sont soit avant un h2, soit un h3, soit un h4 -->
                    <xsl:for-each
                        select="following-sibling::h5 intersect following-sibling::h2[1]/preceding-sibling::h5">
                        <xsl:if
                            test="(preceding-sibling::h3[1] intersect $currentSectionTitle/preceding-sibling::h3[1]) and (preceding-sibling::h4[1] intersect $currentSectionTitle)">
                            <xsl:apply-templates select="."/>
                        </xsl:if>
                    </xsl:for-each>
                </xsl:when>
                <xsl:when test="self::h5">
                    <!-- Apply template aux h6 devant qui sont soit avant un h2, soit un h3, soit un h4 -->
                    <xsl:for-each
                        select="following-sibling::h6 intersect following-sibling::h2[1]/preceding-sibling::h6">
                        <xsl:if
                            test="(preceding-sibling::h3[1] intersect $currentSectionTitle/preceding-sibling::h3[1]) and (preceding-sibling::h4[1] intersect $currentSectionTitle/preceding-sibling::h4[1]) and (preceding-sibling::h5[1] intersect $currentSectionTitle)">
                            <xsl:apply-templates select="."/>
                        </xsl:if>
                    </xsl:for-each>
                </xsl:when>
            </xsl:choose>
        </section>
    </xsl:template>

    <!-- Wikipedia h6 is not a section in hdoc: div with h6 title -->
    <xsl:template match="h6">
        <div>
            <h6>
                <xsl:apply-templates select="node()" mode="titleh6"/>
            </h6>

            <xsl:variable name="currentSectionTitle" select="."/>
            <!-- Text of h6 is contained between current h6 and next section title (next h3/h4/h5/h6) -->
            <xsl:for-each
                select="following-sibling::* intersect following-sibling::h2[1]/preceding-sibling::*">
                <xsl:if
                    test="not(preceding-sibling::h3 intersect $currentSectionTitle/following-sibling::h3) and not($currentSectionTitle/following-sibling::h4 intersect preceding-sibling::h4) and not($currentSectionTitle/following-sibling::h5 intersect preceding-sibling::h5) and not($currentSectionTitle/following-sibling::h6 intersect preceding-sibling::h6) and not(self::h3) and not(self::h4)  and not(self::h5)  and not(self::h6)">
                    <xsl:apply-templates select="." mode="textOnly"/>
                </xsl:if>
            </xsl:for-each>
        </div>
    </xsl:template>

    <!-- Text elements not surrounded by div -->
    <xsl:template match="p|span|i|ul|ol" mode="textOnly">
        <xsl:element name="{local-name()}" namespace="http://www.utc.fr/ics/hdoc/xhtml">
            <xsl:apply-templates select="node()" mode="textOnly"/>
        </xsl:element>
    </xsl:template>

    <!-- Paragraph template -->
    <xsl:template match="p">
        <div>
            <xsl:element name="{local-name()}" namespace="http://www.utc.fr/ics/hdoc/xhtml">
                <xsl:apply-templates select="node()" mode="textOnly"/>
            </xsl:element>
        </div>
    </xsl:template>

    <!-- li -->
    <xsl:template match="li" mode="#all">
        <xsl:element name="{local-name()}" namespace="http://www.utc.fr/ics/hdoc/xhtml">
                   
            <p>
                <xsl:apply-templates select="node()" mode="textOnly"/>
            </p>
        </xsl:element>
    </xsl:template>
    <!--Gestion des images integrées dans le texte-->
    
    <xsl:template match="li[@class='gallerybox']" mode="#all">
        <xsl:element name="{local-name()}" namespace="http://www.utc.fr/ics/hdoc/xhtml">
                <img src="./ressources/{translate(tokenize(.//img/@src, '/')[last()],'?%','_')}" alt="{.//img/@alt}"/>
                 <p>
                     <xsl:value-of select=".//div[@class='galerytext']/p"/>
                 </p>
        </xsl:element>
    </xsl:template>

    <!-- text followed directly by ul not allowed in li -->
    <xsl:template match="li[descendant::ul]" mode="#all">
        <xsl:element name="{local-name()}" namespace="http://www.utc.fr/ics/hdoc/xhtml">
            <p>
                <xsl:apply-templates
                    select="descendant::node() intersect descendant::ul[1]/preceding-sibling::node()"
                    mode="textOnly"/>
            </p>
            <xsl:apply-templates select="descendant::ul" mode="textOnly"/>
        </xsl:element>
    </xsl:template>

    <!-- dl/dt/dd in each in a div -->
    <xsl:template match="dl[descendant::dt]" mode="#all">
        <ul>
            <xsl:apply-templates select="dt"/>
        </ul>
    </xsl:template>

    <xsl:template match="dl[not(descendant::dt)]" mode="#all">
        <xsl:apply-templates select="dd"/>
    </xsl:template>

    <!-- dt contains the title -->
    <xsl:template match="dt">
        <li>
            <p>
                <em>
                    <xsl:apply-templates select="node()" mode="titleh6"/>
                </em>
            </p>

            <!-- Some dt are followed by multiple dd before next dd: make sure to copy text of each dd before next dt -->
            <xsl:variable name="currentElement" select="."/>
            <xsl:apply-templates
                select="following-sibling::dd[preceding-sibling::dt[1] intersect $currentElement]"/>

            <!-- Some wikipedia articles don't use dd after dt... -->
            <xsl:if test="not(following-sibling::dd[1])">
                <xsl:apply-templates select="../following-sibling::p[1]" mode="textOnly"/>
            </xsl:if>
        </li>
    </xsl:template>

    <!-- dd contains the content -->
    <xsl:template match="dd">
        <p>
            <xsl:apply-templates select="node()"/>
        </p>
        <xsl:apply-templates select="ul|ol" mode="textOnly"/>
    </xsl:template>

    <!-- Rules for title elements (h1, h2...) -->
    <xsl:template match="*" mode="title">
        <xsl:apply-templates select="node()" mode="title"/>
    </xsl:template>

    <xsl:template match="*" mode="titleh6" priority="2">
        <xsl:value-of select="."/>
    </xsl:template>

    <xsl:template match="b" mode="textOnly">
        <!-- b is not allowed, however em is allowed: replacing all b by em -->
        <xsl:element name="em" namespace="http://www.utc.fr/ics/hdoc/xhtml">
            <xsl:apply-templates select="node()" mode="textOnly"/>
        </xsl:element>
    </xsl:template>

    <!-- b outside p, wrap it into a p (and convert it to em) -->
    <xsl:template match="b[not(ancestor::p) and not(ancestor::li) and not(ancestor::a)]"
        mode="textOnly">
        <p>
            <em>
                <xsl:apply-templates select="node()" mode="textOnly"/>
            </em>
        </p>
    </xsl:template>

    <!-- Link elements -->
    <!-- a in title not allowed, only keeping text -->
    <xsl:template match="h2/span/a | h3/span/a | h4/span/a | h5/span/a | h6/span/a" mode="#all">
        <xsl:value-of select="."/>
    </xsl:template>

    <xsl:template match="a" mode="#all">
        <xsl:element name="{local-name()}" namespace="http://www.utc.fr/ics/hdoc/xhtml">
            <xsl:attribute name="href" select="concat('http://wikipedia.org', @href)"/>
            <xsl:value-of select="."/>
        </xsl:element>
    </xsl:template>

    <!-- Link elements without actual wikipedia page: keeping only their text -->
    <xsl:template match="a[contains(@class, 'new')]" mode="#all">
        <xsl:value-of select="."/>
    </xsl:template>

    <!-- Handle Listing -->
    <xsl:template match="div[@class='mw-highlight mw-content-ltr']" mode="textOnly">
        <xsl:apply-templates select="pre" mode="textOnly"/>
    </xsl:template>

    <xsl:template match="pre" mode="textOnly">
        <xsl:variable name="vUid">
            <xsl:number level="any" count="pre"/>
        </xsl:variable>

        <xsl:variable name="listingContent">
            <xsl:copy-of select="@*|node()"/>
        </xsl:variable>

        <xsl:variable name="listingContentWithoutBadCharacters">
            <xsl:copy-of select="replace($listingContent,'&amp;','&amp;amp;')"/>
        </xsl:variable>

        <xsl:variable name="listingContent2">
            <xsl:value-of select="substring-after($listingContentWithoutBadCharacters,'&#10;')"/>
        </xsl:variable>

        <xsl:variable name="basename">
            <xsl:call-template name="substring-before-last">
                <xsl:with-param name="string1" select="$listingContent2"/>
                <xsl:with-param name="string2" select="'&#10;'"/>
            </xsl:call-template>
        </xsl:variable>
        <object type="application/x-listing" data="listing/listing{$vUid}.xml">
            <!--Modifier le application/pdf lorsque le rng sera changé-->
            <xsl:result-document href="listing/listing{$vUid}.xml" method="text">
                <xsl:text disable-output-escaping="no">
&lt;sc:item xmlns:sc="http://www.utc.fr/ics/scenari/v3/core"&gt;
	&lt;op:code xmlns:sp="http://www.utc.fr/ics/scenari/v3/primitive" xmlns:op="utc.fr:ics/opale3"&gt;
		&lt;sc:code mimeType="text/plain"&gt;</xsl:text>
                <xsl:value-of select="$basename"/>
                <xsl:text disable-output-escaping="no">&lt;/sc:code&gt;
	&lt;/op:code&gt;
&lt;/sc:item&gt;
                </xsl:text>
            </xsl:result-document>
        </object>
    </xsl:template>

    <xsl:template name="substring-before-last">
        <xsl:param name="string1" select="''"/>
        <xsl:param name="string2" select="''"/>

        <xsl:if test="$string1 != '' and $string2 != ''">
            <xsl:variable name="head" select="substring-before($string1, $string2)"/>
            <xsl:variable name="tail" select="substring-after($string1, $string2)"/>
            <xsl:value-of select="$head"/>
            <xsl:if test="contains($tail, $string2)">
                <xsl:value-of select="$string2"/>
                <xsl:call-template name="substring-before-last">
                    <xsl:with-param name="string1" select="$tail"/>
                    <xsl:with-param name="string2" select="$string2"/>
                </xsl:call-template>
            </xsl:if>
        </xsl:if>
    </xsl:template>


    <!-- Simple Tables -->
    <xsl:template match="table" mode="#all">
        <xsl:choose>
            <!-- Tables that doesn't contain colspan, rowspan or included tables -->
            <xsl:when
                test="not(descendant::td/@colspan | descendant::td/@rowspan | descendant::table)">
                <table>
                    <xsl:apply-templates select="node()" mode="textOnly"/>
                </table>
            </xsl:when>
            <xsl:otherwise>
                <xsl:call-template name="TableComplexe"/>
            </xsl:otherwise>
        </xsl:choose>
    </xsl:template>

    <!-- Complexe Tables : create extern Ods files and copy content in it-->
    <xsl:template name="TableComplexe">
        <xsl:variable name="tableId">
            <xsl:number level="any" count="table"/>
        </xsl:variable>
        <object type="application/vnd.oasis.opendocument.spreadsheet"
            data="tables/table{$tableId}.ods">
            <xsl:result-document href="tables/table{$tableId}.ods" method="html" encoding="utf-8">
                <xsl:element name="table">
                    <xsl:copy-of select="@*|node()"/>
                </xsl:element>
            </xsl:result-document>
        </object>
    </xsl:template>

    <xsl:template match="caption" mode="textOnly">
        <caption>
            <xsl:value-of select="node()"/>
        </caption>
    </xsl:template>

    <xsl:template match="tr" mode="textOnly">
        <tr>
            <xsl:apply-templates select="node()" mode="textOnly"/>
        </tr>
    </xsl:template>

    <xsl:template match="td | th" mode="textOnly">
        <xsl:choose>
            <xsl:when test="not(ul|ol)">
                <td>
                    <p>
                        <xsl:apply-templates select="." mode="table"/>
                    </p>
                </td>
            </xsl:when>
            <xsl:otherwise>
                <td>
                    <xsl:apply-templates select="." mode="table"/>
                </td>
            </xsl:otherwise>
        </xsl:choose>
    </xsl:template>

    <xsl:template match="a[@class='image']" mode="table"/>

    <xsl:template match="b" mode="table" priority="1">
        <!-- b is not allowed, however em is allowed: replacing all b by em -->
        <xsl:element name="em" namespace="http://www.utc.fr/ics/hdoc/xhtml">
            <xsl:apply-templates select="node()" mode="textOnly"/>
        </xsl:element>
    </xsl:template>

    <xsl:template match="td/span[contains(@class, 'mw-edit')]" mode="table"/>

    <!-- Balises pour les abbréviations -->
    <xsl:template match="abbr" mode="table" priority="1">
        <xsl:apply-templates select="node()" mode="textOnly"/>
    </xsl:template>

    <xsl:template match="abbr/sup" mode="table" priority="1">
        <sup>
            <xsl:value-of select="."/>
        </sup>
    </xsl:template>

    <xsl:template match="ul|ol" mode="table" priority="1">
        <xsl:element name="{local-name()}">
            <xsl:apply-templates select="node()" mode="textOnly"/>
        </xsl:element>
    </xsl:template>


    <!-- Images -->
    <xsl:template match="div[contains(@class,'thumb')]">
        <div>
            <xsl:apply-templates select=".//img" mode="textOnly"/>
            <xsl:apply-templates select=".//p[@class='thumbcaption']"/>
        </div>
    </xsl:template>

    <xsl:template match="img" mode="textOnly">
        <img src="./ressources/{translate(tokenize(@src, '/')[last()],'?%','_')}" alt="{@alt}"/>
    </xsl:template>


    <xsl:template match="p[@class='thumbcaption']">
        <p>
            <xsl:value-of select="text()"/>
        </p>
    </xsl:template>


    <!-- ===== Ignored content ===== -->

    <!-- Only keeping a with information: we give up page references -->
    <xsl:template match="a[starts-with(@href, '#')]" mode="#all"/>

    <!-- Removing Wikipedia internal sup, they are not useful to us (sup are "cite source / reference" etc...) -->
    <xsl:template match="sup" mode="#all"/>

    <!-- Ignoring empty text elements only relevant to Wikipedia -->
    <xsl:template match="p[empty(node())]" mode="#all" priority="2"/>
    <xsl:template match="span[contains(@class, 'mw-edit')]" mode="#all" priority="2"/>

    <!-- Ignoring table of contents h2 -->
    <xsl:template match="div[@id='toctitle']/h2" priority="1"/>

    <!-- Ignoring end of file h2: see also, etc... -->
    <!--    <xsl:template match="h2[not(following-sibling::p intersect following-sibling::h2[1]/preceding-sibling::p)]"/>-->
    <xsl:template
        match=" h2[following-sibling::ul[li/a[@class='external text'] and preceding-sibling::h2]][last()]"/>
    <xsl:template
        match=" h2[following-sibling::div[@class='references-small decimal'] and preceding-sibling::h2][last()]"/>
    <xsl:template match="h2[following-sibling::h3[span[@id='Notes']]][last()]"/>
    <xsl:template match="div[@id='mw-navigation']/h2"/>
    <xsl:template match="h2[span[@id='Voir_aussi']]"/>
    <xsl:template match="h2[span[@id='Notes_et_r.C3.A9f.C3.A9rences']]"/>

    <!-- Ignoring divs by default: they are not relevant to us -->
    <xsl:template match="div[not(descendant::table)]" mode="#all"/>
</xsl:stylesheet>