Merge branch 'master' of https://gitlab.utc.fr/crozatst/hdoc

framapad_to_hdoc/README.md

@@ -6,8 +6,8 @@ http://www.gnu.org/licenses/gpl-3.0.txt
 
 ## Credits
 - 2016
--   - Etienne Chognard
--   - Fabien Boucaud
+    - Etienne Chognard
+    - Fabien Boucaud
 - 2015
     - Jean-Côme Douteau
     - Gabrielle Rit
@@ -85,4 +85,4 @@ Currently available on: https://framemo.org/framapad_to_opale
 - br -> p
 
 ## Capitalisation
-Using regular expression with xsl is a good way to parse a non xml file.
\ No newline at end of file
+Using regular expression with xsl is a good way to parse a non xml file.

framapad_to_hdoc/etherpad_to_hdoc.ant → framapad_to_hdoc/framapad_to_hdoc.ant

 <?xml version="1.0" encoding="UTF-8"?>
 <project basedir="." name="myantce" default="main">
-    <property file="etherpad_to_hdoc.properties"/>
+    <property file="framapad_to_hdoc.properties"/>
     <!-- import classes -->
     <taskdef resource="net/sf/antcontrib/antlib.xml"/>
     <taskdef name="htmlcleaner" classname="org.htmlcleaner.HtmlCleanerForAnt"/>

framapad_to_hdoc/run.bat

 @echo off
 set lib=lib
-set ant=etherpad_to_hdoc.ant
+set ant=framapad_to_hdoc.ant
 set antparam=-Dprogram.param=%1
 
 set scJarList=%lib%\*

framapad_to_hdoc/run.sh

 #!/bin/sh
 lib="lib"
-ant="etherpad_to_hdoc.ant"
+ant="framapad_to_hdoc.ant"
 antparam="-Dprogram.param=$1"
 
 #Recherche de java et controle que se soit une version SUN

framapad_to_opale/etherpad_to_opale.ant → framapad_to_opale/framapad_to_opale.ant

 <?xml version="1.0" encoding="UTF-8"?>
 <project basedir="." name="myantce" default="main">
-    <property file="etherpad_to_opale.properties"/>    
+    <property file="framapad_to_opale.properties"/>    
     <!-- CHECK FOR OS FAMILY -->
     <condition property="is_windows">
         <os family="windows"/>
@@ -24,7 +24,7 @@
                 <include name="*.html"/>
             </fileset>
         </copy> 
-        <exec dir="../etherpad_to_hdoc" executable="run.bat"/>
+        <exec dir="../framapad_to_hdoc" executable="run.bat"/>
     </target>
     
     <target name="to_hdoc_unix" if="${is_unix}">
@@ -38,8 +38,8 @@
                 <include name="*.html"/>
             </fileset>
         </copy> 
-        <exec executable="/bin/bash" dir="../etherpad_to_hdoc">
-            <arg value="../etherpad_to_hdoc/run.sh"/>
+        <exec executable="/bin/bash" dir="../framapad_to_hdoc">
+            <arg value="../framapad_to_hdoc/run.sh"/>
         </exec>
     </target>
     
@@ -90,4 +90,4 @@
         <antcall target="to_opale_windows"/>
         <antcall target="to_opale_end"/>        
     </target>
-</project>
\ No newline at end of file
+</project>

framapad_to_opale/etherpad_to_opale.properties → framapad_to_opale/framapad_to_opale.properties

@@ -5,8 +5,8 @@ xsl = ${basedir}/xsl
 lib = ${basedir}/lib
 log = ${basedir}/log
 
-eth_in = ../etherpad_to_hdoc/input
-eth_out = ../etherpad_to_hdoc/output
+eth_in = ../framapad_to_hdoc/input
+eth_out = ../framapad_to_hdoc/output
 opa_in = ../hdoc_to_opale/input
 opa_out = ../hdoc_to_opale/output
-rootfilename = content.xml
\ No newline at end of file
+rootfilename = content.xml

framapad_to_opale/run.bat

 @echo off
 set lib=lib
-set ant=etherpad_to_opale.ant
+set ant=framapad_to_opale.ant
 set antparam=-Dprogram.param=%1
 
 set scJarList=%lib%\*

framapad_to_opale/run.sh

 #!/bin/sh
 lib="lib"
-ant="etherpad_to_opale.ant"
+ant="framapad_to_opale.ant"
 antparam="-Dprogram.param=$1"
 
 #Recherche de java et controle que se soit une version SUN

hdoc_to_basex/build.properties

-libdir=${basedir}/lib
-rngdir=${basedir}/rng
-xsldir=${basedir}/xsl
-srcdir=${basedir}/in
-outdir=${basedir}/out
-InputPath=${basedir}/input
-OutputPath=${basedir}/output
+lib=${basedir}/lib
+log=${basedir}/log
+xsl=${basedir}/xsl
+in=${basedir}/input
+out=${basedir}/output
+tmp=${basedir}/tmp
\ No newline at end of file

hdoc_to_basex/hdoc_to_basex.ant

@@ -3,49 +3,42 @@
   
   <taskdef resource="net/sf/antcontrib/antlib.xml"/>
   
-    <property name="in" location="${basedir}/input"/>
-    <property name="out" location="${basedir}/output"/>
-    <property name="tmp" location="${basedir}/tmp"/>
-    <property name="xsl" location="${basedir}/xsl"/>
-    <property name="lib" location="${basedir}/lib"/>
-    <property name="log" location="${basedir}/log"/>
-        
-        <target name="convert">
-          <echo message="DEBUT"/>
-
-     
-
-      <mkdir dir="${tmp}"/>
-      <delete dir="${out}" failonerror="false"/>
-
-      <mkdir dir="${out}"/>
-      <delete dir="${log}" failonerror="false"/>
-
-      <mkdir dir="${log}"/>
-          
-
-          <for param="inputFile">
-        <path>
-          <fileset dir="${in}" includes="**/*.hdoc"/>
-        </path>
-
-
-        <sequential>
-          <local name="filename"/>
-          <basename property="filename" file="@{inputFile}"/>
-
-            <antcall target="UnzipHdocFile">
-            <param name="filename" value="${filename}"/>
-          </antcall>
-          
-          <antcall target="content">
-            <param name="filename" value="${filename}"/>
-          </antcall>
-          <echo message="FIN"/>
-        </sequential>
-          
-        </for>
-    </target>
+  <property file="build.properties"/>
+		
+	<target name="convert">
+		<!-- Preparation for the file transformation : delete old folders and create new folders -->
+		<mkdir dir="${tmp}"/>
+		<delete dir="${out}" failonerror="false"/>
+		<mkdir dir="${out}"/>
+		<delete dir="${log}" failonerror="false"/>
+		<mkdir dir="${log}"/>
+		<echo message="DEBUT"/>
+		
+		<!-- Convert all the hdoc files in the directory ${in} to data xml files that will be imported to basex later.
+		Fonctions "UnzipHdocFile" and "content" will be called. -->
+		<for param="inputFile">
+			<path>
+			  <fileset dir="${in}" includes="**/*.hdoc"/>
+			</path>
+
+			<sequential>
+			  <local name="filename"/>
+			  <basename property="filename" file="@{inputFile}"/>
+
+				<antcall target="UnzipHdocFile">
+				<param name="filename" value="${filename}"/>
+			  </antcall>
+			  
+			  <antcall target="content">
+				<param name="filename" value="${filename}"/>
+			  </antcall>
+			  <echo message="FIN"/>
+			</sequential>
+		</for>
+		
+		<!-- Clean-->
+		<delete dir="${tmp}" failonerror="false"/>
+	</target>
 
     <target name="UnzipHdocFile">
         <!-- Unzip the input hdoc file. Decompressed folder is named "decompressedHdoc" : this name is the only one which
@@ -56,15 +49,13 @@
     </target>
 
      <target name="content" >
-        <xslt in="${tmp}/${filename}/decompressedHdoc/content.xml" out="${out}/${filename}_data.xml" style="${xsl}/transformation.xsl" processor="org.apache.tools.ant.taskdefs.optional.TraXLiaison">
-
-            <param name="filename" expression="${filename}"/>
-            <param name="lib" expression="${lib}"/>
-        </xslt>
-         <echo message="${filename}" />
-         <delete dir="${tmp}" failonerror="false"/>
-
-
+		<!-- Transformation of a xml file decompressed from hdoc file to data xml file to be imported to basex.
+		The transformation will be done in terms of the xsl file in ${xsl}.-->
+		<xslt in="${tmp}/${filename}/decompressedHdoc/content.xml" out="${out}/${filename}_data.xml" style="${xsl}/transformation.xsl" processor="org.apache.tools.ant.taskdefs.optional.TraXLiaison">
+			<param name="filename" expression="${filename}"/>
+			<param name="lib" expression="${lib}"/>
+		</xslt>
+		<echo message="${filename}" />
     </target> 
     
 </project>
\ No newline at end of file

opale_to_hdoc/README.md

@@ -149,3 +149,11 @@ The opale_to_hdoc converter transcribes an Opale publication to a hdoc format. O
 
 The source code converter is available and can be completed in order to 
 improve it and manage more Opale elements.
+
+
+Modified by Benoit Villain (12/12/2016)
+------------------
+Les divisions et les parties d'un grain n'étaient pas traitées... Contrairement à ce que mentionne le README...
+J'ai donc apporté les modifications nécessaires pour que les divisions et les parties (récursives) soient bel et bien transformées en "section" hdoc.
+
+

wikipedia_to_hdoc/xslt/wiki_to_hdoc.xsl

@@ -423,8 +423,10 @@
     <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[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="div[@id='mw-navigation']/h2"/>
     
     <!-- Ignoring divs by default: they are not relevant to us -->
     <xsl:template match="div" mode="#all"/>