From 4f553caf544d85503c40cc375b619f887efac2ca Mon Sep 17 00:00:00 2001
From: eg3p <eg3p>
Date: Wed, 16 May 2007 23:49:23 +0000
Subject: [PATCH] added task to pull down all THDL transcripts from orion

---
 lucene-thdl-build.xml | 50 +++++++++++++++++++++----------------------
 1 file changed, 24 insertions(+), 26 deletions(-)

diff --git a/lucene-thdl-build.xml b/lucene-thdl-build.xml
index 1191b7b..4fe93ce 100644
--- a/lucene-thdl-build.xml
+++ b/lucene-thdl-build.xml
@@ -44,16 +44,6 @@
     -->
 
 <!--
-    <target name="archive-get-transcripts" depends="archive-compile">
-		<java classname="org.thdl.archive.GetTranscripts" fork="yes">
-            <arg value="${archive.metadata.file}"/>
-            <arg value="${wylie.transcript.dir}"/>
-			<classpath>
-				<pathelement location="${bin.dir}"/>
-				<path refid="classpath"/>
-			</classpath>
-		</java>
-    </target>
     
     <target name="archive-transcripts-to-unicode">
 		<java classname="net.sf.saxon.Transform" fork="yes">
@@ -70,34 +60,42 @@
     -->
     
     <!-- archive tasks -->
+    
+    <!-- title_metadata.xml is not being retrieved as UTF-8. need to fix this. 
+        perhaps we don't need to decodeUTF the transcript file name???
+    -->
     <target name="archive-get-metadata">
         <mkdir dir="${archive}"/>
         <get src="${get.title.metadata}" dest="${archive}/title_metadata.xml" verbose="on"/>
     </target>
-    
-    <target name="archive-get-transcripts">
+
+    <target name="archive-get-list-of-transcripts">
     	<mkdir dir="${wylie}"/>
     	<xslt in="${archive}/title_metadata.xml" out="${archive}/titles_as_list.txt" style="${styles}/get-list-of-transcripts.xsl"/>
-	<loadfile property="transcript-list" srcFile="${archive}/titles_as_list.txt"/>
-	<foreach list="${transcript-list}" delimiter=" " param="filename" target="archive-get-one-transcript"/>
+    </target>
+
+    <!-- must delete following transcripts from titles_as_list.txt
+            02069_clip-21-husked-barle_00.xml
+            02116_nasal-congestion_00.xml
+            A_New_Script02.xml
+            
+            00007_06-dawa-and-purdrön_08.xml 
+         or, should put in protection against URLs that cannot be "got"
+    -->    
+    <target name="archive-get-transcripts">
+        <loadfile property="transcript-list" srcFile="${archive}/titles_as_list.txt"/>
+        <!--<loadfile property="transcript-list" srcFile="${archive}/titles_as_list.txt" encoding="UTF-8"/>-->
+        <foreach list="${transcript-list}" delimiter=" " param="filename" target="archive-get-one-transcript"/>
     </target>
     
+    <!-- problem: destination filename should not be URLEncoded -->
     <target name="archive-get-one-transcript">
+        <!--<urlencode property="file.location" value="00007_06-dawa-and-purdrön_08.xml" />
+        <get src="${url.to.transcripts}/${file.location}" dest="${wylie}/TEST.xml"/>-->
     	<get src="${url.to.transcripts}/${filename}" dest="${wylie}/${filename}"/>
     </target>
- 
-    <!--
-    <filelist 
-    id="docfiles" 
-    dir="${doc.src}">
-    <file name="foo.xml"/>
-    <file name="bar.xml"/>
-    </filelist>
 
-<filterreader classname="org.apache.tools.ant.filters.PrefixLines">
-  <param name="prefix" value="Foo"/>
-</filterreader>
--->
+    
     <!-- concordance program -->
 	<target name="lucene-thdl-compile" depends="init">
         <mkdir dir="${lucene-thdl.bin}"/>