fix CPAN RT bug 71711

tla · Oct 16, 2011 · 75732dd · 75732dd
1 parent a950fda
commit 75732dd
Show file tree

Hide file tree

Showing 4 changed files with 68 additions and 8 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1 +1,3 @@
 *~
+*.bbprojectd
+catalyst/lib/Text
diff --git a/cpan/Text-TEI-Collate/lib/Text/TEI/Collate/Manuscript.pm b/cpan/Text-TEI-Collate/lib/Text/TEI/Collate/Manuscript.pm
@@ -1,6 +1,6 @@
 package Text::TEI::Collate::Manuscript;
 
-use vars qw( $VERSION %assigned_sigla );
+use vars qw( $VERSION %assigned_sigla %tags );
 use Moose;
 use Moose::Util::TypeConstraints;
 use Text::TEI::Collate::Word;
@@ -116,17 +116,26 @@ sub _init_from_xmldesc {
 		warn "Manuscript initialization needs a TEI document!";
 		return;
 	}
-	# Get the identifier
+
+	# Set up the tags we need, with or without namespaces.
+	map { $tags{$_} = "//$_" } qw/ msDesc settlement repository idno p lg /;
+	# Set up our XPath object
 	my $xpc = XML::LibXML::XPathContext->new( $xmlobj );
-	$xpc->registerNs( 'tei', $xmlobj->namespaceURI );
+	# Use namespace-aware tags if we have to 
+	if( $xmlobj->namespaceURI ) {
+	    $xpc->registerNs( 'tei', $xmlobj->namespaceURI );
+	    map { $tags{$_} = "//tei:$_" } keys %tags;
+	}
 	$self->_set_xpc( $xpc );
-	if( my $desc = $xpc->find( '//tei:msDesc' ) ) {
+
+	# Get the identifier
+	if( my $desc = $xpc->find( $tags{msDesc} ) ) {
 		my $descnode = $desc->get_node(1);
 		$self->_save_msdesc( $descnode );
 		my( $setNode, $reposNode, $idNode ) =
-			( $xpc->find( '//tei:settlement' )->get_node(1),
-			  $xpc->find( '//tei:repository' )->get_node(1),
-			  $xpc->find( '//tei:idno' )->get_node(1) );
+			( $xpc->find( $tags{settlement} )->get_node(1),
+			  $xpc->find( $tags{repository} )->get_node(1),
+			  $xpc->find( $tags{idno} )->get_node(1) );
 		$self->settlement( $setNode ? $setNode->textContent : '' );
 		$self->repository( $reposNode ? $reposNode->textContent : '' );
 		$self->idno( $idNode ? $idNode->textContent : '' );
@@ -187,7 +196,8 @@ sub _read_paragraphs_or_lines {
 
 	my @words;
 	my $xpc = $self->_xpc;
- 	my @pgraphs = $xpc->findnodes( './/tei:p | .//tei:lg', $element );
+	my $xpexpr = '.' . $tags{p} . '|.' . $tags{lg};
+ 	my @pgraphs = $xpc->findnodes( $xpexpr, $element );
     return () unless @pgraphs;
 	foreach my $pg( @pgraphs ) {
 		# If this paragraph is the descendant of a note element,

diff --git a/cpan/Text-TEI-Collate/t/02ms_open.t b/cpan/Text-TEI-Collate/t/02ms_open.t
@@ -142,3 +142,17 @@ foreach my $idx ( 0 .. $#wordcount ) {
 	is( join( ' ', map { $_->{t} } @{$json[0]->{witnesses}->[$idx]->{tokens}} ), $t, "Got correct words in plaintext JSON tokens" );
 }
 
+# Test opening a TEI file with no namespace 
+my $xmlparser = XML::LibXML->new();
+my $doc;
+eval { $doc = $xmlparser->parse_file( "$dirname/data/tei_no_ns.xml" )->documentElement(); };
+ok( defined $doc, "parsed the XML file tei_no_ns.xml" );
+my $ms_obj = Text::TEI::Collate::Manuscript->new( 
+	'sourcetype' => 'xmldesc',
+	'source' => $doc,
+	);
+my $words = scalar @{$ms_obj->words};
+my @placeholders = grep { $_->placeholders } @{$ms_obj->words};
+is( $words, 30, "Got correct number of total words" );
+is( scalar @placeholders, 2, "Got correct number of placeholder words" );
+
diff --git a/cpan/Text-TEI-Collate/t/data/tei_no_ns.xml b/cpan/Text-TEI-Collate/t/data/tei_no_ns.xml
@@ -0,0 +1,34 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<?xml-stylesheet type="text/css" href="format.css" ?>
+<!DOCTYPE TEI PUBLIC "-//TEI//DTD TEI P5//EN" "tei.dtd" >
+<TEI>
+  <teiHeader>
+    <fileDesc>
+      <titleStmt>
+        <title>JK test_A.xml</title>
+        <author>JoelK</author>
+      </titleStmt>
+      <publicationStmt>
+        <p> </p>
+      </publicationStmt>
+      <notesStmt>
+        <note>Created by the <ref target="http://www.oeaw.ac.at/kvk/cte/">Classical Text
+            Editor</ref></note>
+      </notesStmt>
+    </fileDesc>
+    <encodingDesc>
+      <variantEncoding method="double-end-point" location="internal"/>
+      <tagsDecl>
+        <rendition xml:id="rd-Text" scheme="css"
+          >margin-left:0mm;margin-right:0mm;text-indent:8.5mm;line-height:14pt;margin-top:0pt;margin-bottom:0pt;text-align:justify;font-family:Times
+          New
+          Roman;font-size:11.5pt;color:#000000;font-style:normal;font-weight:normal;text-decoration:none;text-underline-position:0pt;text-decoration:none;text-transform:none;font-variant:normal;display:block;vertical-align:baseline;letter-spacing:0pt;direction:ltr;position:relative;top:0pt;</rendition>
+        <rendition xml:id="rd-N1" scheme="css"
+          >text-indent:0mm;line-height:12pt;font-size:9.5pt;</rendition>
+        <rendition xml:id="rd-A1" scheme="css"
+          >text-indent:0mm;line-height:12pt;font-size:9.5pt;</rendition>
+      </tagsDecl>
+    </encodingDesc>
+  </teiHeader>
+  <text xml:space="preserve" rendition="#rd-Text"><body><p rendition="#rd-Text">First paragraph</p><p rendition="#rd-Text">This is a test to try to figure out how the program can be constructed so as to generate an XML file (TEI-compliant) that respects the manuscript tradition.</p></body></text>
+</TEI>