From: Tara L Andrews <tla@mit.edu>
Date: Thu, 30 Aug 2012 21:38:06 +0000 (+0200)
Subject: add exception handling
X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=commitdiff_plain;h=003113289b66dfdde424a3f5f9127c79fa9f5dda;p=scpubgit%2Fstemmatology.git

add exception handling
---

diff --git a/lib/Text/Tradition/Parser/CTE.pm b/lib/Text/Tradition/Parser/CTE.pm
index be6adfc..69acdb5 100644
--- a/lib/Text/Tradition/Parser/CTE.pm
+++ b/lib/Text/Tradition/Parser/CTE.pm
@@ -3,6 +3,7 @@ package Text::Tradition::Parser::CTE;
 use strict;
 use warnings;
 use Encode qw/ decode /;
+use Text::Tradition::Error;
 use Text::Tradition::Parser::Util qw/ collate_variants /;
 use XML::LibXML;
 use XML::LibXML::XPathContext;
@@ -131,7 +132,7 @@ sub _remove_formatting {
         warn "Could not find string or file option to parse";
         return;
     }
-    
+
     # Second, remove the formatting
 	my $xpc = XML::LibXML::XPathContext->new( $doc->documentElement );
 	my @useless = $xpc->findnodes( '//hi' );
@@ -147,8 +148,12 @@ sub _remove_formatting {
 	}
 	
 	# Third, write out and reparse to merge the text nodes.
-	my $result = decode( $doc->encoding, $doc->toString() );
+	my $enc = $doc->encoding || 'UTF-8';
+	my $result = decode( $enc, $doc->toString() );
 	my $tei = $parser->parse_string( $result )->documentElement;
+	unless( $tei->nodeName =~ /^tei(corpus)?$/i ) {
+		throw( "Parsed document has non-TEI root element " . $tei->nodeName );
+	}
 	$xpc = XML::LibXML::XPathContext->new( $tei );
 	return( $tei, $xpc );
 }
@@ -416,6 +421,13 @@ sub _add_wit_path {
     }
 }
 
+sub throw {
+	Text::Tradition::Error->throw( 
+		'ident' => 'Parser::CTE error',
+		'message' => $_[0],
+		);
+}
+
 =back
 
 =head1 LICENSE
diff --git a/lib/Text/Tradition/Parser/GraphML.pm b/lib/Text/Tradition/Parser/GraphML.pm
index a4d5717..63305ac 100644
--- a/lib/Text/Tradition/Parser/GraphML.pm
+++ b/lib/Text/Tradition/Parser/GraphML.pm
@@ -4,7 +4,7 @@ use strict;
 use warnings;
 use Exporter 'import';
 use vars qw/ @EXPORT_OK $xpc /;
-
+use Text::Tradition::Error;
 use XML::LibXML;
 use XML::LibXML::XPathContext;
 
@@ -74,9 +74,14 @@ sub graphml_parse {
             $edgedata->{$keyid} = $keyname;
         }
     }
+    
+    my @graph_elements = $xpc->findnodes( '/g:graphml/g:graph' );
+	unless( @graph_elements ) {
+		throw( "No graph elements found in graph XML - is this really GraphML?" );
+	}
 
     my @returned_graphs;
-    foreach my $graph_el ( $xpc->findnodes( '/g:graphml/g:graph' ) ) {
+    foreach my $graph_el ( @graph_elements ) {
         my $graph_hash = { 'nodes' => [],
 						   'edges' => [],
 						   'name'  => $graph_el->getAttribute( 'id' ) };
@@ -146,6 +151,13 @@ sub _lookup_node_data {
     return $data;
 }
     
+sub throw {
+	Text::Tradition::Error->throw( 
+		'ident' => 'Parser::GraphML error',
+		'message' => $_[0],
+		);
+}
+
 =head1 LICENSE
 
 This package is free software and is provided "as is" without express
diff --git a/lib/Text/Tradition/Parser/TEI.pm b/lib/Text/Tradition/Parser/TEI.pm
index 78ae542..8c2c3f0 100644
--- a/lib/Text/Tradition/Parser/TEI.pm
+++ b/lib/Text/Tradition/Parser/TEI.pm
@@ -2,6 +2,7 @@ package Text::Tradition::Parser::TEI;
 
 use strict;
 use warnings;
+use Text::Tradition::Error;
 use Text::Tradition::Parser::Util qw( collate_variants );
 use XML::LibXML;
 use XML::LibXML::XPathContext;
@@ -112,6 +113,9 @@ sub parse {
         return;
     }
     my $tei = $doc->documentElement();
+	unless( $tei->nodeName =~ /^tei(corpus)?$/i ) {
+		throw( "Parsed document has non-TEI root element " . $tei->nodeName );
+	}
     my $xpc = XML::LibXML::XPathContext->new( $tei );
     my $ns;
     if( $tei->namespaceURI ) {
@@ -482,6 +486,13 @@ sub _get_sigla {
 
 1;
 
+sub throw {
+	Text::Tradition::Error->throw( 
+		'ident' => 'Parser::TEI error',
+		'message' => $_[0],
+		);
+}
+
 =head1 BUGS / TODO
 
 =over