Changeset 72

Timestamp:

2007年07月09日 22時57分28秒 (1 year ago)

Author:

rintaro

Message:

Nenshi:

• HTMLFormFiller fix and testcase use Test::Base;
  
• Makefile.PL changed: use_test_base
  
• HTMLParser and test.
  

Location:

Nenshi/trunk

Files:

• Makefile.PL (modified) (1 diff)
• lib/Nenshi/Filter/HTMLFormFiller.pm (modified) (9 diffs)
• lib/Nenshi/Input.pm (modified) (4 diffs)
• t/12_input_xml.t (moved) (moved from Nenshi/trunk/t/12_input.t)
• t/41_builder.t (added)
• t/51_filter_fill.t (added)

Nenshi/trunk/Makefile.PL

@@ -26 +26 @@
 build_requires 'Test::More';
 
-#use_test_base;
+use_test_base;
 auto_include;
 WriteAll;

Nenshi/trunk/lib/Nenshi/Filter/HTMLFormFiller.pm

@@ -4 +4 @@
 use warnings;
 
-use Iterator::Simple qw(ifilter);
+use Iterator::Simple qw(ifilter iter);
 use List::Util qw(first);
 use Nenshi::Core qw(:eventkind Attrs QName);
 
+use overload 
+        '&{}' => sub { my $self = shift; sub { $self->__call__(@_) } };
 sub new {
         my $self = bless {}, shift;
@@ -14 +16 @@
         $self->{name} = $o{name};
         $self->{data} = $o{data} || {};
+        $self;
 }
 
@@ -31 +34 @@
 
                         if($tagname eq 'form' and (
-                                $self->{name} and $attrs->get('name') == $self->{name} or
-                                $self->{id} and $attrs->get('id') = $self->{id} or
+                                $self->{name} and $attrs->get('name') eq $self->{name} or
+                                $self->{id} and $attrs->get('id') eq $self->{id} or
                                 not($self->{id} or $self->{name}))
                         ) {
@@ -51 +54 @@
                                                                 }
                                                                 else {
-                                                                        $checked = scalar @{(grep { $_ } @$value)}
+                                                                        $checked = scalar (grep { $_ } @$value)
                                                                 }
                                                         }
@@ -63 +66 @@
                                                         }
                                                         if($checked) {
-                                                                $attrs |= [[checked => 'checked']];
+                                                                $attrs |= [[QName('checked') => 'checked']];
                                                         }
-                                                        elsif($attrs->has('checked') {
+                                                        elsif($attrs->has('checked')) {
                                                                 $attrs -= 'checked';
                                                         }
@@ -78 +81 @@
                                                         }
                                                         if(defined $value) {
-                                                                $attrs |= [[ value => "$value" ]];
+                                                                $attrs |= [[ QName('value'),  "$value" ]];
                                                         }
                                                 }
@@ -90 +93 @@
                                         }
                                 }
-                                elsif($tagname == 'textarea') {
+                                elsif($tagname eq 'textarea') {
                                         my $name = $attrs->get('name');
                                         if(exists $self->{data}{$name}) {
@@ -143 +146 @@
                                 my($tag, $attrs) = @$odata;
                                 if($selected) {
-                                        $attrs |= [[selected => 'selected']];
+                                        $attrs |= [[QName('selected') => 'selected']];
                                 }
                                 else {
@@ -163 +166 @@
                                 $in_textarea = 0;
                         }
-                        return iter([@buf, [$kind, $data], $pos]);
+                        return iter([@buf, [$kind, $data, $pos]]);
                 }
                 else {

Nenshi/trunk/lib/Nenshi/Input.pm

@@ -14 +14 @@
 use base qw(Exporter);
 
-our @EXPORT = qw(XML HTML XMLParser);
+our @EXPORT = qw(XML HTML XMLParser HTMLParser);
 
 sub XMLParser { Nenshi::XMLParser->new(@_); }
+sub HTMLParser { Nenshi::HTMLParser->new(@_); }
 sub XML { Stream(list(XMLParser(IO::String->new($_[0])))); }
 sub HTML { Stream(list(HTMLParser(IO::String->new(@_)))); }
@@ -165 +166 @@
                                                 $parser->parse_done;
                                                 undef $parser;
-                                                $self->{source}->close;
                                                 delete $self->{parser};
                                                 delete  $self->{source};
@@ -195 +195 @@
 }
 
+{
+        package Nenshi::HTMLParser;
+
+        use base qw(HTML::Parser);
+        use Iterator::Simple qw( iterator iter );
+        use Nenshi::Core qw( :eventkind Stream Attrs QName );
+        use Nenshi::Util qw( set );
+        use HTML::Entities;
+
+        use constant _EMPTY_ELEMS => set(
+                qw( area base basefont br col frame hr img input isindex link meta param )
+        );
+
+        sub init {
+                my($self, $source, $o) = @_;
+                $self->SUPER::init(
+                        api_version => 3,
+                        handlers => {
+                                start => ['handle_start', 'self,tagname,attr,attrseq,line,column'],
+                                end => ['handle_end', 'self,tagname,line,column'],
+                                text => ['handle_text', 'self,text,line,column'],
+                                comment => ['handle_comment', 'self,text,line,column'],
+                                declaration => ['handle_doctype', 'self,tokens,line,column'],
+                                process => ['handle_process', 'self,token0,line,column'],
+                        },
+                        unbroken_text => 1,
+                        attr_encoded => 1,
+                        #utf8_mode => 1,
+                );
+                $self->{source} = $source;
+                $self->{encoding} = $o->{encoding} || 'utf8';
+                $self->{filename} = $o->{filename};
+                $self->{_queue} = [];
+                $self->{_open_tags} = [];
+                $self;
+        }
+        
+        sub _enqueue {
+                my $self = shift;
+                push @{$self->{_queue}}, [ @_[0,1], [$self->{filename}, @{$_[2]}] ]
+        }
+
+        sub handle_start {
+                my $self = shift;
+                my($tag, $attr, $attrseq, $line, $col) = @_;
+                my @fixed_attr;
+                for my $name (@$attrseq) {
+                        my $val = $attr->{$name};
+                        if(not Encode::is_utf8($val)) {
+                                $val = Encode::decode($self->{encoding}, $val);
+                        }
+                        decode_entities($val);
+                        push @fixed_attr, [QName($name), $val],
+                }
+                $self->_enqueue(START_, [ QName($tag), Attrs(\@fixed_attr) ], [$line, $col] );
+                if(_EMPTY_ELEMS->contains($tag)) {
+                        $self->_enqueue(END_, QName($tag), [$line, $col]);
+                }
+                else {
+                        push @{$self->{_open_tags}}, $tag;
+                }
+        }
+
+        sub handle_end {
+                my $self = shift;
+                my($tag, $line, $col) = @_;
+                if(not _EMPTY_ELEMS->contains($tag)) {
+                        while(@{$self->{_open_tags}}) {
+                                my $open_tag = pop @{$self->{_open_tags}};
+                                $self->_enqueue(END_, QName($open_tag), [$line, $col]);
+                                last if $open_tag eq $tag;
+                        }
+                }
+        }
+
+        sub handle_text {
+                my $self = shift;
+                my($text, $line, $col) =@_;
+                if(not Encode::is_utf8($text)) {
+                        $text = Encode::decode($self->{encoding}, $text);
+                }
+                decode_entities($text);
+                $self->_enqueue(TEXT, $text, [$line, $col]);
+        }
+
+        sub handle_comment {
+                my $self = shift;
+                my($text, $line, $col) = @_;
+                $self->_enqueue(COMMENT, $text, [$line, $col]);
+        }
+
+        sub handle_process {
+                my $self = shift;
+                my($data, $line, $col) = @_;
+                my $target;
+                ($target, $data) = split(' ', $data, 2);
+                $data =~ s/\s*\??$//s; # strip last '?'
+                $self->_enqueue(PI, [$target, $data], [$line, $col]);
+        }
+
+        sub handle_declaration {
+                my $self = shift;
+                my($data, $line, $col);
+                warn $data;
+        }
+
+        sub parse {
+                my($self) = @_;
+                my $bufsize = 4096;
+                my $_queue = $self->{_queue};
+                my $finished = 0;
+
+                my $generator = iterator {
+                        while(1) {
+                                if(@$_queue) {
+                                        return shift @$_queue;
+                                }
+                                return if $finished;
+                                my $data;
+                                if($self->{source}->read($data, $bufsize)) {
+                                        $self->SUPER::parse($data);
+                                        next;
+                                }
+                                else {
+                                        $self->eof;
+                                        delete $self->{source};
+                                        $finished = 1;
+                                        next;
+                                }
+                                
+                                if($@) {
+                                        Nenshi::ParseError->throw("$@",$self->{filename},$self->current_line, $self->current_column);
+                                }
+                        }
+                        return;
+                };
+                return Stream($generator);
+        }
+
+        sub __iter__ {
+                my($self) = @_;
+                return iter($self->parse())
+        }
+}
 1;
+
 __END__
-        sub _coalesce {
+
+sub _coalesce {
                 my($stream) = @_;
                 my @textbuf;
@@ -232 +378 @@
 1;
 
-__END__
-package Nenshi::HTMLParser;
-{
-
-use base qw(HTML::Parser);
-use Nenshi::Stream qw( :eventkind );
-
-my %_EMPTY_ELEMS = (
-        map {$_ => 1} qw( area base basefont br col frame hr img input isindex link meta param )
-);
-
-sub new {
-        my $class = shift;
-        my $file = shift;
-        my $options = shift;
-
-        my $self = $self->SUPER::new();
-}
-
-sub _enqueue {
-        my $self = shift;
-        push @{$self->{_queue}}, [ @_[0,1], [$self->{filename}, $_[3][0], $_[3][1]] ]
-}
-
-sub handle_text {
-        my $self = shift;
-}
-
-sub handle_starttag {
-        my $self = shift;
-        my($tagname, $attr, $attrseq, $line, $pos) = @_;
-        my @fixed_attr;
-        for(@$attrseq) {
-                my $val = $attr->{$_}
-                if(defined $val) {
-                        $val = $_;
-                }
-                push @fixed_attr, [$_, $val];
-        }
-        $self->_enqueue(START, [ QName($tagname), Attrs(\@fixed_attr) ], [$line, $pos] );
-
-sub handle_end {
-}
-
-sub handle_declaration {
-}
-
-comment, process
-}
+