whatpm/Whatpm/mkhtmlparser.pl

#!/usr/bin/perl 
use strict;

my $DEBUG = $ENV{DEBUG};

while (<>) {
  s/!!!emit\b/return /;
  s{!!!next-input-character;}{q{
      if (@{$self->{char}}) {
        $self->{next_char} = shift @{$self->{char}};
      } else {
        $self->{set_next_char}->($self);
      }
  }}ge;
  s{!!!back-next-input-character\b}{q{unshift @{$self->{char}}, }}ge;
  s{!!!nack\s*\(\s*'([^']+)'\s*\)\s*;}{
    ($DEBUG ? qq{
      if (\$self->{self_closing}) {
        !!!cp ('$1.2');
      } else {
        !!!cp ('$1.3');
      }
    } : '')
  }ge;
  s{!!!ack\s*(?>\([^)]*\)\s*)?;}{q{delete $self->{self_closing};}}ge;
  s{!!!ack-later\s*(?>\([^)]*\)\s*)?;}{}ge;
  s{!!!parse-error;}{q{$self->{parse_error}->();}}ge;
  s{!!!parse-error\s*\(}{
    q{$self->{parse_error}->(level => $self->{must_level}, }
  }ge;
  s{!!!insert-element-f\s*\(([^(),]+),([^(),]+)\)\s*;}{qq{
    {
      my \$el;
      !!!create-element (\$el, $1, $2\->{tag_name}, $2\->{attributes}, $2);
      \$insert->(\$el);
      push \@{\$self->{open_elements}}, [\$el, (\$el_category_f->{$1}->{$2\->{tag_name}} || 0) | FOREIGN_EL];
    }
  }}ge;
  s{!!!insert-element-t\s*\(([^(),]+),([^(),]+),([^(),]+)\)\s*;}{qq{
    {
      my \$el;
      !!!create-element (\$el, \$HTML_NS, $1, $2, $3);
      \$insert->(\$el);
      push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} || 0];
    }
  }}ge;
  s{!!!insert-element\s*\(([^(),]+),\s*,([^(),]+)\)\s*;}{qq{
    {
      my \$el;
      !!!create-element (\$el, \$HTML_NS, $1,, $2);
      \$self->{open_elements}->[-1]->[0]->append_child (\$el);
      push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} || 0];
    }
  }}ge;
  s{!!!insert-element\s*\(([^(),]+),([^(),]+),([^(),]+)\)\s*;}{qq{
    {
      my \$el;
      !!!create-element (\$el, \$HTML_NS, $1, $2, $3);
      \$self->{open_elements}->[-1]->[0]->append_child (\$el);
      push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} || 0];
    }
  }}ge;
  s{!!!create-element\s*\(([^(),]+),([^(),]+),([^(),]+)(?:,([^(),]*)(?>,([^(),]+))?)?\)\s*;}{
    my $l_var = $1;
    my $r = qq{
      $l_var = \$self->{document}->create_element_ns
        ($2, [undef, $3]);
    };
    if (defined $4 and length $4) {
      $r .= qq{
        for my \$attr_name (keys %{$4}) {
          my \$attr_t = $4\->{\$attr_name};
          my \$attr = \$self->{document}->create_attribute_ns
              (undef, [undef, \$attr_name]);
          \$attr->value (\$attr_t->{value});
          \$attr->set_user_data (manakai_source_line => \$attr_t->{line});
          \$attr->set_user_data (manakai_source_column => \$attr_t->{column});
          $l_var->set_attribute_node_ns (\$attr);
        }
      };
    }
    if (defined $5) {
      my $token_var = $5;
      $token_var =~ s/^\s+//;
      $token_var =~ s/\s+$//;
      $r .= qq{
        $l_var->set_user_data (manakai_source_line => $token_var\->{line})
            if defined $token_var\->{line};
        $l_var->set_user_data (manakai_source_column => $token_var\->{column})
            if defined $token_var\->{column};
      };
      ## TODO: In future version, it should be allowed for an application
      ## developer to choose whether these information should be kept
      ## for tracking or not for performance by some means.
    }
    $r;
  }ge; # MUST
  s{!!!next-token;}{q{$token = $self->_get_next_token;}}ge;
  s{!!!back-token;}{
    q{
      $token->{self_closing} = $self->{self_closing};
      unshift @{$self->{token}}, $token;
      delete $self->{self_closing};
    }
  }ge;
  s{!!!back-token\s*\(}{q{unshift @{$self->{token}}, (}}ge;
  s{!!!cp\s*\(\s*(\S+)\s*\)\s*;}{
    $DEBUG ? qq{
      #print STDERR "$1, ";
      \$Whatpm::HTML::Debug::cp_pass->($1) if \$Whatpm::HTML::Debug::cp_pass;
      BEGIN {
        \$Whatpm::HTML::Debug::cp->{$1} = 1;
      }
    } : ''
  }ge;
  print;
}
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4	wakaba	1.4	my $DEBUG = $ENV{DEBUG};
5
6	wakaba	1.1	while (<>) {
7			s/!!!emit\b/return /;
8			s{!!!next-input-character;}{q{
9			if (@{$self->{char}}) {
10	wakaba	1.3	$self->{next_char} = shift @{$self->{char}};
11	wakaba	1.1	} else {
12	wakaba	1.3	$self->{set_next_char}->($self);
13	wakaba	1.1	}
14			}}ge;
15			s{!!!back-next-input-character\b}{q{unshift @{$self->{char}}, }}ge;
16	wakaba	1.10	s{!!!nack\s\(\s'([^']+)'\s\)\s;}{
17			($DEBUG ? qq{
18			if (\$self->{self_closing}) {
19			!!!cp ('$1.2');
20			} else {
21			!!!cp ('$1.3');
22			}
23			} : '')
24			}ge;
25			s{!!!ack\s(?>\([^)]\)\s*)?;}{q{delete $self->{self_closing};}}ge;
26			s{!!!ack-later\s(?>\([^)]\)\s*)?;}{}ge;
27	wakaba	1.1	s{!!!parse-error;}{q{$self->{parse_error}->();}}ge;
28	wakaba	1.6	s{!!!parse-error\s*\(}{
29			q{$self->{parse_error}->(level => $self->{must_level}, }
30			}ge;
31	wakaba	1.11	s{!!!insert-element-f\s\(([^(),]+),([^(),]+)\)\s;}{qq{
32	wakaba	1.7	{
33			my \$el;
34	wakaba	1.11	!!!create-element (\$el, $1, $2\->{tag_name}, $2\->{attributes}, $2);
35	wakaba	1.7	\$insert->(\$el);
36	wakaba	1.11	push \@{\$self->{open_elements}}, [\$el, (\$el_category_f->{$1}->{$2\->{tag_name}} \|\| 0) \| FOREIGN_EL];
37	wakaba	1.7	}
38			}}ge;
39	wakaba	1.11	s{!!!insert-element-t\s\(([^(),]+),([^(),]+),([^(),]+)\)\s;}{qq{
40	wakaba	1.1	{
41			my \$el;
42	wakaba	1.11	!!!create-element (\$el, \$HTML_NS, $1, $2, $3);
43	wakaba	1.1	\$insert->(\$el);
44	wakaba	1.9	push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} \|\| 0];
45	wakaba	1.1	}
46			}}ge;
47	wakaba	1.7	s{!!!insert-element\s\(([^(),]+),\s,([^(),]+)\)\s*;}{qq{
48			{
49			my \$el;
50	wakaba	1.11	!!!create-element (\$el, \$HTML_NS, $1,, $2);
51	wakaba	1.7	\$self->{open_elements}->[-1]->[0]->append_child (\$el);
52	wakaba	1.9	push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} \|\| 0];
53	wakaba	1.7	}
54			}}ge;
55			s{!!!insert-element\s\(([^(),]+),([^(),]+),([^(),]+)\)\s;}{qq{
56			{
57			my \$el;
58	wakaba	1.11	!!!create-element (\$el, \$HTML_NS, $1, $2, $3);
59	wakaba	1.2	\$self->{open_elements}->[-1]->[0]->append_child (\$el);
60	wakaba	1.9	push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} \|\| 0];
61	wakaba	1.1	}
62			}}ge;
63	wakaba	1.11	s{!!!create-element\s\(([^(),]+),([^(),]+),([^(),]+)(?:,([^(),])(?>,([^(),]+))?)?\)\s*;}{
64	wakaba	1.7	my $l_var = $1;
65	wakaba	1.1	my $r = qq{
66	wakaba	1.7	$l_var = \$self->{document}->create_element_ns
67	wakaba	1.11	($2, [undef, $3]);
68	wakaba	1.1	};
69	wakaba	1.11	if (defined $4 and length $4) {
70	wakaba	1.1	$r .= qq{
71	wakaba	1.11	for my \$attr_name (keys %{$4}) {
72			my \$attr_t = $4\->{\$attr_name};
73	wakaba	1.8	my \$attr = \$self->{document}->create_attribute_ns
74			(undef, [undef, \$attr_name]);
75			\$attr->value (\$attr_t->{value});
76			\$attr->set_user_data (manakai_source_line => \$attr_t->{line});
77			\$attr->set_user_data (manakai_source_column => \$attr_t->{column});
78			$l_var->set_attribute_node_ns (\$attr);
79	wakaba	1.1	}
80			};
81	wakaba	1.7	}
82	wakaba	1.11	if (defined $5) {
83			my $token_var = $5;
84	wakaba	1.7	$token_var =~ s/^\s+//;
85			$token_var =~ s/\s+$//;
86			$r .= qq{
87			$l_var->set_user_data (manakai_source_line => $token_var\->{line})
88			if defined $token_var\->{line};
89			$l_var->set_user_data (manakai_source_column => $token_var\->{column})
90			if defined $token_var\->{column};
91			};
92			## TODO: In future version, it should be allowed for an application
93			## developer to choose whether these information should be kept
94			## for tracking or not for performance by some means.
95	wakaba	1.1	}
96			$r;
97			}ge; # MUST
98			s{!!!next-token;}{q{$token = $self->_get_next_token;}}ge;
99	wakaba	1.10	s{!!!back-token;}{
100			q{
101			$token->{self_closing} = $self->{self_closing};
102			unshift @{$self->{token}}, $token;
103			delete $self->{self_closing};
104			}
105			}ge;
106	wakaba	1.1	s{!!!back-token\s*\(}{q{unshift @{$self->{token}}, (}}ge;
107	wakaba	1.4	s{!!!cp\s\(\s(\S+)\s\)\s;}{
108			$DEBUG ? qq{
109	wakaba	1.5	#print STDERR "$1, ";
110	wakaba	1.4	\$Whatpm::HTML::Debug::cp_pass->($1) if \$Whatpm::HTML::Debug::cp_pass;
111			BEGIN {
112			\$Whatpm::HTML::Debug::cp->{$1} = 1;
113			}
114			} : ''
115			}ge;
116	wakaba	1.1	print;
117			}