whatpm/Whatpm/mkhtmlparser.pl

#!/usr/bin/perl 
use strict;

my $DEBUG = $ENV{DEBUG};

while (<>) {
  s/!!!emit\b/return /;
  s{!!!next-input-character;}{q{
    pop @{$self->{prev_char}};
    unshift @{$self->{prev_char}}, $self->{next_char};

    if ($self->{char_buffer_pos} < length $self->{char_buffer}) {
      $self->{line_prev} = $self->{line};
      $self->{column_prev} = $self->{column};
      $self->{column}++;
      $self->{next_char}
          = ord substr ($self->{char_buffer}, $self->{char_buffer_pos}++, 1);
    } else {
      $self->{set_next_char}->($self);
    }
  }}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{!!!insert-element-f\s*\(([^(),]+),([^(),]+),([^(),]+),([^(),]+)\)\s*;}{qq{
    {
      my \$el;
      !!!create-element (\$el, $1, $2, $3, $4);
      \$insert->(\$el);
      push \@{\$self->{open_elements}}, [\$el, (\$el_category_f->{$1}->{$2} || 0) | FOREIGN_EL];

      if ($3\->{xmlns} and $3\->{xmlns}->{value} ne ($1)) {
        !!!parse-error (type => 'bad namespace', token => $4);
## TODO: Error type documentation
      }
    }
  }}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*\(([^(),]+),\s*([^(),]+),([^(),]+)(?:,([^(),]*)(?>,([^(),]+))?)?\)\s*;}{
    my ($l_var, $nsuri, $lname, $attrs, $token_var) = ($1, $2, $3, $4, $5);
    $nsuri =~ s/^\s+//;
    $nsuri =~ s/\s+\z//;
    my $r = qq{
      $l_var = \$self->{document}->create_element_ns
        ($nsuri, [undef, $lname]);
    };
    if (defined $attrs and length $attrs) {
      my $attr_xname;
      if ($nsuri eq q<$HTML_NS>) {
        $attr_xname = q[undef, [undef, $attr_name]];
      } else {
        ## NOTE: "Adjust SVG attributes" (SVG only),
        ## "adjust MathML attributes" (MathML only), and
        ## "adjust foreign attributes".
        $attr_xname = qq[
          \@{
            \$foreign_attr_xname->{\$attr_name} ||
            [undef, [undef,
                     ($nsuri) eq \$SVG_NS ?
                         (\$svg_attr_name->{\$attr_name} || \$attr_name) :
                     ($nsuri) eq \$MML_NS ?
                         (\$attr_name eq 'definitionurl' ?
                             'definitionURL' : \$attr_name) :
                         \$attr_name]]
          }
        ];
      }
      $r .= qq{
        for my \$attr_name (keys %{$attrs}) {
          my \$attr_t = $attrs\->{\$attr_name};
          my \$attr = \$self->{document}->create_attribute_ns ($attr_xname);
          \$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 $token_var) {
      $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.
    }
    $r;
  }ge; # MUST
  s{!!!parse-error;}{q{$self->{parse_error}->();}}ge;
  s{!!!parse-error\s*\(}{
    q{$self->{parse_error}->(level => $self->{level}->{must}, }
  }ge;
  s{!!!next-token;}{q{$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 = $ENV{DEBUG};
5
6	wakaba	1.1	while (<>) {
7			s/!!!emit\b/return /;
8			s{!!!next-input-character;}{q{
9	wakaba	1.16	pop @{$self->{prev_char}};
10			unshift @{$self->{prev_char}}, $self->{next_char};
11
12			if ($self->{char_buffer_pos} < length $self->{char_buffer}) {
13			$self->{line_prev} = $self->{line};
14			$self->{column_prev} = $self->{column};
15			$self->{column}++;
16			$self->{next_char}
17			= ord substr ($self->{char_buffer}, $self->{char_buffer_pos}++, 1);
18			} else {
19			$self->{set_next_char}->($self);
20			}
21	wakaba	1.1	}}ge;
22	wakaba	1.10	s{!!!nack\s\(\s'([^']+)'\s\)\s;}{
23			($DEBUG ? qq{
24			if (\$self->{self_closing}) {
25			!!!cp ('$1.2');
26			} else {
27			!!!cp ('$1.3');
28			}
29			} : '')
30			}ge;
31			s{!!!ack\s(?>\([^)]\)\s*)?;}{q{delete $self->{self_closing};}}ge;
32			s{!!!ack-later\s(?>\([^)]\)\s*)?;}{}ge;
33	wakaba	1.12	s{!!!insert-element-f\s\(([^(),]+),([^(),]+),([^(),]+),([^(),]+)\)\s;}{qq{
34	wakaba	1.7	{
35			my \$el;
36	wakaba	1.12	!!!create-element (\$el, $1, $2, $3, $4);
37	wakaba	1.7	\$insert->(\$el);
38	wakaba	1.12	push \@{\$self->{open_elements}}, [\$el, (\$el_category_f->{$1}->{$2} \|\| 0) \| FOREIGN_EL];
39
40			if ($3\->{xmlns} and $3\->{xmlns}->{value} ne ($1)) {
41			!!!parse-error (type => 'bad namespace', token => $4);
42			## TODO: Error type documentation
43			}
44	wakaba	1.7	}
45			}}ge;
46	wakaba	1.11	s{!!!insert-element-t\s\(([^(),]+),([^(),]+),([^(),]+)\)\s;}{qq{
47	wakaba	1.1	{
48			my \$el;
49	wakaba	1.11	!!!create-element (\$el, \$HTML_NS, $1, $2, $3);
50	wakaba	1.1	\$insert->(\$el);
51	wakaba	1.9	push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} \|\| 0];
52	wakaba	1.1	}
53			}}ge;
54	wakaba	1.7	s{!!!insert-element\s\(([^(),]+),\s,([^(),]+)\)\s*;}{qq{
55			{
56			my \$el;
57	wakaba	1.11	!!!create-element (\$el, \$HTML_NS, $1,, $2);
58	wakaba	1.7	\$self->{open_elements}->[-1]->[0]->append_child (\$el);
59	wakaba	1.9	push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} \|\| 0];
60	wakaba	1.7	}
61			}}ge;
62			s{!!!insert-element\s\(([^(),]+),([^(),]+),([^(),]+)\)\s;}{qq{
63			{
64			my \$el;
65	wakaba	1.11	!!!create-element (\$el, \$HTML_NS, $1, $2, $3);
66	wakaba	1.2	\$self->{open_elements}->[-1]->[0]->append_child (\$el);
67	wakaba	1.9	push \@{\$self->{open_elements}}, [\$el, \$el_category->{$1} \|\| 0];
68	wakaba	1.1	}
69			}}ge;
70	wakaba	1.12	s{!!!create-element\s\(([^(),]+),\s([^(),]+),([^(),]+)(?:,([^(),])(?>,([^(),]+))?)?\)\s;}{
71			my ($l_var, $nsuri, $lname, $attrs, $token_var) = ($1, $2, $3, $4, $5);
72			$nsuri =~ s/^\s+//;
73			$nsuri =~ s/\s+\z//;
74	wakaba	1.1	my $r = qq{
75	wakaba	1.7	$l_var = \$self->{document}->create_element_ns
76	wakaba	1.12	($nsuri, [undef, $lname]);
77	wakaba	1.1	};
78	wakaba	1.12	if (defined $attrs and length $attrs) {
79			my $attr_xname;
80			if ($nsuri eq q<$HTML_NS>) {
81			$attr_xname = q[undef, [undef, $attr_name]];
82			} else {
83	wakaba	1.14	## NOTE: "Adjust SVG attributes" (SVG only),
84			## "adjust MathML attributes" (MathML only), and
85	wakaba	1.12	## "adjust foreign attributes".
86			$attr_xname = qq[
87			\@{
88			\$foreign_attr_xname->{\$attr_name} \|\|
89			[undef, [undef,
90	wakaba	1.14	($nsuri) eq \$SVG_NS ?
91	wakaba	1.12	(\$svg_attr_name->{\$attr_name} \|\| \$attr_name) :
92	wakaba	1.14	($nsuri) eq \$MML_NS ?
93			(\$attr_name eq 'definitionurl' ?
94			'definitionURL' : \$attr_name) :
95	wakaba	1.12	\$attr_name]]
96			}
97			];
98			}
99	wakaba	1.1	$r .= qq{
100	wakaba	1.12	for my \$attr_name (keys %{$attrs}) {
101			my \$attr_t = $attrs\->{\$attr_name};
102			my \$attr = \$self->{document}->create_attribute_ns ($attr_xname);
103	wakaba	1.8	\$attr->value (\$attr_t->{value});
104			\$attr->set_user_data (manakai_source_line => \$attr_t->{line});
105			\$attr->set_user_data (manakai_source_column => \$attr_t->{column});
106			$l_var->set_attribute_node_ns (\$attr);
107	wakaba	1.1	}
108			};
109	wakaba	1.7	}
110	wakaba	1.12	if (defined $token_var) {
111	wakaba	1.7	$token_var =~ s/^\s+//;
112			$token_var =~ s/\s+$//;
113			$r .= qq{
114			$l_var->set_user_data (manakai_source_line => $token_var\->{line})
115			if defined $token_var\->{line};
116			$l_var->set_user_data (manakai_source_column => $token_var\->{column})
117			if defined $token_var\->{column};
118			};
119			## TODO: In future version, it should be allowed for an application
120			## developer to choose whether these information should be kept
121			## for tracking or not for performance by some means.
122	wakaba	1.1	}
123			$r;
124			}ge; # MUST
125	wakaba	1.12	s{!!!parse-error;}{q{$self->{parse_error}->();}}ge;
126			s{!!!parse-error\s*\(}{
127	wakaba	1.13	q{$self->{parse_error}->(level => $self->{level}->{must}, }
128	wakaba	1.12	}ge;
129	wakaba	1.1	s{!!!next-token;}{q{$token = $self->_get_next_token;}}ge;
130	wakaba	1.10	s{!!!back-token;}{
131			q{
132			$token->{self_closing} = $self->{self_closing};
133			unshift @{$self->{token}}, $token;
134			delete $self->{self_closing};
135			}
136			}ge;
137	wakaba	1.1	s{!!!back-token\s*\(}{q{unshift @{$self->{token}}, (}}ge;
138	wakaba	1.4	s{!!!cp\s\(\s(\S+)\s\)\s;}{
139			$DEBUG ? qq{
140	wakaba	1.5	#print STDERR "$1, ";
141	wakaba	1.4	\$Whatpm::HTML::Debug::cp_pass->($1) if \$Whatpm::HTML::Debug::cp_pass;
142			BEGIN {
143			\$Whatpm::HTML::Debug::cp->{$1} = 1;
144			}
145			} : ''
146			}ge;
147	wakaba	1.1	print;
148			}