To check out this repository please hg clone the following URL, or open the URL using EasyMercurial or your preferred Mercurial client.
root / vendor / gems / coderay-0.9.7 / lib / coderay / scanners / html.rb @ 442:753f1380d6bc
History | View | Annotate | Download (4.15 KB)
| 1 |
module CodeRay |
|---|---|
| 2 |
module Scanners |
| 3 |
|
| 4 |
# HTML Scanner
|
| 5 |
class HTML < Scanner |
| 6 |
|
| 7 |
include Streamable
|
| 8 |
register_for :html
|
| 9 |
|
| 10 |
KINDS_NOT_LOC = [
|
| 11 |
:comment, :doctype, :preprocessor, |
| 12 |
:tag, :attribute_name, :operator, |
| 13 |
:attribute_value, :delimiter, :content, |
| 14 |
:plain, :entity, :error |
| 15 |
] |
| 16 |
|
| 17 |
ATTR_NAME = /[\w.:-]+/ |
| 18 |
ATTR_VALUE_UNQUOTED = ATTR_NAME |
| 19 |
TAG_END = /\/?>/ |
| 20 |
HEX = /[0-9a-fA-F]/ |
| 21 |
ENTITY = / |
| 22 |
&
|
| 23 |
(?:
|
| 24 |
\w+
|
| 25 |
|
|
| 26 |
\#
|
| 27 |
(?:
|
| 28 |
\d+
|
| 29 |
|
|
| 30 |
x#{HEX}+
|
| 31 |
)
|
| 32 |
)
|
| 33 |
;
|
| 34 |
/ox
|
| 35 |
|
| 36 |
PLAIN_STRING_CONTENT = {
|
| 37 |
"'" => /[^&'>\n]+/, |
| 38 |
'"' => /[^&">\n]+/, |
| 39 |
} |
| 40 |
|
| 41 |
def reset |
| 42 |
super
|
| 43 |
@state = :initial |
| 44 |
end
|
| 45 |
|
| 46 |
private |
| 47 |
def setup |
| 48 |
@state = :initial |
| 49 |
@plain_string_content = nil |
| 50 |
end
|
| 51 |
|
| 52 |
def scan_tokens tokens, options |
| 53 |
|
| 54 |
state = @state
|
| 55 |
plain_string_content = @plain_string_content
|
| 56 |
|
| 57 |
until eos?
|
| 58 |
|
| 59 |
kind = nil
|
| 60 |
match = nil
|
| 61 |
|
| 62 |
if scan(/\s+/m) |
| 63 |
kind = :space
|
| 64 |
|
| 65 |
else
|
| 66 |
|
| 67 |
case state
|
| 68 |
|
| 69 |
when :initial |
| 70 |
if scan(/<!--.*?-->/m) |
| 71 |
kind = :comment
|
| 72 |
elsif scan(/<!DOCTYPE.*?>/m) |
| 73 |
kind = :doctype
|
| 74 |
elsif scan(/<\?xml.*?\?>/m) |
| 75 |
kind = :preprocessor
|
| 76 |
elsif scan(/<\?.*?\?>|<%.*?%>/m) |
| 77 |
kind = :comment
|
| 78 |
elsif scan(/<\/[-\w.:]*>/m) |
| 79 |
kind = :tag
|
| 80 |
elsif match = scan(/<[-\w.:]+>?/m) |
| 81 |
kind = :tag
|
| 82 |
state = :attribute unless match[-1] == ?> |
| 83 |
elsif scan(/[^<>&]+/) |
| 84 |
kind = :plain
|
| 85 |
elsif scan(/#{ENTITY}/ox) |
| 86 |
kind = :entity
|
| 87 |
elsif scan(/[<>&]/) |
| 88 |
kind = :error
|
| 89 |
else
|
| 90 |
raise_inspect '[BUG] else-case reached with state %p' % [state], tokens
|
| 91 |
end
|
| 92 |
|
| 93 |
when :attribute |
| 94 |
if scan(/#{TAG_END}/o) |
| 95 |
kind = :tag
|
| 96 |
state = :initial
|
| 97 |
elsif scan(/#{ATTR_NAME}/o) |
| 98 |
kind = :attribute_name
|
| 99 |
state = :attribute_equal
|
| 100 |
else
|
| 101 |
kind = :error
|
| 102 |
getch |
| 103 |
end
|
| 104 |
|
| 105 |
when :attribute_equal |
| 106 |
if scan(/=/) |
| 107 |
kind = :operator
|
| 108 |
state = :attribute_value
|
| 109 |
elsif scan(/#{ATTR_NAME}/o) |
| 110 |
kind = :attribute_name
|
| 111 |
elsif scan(/#{TAG_END}/o) |
| 112 |
kind = :tag
|
| 113 |
state = :initial
|
| 114 |
elsif scan(/./) |
| 115 |
kind = :error
|
| 116 |
state = :attribute
|
| 117 |
end
|
| 118 |
|
| 119 |
when :attribute_value |
| 120 |
if scan(/#{ATTR_VALUE_UNQUOTED}/o) |
| 121 |
kind = :attribute_value
|
| 122 |
state = :attribute
|
| 123 |
elsif match = scan(/["']/) |
| 124 |
tokens << [:open, :string] |
| 125 |
state = :attribute_value_string
|
| 126 |
plain_string_content = PLAIN_STRING_CONTENT[match]
|
| 127 |
kind = :delimiter
|
| 128 |
elsif scan(/#{TAG_END}/o) |
| 129 |
kind = :tag
|
| 130 |
state = :initial
|
| 131 |
else
|
| 132 |
kind = :error
|
| 133 |
getch |
| 134 |
end
|
| 135 |
|
| 136 |
when :attribute_value_string |
| 137 |
if scan(plain_string_content)
|
| 138 |
kind = :content
|
| 139 |
elsif scan(/['"]/) |
| 140 |
tokens << [matched, :delimiter]
|
| 141 |
tokens << [:close, :string] |
| 142 |
state = :attribute
|
| 143 |
next
|
| 144 |
elsif scan(/#{ENTITY}/ox) |
| 145 |
kind = :entity
|
| 146 |
elsif scan(/&/) |
| 147 |
kind = :content
|
| 148 |
elsif scan(/[\n>]/) |
| 149 |
tokens << [:close, :string] |
| 150 |
kind = :error
|
| 151 |
state = :initial
|
| 152 |
end
|
| 153 |
|
| 154 |
else
|
| 155 |
raise_inspect 'Unknown state: %p' % [state], tokens
|
| 156 |
|
| 157 |
end
|
| 158 |
|
| 159 |
end
|
| 160 |
|
| 161 |
match ||= matched |
| 162 |
if $CODERAY_DEBUG and not kind |
| 163 |
raise_inspect 'Error token %p in line %d' %
|
| 164 |
[[match, kind], line], tokens, state |
| 165 |
end
|
| 166 |
raise_inspect 'Empty token', tokens unless match |
| 167 |
|
| 168 |
tokens << [match, kind] |
| 169 |
end
|
| 170 |
|
| 171 |
if options[:keep_state] |
| 172 |
@state = state
|
| 173 |
@plain_string_content = plain_string_content
|
| 174 |
end
|
| 175 |
|
| 176 |
tokens |
| 177 |
end
|
| 178 |
|
| 179 |
end
|
| 180 |
|
| 181 |
end
|
| 182 |
end
|