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 / ruby.rb @ 442:753f1380d6bc
History | View | Annotate | Download (15 KB)
| 1 |
# encoding: utf-8
|
|---|---|
| 2 |
module CodeRay |
| 3 |
module Scanners |
| 4 |
|
| 5 |
# This scanner is really complex, since Ruby _is_ a complex language!
|
| 6 |
#
|
| 7 |
# It tries to highlight 100% of all common code,
|
| 8 |
# and 90% of strange codes.
|
| 9 |
#
|
| 10 |
# It is optimized for HTML highlighting, and is not very useful for
|
| 11 |
# parsing or pretty printing.
|
| 12 |
#
|
| 13 |
# For now, I think it's better than the scanners in VIM or Syntax, or
|
| 14 |
# any highlighter I was able to find, except Caleb's RubyLexer.
|
| 15 |
#
|
| 16 |
# I hope it's also better than the rdoc/irb lexer.
|
| 17 |
class Ruby < Scanner |
| 18 |
|
| 19 |
include Streamable
|
| 20 |
|
| 21 |
register_for :ruby
|
| 22 |
file_extension 'rb'
|
| 23 |
|
| 24 |
helper :patterns
|
| 25 |
|
| 26 |
if not defined? EncodingError |
| 27 |
EncodingError = Class.new Exception |
| 28 |
end
|
| 29 |
|
| 30 |
private |
| 31 |
def scan_tokens tokens, options |
| 32 |
if string.respond_to?(:encoding) |
| 33 |
unless string.encoding == Encoding::UTF_8 |
| 34 |
self.string = string.encode Encoding::UTF_8, |
| 35 |
:invalid => :replace, :undef => :replace, :replace => '?' |
| 36 |
end
|
| 37 |
unicode = false
|
| 38 |
else
|
| 39 |
unicode = exist?(/[^\x00-\x7f]/)
|
| 40 |
end
|
| 41 |
|
| 42 |
last_token_dot = false
|
| 43 |
value_expected = true
|
| 44 |
heredocs = nil
|
| 45 |
last_state = nil
|
| 46 |
state = :initial
|
| 47 |
depth = nil
|
| 48 |
inline_block_stack = [] |
| 49 |
|
| 50 |
|
| 51 |
patterns = Patterns # avoid constant lookup |
| 52 |
|
| 53 |
until eos?
|
| 54 |
match = nil
|
| 55 |
kind = nil
|
| 56 |
|
| 57 |
if state.instance_of? patterns::StringState |
| 58 |
# {{{
|
| 59 |
match = scan_until(state.pattern) || scan_until(/\z/)
|
| 60 |
tokens << [match, :content] unless match.empty? |
| 61 |
break if eos? |
| 62 |
|
| 63 |
if state.heredoc and self[1] # end of heredoc |
| 64 |
match = getch.to_s |
| 65 |
match << scan_until(/$/) unless eos? |
| 66 |
tokens << [match, :delimiter]
|
| 67 |
tokens << [:close, state.type]
|
| 68 |
state = state.next_state |
| 69 |
next
|
| 70 |
end
|
| 71 |
|
| 72 |
case match = getch
|
| 73 |
|
| 74 |
when state.delim
|
| 75 |
if state.paren
|
| 76 |
state.paren_depth -= 1
|
| 77 |
if state.paren_depth > 0 |
| 78 |
tokens << [match, :nesting_delimiter]
|
| 79 |
next
|
| 80 |
end
|
| 81 |
end
|
| 82 |
tokens << [match, :delimiter]
|
| 83 |
if state.type == :regexp and not eos? |
| 84 |
modifiers = scan(/#{patterns::REGEXP_MODIFIERS}/ox)
|
| 85 |
tokens << [modifiers, :modifier] unless modifiers.empty? |
| 86 |
end
|
| 87 |
tokens << [:close, state.type]
|
| 88 |
value_expected = false
|
| 89 |
state = state.next_state |
| 90 |
|
| 91 |
when '\\' |
| 92 |
if state.interpreted
|
| 93 |
if esc = scan(/ #{patterns::ESCAPE} /ox) |
| 94 |
tokens << [match + esc, :char]
|
| 95 |
else
|
| 96 |
tokens << [match, :error]
|
| 97 |
end
|
| 98 |
else
|
| 99 |
case m = getch
|
| 100 |
when state.delim, '\\' |
| 101 |
tokens << [match + m, :char]
|
| 102 |
when nil |
| 103 |
tokens << [match, :error]
|
| 104 |
else
|
| 105 |
tokens << [match + m, :content]
|
| 106 |
end
|
| 107 |
end
|
| 108 |
|
| 109 |
when '#' |
| 110 |
case peek(1) |
| 111 |
when '{' |
| 112 |
inline_block_stack << [state, depth, heredocs] |
| 113 |
value_expected = true
|
| 114 |
state = :initial
|
| 115 |
depth = 1
|
| 116 |
tokens << [:open, :inline] |
| 117 |
tokens << [match + getch, :inline_delimiter]
|
| 118 |
when '$', '@' |
| 119 |
tokens << [match, :escape]
|
| 120 |
last_state = state # scan one token as normal code, then return here
|
| 121 |
state = :initial
|
| 122 |
else
|
| 123 |
raise_inspect 'else-case # reached; #%p not handled' % peek(1), tokens |
| 124 |
end
|
| 125 |
|
| 126 |
when state.paren
|
| 127 |
state.paren_depth += 1
|
| 128 |
tokens << [match, :nesting_delimiter]
|
| 129 |
|
| 130 |
when /#{patterns::REGEXP_SYMBOLS}/ox |
| 131 |
tokens << [match, :function]
|
| 132 |
|
| 133 |
else
|
| 134 |
raise_inspect 'else-case " reached; %p not handled, state = %p' % [match, state], tokens
|
| 135 |
|
| 136 |
end
|
| 137 |
next
|
| 138 |
# }}}
|
| 139 |
else
|
| 140 |
# {{{
|
| 141 |
if match = scan(/[ \t\f]+/) |
| 142 |
kind = :space
|
| 143 |
match << scan(/\s*/) unless eos? || heredocs |
| 144 |
value_expected = true if match.index(?\n) |
| 145 |
tokens << [match, kind] |
| 146 |
next
|
| 147 |
|
| 148 |
elsif match = scan(/\\?\n/) |
| 149 |
kind = :space
|
| 150 |
if match == "\n" |
| 151 |
value_expected = true
|
| 152 |
state = :initial if state == :undef_comma_expected |
| 153 |
end
|
| 154 |
if heredocs
|
| 155 |
unscan # heredoc scanning needs \n at start
|
| 156 |
state = heredocs.shift |
| 157 |
tokens << [:open, state.type]
|
| 158 |
heredocs = nil if heredocs.empty? |
| 159 |
next
|
| 160 |
else
|
| 161 |
match << scan(/\s*/) unless eos? |
| 162 |
end
|
| 163 |
tokens << [match, kind] |
| 164 |
next
|
| 165 |
|
| 166 |
elsif bol? && match = scan(/\#!.*/) |
| 167 |
tokens << [match, :doctype]
|
| 168 |
next
|
| 169 |
|
| 170 |
elsif match = scan(/\#.*/) or |
| 171 |
( bol? and match = scan(/#{patterns::RUBYDOC_OR_DATA}/o) ) |
| 172 |
kind = :comment
|
| 173 |
tokens << [match, kind] |
| 174 |
next
|
| 175 |
|
| 176 |
elsif state == :initial |
| 177 |
|
| 178 |
# IDENTS #
|
| 179 |
if match = scan(unicode ? /#{patterns::METHOD_NAME}/uo : |
| 180 |
/#{patterns::METHOD_NAME}/o)
|
| 181 |
if last_token_dot
|
| 182 |
kind = if match[/^[A-Z]/] and not match?(/\(/) then :constant else :ident end |
| 183 |
else
|
| 184 |
if value_expected != :expect_colon && scan(/:(?= )/) |
| 185 |
tokens << [match, :key]
|
| 186 |
match = ':'
|
| 187 |
kind = :operator
|
| 188 |
else
|
| 189 |
kind = patterns::IDENT_KIND[match]
|
| 190 |
if kind == :ident |
| 191 |
if match[/\A[A-Z]/] and not match[/[!?]$/] and not match?(/\(/) |
| 192 |
kind = :constant
|
| 193 |
end
|
| 194 |
elsif kind == :reserved |
| 195 |
state = patterns::DEF_NEW_STATE[match]
|
| 196 |
value_expected = :set if patterns::KEYWORDS_EXPECTING_VALUE[match] |
| 197 |
end
|
| 198 |
end
|
| 199 |
end
|
| 200 |
value_expected = :set if check(/#{patterns::VALUE_FOLLOWS}/o) |
| 201 |
|
| 202 |
elsif last_token_dot and match = scan(/#{patterns::METHOD_NAME_OPERATOR}|\(/o) |
| 203 |
kind = :ident
|
| 204 |
value_expected = :set if check(unicode ? /#{patterns::VALUE_FOLLOWS}/uo : |
| 205 |
/#{patterns::VALUE_FOLLOWS}/o)
|
| 206 |
|
| 207 |
# OPERATORS #
|
| 208 |
elsif not last_token_dot and match = scan(/ \.\.\.? | (?:\.|::)() | [,\(\)\[\]\{\}] | ==?=? /x) |
| 209 |
if match !~ / [.\)\]\}] /x or match =~ /\.\.\.?/ |
| 210 |
value_expected = :set
|
| 211 |
end
|
| 212 |
last_token_dot = :set if self[1] |
| 213 |
kind = :operator
|
| 214 |
unless inline_block_stack.empty?
|
| 215 |
case match
|
| 216 |
when '{' |
| 217 |
depth += 1
|
| 218 |
when '}' |
| 219 |
depth -= 1
|
| 220 |
if depth == 0 # closing brace of inline block reached |
| 221 |
state, depth, heredocs = inline_block_stack.pop |
| 222 |
heredocs = nil if heredocs && heredocs.empty? |
| 223 |
tokens << [match, :inline_delimiter]
|
| 224 |
kind = :inline
|
| 225 |
match = :close
|
| 226 |
end
|
| 227 |
end
|
| 228 |
end
|
| 229 |
|
| 230 |
elsif match = scan(/ ['"] /mx) |
| 231 |
tokens << [:open, :string] |
| 232 |
kind = :delimiter
|
| 233 |
state = patterns::StringState.new :string, match == '"', match # important for streaming |
| 234 |
|
| 235 |
elsif match = scan(unicode ? /#{patterns::INSTANCE_VARIABLE}/uo : |
| 236 |
/#{patterns::INSTANCE_VARIABLE}/o)
|
| 237 |
kind = :instance_variable
|
| 238 |
|
| 239 |
elsif value_expected and match = scan(/\//) |
| 240 |
tokens << [:open, :regexp] |
| 241 |
kind = :delimiter
|
| 242 |
interpreted = true
|
| 243 |
state = patterns::StringState.new :regexp, interpreted, match |
| 244 |
|
| 245 |
# elsif match = scan(/[-+]?#{patterns::NUMERIC}/o)
|
| 246 |
elsif match = value_expected ? scan(/[-+]?#{patterns::NUMERIC}/o) : scan(/#{patterns::NUMERIC}/o) |
| 247 |
kind = self[1] ? :float : :integer |
| 248 |
|
| 249 |
elsif match = scan(unicode ? /#{patterns::SYMBOL}/uo : |
| 250 |
/#{patterns::SYMBOL}/o)
|
| 251 |
case delim = match[1] |
| 252 |
when ?', ?" |
| 253 |
tokens << [:open, :symbol] |
| 254 |
tokens << [':', :symbol] |
| 255 |
match = delim.chr |
| 256 |
kind = :delimiter
|
| 257 |
state = patterns::StringState.new :symbol, delim == ?", match |
| 258 |
else
|
| 259 |
kind = :symbol
|
| 260 |
end
|
| 261 |
|
| 262 |
elsif match = scan(/ -[>=]? | [+!~^]=? | [*|&]{1,2}=? | >>? /x) |
| 263 |
value_expected = :set
|
| 264 |
kind = :operator
|
| 265 |
|
| 266 |
elsif value_expected and match = scan(unicode ? /#{patterns::HEREDOC_OPEN}/uo : |
| 267 |
/#{patterns::HEREDOC_OPEN}/o)
|
| 268 |
indented = self[1] == '-' |
| 269 |
quote = self[3] |
| 270 |
delim = self[quote ? 4 : 2] |
| 271 |
kind = patterns::QUOTE_TO_TYPE[quote]
|
| 272 |
tokens << [:open, kind]
|
| 273 |
tokens << [match, :delimiter]
|
| 274 |
match = :close
|
| 275 |
heredoc = patterns::StringState.new kind, quote != '\'', delim, (indented ? :indented : :linestart ) |
| 276 |
heredocs ||= [] # create heredocs if empty
|
| 277 |
heredocs << heredoc |
| 278 |
|
| 279 |
elsif value_expected and match = scan(/#{patterns::FANCY_START_CORRECT}/o) |
| 280 |
kind, interpreted = *patterns::FancyStringType.fetch(self[1]) do |
| 281 |
raise_inspect 'Unknown fancy string: %%%p' % k, tokens
|
| 282 |
end
|
| 283 |
tokens << [:open, kind]
|
| 284 |
state = patterns::StringState.new kind, interpreted, self[2] |
| 285 |
kind = :delimiter
|
| 286 |
|
| 287 |
elsif value_expected and match = scan(unicode ? /#{patterns::CHARACTER}/uo : |
| 288 |
/#{patterns::CHARACTER}/o)
|
| 289 |
kind = :integer
|
| 290 |
|
| 291 |
elsif match = scan(/ [\/%]=? | <(?:<|=>?)? | [?:;] /x) |
| 292 |
value_expected = :set
|
| 293 |
kind = :operator
|
| 294 |
|
| 295 |
elsif match = scan(/`/) |
| 296 |
if last_token_dot
|
| 297 |
kind = :operator
|
| 298 |
else
|
| 299 |
tokens << [:open, :shell] |
| 300 |
kind = :delimiter
|
| 301 |
state = patterns::StringState.new :shell, true, match |
| 302 |
end
|
| 303 |
|
| 304 |
elsif match = scan(unicode ? /#{patterns::GLOBAL_VARIABLE}/uo : |
| 305 |
/#{patterns::GLOBAL_VARIABLE}/o)
|
| 306 |
kind = :global_variable
|
| 307 |
|
| 308 |
elsif match = scan(unicode ? /#{patterns::CLASS_VARIABLE}/uo : |
| 309 |
/#{patterns::CLASS_VARIABLE}/o)
|
| 310 |
kind = :class_variable
|
| 311 |
|
| 312 |
else
|
| 313 |
if !unicode && !string.respond_to?(:encoding) |
| 314 |
# check for unicode
|
| 315 |
debug, $DEBUG = $DEBUG, false |
| 316 |
begin
|
| 317 |
if check(/./mu).size > 1 |
| 318 |
# seems like we should try again with unicode
|
| 319 |
unicode = true
|
| 320 |
end
|
| 321 |
rescue
|
| 322 |
# bad unicode char; use getch
|
| 323 |
ensure
|
| 324 |
$DEBUG = debug
|
| 325 |
end
|
| 326 |
next if unicode |
| 327 |
end
|
| 328 |
kind = :error
|
| 329 |
match = scan(unicode ? /./mu : /./m) |
| 330 |
|
| 331 |
end
|
| 332 |
|
| 333 |
elsif state == :def_expected |
| 334 |
state = :initial
|
| 335 |
if scan(/self\./) |
| 336 |
tokens << ['self', :pre_constant] |
| 337 |
tokens << ['.', :operator] |
| 338 |
end
|
| 339 |
if match = scan(unicode ? /(?>#{patterns::METHOD_NAME_EX})(?!\.|::)/uo : |
| 340 |
/(?>#{patterns::METHOD_NAME_EX})(?!\.|::)/o)
|
| 341 |
kind = :method
|
| 342 |
else
|
| 343 |
next
|
| 344 |
end
|
| 345 |
|
| 346 |
elsif state == :module_expected |
| 347 |
if match = scan(/<</) |
| 348 |
kind = :operator
|
| 349 |
else
|
| 350 |
state = :initial
|
| 351 |
if match = scan(unicode ? /(?:#{patterns::IDENT}::)*#{patterns::IDENT}/uo : |
| 352 |
/(?:#{patterns::IDENT}::)*#{patterns::IDENT}/o)
|
| 353 |
kind = :class
|
| 354 |
else
|
| 355 |
next
|
| 356 |
end
|
| 357 |
end
|
| 358 |
|
| 359 |
elsif state == :undef_expected |
| 360 |
state = :undef_comma_expected
|
| 361 |
if match = scan(unicode ? /#{patterns::METHOD_NAME_EX}/uo : |
| 362 |
/#{patterns::METHOD_NAME_EX}/o)
|
| 363 |
kind = :method
|
| 364 |
elsif match = scan(unicode ? /#{patterns::SYMBOL}/uo : |
| 365 |
/#{patterns::SYMBOL}/o)
|
| 366 |
case delim = match[1] |
| 367 |
when ?', ?" |
| 368 |
tokens << [:open, :symbol] |
| 369 |
tokens << [':', :symbol] |
| 370 |
match = delim.chr |
| 371 |
kind = :delimiter
|
| 372 |
state = patterns::StringState.new :symbol, delim == ?", match |
| 373 |
state.next_state = :undef_comma_expected
|
| 374 |
else
|
| 375 |
kind = :symbol
|
| 376 |
end
|
| 377 |
else
|
| 378 |
state = :initial
|
| 379 |
next
|
| 380 |
end
|
| 381 |
|
| 382 |
elsif state == :alias_expected |
| 383 |
match = scan(unicode ? /(#{patterns::METHOD_NAME_OR_SYMBOL})([ \t]+)(#{patterns::METHOD_NAME_OR_SYMBOL})/uo :
|
| 384 |
/(#{patterns::METHOD_NAME_OR_SYMBOL})([ \t]+)(#{patterns::METHOD_NAME_OR_SYMBOL})/o)
|
| 385 |
|
| 386 |
if match
|
| 387 |
tokens << [self[1], (self[1][0] == ?: ? :symbol : :method)] |
| 388 |
tokens << [self[2], :space] |
| 389 |
tokens << [self[3], (self[3][0] == ?: ? :symbol : :method)] |
| 390 |
end
|
| 391 |
state = :initial
|
| 392 |
next
|
| 393 |
|
| 394 |
elsif state == :undef_comma_expected |
| 395 |
if match = scan(/,/) |
| 396 |
kind = :operator
|
| 397 |
state = :undef_expected
|
| 398 |
else
|
| 399 |
state = :initial
|
| 400 |
next
|
| 401 |
end
|
| 402 |
|
| 403 |
end
|
| 404 |
# }}}
|
| 405 |
|
| 406 |
unless kind == :error |
| 407 |
if value_expected = value_expected == :set |
| 408 |
value_expected = :expect_colon if match == '?' || match == 'when' |
| 409 |
end
|
| 410 |
last_token_dot = last_token_dot == :set
|
| 411 |
end
|
| 412 |
|
| 413 |
if $CODERAY_DEBUG and not kind |
| 414 |
raise_inspect 'Error token %p in line %d' %
|
| 415 |
[[match, kind], line], tokens, state |
| 416 |
end
|
| 417 |
raise_inspect 'Empty token', tokens unless match |
| 418 |
|
| 419 |
tokens << [match, kind] |
| 420 |
|
| 421 |
if last_state
|
| 422 |
state = last_state |
| 423 |
last_state = nil
|
| 424 |
end
|
| 425 |
end
|
| 426 |
end
|
| 427 |
|
| 428 |
inline_block_stack << [state] if state.is_a? patterns::StringState |
| 429 |
until inline_block_stack.empty?
|
| 430 |
this_block = inline_block_stack.pop |
| 431 |
tokens << [:close, :inline] if this_block.size > 1 |
| 432 |
state = this_block.first |
| 433 |
tokens << [:close, state.type]
|
| 434 |
end
|
| 435 |
|
| 436 |
tokens |
| 437 |
end
|
| 438 |
|
| 439 |
end
|
| 440 |
|
| 441 |
end
|
| 442 |
end
|
| 443 |
|
| 444 |
# vim:fdm=marker
|