To check out this repository please hg clone the following URL, or open the URL using EasyMercurial or your preferred Mercurial client.
root / .svn / pristine / d2 / d27acaba5e51d3baf547fd1e551a5deb117304a0.svn-base @ 1297:0a574315af3e
History | View | Annotate | Download (16.5 KB)
| 1 |
module CodeRay |
|---|---|
| 2 |
module Scanners |
| 3 |
|
| 4 |
# This scanner is really complex, since Ruby _is_ a complex language! |
| 5 |
# |
| 6 |
# It tries to highlight 100% of all common code, |
| 7 |
# and 90% of strange codes. |
| 8 |
# |
| 9 |
# It is optimized for HTML highlighting, and is not very useful for |
| 10 |
# parsing or pretty printing. |
| 11 |
class Ruby < Scanner |
| 12 |
|
| 13 |
register_for :ruby |
| 14 |
file_extension 'rb' |
| 15 |
|
| 16 |
autoload :Patterns, 'coderay/scanners/ruby/patterns' |
| 17 |
autoload :StringState, 'coderay/scanners/ruby/string_state' |
| 18 |
|
| 19 |
def interpreted_string_state |
| 20 |
StringState.new :string, true, '"' |
| 21 |
end |
| 22 |
|
| 23 |
protected |
| 24 |
|
| 25 |
def setup |
| 26 |
@state = :initial |
| 27 |
end |
| 28 |
|
| 29 |
def scan_tokens encoder, options |
| 30 |
state, heredocs = options[:state] || @state |
| 31 |
heredocs = heredocs.dup if heredocs.is_a?(Array) |
| 32 |
|
| 33 |
if state && state.instance_of?(StringState) |
| 34 |
encoder.begin_group state.type |
| 35 |
end |
| 36 |
|
| 37 |
last_state = nil |
| 38 |
|
| 39 |
method_call_expected = false |
| 40 |
value_expected = true |
| 41 |
|
| 42 |
inline_block_stack = nil |
| 43 |
inline_block_curly_depth = 0 |
| 44 |
|
| 45 |
if heredocs |
| 46 |
state = heredocs.shift |
| 47 |
encoder.begin_group state.type |
| 48 |
heredocs = nil if heredocs.empty? |
| 49 |
end |
| 50 |
|
| 51 |
# def_object_stack = nil |
| 52 |
# def_object_paren_depth = 0 |
| 53 |
|
| 54 |
patterns = Patterns # avoid constant lookup |
| 55 |
|
| 56 |
unicode = string.respond_to?(:encoding) && string.encoding.name == 'UTF-8' |
| 57 |
|
| 58 |
until eos? |
| 59 |
|
| 60 |
if state.instance_of? ::Symbol |
| 61 |
|
| 62 |
if match = scan(/[ \t\f\v]+/) |
| 63 |
encoder.text_token match, :space |
| 64 |
|
| 65 |
elsif match = scan(/\n/) |
| 66 |
if heredocs |
| 67 |
unscan # heredoc scanning needs \n at start |
| 68 |
state = heredocs.shift |
| 69 |
encoder.begin_group state.type |
| 70 |
heredocs = nil if heredocs.empty? |
| 71 |
else |
| 72 |
state = :initial if state == :undef_comma_expected |
| 73 |
encoder.text_token match, :space |
| 74 |
value_expected = true |
| 75 |
end |
| 76 |
|
| 77 |
elsif match = scan(bol? ? / \#(!)?.* | #{patterns::RUBYDOC_OR_DATA} /ox : /\#.*/)
|
| 78 |
encoder.text_token match, self[1] ? :doctype : :comment |
| 79 |
|
| 80 |
elsif match = scan(/\\\n/) |
| 81 |
if heredocs |
| 82 |
unscan # heredoc scanning needs \n at start |
| 83 |
encoder.text_token scan(/\\/), :space |
| 84 |
state = heredocs.shift |
| 85 |
encoder.begin_group state.type |
| 86 |
heredocs = nil if heredocs.empty? |
| 87 |
else |
| 88 |
encoder.text_token match, :space |
| 89 |
end |
| 90 |
|
| 91 |
elsif state == :initial |
| 92 |
|
| 93 |
# IDENTS # |
| 94 |
if !method_call_expected && |
| 95 |
match = scan(unicode ? /#{patterns::METHOD_NAME}/uo :
|
| 96 |
/#{patterns::METHOD_NAME}/o)
|
| 97 |
value_expected = false |
| 98 |
kind = patterns::IDENT_KIND[match] |
| 99 |
if kind == :ident |
| 100 |
if match[/\A[A-Z]/] && !(match[/[!?]$/] || match?(/\(/)) |
| 101 |
kind = :constant |
| 102 |
end |
| 103 |
elsif kind == :keyword |
| 104 |
state = patterns::KEYWORD_NEW_STATE[match] |
| 105 |
value_expected = true if patterns::KEYWORDS_EXPECTING_VALUE[match] |
| 106 |
end |
| 107 |
value_expected = true if !value_expected && check(/#{patterns::VALUE_FOLLOWS}/o)
|
| 108 |
encoder.text_token match, kind |
| 109 |
|
| 110 |
elsif method_call_expected && |
| 111 |
match = scan(unicode ? /#{patterns::METHOD_AFTER_DOT}/uo :
|
| 112 |
/#{patterns::METHOD_AFTER_DOT}/o)
|
| 113 |
if method_call_expected == '::' && match[/\A[A-Z]/] && !match?(/\(/) |
| 114 |
encoder.text_token match, :constant |
| 115 |
else |
| 116 |
encoder.text_token match, :ident |
| 117 |
end |
| 118 |
method_call_expected = false |
| 119 |
value_expected = check(/#{patterns::VALUE_FOLLOWS}/o)
|
| 120 |
|
| 121 |
# OPERATORS # |
| 122 |
elsif !method_call_expected && match = scan(/ (\.(?!\.)|::) | (?: \.\.\.? | ==?=? | [,\(\[\{] )() | [\)\]\}] /x)
|
| 123 |
method_call_expected = self[1] |
| 124 |
value_expected = !method_call_expected && self[2] |
| 125 |
if inline_block_stack |
| 126 |
case match |
| 127 |
when '{'
|
| 128 |
inline_block_curly_depth += 1 |
| 129 |
when '}' |
| 130 |
inline_block_curly_depth -= 1 |
| 131 |
if inline_block_curly_depth == 0 # closing brace of inline block reached |
| 132 |
state, inline_block_curly_depth, heredocs = inline_block_stack.pop |
| 133 |
inline_block_stack = nil if inline_block_stack.empty? |
| 134 |
heredocs = nil if heredocs && heredocs.empty? |
| 135 |
encoder.text_token match, :inline_delimiter |
| 136 |
encoder.end_group :inline |
| 137 |
next |
| 138 |
end |
| 139 |
end |
| 140 |
end |
| 141 |
encoder.text_token match, :operator |
| 142 |
|
| 143 |
elsif match = scan(unicode ? /#{patterns::SYMBOL}/uo :
|
| 144 |
/#{patterns::SYMBOL}/o)
|
| 145 |
case delim = match[1] |
| 146 |
when ?', ?" |
| 147 |
encoder.begin_group :symbol |
| 148 |
encoder.text_token ':', :symbol |
| 149 |
match = delim.chr |
| 150 |
encoder.text_token match, :delimiter |
| 151 |
state = self.class::StringState.new :symbol, delim == ?", match |
| 152 |
else |
| 153 |
encoder.text_token match, :symbol |
| 154 |
value_expected = false |
| 155 |
end |
| 156 |
|
| 157 |
elsif match = scan(/ ' (?:(?>[^'\\]*) ')? | " (?:(?>[^"\\\#]*) ")? /mx) |
| 158 |
encoder.begin_group :string |
| 159 |
if match.size == 1 |
| 160 |
encoder.text_token match, :delimiter |
| 161 |
state = self.class::StringState.new :string, match == '"', match # important for streaming |
| 162 |
else |
| 163 |
encoder.text_token match[0,1], :delimiter |
| 164 |
encoder.text_token match[1..-2], :content if match.size > 2 |
| 165 |
encoder.text_token match[-1,1], :delimiter |
| 166 |
encoder.end_group :string |
| 167 |
value_expected = false |
| 168 |
end |
| 169 |
|
| 170 |
elsif match = scan(unicode ? /#{patterns::INSTANCE_VARIABLE}/uo :
|
| 171 |
/#{patterns::INSTANCE_VARIABLE}/o)
|
| 172 |
value_expected = false |
| 173 |
encoder.text_token match, :instance_variable |
| 174 |
|
| 175 |
elsif value_expected && match = scan(/\//) |
| 176 |
encoder.begin_group :regexp |
| 177 |
encoder.text_token match, :delimiter |
| 178 |
state = self.class::StringState.new :regexp, true, '/' |
| 179 |
|
| 180 |
elsif match = scan(value_expected ? /[-+]?#{patterns::NUMERIC}/o : /#{patterns::NUMERIC}/o)
|
| 181 |
if method_call_expected |
| 182 |
encoder.text_token match, :error |
| 183 |
method_call_expected = false |
| 184 |
else |
| 185 |
encoder.text_token match, self[1] ? :float : :integer # TODO: send :hex/:octal/:binary |
| 186 |
end |
| 187 |
value_expected = false |
| 188 |
|
| 189 |
elsif match = scan(/ [-+!~^\/]=? | [:;] | [*|&]{1,2}=? | >>? /x)
|
| 190 |
value_expected = true |
| 191 |
encoder.text_token match, :operator |
| 192 |
|
| 193 |
elsif value_expected && match = scan(/#{patterns::HEREDOC_OPEN}/o)
|
| 194 |
quote = self[3] |
| 195 |
delim = self[quote ? 4 : 2] |
| 196 |
kind = patterns::QUOTE_TO_TYPE[quote] |
| 197 |
encoder.begin_group kind |
| 198 |
encoder.text_token match, :delimiter |
| 199 |
encoder.end_group kind |
| 200 |
heredocs ||= [] # create heredocs if empty |
| 201 |
heredocs << self.class::StringState.new(kind, quote != "'", delim, |
| 202 |
self[1] == '-' ? :indented : :linestart) |
| 203 |
value_expected = false |
| 204 |
|
| 205 |
elsif value_expected && match = scan(/#{patterns::FANCY_STRING_START}/o)
|
| 206 |
kind = patterns::FANCY_STRING_KIND[self[1]] |
| 207 |
encoder.begin_group kind |
| 208 |
state = self.class::StringState.new kind, patterns::FANCY_STRING_INTERPRETED[self[1]], self[2] |
| 209 |
encoder.text_token match, :delimiter |
| 210 |
|
| 211 |
elsif value_expected && match = scan(/#{patterns::CHARACTER}/o)
|
| 212 |
value_expected = false |
| 213 |
encoder.text_token match, :integer |
| 214 |
|
| 215 |
elsif match = scan(/ %=? | <(?:<|=>?)? | \? /x) |
| 216 |
value_expected = true |
| 217 |
encoder.text_token match, :operator |
| 218 |
|
| 219 |
elsif match = scan(/`/) |
| 220 |
encoder.begin_group :shell |
| 221 |
encoder.text_token match, :delimiter |
| 222 |
state = self.class::StringState.new :shell, true, match |
| 223 |
|
| 224 |
elsif match = scan(unicode ? /#{patterns::GLOBAL_VARIABLE}/uo :
|
| 225 |
/#{patterns::GLOBAL_VARIABLE}/o)
|
| 226 |
encoder.text_token match, :global_variable |
| 227 |
value_expected = false |
| 228 |
|
| 229 |
elsif match = scan(unicode ? /#{patterns::CLASS_VARIABLE}/uo :
|
| 230 |
/#{patterns::CLASS_VARIABLE}/o)
|
| 231 |
encoder.text_token match, :class_variable |
| 232 |
value_expected = false |
| 233 |
|
| 234 |
elsif match = scan(/\\\z/) |
| 235 |
encoder.text_token match, :space |
| 236 |
|
| 237 |
else |
| 238 |
if method_call_expected |
| 239 |
method_call_expected = false |
| 240 |
next |
| 241 |
end |
| 242 |
unless unicode |
| 243 |
# check for unicode |
| 244 |
$DEBUG_BEFORE, $DEBUG = $DEBUG, false |
| 245 |
begin |
| 246 |
if check(/./mu).size > 1 |
| 247 |
# seems like we should try again with unicode |
| 248 |
unicode = true |
| 249 |
end |
| 250 |
rescue |
| 251 |
# bad unicode char; use getch |
| 252 |
ensure |
| 253 |
$DEBUG = $DEBUG_BEFORE |
| 254 |
end |
| 255 |
next if unicode |
| 256 |
end |
| 257 |
|
| 258 |
encoder.text_token getch, :error |
| 259 |
|
| 260 |
end |
| 261 |
|
| 262 |
if last_state |
| 263 |
state = last_state |
| 264 |
last_state = nil |
| 265 |
end |
| 266 |
|
| 267 |
elsif state == :def_expected |
| 268 |
if match = scan(unicode ? /(?>#{patterns::METHOD_NAME_EX})(?!\.|::)/uo :
|
| 269 |
/(?>#{patterns::METHOD_NAME_EX})(?!\.|::)/o)
|
| 270 |
encoder.text_token match, :method |
| 271 |
state = :initial |
| 272 |
else |
| 273 |
last_state = :dot_expected |
| 274 |
state = :initial |
| 275 |
end |
| 276 |
|
| 277 |
elsif state == :dot_expected |
| 278 |
if match = scan(/\.|::/) |
| 279 |
# invalid definition |
| 280 |
state = :def_expected |
| 281 |
encoder.text_token match, :operator |
| 282 |
else |
| 283 |
state = :initial |
| 284 |
end |
| 285 |
|
| 286 |
elsif state == :module_expected |
| 287 |
if match = scan(/<</) |
| 288 |
encoder.text_token match, :operator |
| 289 |
else |
| 290 |
state = :initial |
| 291 |
if match = scan(unicode ? / (?:#{patterns::IDENT}::)* #{patterns::IDENT} /oux :
|
| 292 |
/ (?:#{patterns::IDENT}::)* #{patterns::IDENT} /ox)
|
| 293 |
encoder.text_token match, :class |
| 294 |
end |
| 295 |
end |
| 296 |
|
| 297 |
elsif state == :undef_expected |
| 298 |
state = :undef_comma_expected |
| 299 |
if match = scan(unicode ? /(?>#{patterns::METHOD_NAME_EX})(?!\.|::)/uo :
|
| 300 |
/(?>#{patterns::METHOD_NAME_EX})(?!\.|::)/o)
|
| 301 |
encoder.text_token match, :method |
| 302 |
elsif match = scan(/#{patterns::SYMBOL}/o)
|
| 303 |
case delim = match[1] |
| 304 |
when ?', ?" |
| 305 |
encoder.begin_group :symbol |
| 306 |
encoder.text_token ':', :symbol |
| 307 |
match = delim.chr |
| 308 |
encoder.text_token match, :delimiter |
| 309 |
state = self.class::StringState.new :symbol, delim == ?", match |
| 310 |
state.next_state = :undef_comma_expected |
| 311 |
else |
| 312 |
encoder.text_token match, :symbol |
| 313 |
end |
| 314 |
else |
| 315 |
state = :initial |
| 316 |
end |
| 317 |
|
| 318 |
elsif state == :undef_comma_expected |
| 319 |
if match = scan(/,/) |
| 320 |
encoder.text_token match, :operator |
| 321 |
state = :undef_expected |
| 322 |
else |
| 323 |
state = :initial |
| 324 |
end |
| 325 |
|
| 326 |
elsif state == :alias_expected |
| 327 |
match = scan(unicode ? /(#{patterns::METHOD_NAME_OR_SYMBOL})([ \t]+)(#{patterns::METHOD_NAME_OR_SYMBOL})/uo :
|
| 328 |
/(#{patterns::METHOD_NAME_OR_SYMBOL})([ \t]+)(#{patterns::METHOD_NAME_OR_SYMBOL})/o)
|
| 329 |
|
| 330 |
if match |
| 331 |
encoder.text_token self[1], (self[1][0] == ?: ? :symbol : :method) |
| 332 |
encoder.text_token self[2], :space |
| 333 |
encoder.text_token self[3], (self[3][0] == ?: ? :symbol : :method) |
| 334 |
end |
| 335 |
state = :initial |
| 336 |
|
| 337 |
else |
| 338 |
#:nocov: |
| 339 |
raise_inspect 'Unknown state: %p' % [state], encoder |
| 340 |
#:nocov: |
| 341 |
end |
| 342 |
|
| 343 |
else # StringState |
| 344 |
|
| 345 |
match = scan_until(state.pattern) || scan_rest |
| 346 |
unless match.empty? |
| 347 |
encoder.text_token match, :content |
| 348 |
break if eos? |
| 349 |
end |
| 350 |
|
| 351 |
if state.heredoc && self[1] # end of heredoc |
| 352 |
match = getch |
| 353 |
match << scan_until(/$/) unless eos? |
| 354 |
encoder.text_token match, :delimiter unless match.empty? |
| 355 |
encoder.end_group state.type |
| 356 |
state = state.next_state |
| 357 |
next |
| 358 |
end |
| 359 |
|
| 360 |
case match = getch |
| 361 |
|
| 362 |
when state.delim |
| 363 |
if state.paren_depth |
| 364 |
state.paren_depth -= 1 |
| 365 |
if state.paren_depth > 0 |
| 366 |
encoder.text_token match, :content |
| 367 |
next |
| 368 |
end |
| 369 |
end |
| 370 |
encoder.text_token match, :delimiter |
| 371 |
if state.type == :regexp && !eos? |
| 372 |
match = scan(/#{patterns::REGEXP_MODIFIERS}/o)
|
| 373 |
encoder.text_token match, :modifier unless match.empty? |
| 374 |
end |
| 375 |
encoder.end_group state.type |
| 376 |
value_expected = false |
| 377 |
state = state.next_state |
| 378 |
|
| 379 |
when '\\' |
| 380 |
if state.interpreted |
| 381 |
if esc = scan(/#{patterns::ESCAPE}/o)
|
| 382 |
encoder.text_token match + esc, :char |
| 383 |
else |
| 384 |
encoder.text_token match, :error |
| 385 |
end |
| 386 |
else |
| 387 |
case esc = getch |
| 388 |
when nil |
| 389 |
encoder.text_token match, :content |
| 390 |
when state.delim, '\\' |
| 391 |
encoder.text_token match + esc, :char |
| 392 |
else |
| 393 |
encoder.text_token match + esc, :content |
| 394 |
end |
| 395 |
end |
| 396 |
|
| 397 |
when '#' |
| 398 |
case peek(1) |
| 399 |
when '{'
|
| 400 |
inline_block_stack ||= [] |
| 401 |
inline_block_stack << [state, inline_block_curly_depth, heredocs] |
| 402 |
value_expected = true |
| 403 |
state = :initial |
| 404 |
inline_block_curly_depth = 1 |
| 405 |
encoder.begin_group :inline |
| 406 |
encoder.text_token match + getch, :inline_delimiter |
| 407 |
when '$', '@' |
| 408 |
encoder.text_token match, :escape |
| 409 |
last_state = state |
| 410 |
state = :initial |
| 411 |
else |
| 412 |
#:nocov: |
| 413 |
raise_inspect 'else-case # reached; #%p not handled' % [peek(1)], encoder |
| 414 |
#:nocov: |
| 415 |
end |
| 416 |
|
| 417 |
when state.opening_paren |
| 418 |
state.paren_depth += 1 |
| 419 |
encoder.text_token match, :content |
| 420 |
|
| 421 |
else |
| 422 |
#:nocov |
| 423 |
raise_inspect 'else-case " reached; %p not handled, state = %p' % [match, state], encoder |
| 424 |
#:nocov: |
| 425 |
|
| 426 |
end |
| 427 |
|
| 428 |
end |
| 429 |
|
| 430 |
end |
| 431 |
|
| 432 |
# cleaning up |
| 433 |
if state.is_a? StringState |
| 434 |
encoder.end_group state.type |
| 435 |
end |
| 436 |
|
| 437 |
if options[:keep_state] |
| 438 |
if state.is_a?(StringState) && state.heredoc |
| 439 |
(heredocs ||= []).unshift state |
| 440 |
state = :initial |
| 441 |
elsif heredocs && heredocs.empty? |
| 442 |
heredocs = nil |
| 443 |
end |
| 444 |
@state = state, heredocs |
| 445 |
end |
| 446 |
|
| 447 |
if inline_block_stack |
| 448 |
until inline_block_stack.empty? |
| 449 |
state, = *inline_block_stack.pop |
| 450 |
encoder.end_group :inline |
| 451 |
encoder.end_group state.type |
| 452 |
end |
| 453 |
end |
| 454 |
|
| 455 |
encoder |
| 456 |
end |
| 457 |
|
| 458 |
end |
| 459 |
|
| 460 |
end |
| 461 |
end |