|
| 1 | +module CodeRay |
| 2 | +module Scanners |
| 3 | + |
| 4 | + # Scanner for Go, copy from c |
| 5 | + class Go < Scanner |
| 6 | + |
| 7 | + register_for :go |
| 8 | + file_extension 'go' |
| 9 | + |
| 10 | + # http://golang.org/ref/spec#Keywords |
| 11 | + KEYWORDS = [ |
| 12 | + 'break', 'default', 'func', 'interface', 'select', |
| 13 | + 'case', 'defer', 'go', 'map', 'struct', |
| 14 | + 'chan', 'else', 'goto', 'package', 'switch', |
| 15 | + 'const', 'fallthrough', 'if', 'range', 'type', |
| 16 | + 'continue', 'for', 'import', 'return', 'var', |
| 17 | + ] # :nodoc: |
| 18 | + |
| 19 | + # http://golang.org/ref/spec#Types |
| 20 | + PREDEFINED_TYPES = [ |
| 21 | + 'bool', |
| 22 | + 'uint8', 'uint16', 'uint32', 'uint64', |
| 23 | + 'int8', 'int16', 'int32', 'int64', |
| 24 | + 'float32', 'float64', |
| 25 | + 'complex64', 'complex128', |
| 26 | + 'byte', 'rune', 'string', 'error', |
| 27 | + 'uint', 'int', 'uintptr', |
| 28 | + ] # :nodoc: |
| 29 | + |
| 30 | + PREDEFINED_CONSTANTS = [ |
| 31 | + 'nil', 'iota', |
| 32 | + 'true', 'false', |
| 33 | + ] # :nodoc: |
| 34 | + |
| 35 | + PREDEFINED_FUNCTIONS = %w[ |
| 36 | + append cap close complex copy delete imag len |
| 37 | + make new panic print println real recover |
| 38 | + ] # :nodoc: |
| 39 | + |
| 40 | + IDENT_KIND = WordList.new(:ident). |
| 41 | + add(KEYWORDS, :keyword). |
| 42 | + add(PREDEFINED_TYPES, :predefined_type). |
| 43 | + add(PREDEFINED_CONSTANTS, :predefined_constant). |
| 44 | + add(PREDEFINED_FUNCTIONS, :predefined) # :nodoc: |
| 45 | + |
| 46 | + ESCAPE = / [rbfntv\n\\'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} /x # :nodoc: |
| 47 | + UNICODE_ESCAPE = / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x # :nodoc: |
| 48 | + |
| 49 | + protected |
| 50 | + |
| 51 | + def scan_tokens encoder, options |
| 52 | + |
| 53 | + state = :initial |
| 54 | + label_expected = true |
| 55 | + case_expected = false |
| 56 | + label_expected_before_preproc_line = nil |
| 57 | + in_preproc_line = false |
| 58 | + |
| 59 | + until eos? |
| 60 | + |
| 61 | + case state |
| 62 | + |
| 63 | + when :initial |
| 64 | + |
| 65 | + if match = scan(/ \s+ | \\\n /x) |
| 66 | + if in_preproc_line && match != "\\\n" && match.index(?\n) |
| 67 | + in_preproc_line = false |
| 68 | + case_expected = false |
| 69 | + label_expected = label_expected_before_preproc_line |
| 70 | + end |
| 71 | + encoder.text_token match, :space |
| 72 | + |
| 73 | + elsif match = scan(%r! // [^\n\\]* (?: \\. [^\n\\]* )* | /\* (?: .*? \*/ | .* ) !mx) |
| 74 | + encoder.text_token match, :comment |
| 75 | + |
| 76 | + elsif match = scan(/ <?- (?![\d.]) | [+*=<>?:;,!&^|()\[\]{}~%]+ | \/=? | \.(?!\d) /x) |
| 77 | + if case_expected |
| 78 | + label_expected = true if match == ':' |
| 79 | + case_expected = false |
| 80 | + end |
| 81 | + encoder.text_token match, :operator |
| 82 | + |
| 83 | + elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x) |
| 84 | + kind = IDENT_KIND[match] |
| 85 | + if kind == :ident && label_expected && !in_preproc_line && scan(/:(?!:)/) |
| 86 | + kind = :label |
| 87 | + label_expected = false |
| 88 | + match << matched |
| 89 | + else |
| 90 | + label_expected = false |
| 91 | + if kind == :keyword |
| 92 | + case match |
| 93 | + when 'case', 'default' |
| 94 | + case_expected = true |
| 95 | + end |
| 96 | + end |
| 97 | + end |
| 98 | + encoder.text_token match, kind |
| 99 | + |
| 100 | + elsif match = scan(/L?"/) |
| 101 | + encoder.begin_group :string |
| 102 | + if match[0] == ?L |
| 103 | + encoder.text_token 'L', :modifier |
| 104 | + match = '"' |
| 105 | + end |
| 106 | + encoder.text_token match, :delimiter |
| 107 | + state = :string |
| 108 | + |
| 109 | + elsif match = scan(/ ` ([^`]+)? (`)? /x) |
| 110 | + encoder.begin_group :shell |
| 111 | + encoder.text_token '`', :delimiter |
| 112 | + encoder.text_token self[1], :content if self[1] |
| 113 | + encoder.text_token self[2], :delimiter if self[2] |
| 114 | + encoder.end_group :shell |
| 115 | + |
| 116 | + elsif match = scan(/ \# \s* if \s* 0 /x) |
| 117 | + match << scan_until(/ ^\# (?:elif|else|endif) .*? $ | \z /xm) unless eos? |
| 118 | + encoder.text_token match, :comment |
| 119 | + |
| 120 | + elsif match = scan(/#[ \t]*(\w*)/) |
| 121 | + encoder.text_token match, :preprocessor |
| 122 | + in_preproc_line = true |
| 123 | + label_expected_before_preproc_line = label_expected |
| 124 | + state = :include_expected if self[1] == 'include' |
| 125 | + |
| 126 | + elsif match = scan(/ L?' (?: [^\'\n\\] | \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) )? '? /ox) |
| 127 | + label_expected = false |
| 128 | + encoder.text_token match, :char |
| 129 | + |
| 130 | + elsif match = scan(/\$/) |
| 131 | + encoder.text_token match, :ident |
| 132 | + |
| 133 | + elsif match = scan(/-?\d*(\.\d*)?([eE][+-]?\d+)?i/) |
| 134 | + label_expected = false |
| 135 | + encoder.text_token match, :imaginary |
| 136 | + |
| 137 | + elsif match = scan(/-?0[xX][0-9A-Fa-f]+/) |
| 138 | + label_expected = false |
| 139 | + encoder.text_token match, :hex |
| 140 | + |
| 141 | + elsif match = scan(/-?(?:0[0-7]+)(?![89.eEfF])/) |
| 142 | + label_expected = false |
| 143 | + encoder.text_token match, :octal |
| 144 | + |
| 145 | + elsif match = scan(/-?(?:\d*\.\d+|\d+\.)(?:[eE][+-]?\d+)?|\d+[eE][+-]?\d+/) |
| 146 | + label_expected = false |
| 147 | + encoder.text_token match, :float |
| 148 | + |
| 149 | + elsif match = scan(/-?(?:\d+)(?![.eEfF])L?L?/) |
| 150 | + label_expected = false |
| 151 | + encoder.text_token match, :integer |
| 152 | + |
| 153 | + else |
| 154 | + encoder.text_token getch, :error |
| 155 | + |
| 156 | + end |
| 157 | + |
| 158 | + when :string |
| 159 | + if match = scan(/[^\\\n"]+/) |
| 160 | + encoder.text_token match, :content |
| 161 | + elsif match = scan(/"/) |
| 162 | + encoder.text_token match, :delimiter |
| 163 | + encoder.end_group :string |
| 164 | + state = :initial |
| 165 | + label_expected = false |
| 166 | + elsif match = scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox) |
| 167 | + encoder.text_token match, :char |
| 168 | + elsif match = scan(/ \\ /x) |
| 169 | + encoder.text_token match, :error |
| 170 | + elsif match = scan(/$/) |
| 171 | + encoder.end_group :string |
| 172 | + state = :initial |
| 173 | + label_expected = false |
| 174 | + else |
| 175 | + raise_inspect "else case \" reached; %p not handled." % peek(1), encoder |
| 176 | + end |
| 177 | + |
| 178 | + when :include_expected |
| 179 | + if match = scan(/<[^>\n]+>?|"[^"\n\\]*(?:\\.[^"\n\\]*)*"?/) |
| 180 | + encoder.text_token match, :include |
| 181 | + state = :initial |
| 182 | + |
| 183 | + elsif match = scan(/\s+/) |
| 184 | + encoder.text_token match, :space |
| 185 | + state = :initial if match.index ?\n |
| 186 | + |
| 187 | + else |
| 188 | + state = :initial |
| 189 | + |
| 190 | + end |
| 191 | + |
| 192 | + else |
| 193 | + raise_inspect 'Unknown state', encoder |
| 194 | + |
| 195 | + end |
| 196 | + |
| 197 | + end |
| 198 | + |
| 199 | + if state == :string |
| 200 | + encoder.end_group :string |
| 201 | + end |
| 202 | + |
| 203 | + encoder |
| 204 | + end |
| 205 | + |
| 206 | + end |
| 207 | + |
| 208 | +end |
| 209 | +end |
0 commit comments