syntax/c.jsf.in

   1 # $MirOS: contrib/code/jupp/syntax/c.jsf.in,v 1.8 2017/12/02 05:38:05 tg Exp $
   2 #-
   3 # JOE syntax highlight file for C and C++
   4
   5 # A (deterministic) state machine which performs lexical analysis of C.
   6 # (This is the "assembly language" of syntax highlighting.  A separate
   7 # program could be used to convert a regular expression NFA syntax into this
   8 # format).
   9
  10 # Each state begins with ':<name> <color-name>'
  11 # <color-name> is the color used for characters eaten by the state
  12 # (really a symbol for a user definable color).
  13
  14 # The first state defined is the initial state.
  15
  16 # Within a state, define transitions (jumps) to other states.  Each
  17 # jump has the form: <character-list> <target-state> [<option>s]
  18
  19 # There are two ways to specify <character-list>s, either * for any
  20 # character not otherwise specified, or a literal list of characters within
  21 # quotes (ranges and escape sequences allows).  When the next character
  22 # matches any in the list, a jump to the target-state is taken and the
  23 # character is eaten (we advance to the next character of the file to be
  24 # colored).
  25 #
  26 # The * transition should be the first transition specified in the state.
  27 #
  28 # There are several options:
  29 #   noeat       do not eat the character, instead feed it to the next state
  30 #               (this tends to make the states smaller, but be careful: you
  31 #               can make infinite loops).  'noeat' implies 'recolor=-1'.
  32 #
  33 #   recolor=-N  Recolor the past N characters with the color of the
  34 #               target-state.  For example once /* is recognized as the
  35 #               start of C comment, you want to color the /* with the C
  36 #               comment color.
  37 #
  38 #   buffer      start copying characters to a buffer, beginning with this
  39 #               one (it's ok to not terminate buffering with a matching
  40 #               'strings' option- the buffer is limited to leading 19
  41 #               characters).
  42 #
  43 #   strings     A list of strings follows.  If the buffer matches any of the
  44 #               given strings, a jump to the target-state in the string list
  45 #               is taken instead of the normal jump.
  46 #
  47 #   istrings    Same as strings, but case is ignored.
  48 #
  49 #   hold        Stop buffering string- a future 'strings' or 'istrings' will
  50 #               look at contents of buffer at this point.  Useful for distinguishing
  51 #               commands and function calls in some languages 'write 7' is a command
  52 #               'write (' is a function call- hold lets us stop at the space and delay
  53 #               the string lookup until the ( or 7.
  54 #
  55 #   The format of the string list is:
  56 #
  57 #      "string"   <target-state> [<options>s]
  58 #      "string"   <target-state> [<options>s]
  59 #      done
  60 #
  61 #   (all of the options above are allowed except "strings", "istrings" and "noeat".  noeat is
  62 #    always implied after a matched string).
  63 #
  64 # Weirdness: only states have colors, not transitions.  This means that you
  65 # sometimes have to make dummy states with '* next-state noeat' just to get
  66 # a color specification.
  67 #
  68 # Encoding: everything is handled as UTF-8 by octets, that is, you can have
  69 # individual octets change states (the colour of the last octet applies).
  70 # The JSF files themselves must accordingly be encoded in UTF-8. When a raw
  71 # octet or an invalid UTF-8 sequence is read, a number of \xFF octets are
  72 # matched on.
  73
  74
  75 # Define no. sync lines
  76 # You can say:
  77 # -200     means 200 lines
  78 # -        means always start parsing from beginning of file when we lose sync
  79 #          if nothing is specified, the default is -120
  80
  81 # Always reparse from beginning
  82 -
  83
  84
  85 # Define colors
  86 #
  87 # bold inverse blink dim underline
  88 # white cyan magenta blue yellow green red black
  89 # bg_white bg_cyan bg_magenta bg_blue bg_yellow bg_green bg_red bg_black
  90
  91 =Idle
  92 =Bad            bold red
  93 =Preproc        yellow
  94 =Define         bold blue
  95 =Comment        green
  96 =IncLocal       cyan
  97 =IncSystem      bold cyan
  98 =Constant       cyan
  99 =Escape         bold cyan
 100 =Type           bold
 101 =Keyword        bold
 102 =CppKeyword     bold
 103 =Brace          magenta
 104 =Control
 105
 106 :reset Idle
 107         *               first           noeat
 108         " \t"           reset
 109
 110 :first Idle
 111         *               idle            noeat
 112         "#"             pre             recolor=-1
 113
 114 :pre Preproc
 115         *               preproc         noeat
 116         " \t"           pre
 117         "a-z"           preident        recolor=-1 buffer
 118
 119 :preident Preproc
 120         *               preproc         noeat strings
 121         "define"        predef
 122         "include"       preinc
 123 done
 124         "a-z"           preident
 125
 126 :preinc Preproc
 127         *               preinc
 128         " \t"           preinc_ws
 129         "\n"            reset
 130
 131 :preinc_ws Preproc
 132         *               prebad          recolor=-1
 133         " \t"           preinc_ws
 134         "\""            preinc_local    recolor=-1
 135         "<"             preinc_system   recolor=-1
 136
 137 :preinc_local IncLocal
 138         *               preinc_local
 139         "\"\n"          reset
 140
 141 :preinc_system IncSystem
 142         *               preinc_system
 143         ">\n"           reset
 144
 145 :prebad Bad
 146         *               prebad
 147         "\n"            reset
 148
 149
 150 :predef Preproc
 151         *               predef
 152         " \t"           predef_ws
 153         "\n"            reset
 154
 155 :predef_ws Preproc
 156         *               prebad          recolor=-1
 157         " \t"           predef_ws
 158         "a-zA-Z0-9_"    predef_ident    recolor=-1
 159
 160 :predef_ident Define
 161         *               idle            noeat
 162         "a-zA-Z0-9_"    predef_ident
 163
 164
 165 :preproc Preproc
 166         *               preproc
 167         "\n"            reset
 168         "\\"            preproc_cont
 169         "/"             preproc_slash
 170
 171 :preproc_slash Preproc
 172         *               preproc         noeat
 173         "*"             comment         recolor=-2
 174         "/"             line_comment    recolor=-2
 175
 176 :preproc_cont Preproc
 177         *               preproc_cont
 178         "\n"            preproc
 179
 180 # All following states are for when we're not in a preprocessor line
 181
 182 :idle Idle
 183         *               idle
 184         "\n"            reset
 185         "/"             slash
 186         "0"             first_digit     recolor=-1
 187         "1-9"           decimal recolor=-1
 188         "."             maybe_float
 189         "\""            string          recolor=-1
 190         "'"             char            recolor=-1
 191         "a-zA-Z_"       ident           buffer
 192         "{}"            brace           recolor=-1
 193         ",:;=()><[]*&|!~+\-%^"  control         recolor=-1
 194
 195 :brace Brace
 196         *       idle    noeat
 197
 198 :control Control
 199         *       idle    noeat
 200
 201 :slash Idle
 202         *               idle            noeat recolor=-2        # Not sure about this
 203         "*"             comment         recolor=-2
 204         "/"             line_comment    recolor=-2
 205
 206 :comment Comment
 207         *               comment
 208         "*"             maybe_end_comment
 209
 210 :maybe_end_comment Comment
 211         *               comment
 212         "/"             idle
 213         "*"             maybe_end_comment
 214
 215 :line_comment Comment
 216         *               line_comment
 217         "\n"            reset
 218
 219 :first_digit Constant
 220         *               idle    noeat
 221         "xX"            hex
 222         "."             float
 223         "eE"            epart
 224         "0-7"           octal
 225         "89"            bad_number      recolor=-1
 226
 227 :bad_number Bad
 228         *               idle    noeat
 229         "0-9"           bad_number
 230
 231 :octal Constant
 232         *               idle    noeat
 233         "0-7"           octal
 234         "89"            bad_number      recolor=-1
 235
 236 :hex Constant
 237         *               idle    noeat
 238         "0-9A-Fa-f"     hex
 239
 240 :decimal Constant
 241         *               idle    noeat
 242         "0-9"           decimal
 243         "eE"            epart
 244         "."             float
 245
 246 :maybe_float Constant
 247         *               idle    recolor=-2      noeat
 248         "0-9"           float           recolor=-2
 249
 250 :float Constant
 251         *               idle    noeat
 252         "eE"            epart
 253         "0-9"           float
 254
 255 :epart Constant
 256         *               idle    noeat
 257         "0-9+\-"        enum
 258
 259 :enum Constant
 260         *               idle    noeat
 261         "0-9"           enum
 262
 263 :string Constant
 264         *               string
 265         "\""            idle
 266         "\\"            string_escape   recolor=-1
 267         "%"             string_control  recolor=-1
 268
 269 :string_escape Escape
 270         *               string
 271         "x"             string_hex1
 272         "0-7"           string_octal2
 273         "\n"            string          recolor=-2
 274
 275 :string_hex1 Escape
 276         *               string          noeat
 277         "0-9a-fA-F"     string_hex2
 278
 279 :string_hex2 Escape
 280         *               string          noeat
 281         "0-9a-fA-F"     string
 282
 283 :string_octal2 Escape
 284         *               string          noeat
 285         "0-7"           string_octal3
 286
 287 :string_octal3 Escape
 288         *               string          noeat
 289         "0-7"           string
 290
 291 :string_control Escape
 292         *               string_control
 293         "\""            reset
 294         "diouxXeEfFgGaAcspn%SC\n"       string
 295         "\\"            string_escape   recolor=-1
 296
 297 :char Constant
 298         *               char
 299         "\n"            reset
 300         "'"             idle
 301         "\\"            char_escape     recolor=-1
 302
 303 :char_escape    Escape
 304         *               char
 305         "x"             char_hex1
 306         "0-7"           char_octal2
 307         "\n"            char            recolor=-2
 308
 309 :char_hex1 Escape
 310         *               char            noeat
 311         "0-9a-fA-F"     char_hex2
 312
 313 :char_hex2 Escape
 314         *               char            noeat
 315         "0-9a-fA-F"     char
 316
 317 :char_octal2 Escape
 318         *               char            noeat
 319         "0-7"           char_octal3
 320
 321 :char_octal3 Escape
 322         *               char            noeat
 323         "0-7"           char
 324
 325 :ident Idle
 326         *               idle            noeat strings
 327         "int"           type
 328         "float"         type
 329         "long"          type
 330         "short"         type
 331         "char"          type
 332         "double"        type
 333         "signed"        type
 334         "unsigned"      type
 335         "void"          type
 336         "static"        type
 337         "register"      type
 338         "extern"        type
 339         "inline"        type
 340         "auto"          type
 341         "const"         type
 342         "volatile"      type
 343         "if"            kw
 344         "else"          kw
 345         "while"         kw
 346         "for"           kw
 347         "break"         kw
 348         "continue"      kw
 349         "do"            kw
 350         "case"          kw
 351         "default"       kw
 352         "switch"        kw
 353         "goto"          kw
 354         "struct"        kw
 355         "enum"          kw
 356         "return"        kw
 357         "sizeof"        kw
 358         "typedef"       kw
 359         "union"         kw
 360         "asm"           kw
 361 # C++ keywords
 362         "class"         cppkw
 363         "public"        cppkw
 364         "private"       cppkw
 365         "protected"     cppkw
 366         "friend"        cppkw
 367         "new"           cppkw
 368         "delete"        cppkw
 369         "bool"          cppkw
 370         "namespace"     cppkw
 371         "operator"      cppkw
 372         "template"      cppkw
 373         "catch"         cppkw
 374         "throw"         cppkw
 375         "try"           cppkw
 376         "typeof"        cppkw
 377 done
 378         "a-zA-Z0-9_"    ident
 379
 380 :type Type
 381         *       idle    noeat
 382
 383 :kw Keyword
 384         *       idle    noeat
 385
 386 :cppkw CppKeyword
 387         *       idle    noeat