src/stc/scintilla/src/LexErlang.cxx

   1 // Scintilla source code edit control
   2 /** @file LexErlang.cxx
   3  ** Lexer for Erlang.
   4  ** Written by Peter-Henry Mander, based on Matlab lexer by Jose' Fonseca
   5  **/
   6 // Copyright 1998-2001 by Neil Hodgson <neilh@scintilla.org>
   7 // The License.txt file describes the conditions under which this software may be distributed.
   8
   9 #include <stdlib.h>
  10 #include <string.h>
  11 #include <ctype.h>
  12 #include <stdio.h>
  13 #include <stdarg.h>
  14
  15 #include "Platform.h"
  16
  17 #include "PropSet.h"
  18 #include "Accessor.h"
  19 #include "StyleContext.h"
  20 #include "KeyWords.h"
  21 #include "Scintilla.h"
  22 #include "SciLexer.h"
  23
  24 /*
  25    TODO:
  26    o  _Param should be a new lexical type
  27 */
  28
  29 static int is_radix(int radix, int ch) {
  30    int digit;
  31    if ( 16 < radix || 2 > radix ) {
  32       return 0;
  33    }
  34    if ( isdigit(ch) ) {
  35       digit = ch - '0';
  36    } else if ( isxdigit(ch) ) {
  37       digit = toupper(ch) - 'A' + 10;
  38    } else {
  39       return 0;
  40    }
  41    if ( digit < radix ) {
  42       return 1;
  43    } else {
  44       return 0;
  45    }
  46 }
  47
  48 typedef enum {
  49    STATE_NULL,
  50    ATOM_UNQUOTED,
  51    ATOM_QUOTED,
  52    ATOM_FUN_NAME,
  53    NODE_NAME_UNQUOTED,
  54    NODE_NAME_QUOTED,
  55    MACRO_START,
  56    MACRO_UNQUOTED,
  57    MACRO_QUOTED,
  58    RECORD_START,
  59    RECORD_UNQUOTED,
  60    RECORD_QUOTED,
  61    NUMERAL_START,
  62    NUMERAL_SIGNED,
  63    NUMERAL_RADIX_LITERAL,
  64    NUMERAL_SPECULATIVE_MANTISSA,
  65    NUMERAL_FLOAT_MANTISSA,
  66    NUMERAL_FLOAT_EXPONENT,
  67    NUMERAL_FLOAT_SIGNED_EXPONENT,
  68    PARSE_ERROR
  69 } atom_parse_state_t;
  70
  71 static void ColouriseErlangDoc(unsigned int startPos, int length, int initStyle,
  72                                WordList *keywordlists[], Accessor &styler) {
  73
  74         WordList &keywords = *keywordlists[0];
  75
  76         styler.StartAt(startPos);
  77
  78         StyleContext sc(startPos, length, initStyle, styler);
  79    atom_parse_state_t parse_state = STATE_NULL;
  80    int radix_digits = 0;
  81    int exponent_digits = 0;
  82         for (; sc.More(); sc.Forward()) {
  83       if ( STATE_NULL != parse_state ) {
  84          switch (parse_state) {
  85          case STATE_NULL:
  86                                 sc.SetState(SCE_ERLANG_DEFAULT);
  87             break;
  88          case ATOM_UNQUOTED:
  89             if ( '@' == sc.ch ){
  90                parse_state = NODE_NAME_UNQUOTED;
  91             } else if ( !isalnum(sc.ch) && sc.ch != '_' ) {
  92                char s[100];
  93                sc.GetCurrent(s, sizeof(s));
  94                if (keywords.InList(s)) {
  95                   sc.ChangeState(SCE_ERLANG_KEYWORD);
  96                   sc.SetState(SCE_ERLANG_DEFAULT);
  97                   parse_state = STATE_NULL;
  98                } else {
  99                   if ( '/' == sc.ch ) {
 100                      parse_state = ATOM_FUN_NAME;
 101                   } else {
 102                      sc.ChangeState(SCE_ERLANG_ATOM);
 103                      sc.SetState(SCE_ERLANG_DEFAULT);
 104                      parse_state = STATE_NULL;
 105                   }
 106                }
 107             }
 108             break;
 109          case ATOM_QUOTED:
 110             if ( '@' == sc.ch ){
 111                parse_state = NODE_NAME_QUOTED;
 112             } else if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 113                sc.ChangeState(SCE_ERLANG_ATOM);
 114                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 115                parse_state = STATE_NULL;
 116             }
 117             break;
 118          case ATOM_FUN_NAME:
 119             if ( !isdigit(sc.ch) ) {
 120                sc.ChangeState(SCE_ERLANG_FUNCTION_NAME);
 121                sc.SetState(SCE_ERLANG_DEFAULT);
 122                parse_state = STATE_NULL;
 123             }
 124             break;
 125          case NODE_NAME_QUOTED:
 126             if ( '@' == sc.ch ) {
 127                sc.SetState(SCE_ERLANG_DEFAULT);
 128                parse_state = STATE_NULL;
 129             } else if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 130                sc.ChangeState(SCE_ERLANG_NODE_NAME);
 131                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 132                parse_state = STATE_NULL;
 133             }
 134             break;
 135          case NODE_NAME_UNQUOTED:
 136             if ( '@' == sc.ch ) {
 137                sc.SetState(SCE_ERLANG_DEFAULT);
 138                parse_state = STATE_NULL;
 139             } else if ( !isalnum(sc.ch) && sc.ch != '_' ) {
 140                sc.ChangeState(SCE_ERLANG_NODE_NAME);
 141                sc.SetState(SCE_ERLANG_DEFAULT);
 142                parse_state = STATE_NULL;
 143             }
 144             break;
 145          case RECORD_START:
 146             if ( '\'' == sc.ch ) {
 147                parse_state = RECORD_QUOTED;
 148             } else if (isalpha(sc.ch) && islower(sc.ch)) {
 149                parse_state = RECORD_UNQUOTED;
 150             } else { // error
 151                sc.SetState(SCE_ERLANG_DEFAULT);
 152                parse_state = STATE_NULL;
 153             }
 154             break;
 155          case RECORD_QUOTED:
 156             if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 157                sc.ChangeState(SCE_ERLANG_RECORD);
 158                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 159                parse_state = STATE_NULL;
 160             }
 161             break;
 162          case RECORD_UNQUOTED:
 163             if ( !isalpha(sc.ch) && '_' != sc.ch ) {
 164                sc.ChangeState(SCE_ERLANG_RECORD);
 165                sc.SetState(SCE_ERLANG_DEFAULT);
 166                parse_state = STATE_NULL;
 167             }
 168             break;
 169          case MACRO_START:
 170             if ( '\'' == sc.ch ) {
 171                parse_state = MACRO_QUOTED;
 172             } else if (isalpha(sc.ch)) {
 173                parse_state = MACRO_UNQUOTED;
 174             } else { // error
 175                sc.SetState(SCE_ERLANG_DEFAULT);
 176                parse_state = STATE_NULL;
 177             }
 178             break;
 179          case MACRO_UNQUOTED:
 180             if ( !isalpha(sc.ch) && '_' != sc.ch ) {
 181                sc.ChangeState(SCE_ERLANG_MACRO);
 182                sc.SetState(SCE_ERLANG_DEFAULT);
 183                parse_state = STATE_NULL;
 184             }
 185             break;
 186          case MACRO_QUOTED:
 187             if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 188                sc.ChangeState(SCE_ERLANG_MACRO);
 189                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 190                parse_state = STATE_NULL;
 191             }
 192             break;
 193          case NUMERAL_START:
 194             if ( isdigit(sc.ch) ) {
 195                radix_digits *= 10;
 196                radix_digits += sc.ch - '0'; // Assuming ASCII here!
 197             } else if ( '#' == sc.ch ) {
 198                if ( 2 > radix_digits || 16 < radix_digits) {
 199                   sc.SetState(SCE_ERLANG_DEFAULT);
 200                   parse_state = STATE_NULL;
 201                } else {
 202                   parse_state = NUMERAL_RADIX_LITERAL;
 203                }
 204             } else if ( '.' == sc.ch && isdigit(sc.chNext)) {
 205                radix_digits = 0;
 206                parse_state = NUMERAL_FLOAT_MANTISSA;
 207             } else if ( 'e' == sc.ch || 'E' == sc.ch ) {
 208                exponent_digits = 0;
 209                parse_state = NUMERAL_FLOAT_EXPONENT;
 210             } else {
 211                radix_digits = 0;
 212                sc.ChangeState(SCE_ERLANG_NUMBER);
 213                sc.SetState(SCE_ERLANG_DEFAULT);
 214                parse_state = STATE_NULL;
 215             }
 216             break;
 217          case NUMERAL_RADIX_LITERAL:
 218             if ( !is_radix(radix_digits,sc.ch) ) {
 219                radix_digits = 0;
 220                if ( !isalnum(sc.ch) ) {
 221                   sc.ChangeState(SCE_ERLANG_NUMBER);
 222                }
 223                sc.SetState(SCE_ERLANG_DEFAULT);
 224                parse_state = STATE_NULL;
 225             }
 226             break;
 227          case NUMERAL_FLOAT_MANTISSA:
 228             if ( 'e' == sc.ch || 'E' == sc.ch ) {
 229                exponent_digits = 0;
 230                parse_state = NUMERAL_FLOAT_EXPONENT;
 231             } else if ( !isdigit(sc.ch) ) {
 232                sc.ChangeState(SCE_ERLANG_NUMBER);
 233                sc.SetState(SCE_ERLANG_DEFAULT);
 234                parse_state = STATE_NULL;
 235             }
 236             break;
 237          case NUMERAL_FLOAT_EXPONENT:
 238             if ( '-' == sc.ch || '+' == sc.ch ) {
 239                parse_state = NUMERAL_FLOAT_SIGNED_EXPONENT;
 240             } else if ( !isdigit(sc.ch) ) {
 241                if ( 0 < exponent_digits ) {
 242                   sc.ChangeState(SCE_ERLANG_NUMBER);
 243                }
 244                sc.SetState(SCE_ERLANG_DEFAULT);
 245                parse_state = STATE_NULL;
 246             } else {
 247                ++exponent_digits;
 248             }
 249             break;
 250          case NUMERAL_FLOAT_SIGNED_EXPONENT:
 251             if ( !isdigit(sc.ch) ) {
 252                if ( 0 < exponent_digits ) {
 253                   sc.ChangeState(SCE_ERLANG_NUMBER);
 254                }
 255                sc.SetState(SCE_ERLANG_DEFAULT);
 256                parse_state = STATE_NULL;
 257             } else {
 258                ++exponent_digits;
 259             }
 260             break;
 261          case NUMERAL_SIGNED:
 262             if ( !isdigit(sc.ch) ) {
 263                sc.ChangeState(SCE_ERLANG_NUMBER);
 264                sc.SetState(SCE_ERLANG_DEFAULT);
 265                parse_state = STATE_NULL;
 266             } else if ( '.' == sc.ch ) {
 267                parse_state = NUMERAL_FLOAT_MANTISSA;
 268             }
 269             break;
 270          case NUMERAL_SPECULATIVE_MANTISSA:
 271             if ( !isdigit(sc.ch) ) {
 272                sc.ChangeState(SCE_ERLANG_OPERATOR);
 273                sc.SetState(SCE_ERLANG_DEFAULT);
 274                parse_state = STATE_NULL;
 275             } else {
 276                parse_state = NUMERAL_FLOAT_MANTISSA;
 277             }
 278             break;
 279          case PARSE_ERROR:
 280                                 sc.SetState(SCE_ERLANG_DEFAULT);
 281             parse_state = STATE_NULL;
 282             break;
 283          }
 284       } else if (sc.state == SCE_ERLANG_OPERATOR) {
 285                         if (sc.chPrev == '.') {
 286                                 if (sc.ch == '*' || sc.ch == '/' || sc.ch == '\\' || sc.ch == '^') {
 287                                         sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 288                                 } else if (sc.ch == '\'') {
 289                                         sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 290                                 } else {
 291                                         sc.SetState(SCE_ERLANG_DEFAULT);
 292                                 }
 293                         } else {
 294                                 sc.SetState(SCE_ERLANG_DEFAULT);
 295                         }
 296                 } else if (sc.state == SCE_ERLANG_VARIABLE) {
 297                         if (!isalnum(sc.ch) && sc.ch != '_') {
 298             sc.SetState(SCE_ERLANG_DEFAULT);
 299                         }
 300                 } else if (sc.state == SCE_ERLANG_STRING) {
 301                         if (sc.ch == '\"' && sc.chPrev != '\\') {
 302                                 sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 303                         }
 304                 } else if (sc.state == SCE_ERLANG_COMMENT ) {
 305                         if (sc.atLineEnd) {
 306                                 sc.SetState(SCE_ERLANG_DEFAULT);
 307                         }
 308       } else if (sc.state == SCE_ERLANG_CHARACTER ) {
 309          if ( sc.chPrev == '\\' ) {
 310             sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 311          } else if ( sc.ch != '\\' ) {
 312             sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 313          }
 314       }
 315
 316                 if (sc.state == SCE_ERLANG_DEFAULT) {
 317                         if (sc.ch == '%') {
 318                                 sc.SetState(SCE_ERLANG_COMMENT);
 319                         } else if (sc.ch == '\"') {
 320             sc.SetState(SCE_ERLANG_STRING);
 321          } else if (sc.ch == '#') {
 322             parse_state = RECORD_START;
 323                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 324          } else if (sc.ch == '?') {
 325             parse_state = MACRO_START;
 326                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 327          } else if (sc.ch == '$') {
 328                                 sc.SetState(SCE_ERLANG_CHARACTER);
 329          } else if (sc.ch == '\'') {
 330             parse_state = ATOM_QUOTED;
 331                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 332                         } else if ( isdigit(sc.ch) ) {
 333             parse_state = NUMERAL_START;
 334             radix_digits = sc.ch - '0';
 335                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 336          } else if ( '.' == sc.ch ) {
 337             parse_state = NUMERAL_SPECULATIVE_MANTISSA;
 338                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 339                         } else if (isalpha(sc.ch) && isupper(sc.ch)) {
 340                                 sc.SetState(SCE_ERLANG_VARIABLE);
 341                         } else if (isalpha(sc.ch)) {
 342             parse_state = ATOM_UNQUOTED;
 343                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 344                         } else if (isoperator(static_cast<char>(sc.ch)) || sc.ch == '\\') {
 345                                 sc.SetState(SCE_ERLANG_OPERATOR);
 346                         }
 347                 }
 348         }
 349         sc.Complete();
 350 }
 351
 352 static int ClassifyFoldPointErlang(
 353    Accessor &styler,
 354    int styleNext,
 355    int keyword_start
 356 ) {
 357         int lev = 0;
 358    if ( styler.Match(keyword_start,"case")
 359       || (
 360             styler.Match(keyword_start,"fun")
 361          && SCE_ERLANG_FUNCTION_NAME != styleNext)
 362       || styler.Match(keyword_start,"if")
 363       || styler.Match(keyword_start,"query")
 364       || styler.Match(keyword_start,"receive")
 365    ) {
 366       ++lev;
 367    } else if ( styler.Match(keyword_start,"end") ) {
 368       --lev;
 369    }
 370         return lev;
 371 }
 372
 373
 374 static void FoldErlangDoc(
 375    unsigned int startPos, int length, int initStyle,
 376    WordList** /*keywordlists*/, Accessor &styler
 377 ) {
 378         unsigned int endPos = startPos + length;
 379         //~ int visibleChars = 0;
 380         int lineCurrent = styler.GetLine(startPos);
 381         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 382         int levelCurrent = levelPrev;
 383         char chNext = styler.SafeGetCharAt(startPos);
 384         int styleNext = styler.StyleAt(startPos);
 385         int style = initStyle;
 386         int keyword_start = 0;
 387
 388    bool fold_keywords = true;
 389    bool fold_comments = true;
 390    bool fold_braces = true;
 391    bool fold_function_clauses = false;
 392    bool fold_clauses = false;
 393
 394    //int clause_level = 0;
 395
 396         for (unsigned int i = startPos; i < endPos; i++) {
 397                 char ch = chNext;
 398                 chNext = styler.SafeGetCharAt(i + 1);
 399                 int stylePrev = style;
 400                 style = styleNext;
 401                 styleNext = styler.StyleAt(i + 1);
 402                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 403
 404       if ( (stylePrev != SCE_ERLANG_KEYWORD) && (style == SCE_ERLANG_KEYWORD) ) {
 405          keyword_start = i;
 406       }
 407       if ( fold_keywords ) {
 408          if ( (stylePrev == SCE_ERLANG_KEYWORD)
 409             && (style != SCE_ERLANG_KEYWORD)
 410             && (style != SCE_ERLANG_ATOM)
 411          ) {
 412             levelCurrent += ClassifyFoldPointErlang(styler,styleNext,keyword_start);
 413          }
 414       }
 415
 416       if ( fold_comments ) {
 417          if (style == SCE_ERLANG_COMMENT) {
 418             if ((ch == '%') && (chNext == '{')) {
 419                levelCurrent++;
 420             } else if ((ch == '%') && (chNext == '}')) {
 421                levelCurrent--;
 422             }
 423          }
 424       }
 425
 426       if ( fold_function_clauses ) {
 427          if ( (SC_FOLDLEVELBASE == levelCurrent) /*&& (style == SCE_ERLANG_OPERATOR)*/ ) {
 428             if ( (ch == '-') && (chNext == '>')) {
 429                //~ fprintf(stderr,"levelCurrent=%d\n", levelCurrent);
 430                //++clause_level;
 431                //~ if ( 0 < clause_level )
 432                   ++levelCurrent;
 433             }
 434          }
 435          //~ if (  (stylePrev != SCE_ERLANG_RECORD)
 436             //~ && (style != SCE_ERLANG_NUMBER)
 437             //~ && (style != SCE_ERLANG_STRING)
 438             //~ && (style != SCE_ERLANG_COMMENT)
 439          //~ ) {
 440             if ( (SC_FOLDLEVELBASE+1 == levelCurrent) && (ch == '.') ) {
 441                //--clause_level;
 442                //~ if ( 0 == clause_level )
 443                   --levelCurrent;
 444             }
 445          //~ }
 446       }
 447
 448       if ( fold_clauses ) {
 449          if ( (0 < levelCurrent) && (style == SCE_ERLANG_OPERATOR) ) {
 450             if ((ch == '-') && (chNext == '>')) {
 451                levelCurrent++;
 452             }
 453             if ( (ch == ';') ) {
 454                levelCurrent--;
 455             }
 456          }
 457          if ( (stylePrev != SCE_ERLANG_RECORD)
 458             && (style != SCE_ERLANG_NUMBER)
 459             && (style != SCE_ERLANG_STRING)
 460             && (style != SCE_ERLANG_COMMENT)
 461          ) {
 462             if ( (ch == '.') ) {
 463                levelCurrent--;
 464             }
 465          }
 466          if (  (stylePrev == SCE_ERLANG_KEYWORD)
 467             && (style != SCE_ERLANG_KEYWORD)
 468             && (style != SCE_ERLANG_ATOM)
 469             && (
 470                styler.Match(keyword_start,"end") // 'end' counted twice if fold_keywords too
 471                || styler.Match(keyword_start,"after") )
 472          ) {
 473             levelCurrent--;
 474          }
 475       }
 476
 477       if ( fold_braces ) {
 478          if (style == SCE_ERLANG_OPERATOR) {
 479             if ( (ch == '{') || (ch == '(') || (ch == '[') ) {
 480                levelCurrent++;
 481             } else if ( (ch == '}') || (ch == ')') || (ch == ']') ) {
 482                levelCurrent--;
 483             }
 484          }
 485       }
 486
 487                 if (atEOL) {
 488                         int lev = levelPrev;
 489                         //~ if (visibleChars == 0 && foldCompact)
 490                                 //~ lev |= SC_FOLDLEVELWHITEFLAG;
 491                         //~ if ((levelCurrent > levelPrev) && (visibleChars > 0))
 492                         if ((levelCurrent > levelPrev)) {
 493                                 lev |= SC_FOLDLEVELHEADERFLAG;
 494          }
 495                         if (lev != styler.LevelAt(lineCurrent)) {
 496                                 styler.SetLevel(lineCurrent, lev);
 497                         }
 498                         lineCurrent++;
 499                         levelPrev = levelCurrent;
 500                         //~ visibleChars = 0;
 501                 }
 502                 //~ if (!isspacechar(ch))
 503                         //~ visibleChars++;
 504
 505         }
 506         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 507         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 508         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 509 }
 510
 511 static const char * const erlangWordListDesc[] = {
 512         "Keywords",
 513         0
 514 };
 515
 516 LexerModule lmErlang(
 517    SCLEX_ERLANG,
 518    ColouriseErlangDoc,
 519    "erlang",
 520    FoldErlangDoc,
 521    erlangWordListDesc);
 522