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 #ifdef SCI_NAMESPACE
  25 using namespace Scintilla;
  26 #endif
  27
  28 /*
  29    TODO:
  30    o  _Param should be a new lexical type
  31 */
  32
  33 static int is_radix(int radix, int ch) {
  34    int digit;
  35    if ( 16 < radix || 2 > radix ) {
  36       return 0;
  37    }
  38    if ( isdigit(ch) ) {
  39       digit = ch - '0';
  40    } else if ( isxdigit(ch) ) {
  41       digit = toupper(ch) - 'A' + 10;
  42    } else {
  43       return 0;
  44    }
  45    if ( digit < radix ) {
  46       return 1;
  47    } else {
  48       return 0;
  49    }
  50 }
  51
  52 typedef enum {
  53    STATE_NULL,
  54    ATOM_UNQUOTED,
  55    ATOM_QUOTED,
  56    ATOM_FUN_NAME,
  57    NODE_NAME_UNQUOTED,
  58    NODE_NAME_QUOTED,
  59    MACRO_START,
  60    MACRO_UNQUOTED,
  61    MACRO_QUOTED,
  62    RECORD_START,
  63    RECORD_UNQUOTED,
  64    RECORD_QUOTED,
  65    NUMERAL_START,
  66    NUMERAL_SIGNED,
  67    NUMERAL_RADIX_LITERAL,
  68    NUMERAL_SPECULATIVE_MANTISSA,
  69    NUMERAL_FLOAT_MANTISSA,
  70    NUMERAL_FLOAT_EXPONENT,
  71    NUMERAL_FLOAT_SIGNED_EXPONENT,
  72    PARSE_ERROR
  73 } atom_parse_state_t;
  74
  75 static void ColouriseErlangDoc(unsigned int startPos, int length, int initStyle,
  76                                WordList *keywordlists[], Accessor &styler) {
  77
  78         WordList &keywords = *keywordlists[0];
  79
  80         styler.StartAt(startPos);
  81
  82         StyleContext sc(startPos, length, initStyle, styler);
  83    atom_parse_state_t parse_state = STATE_NULL;
  84    int radix_digits = 0;
  85    int exponent_digits = 0;
  86         for (; sc.More(); sc.Forward()) {
  87       if ( STATE_NULL != parse_state ) {
  88          switch (parse_state) {
  89          case STATE_NULL:
  90                                 sc.SetState(SCE_ERLANG_DEFAULT);
  91             break;
  92          case ATOM_UNQUOTED:
  93             if ( '@' == sc.ch ){
  94                parse_state = NODE_NAME_UNQUOTED;
  95             } else if ( !isalnum(sc.ch) && sc.ch != '_' ) {
  96                char s[100];
  97                sc.GetCurrent(s, sizeof(s));
  98                if (keywords.InList(s)) {
  99                   sc.ChangeState(SCE_ERLANG_KEYWORD);
 100                   sc.SetState(SCE_ERLANG_DEFAULT);
 101                   parse_state = STATE_NULL;
 102                } else {
 103                   if ( '/' == sc.ch ) {
 104                      parse_state = ATOM_FUN_NAME;
 105                   } else {
 106                      sc.ChangeState(SCE_ERLANG_ATOM);
 107                      sc.SetState(SCE_ERLANG_DEFAULT);
 108                      parse_state = STATE_NULL;
 109                   }
 110                }
 111             }
 112             break;
 113          case ATOM_QUOTED:
 114             if ( '@' == sc.ch ){
 115                parse_state = NODE_NAME_QUOTED;
 116             } else if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 117                sc.ChangeState(SCE_ERLANG_ATOM);
 118                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 119                parse_state = STATE_NULL;
 120             }
 121             break;
 122          case ATOM_FUN_NAME:
 123             if ( !isdigit(sc.ch) ) {
 124                sc.ChangeState(SCE_ERLANG_FUNCTION_NAME);
 125                sc.SetState(SCE_ERLANG_DEFAULT);
 126                parse_state = STATE_NULL;
 127             }
 128             break;
 129          case NODE_NAME_QUOTED:
 130             if ( '@' == sc.ch ) {
 131                sc.SetState(SCE_ERLANG_DEFAULT);
 132                parse_state = STATE_NULL;
 133             } else if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 134                sc.ChangeState(SCE_ERLANG_NODE_NAME);
 135                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 136                parse_state = STATE_NULL;
 137             }
 138             break;
 139          case NODE_NAME_UNQUOTED:
 140             if ( '@' == sc.ch ) {
 141                sc.SetState(SCE_ERLANG_DEFAULT);
 142                parse_state = STATE_NULL;
 143             } else if ( !isalnum(sc.ch) && sc.ch != '_' ) {
 144                sc.ChangeState(SCE_ERLANG_NODE_NAME);
 145                sc.SetState(SCE_ERLANG_DEFAULT);
 146                parse_state = STATE_NULL;
 147             }
 148             break;
 149          case RECORD_START:
 150             if ( '\'' == sc.ch ) {
 151                parse_state = RECORD_QUOTED;
 152             } else if (isalpha(sc.ch) && islower(sc.ch)) {
 153                parse_state = RECORD_UNQUOTED;
 154             } else { // error
 155                sc.SetState(SCE_ERLANG_DEFAULT);
 156                parse_state = STATE_NULL;
 157             }
 158             break;
 159          case RECORD_QUOTED:
 160             if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 161                sc.ChangeState(SCE_ERLANG_RECORD);
 162                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 163                parse_state = STATE_NULL;
 164             }
 165             break;
 166          case RECORD_UNQUOTED:
 167             if ( !isalpha(sc.ch) && '_' != sc.ch ) {
 168                sc.ChangeState(SCE_ERLANG_RECORD);
 169                sc.SetState(SCE_ERLANG_DEFAULT);
 170                parse_state = STATE_NULL;
 171             }
 172             break;
 173          case MACRO_START:
 174             if ( '\'' == sc.ch ) {
 175                parse_state = MACRO_QUOTED;
 176             } else if (isalpha(sc.ch)) {
 177                parse_state = MACRO_UNQUOTED;
 178             } else { // error
 179                sc.SetState(SCE_ERLANG_DEFAULT);
 180                parse_state = STATE_NULL;
 181             }
 182             break;
 183          case MACRO_UNQUOTED:
 184             if ( !isalpha(sc.ch) && '_' != sc.ch ) {
 185                sc.ChangeState(SCE_ERLANG_MACRO);
 186                sc.SetState(SCE_ERLANG_DEFAULT);
 187                parse_state = STATE_NULL;
 188             }
 189             break;
 190          case MACRO_QUOTED:
 191             if ( '\'' == sc.ch && '\\' != sc.chPrev ) {
 192                sc.ChangeState(SCE_ERLANG_MACRO);
 193                sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 194                parse_state = STATE_NULL;
 195             }
 196             break;
 197          case NUMERAL_START:
 198             if ( isdigit(sc.ch) ) {
 199                radix_digits *= 10;
 200                radix_digits += sc.ch - '0'; // Assuming ASCII here!
 201             } else if ( '#' == sc.ch ) {
 202                if ( 2 > radix_digits || 16 < radix_digits) {
 203                   sc.SetState(SCE_ERLANG_DEFAULT);
 204                   parse_state = STATE_NULL;
 205                } else {
 206                   parse_state = NUMERAL_RADIX_LITERAL;
 207                }
 208             } else if ( '.' == sc.ch && isdigit(sc.chNext)) {
 209                radix_digits = 0;
 210                parse_state = NUMERAL_FLOAT_MANTISSA;
 211             } else if ( 'e' == sc.ch || 'E' == sc.ch ) {
 212                exponent_digits = 0;
 213                parse_state = NUMERAL_FLOAT_EXPONENT;
 214             } else {
 215                radix_digits = 0;
 216                sc.ChangeState(SCE_ERLANG_NUMBER);
 217                sc.SetState(SCE_ERLANG_DEFAULT);
 218                parse_state = STATE_NULL;
 219             }
 220             break;
 221          case NUMERAL_RADIX_LITERAL:
 222             if ( !is_radix(radix_digits,sc.ch) ) {
 223                radix_digits = 0;
 224                if ( !isalnum(sc.ch) ) {
 225                   sc.ChangeState(SCE_ERLANG_NUMBER);
 226                }
 227                sc.SetState(SCE_ERLANG_DEFAULT);
 228                parse_state = STATE_NULL;
 229             }
 230             break;
 231          case NUMERAL_FLOAT_MANTISSA:
 232             if ( 'e' == sc.ch || 'E' == sc.ch ) {
 233                exponent_digits = 0;
 234                parse_state = NUMERAL_FLOAT_EXPONENT;
 235             } else if ( !isdigit(sc.ch) ) {
 236                sc.ChangeState(SCE_ERLANG_NUMBER);
 237                sc.SetState(SCE_ERLANG_DEFAULT);
 238                parse_state = STATE_NULL;
 239             }
 240             break;
 241          case NUMERAL_FLOAT_EXPONENT:
 242             if ( '-' == sc.ch || '+' == sc.ch ) {
 243                parse_state = NUMERAL_FLOAT_SIGNED_EXPONENT;
 244             } else if ( !isdigit(sc.ch) ) {
 245                if ( 0 < exponent_digits ) {
 246                   sc.ChangeState(SCE_ERLANG_NUMBER);
 247                }
 248                sc.SetState(SCE_ERLANG_DEFAULT);
 249                parse_state = STATE_NULL;
 250             } else {
 251                ++exponent_digits;
 252             }
 253             break;
 254          case NUMERAL_FLOAT_SIGNED_EXPONENT:
 255             if ( !isdigit(sc.ch) ) {
 256                if ( 0 < exponent_digits ) {
 257                   sc.ChangeState(SCE_ERLANG_NUMBER);
 258                }
 259                sc.SetState(SCE_ERLANG_DEFAULT);
 260                parse_state = STATE_NULL;
 261             } else {
 262                ++exponent_digits;
 263             }
 264             break;
 265          case NUMERAL_SIGNED:
 266             if ( !isdigit(sc.ch) ) {
 267                sc.ChangeState(SCE_ERLANG_NUMBER);
 268                sc.SetState(SCE_ERLANG_DEFAULT);
 269                parse_state = STATE_NULL;
 270             } else if ( '.' == sc.ch ) {
 271                parse_state = NUMERAL_FLOAT_MANTISSA;
 272             }
 273             break;
 274          case NUMERAL_SPECULATIVE_MANTISSA:
 275             if ( !isdigit(sc.ch) ) {
 276                sc.ChangeState(SCE_ERLANG_OPERATOR);
 277                sc.SetState(SCE_ERLANG_DEFAULT);
 278                parse_state = STATE_NULL;
 279             } else {
 280                parse_state = NUMERAL_FLOAT_MANTISSA;
 281             }
 282             break;
 283          case PARSE_ERROR:
 284                                 sc.SetState(SCE_ERLANG_DEFAULT);
 285             parse_state = STATE_NULL;
 286             break;
 287          }
 288       } else if (sc.state == SCE_ERLANG_OPERATOR) {
 289                         if (sc.chPrev == '.') {
 290                                 if (sc.ch == '*' || sc.ch == '/' || sc.ch == '\\' || sc.ch == '^') {
 291                                         sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 292                                 } else if (sc.ch == '\'') {
 293                                         sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 294                                 } else {
 295                                         sc.SetState(SCE_ERLANG_DEFAULT);
 296                                 }
 297                         } else {
 298                                 sc.SetState(SCE_ERLANG_DEFAULT);
 299                         }
 300                 } else if (sc.state == SCE_ERLANG_VARIABLE) {
 301                         if (!isalnum(sc.ch) && sc.ch != '_') {
 302             sc.SetState(SCE_ERLANG_DEFAULT);
 303                         }
 304                 } else if (sc.state == SCE_ERLANG_STRING) {
 305                         if (sc.ch == '\"' && sc.chPrev != '\\') {
 306                                 sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 307                         }
 308                 } else if (sc.state == SCE_ERLANG_COMMENT ) {
 309                         if (sc.atLineEnd) {
 310                                 sc.SetState(SCE_ERLANG_DEFAULT);
 311                         }
 312       } else if (sc.state == SCE_ERLANG_CHARACTER ) {
 313          if ( sc.chPrev == '\\' ) {
 314             sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 315          } else if ( sc.ch != '\\' ) {
 316             sc.ForwardSetState(SCE_ERLANG_DEFAULT);
 317          }
 318       }
 319
 320                 if (sc.state == SCE_ERLANG_DEFAULT) {
 321                         if (sc.ch == '%') {
 322                                 sc.SetState(SCE_ERLANG_COMMENT);
 323                         } else if (sc.ch == '\"') {
 324             sc.SetState(SCE_ERLANG_STRING);
 325          } else if (sc.ch == '#') {
 326             parse_state = RECORD_START;
 327                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 328          } else if (sc.ch == '?') {
 329             parse_state = MACRO_START;
 330                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 331          } else if (sc.ch == '$') {
 332                                 sc.SetState(SCE_ERLANG_CHARACTER);
 333          } else if (sc.ch == '\'') {
 334             parse_state = ATOM_QUOTED;
 335                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 336                         } else if ( isdigit(sc.ch) ) {
 337             parse_state = NUMERAL_START;
 338             radix_digits = sc.ch - '0';
 339                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 340          } else if ( '.' == sc.ch ) {
 341             parse_state = NUMERAL_SPECULATIVE_MANTISSA;
 342                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 343                         } else if (isalpha(sc.ch) && isupper(sc.ch)) {
 344                                 sc.SetState(SCE_ERLANG_VARIABLE);
 345                         } else if (isalpha(sc.ch)) {
 346             parse_state = ATOM_UNQUOTED;
 347                                 sc.SetState(SCE_ERLANG_UNKNOWN);
 348                         } else if (isoperator(static_cast<char>(sc.ch)) || sc.ch == '\\') {
 349                                 sc.SetState(SCE_ERLANG_OPERATOR);
 350                         }
 351                 }
 352         }
 353         sc.Complete();
 354 }
 355
 356 static int ClassifyFoldPointErlang(
 357    Accessor &styler,
 358    int styleNext,
 359    int keyword_start
 360 ) {
 361         int lev = 0;
 362    if ( styler.Match(keyword_start,"case")
 363       || (
 364             styler.Match(keyword_start,"fun")
 365          && SCE_ERLANG_FUNCTION_NAME != styleNext)
 366       || styler.Match(keyword_start,"if")
 367       || styler.Match(keyword_start,"query")
 368       || styler.Match(keyword_start,"receive")
 369    ) {
 370       ++lev;
 371    } else if ( styler.Match(keyword_start,"end") ) {
 372       --lev;
 373    }
 374         return lev;
 375 }
 376
 377
 378 static void FoldErlangDoc(
 379    unsigned int startPos, int length, int initStyle,
 380    WordList** /*keywordlists*/, Accessor &styler
 381 ) {
 382         unsigned int endPos = startPos + length;
 383         //~ int visibleChars = 0;
 384         int lineCurrent = styler.GetLine(startPos);
 385         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 386         int levelCurrent = levelPrev;
 387         char chNext = styler.SafeGetCharAt(startPos);
 388         int styleNext = styler.StyleAt(startPos);
 389         int style = initStyle;
 390         int keyword_start = 0;
 391
 392    bool fold_keywords = true;
 393    bool fold_comments = true;
 394    bool fold_braces = true;
 395    bool fold_function_clauses = false;
 396    bool fold_clauses = false;
 397
 398    //int clause_level = 0;
 399
 400         for (unsigned int i = startPos; i < endPos; i++) {
 401                 char ch = chNext;
 402                 chNext = styler.SafeGetCharAt(i + 1);
 403                 int stylePrev = style;
 404                 style = styleNext;
 405                 styleNext = styler.StyleAt(i + 1);
 406                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 407
 408       if ( (stylePrev != SCE_ERLANG_KEYWORD) && (style == SCE_ERLANG_KEYWORD) ) {
 409          keyword_start = i;
 410       }
 411       if ( fold_keywords ) {
 412          if ( (stylePrev == SCE_ERLANG_KEYWORD)
 413             && (style != SCE_ERLANG_KEYWORD)
 414             && (style != SCE_ERLANG_ATOM)
 415          ) {
 416             levelCurrent += ClassifyFoldPointErlang(styler,styleNext,keyword_start);
 417          }
 418       }
 419
 420       if ( fold_comments ) {
 421          if (style == SCE_ERLANG_COMMENT) {
 422             if ((ch == '%') && (chNext == '{')) {
 423                levelCurrent++;
 424             } else if ((ch == '%') && (chNext == '}')) {
 425                levelCurrent--;
 426             }
 427          }
 428       }
 429
 430       if ( fold_function_clauses ) {
 431          if ( (SC_FOLDLEVELBASE == levelCurrent) /*&& (style == SCE_ERLANG_OPERATOR)*/ ) {
 432             if ( (ch == '-') && (chNext == '>')) {
 433                //~ fprintf(stderr,"levelCurrent=%d\n", levelCurrent);
 434                //++clause_level;
 435                //~ if ( 0 < clause_level )
 436                   ++levelCurrent;
 437             }
 438          }
 439          //~ if (  (stylePrev != SCE_ERLANG_RECORD)
 440             //~ && (style != SCE_ERLANG_NUMBER)
 441             //~ && (style != SCE_ERLANG_STRING)
 442             //~ && (style != SCE_ERLANG_COMMENT)
 443          //~ ) {
 444             if ( (SC_FOLDLEVELBASE+1 == levelCurrent) && (ch == '.') ) {
 445                //--clause_level;
 446                //~ if ( 0 == clause_level )
 447                   --levelCurrent;
 448             }
 449          //~ }
 450       }
 451
 452       if ( fold_clauses ) {
 453          if ( (0 < levelCurrent) && (style == SCE_ERLANG_OPERATOR) ) {
 454             if ((ch == '-') && (chNext == '>')) {
 455                levelCurrent++;
 456             }
 457             if ( (ch == ';') ) {
 458                levelCurrent--;
 459             }
 460          }
 461          if ( (stylePrev != SCE_ERLANG_RECORD)
 462             && (style != SCE_ERLANG_NUMBER)
 463             && (style != SCE_ERLANG_STRING)
 464             && (style != SCE_ERLANG_COMMENT)
 465          ) {
 466             if ( (ch == '.') ) {
 467                levelCurrent--;
 468             }
 469          }
 470          if (  (stylePrev == SCE_ERLANG_KEYWORD)
 471             && (style != SCE_ERLANG_KEYWORD)
 472             && (style != SCE_ERLANG_ATOM)
 473             && (
 474                styler.Match(keyword_start,"end") // 'end' counted twice if fold_keywords too
 475                || styler.Match(keyword_start,"after") )
 476          ) {
 477             levelCurrent--;
 478          }
 479       }
 480
 481       if ( fold_braces ) {
 482          if (style == SCE_ERLANG_OPERATOR) {
 483             if ( (ch == '{') || (ch == '(') || (ch == '[') ) {
 484                levelCurrent++;
 485             } else if ( (ch == '}') || (ch == ')') || (ch == ']') ) {
 486                levelCurrent--;
 487             }
 488          }
 489       }
 490
 491                 if (atEOL) {
 492                         int lev = levelPrev;
 493                         //~ if (visibleChars == 0 && foldCompact)
 494                                 //~ lev |= SC_FOLDLEVELWHITEFLAG;
 495                         //~ if ((levelCurrent > levelPrev) && (visibleChars > 0))
 496                         if ((levelCurrent > levelPrev)) {
 497                                 lev |= SC_FOLDLEVELHEADERFLAG;
 498          }
 499                         if (lev != styler.LevelAt(lineCurrent)) {
 500                                 styler.SetLevel(lineCurrent, lev);
 501                         }
 502                         lineCurrent++;
 503                         levelPrev = levelCurrent;
 504                         //~ visibleChars = 0;
 505                 }
 506                 //~ if (!isspacechar(ch))
 507                         //~ visibleChars++;
 508
 509         }
 510         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 511         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 512         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 513 }
 514
 515 static const char * const erlangWordListDesc[] = {
 516         "Keywords",
 517         0
 518 };
 519
 520 LexerModule lmErlang(
 521    SCLEX_ERLANG,
 522    ColouriseErlangDoc,
 523    "erlang",
 524    FoldErlangDoc,
 525    erlangWordListDesc);
 526