scintilla/lexers/LexVHDL.cxx

   1 // Scintilla source code edit control
   2 /** @file LexVHDL.cxx
   3  ** Lexer for VHDL
   4  ** Written by Phil Reid,
   5  ** Based on:
   6  **  - The Verilog Lexer by Avi Yegudin
   7  **  - The Fortran Lexer by Chuan-jian Shen
   8  **  - The C++ lexer by Neil Hodgson
   9  **/
  10 // Copyright 1998-2002 by Neil Hodgson <neilh@scintilla.org>
  11 // The License.txt file describes the conditions under which this software may be distributed.
  12
  13 #include <stdlib.h>
  14 #include <string.h>
  15 #include <stdio.h>
  16 #include <stdarg.h>
  17 #include <assert.h>
  18 #include <ctype.h>
  19
  20 #include "ILexer.h"
  21 #include "Scintilla.h"
  22 #include "SciLexer.h"
  23
  24 #include "WordList.h"
  25 #include "LexAccessor.h"
  26 #include "Accessor.h"
  27 #include "StyleContext.h"
  28 #include "CharacterSet.h"
  29 #include "LexerModule.h"
  30
  31 #ifdef SCI_NAMESPACE
  32 using namespace Scintilla;
  33 #endif
  34
  35 static void ColouriseVHDLDoc(
  36   Sci_PositionU startPos,
  37   Sci_Position length,
  38   int initStyle,
  39   WordList *keywordlists[],
  40   Accessor &styler);
  41
  42
  43 /***************************************/
  44 static inline bool IsAWordChar(const int ch) {
  45   return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_' );
  46 }
  47
  48 /***************************************/
  49 static inline bool IsAWordStart(const int ch) {
  50   return (ch < 0x80) && (isalnum(ch) || ch == '_');
  51 }
  52
  53 /***************************************/
  54 static inline bool IsABlank(unsigned int ch) {
  55     return (ch == ' ') || (ch == 0x09) || (ch == 0x0b) ;
  56 }
  57
  58 /***************************************/
  59 static void ColouriseVHDLDoc(
  60   Sci_PositionU startPos,
  61   Sci_Position length,
  62   int initStyle,
  63   WordList *keywordlists[],
  64   Accessor &styler)
  65 {
  66   WordList &Keywords   = *keywordlists[0];
  67   WordList &Operators  = *keywordlists[1];
  68   WordList &Attributes = *keywordlists[2];
  69   WordList &Functions  = *keywordlists[3];
  70   WordList &Packages   = *keywordlists[4];
  71   WordList &Types      = *keywordlists[5];
  72   WordList &User       = *keywordlists[6];
  73
  74   StyleContext sc(startPos, length, initStyle, styler);
  75   bool isExtendedId = false;    // true when parsing an extended identifier
  76
  77   for (; sc.More(); sc.Forward())
  78   {
  79
  80     // Determine if the current state should terminate.
  81     if (sc.state == SCE_VHDL_OPERATOR) {
  82       sc.SetState(SCE_VHDL_DEFAULT);
  83     } else if (sc.state == SCE_VHDL_NUMBER) {
  84       if (!IsAWordChar(sc.ch) && (sc.ch != '#')) {
  85         sc.SetState(SCE_VHDL_DEFAULT);
  86       }
  87     } else if (sc.state == SCE_VHDL_IDENTIFIER) {
  88       if (!isExtendedId && (!IsAWordChar(sc.ch) || (sc.ch == '.'))) {
  89         char s[100];
  90         sc.GetCurrentLowered(s, sizeof(s));
  91         if (Keywords.InList(s)) {
  92           sc.ChangeState(SCE_VHDL_KEYWORD);
  93         } else if (Operators.InList(s)) {
  94           sc.ChangeState(SCE_VHDL_STDOPERATOR);
  95         } else if (Attributes.InList(s)) {
  96           sc.ChangeState(SCE_VHDL_ATTRIBUTE);
  97         } else if (Functions.InList(s)) {
  98           sc.ChangeState(SCE_VHDL_STDFUNCTION);
  99         } else if (Packages.InList(s)) {
 100           sc.ChangeState(SCE_VHDL_STDPACKAGE);
 101         } else if (Types.InList(s)) {
 102           sc.ChangeState(SCE_VHDL_STDTYPE);
 103         } else if (User.InList(s)) {
 104           sc.ChangeState(SCE_VHDL_USERWORD);
 105         }
 106         sc.SetState(SCE_VHDL_DEFAULT);
 107       } else if (isExtendedId && ((sc.ch == '\\') || sc.atLineEnd)) {
 108         // extended identifiers are terminated by backslash, check for end of line in case we have invalid syntax
 109         isExtendedId = false;
 110         sc.ForwardSetState(SCE_VHDL_DEFAULT);
 111       }
 112     } else if (sc.state == SCE_VHDL_COMMENT || sc.state == SCE_VHDL_COMMENTLINEBANG) {
 113       if (sc.atLineEnd) {
 114         sc.SetState(SCE_VHDL_DEFAULT);
 115       }
 116     } else if (sc.state == SCE_VHDL_STRING) {
 117       if (sc.ch == '\\') {
 118         if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 119           sc.Forward();
 120         }
 121       } else if (sc.ch == '\"') {
 122         sc.ForwardSetState(SCE_VHDL_DEFAULT);
 123       } else if (sc.atLineEnd) {
 124         sc.ChangeState(SCE_VHDL_STRINGEOL);
 125         sc.ForwardSetState(SCE_VHDL_DEFAULT);
 126       }
 127     } else if (sc.state == SCE_VHDL_BLOCK_COMMENT){
 128       if(sc.ch == '*' && sc.chNext == '/'){
 129         sc.Forward();
 130         sc.ForwardSetState(SCE_VHDL_DEFAULT);
 131       }
 132     }
 133
 134     // Determine if a new state should be entered.
 135     if (sc.state == SCE_VHDL_DEFAULT) {
 136       if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 137         sc.SetState(SCE_VHDL_NUMBER);
 138       } else if (IsAWordStart(sc.ch)) {
 139         sc.SetState(SCE_VHDL_IDENTIFIER);
 140       } else if (sc.Match('-', '-')) {
 141         if (sc.Match("--!"))  // Nice to have a different comment style
 142           sc.SetState(SCE_VHDL_COMMENTLINEBANG);
 143         else
 144           sc.SetState(SCE_VHDL_COMMENT);
 145       } else if (sc.Match('/', '*')){
 146         sc.SetState(SCE_VHDL_BLOCK_COMMENT);
 147       } else if (sc.ch == '\"') {
 148         sc.SetState(SCE_VHDL_STRING);
 149       } else if (sc.ch == '\\') {
 150         isExtendedId = true;
 151         sc.SetState(SCE_VHDL_IDENTIFIER);
 152       } else if (isoperator(static_cast<char>(sc.ch))) {
 153         sc.SetState(SCE_VHDL_OPERATOR);
 154       }
 155     }
 156   }
 157   sc.Complete();
 158 }
 159 //=============================================================================
 160 static bool IsCommentLine(Sci_Position line, Accessor &styler) {
 161         Sci_Position pos = styler.LineStart(line);
 162         Sci_Position eol_pos = styler.LineStart(line + 1) - 1;
 163         for (Sci_Position i = pos; i < eol_pos; i++) {
 164                 char ch = styler[i];
 165                 char chNext = styler[i+1];
 166                 if ((ch == '-') && (chNext == '-'))
 167                         return true;
 168                 else if (ch != ' ' && ch != '\t')
 169                         return false;
 170         }
 171         return false;
 172 }
 173 static bool IsCommentBlockStart(Sci_Position line, Accessor &styler)
 174 {
 175     Sci_Position pos = styler.LineStart(line);
 176         Sci_Position eol_pos = styler.LineStart(line + 1) - 1;
 177         for (Sci_Position i = pos; i < eol_pos; i++) {
 178                 char ch = styler[i];
 179                 char chNext = styler[i+1];
 180         char style = styler.StyleAt(i);
 181                 if ((style == SCE_VHDL_BLOCK_COMMENT) && (ch == '/') && (chNext == '*'))
 182                         return true;
 183         }
 184         return false;
 185 }
 186
 187 static bool IsCommentBlockEnd(Sci_Position line, Accessor &styler)
 188 {
 189     Sci_Position pos = styler.LineStart(line);
 190         Sci_Position eol_pos = styler.LineStart(line + 1) - 1;
 191
 192         for (Sci_Position i = pos; i < eol_pos; i++) {
 193                 char ch = styler[i];
 194                 char chNext = styler[i+1];
 195         char style = styler.StyleAt(i);
 196                 if ((style == SCE_VHDL_BLOCK_COMMENT) && (ch == '*') && (chNext == '/'))
 197                         return true;
 198         }
 199         return false;
 200 }
 201
 202 static bool IsCommentStyle(char style)
 203 {
 204     return style == SCE_VHDL_BLOCK_COMMENT || style == SCE_VHDL_COMMENT || style == SCE_VHDL_COMMENTLINEBANG;
 205 }
 206
 207 //=============================================================================
 208 // Folding the code
 209 static void FoldNoBoxVHDLDoc(
 210   Sci_PositionU startPos,
 211   Sci_Position length,
 212   int,
 213   Accessor &styler)
 214 {
 215   // Decided it would be smarter to have the lexer have all keywords included. Therefore I
 216   // don't check if the style for the keywords that I use to adjust the levels.
 217   char words[] =
 218     "architecture begin block case component else elsif end entity generate loop package process record then "
 219     "procedure protected function when units";
 220   WordList keywords;
 221   keywords.Set(words);
 222
 223   bool foldComment      = styler.GetPropertyInt("fold.comment", 1) != 0;
 224   bool foldCompact      = styler.GetPropertyInt("fold.compact", 1) != 0;
 225   bool foldAtElse       = styler.GetPropertyInt("fold.at.else", 1) != 0;
 226   bool foldAtBegin      = styler.GetPropertyInt("fold.at.Begin", 1) != 0;
 227   bool foldAtParenthese = styler.GetPropertyInt("fold.at.Parenthese", 1) != 0;
 228   //bool foldAtWhen       = styler.GetPropertyInt("fold.at.When", 1) != 0;  //< fold at when in case statements
 229
 230   int  visibleChars     = 0;
 231   Sci_PositionU endPos   = startPos + length;
 232
 233   Sci_Position lineCurrent       = styler.GetLine(startPos);
 234   int levelCurrent      = SC_FOLDLEVELBASE;
 235   if(lineCurrent > 0)
 236     levelCurrent        = styler.LevelAt(lineCurrent-1) >> 16;
 237   //int levelMinCurrent   = levelCurrent;
 238   int levelMinCurrentElse = levelCurrent;   //< Used for folding at 'else'
 239   int levelMinCurrentBegin = levelCurrent;  //< Used for folding at 'begin'
 240   int levelNext         = levelCurrent;
 241
 242   /***************************************/
 243   Sci_Position lastStart         = 0;
 244   char prevWord[32]     = "";
 245
 246   /***************************************/
 247   // Find prev word
 248   // The logic for going up or down a level depends on a the previous keyword
 249   // This code could be cleaned up.
 250   Sci_Position end = 0;
 251   Sci_PositionU j;
 252   for(j = startPos; j>0; j--)
 253   {
 254     char ch       = styler.SafeGetCharAt(j);
 255     char chPrev   = styler.SafeGetCharAt(j-1);
 256     int style     = styler.StyleAt(j);
 257     int stylePrev = styler.StyleAt(j-1);
 258     if ((!IsCommentStyle(style)) && (stylePrev != SCE_VHDL_STRING))
 259     {
 260       if(IsAWordChar(chPrev) && !IsAWordChar(ch))
 261       {
 262         end = j-1;
 263       }
 264     }
 265     if ((!IsCommentStyle(style)) && (style != SCE_VHDL_STRING))
 266     {
 267       if(!IsAWordChar(chPrev) && IsAWordStart(ch) && (end != 0))
 268       {
 269         char s[32];
 270         Sci_PositionU k;
 271         for(k=0; (k<31 ) && (k<end-j+1 ); k++) {
 272           s[k] = static_cast<char>(tolower(styler[j+k]));
 273         }
 274         s[k] = '\0';
 275
 276         if(keywords.InList(s)) {
 277           strcpy(prevWord, s);
 278           break;
 279         }
 280       }
 281     }
 282   }
 283   for(j=j+static_cast<Sci_PositionU>(strlen(prevWord)); j<endPos; j++)
 284   {
 285     char ch       = styler.SafeGetCharAt(j);
 286     int style     = styler.StyleAt(j);
 287     if ((!IsCommentStyle(style)) && (style != SCE_VHDL_STRING))
 288     {
 289       if((ch == ';') && (strcmp(prevWord, "end") == 0))
 290       {
 291         strcpy(prevWord, ";");
 292       }
 293     }
 294   }
 295
 296   char  chNext          = styler[startPos];
 297   char  chPrev          = '\0';
 298   char  chNextNonBlank;
 299   int   styleNext       = styler.StyleAt(startPos);
 300   //Platform::DebugPrintf("Line[%04d] Prev[%20s] ************************* Level[%x]\n", lineCurrent+1, prevWord, levelCurrent);
 301
 302   /***************************************/
 303   for (Sci_PositionU i = startPos; i < endPos; i++)
 304   {
 305     char ch         = chNext;
 306     chNext          = styler.SafeGetCharAt(i + 1);
 307     chPrev          = styler.SafeGetCharAt(i - 1);
 308     chNextNonBlank  = chNext;
 309     Sci_PositionU j  = i+1;
 310     while(IsABlank(chNextNonBlank) && j<endPos)
 311     {
 312       j ++ ;
 313       chNextNonBlank = styler.SafeGetCharAt(j);
 314     }
 315     int style           = styleNext;
 316     styleNext       = styler.StyleAt(i + 1);
 317     bool atEOL      = (ch == '\r' && chNext != '\n') || (ch == '\n');
 318
 319     if (foldComment && atEOL)
 320     {
 321       if(IsCommentLine(lineCurrent, styler))
 322       {
 323         if(!IsCommentLine(lineCurrent-1, styler) && IsCommentLine(lineCurrent+1, styler))
 324         {
 325           levelNext++;
 326         }
 327         else if(IsCommentLine(lineCurrent-1, styler) && !IsCommentLine(lineCurrent+1, styler))
 328         {
 329           levelNext--;
 330         }
 331       }
 332       else
 333       {
 334         if (IsCommentBlockStart(lineCurrent, styler) && !IsCommentBlockEnd(lineCurrent, styler))
 335         {
 336           levelNext++;
 337         }
 338         else if (IsCommentBlockEnd(lineCurrent, styler) && !IsCommentBlockStart(lineCurrent, styler))
 339         {
 340           levelNext--;
 341         }
 342       }
 343     }
 344
 345     if ((style == SCE_VHDL_OPERATOR) && foldAtParenthese)
 346     {
 347       if(ch == '(') {
 348         levelNext++;
 349       } else if (ch == ')') {
 350         levelNext--;
 351       }
 352     }
 353
 354     if ((!IsCommentStyle(style)) && (style != SCE_VHDL_STRING))
 355     {
 356       if((ch == ';') && (strcmp(prevWord, "end") == 0))
 357       {
 358         strcpy(prevWord, ";");
 359       }
 360
 361       if(!IsAWordChar(chPrev) && IsAWordStart(ch))
 362       {
 363         lastStart = i;
 364       }
 365
 366       if(IsAWordChar(ch) && !IsAWordChar(chNext)) {
 367         char s[32];
 368         Sci_PositionU k;
 369         for(k=0; (k<31 ) && (k<i-lastStart+1 ); k++) {
 370           s[k] = static_cast<char>(tolower(styler[lastStart+k]));
 371         }
 372         s[k] = '\0';
 373
 374         if(keywords.InList(s))
 375         {
 376           if (
 377             strcmp(s, "architecture") == 0  ||
 378             strcmp(s, "case") == 0          ||
 379             strcmp(s, "generate") == 0      ||
 380             strcmp(s, "block") == 0         ||
 381             strcmp(s, "loop") == 0          ||
 382             strcmp(s, "package") ==0        ||
 383             strcmp(s, "process") == 0       ||
 384             strcmp(s, "protected") == 0     ||
 385             strcmp(s, "record") == 0        ||
 386             strcmp(s, "then") == 0          ||
 387             strcmp(s, "units") == 0)
 388           {
 389             if (strcmp(prevWord, "end") != 0)
 390             {
 391               if (levelMinCurrentElse > levelNext) {
 392                 levelMinCurrentElse = levelNext;
 393               }
 394               levelNext++;
 395             }
 396           } else if (
 397             strcmp(s, "component") == 0      ||
 398             strcmp(s, "entity") == 0         ||
 399             strcmp(s, "configuration") == 0 )
 400           {
 401             if (strcmp(prevWord, "end") != 0 && lastStart)
 402             { // check for instantiated unit by backward searching for the colon.
 403               Sci_PositionU pos = lastStart;
 404               char chAtPos, styleAtPos;
 405               do{// skip white spaces
 406                 pos--;
 407                 styleAtPos = styler.StyleAt(pos);
 408                 chAtPos = styler.SafeGetCharAt(pos);
 409               }while(pos>0 &&
 410                      (chAtPos == ' ' || chAtPos == '\t' ||
 411                       chAtPos == '\n' || chAtPos == '\r' ||
 412                       IsCommentStyle(styleAtPos)));
 413
 414               // check for a colon (':') before the instantiated units "entity", "component" or "configuration". Don't fold thereafter.
 415               if (chAtPos != ':')
 416               {
 417                 if (levelMinCurrentElse > levelNext) {
 418                   levelMinCurrentElse = levelNext;
 419                 }
 420                 levelNext++;
 421               }
 422             }
 423           } else if (
 424             strcmp(s, "procedure") == 0     ||
 425             strcmp(s, "function") == 0)
 426           {
 427             if (strcmp(prevWord, "end") != 0) // check for "end procedure" etc.
 428             { // This code checks to see if the procedure / function is a definition within a "package"
 429               // rather than the actual code in the body.
 430               int BracketLevel = 0;
 431               for(Sci_Position pos=i+1; pos<styler.Length(); pos++)
 432               {
 433                 int styleAtPos = styler.StyleAt(pos);
 434                 char chAtPos = styler.SafeGetCharAt(pos);
 435                 if(chAtPos == '(') BracketLevel++;
 436                 if(chAtPos == ')') BracketLevel--;
 437                 if(
 438                   (BracketLevel == 0) &&
 439                   (!IsCommentStyle(styleAtPos)) &&
 440                   (styleAtPos != SCE_VHDL_STRING) &&
 441                   !iswordchar(styler.SafeGetCharAt(pos-1)) &&
 442                   (chAtPos|' ')=='i' && (styler.SafeGetCharAt(pos+1)|' ')=='s' &&
 443                   !iswordchar(styler.SafeGetCharAt(pos+2)))
 444                 {
 445                   if (levelMinCurrentElse > levelNext) {
 446                     levelMinCurrentElse = levelNext;
 447                   }
 448                   levelNext++;
 449                   break;
 450                 }
 451                 if((BracketLevel == 0) && (chAtPos == ';'))
 452                 {
 453                   break;
 454                 }
 455               }
 456             }
 457
 458           } else if (strcmp(s, "end") == 0) {
 459             levelNext--;
 460           }  else if(strcmp(s, "elsif") == 0) { // elsif is followed by then so folding occurs correctly
 461             levelNext--;
 462           } else if (strcmp(s, "else") == 0) {
 463             if(strcmp(prevWord, "when") != 0)  // ignore a <= x when y else z;
 464             {
 465               levelMinCurrentElse = levelNext - 1;  // VHDL else is all on its own so just dec. the min level
 466             }
 467           } else if(
 468             ((strcmp(s, "begin") == 0) && (strcmp(prevWord, "architecture") == 0)) ||
 469             ((strcmp(s, "begin") == 0) && (strcmp(prevWord, "function") == 0)) ||
 470             ((strcmp(s, "begin") == 0) && (strcmp(prevWord, "procedure") == 0)))
 471           {
 472             levelMinCurrentBegin = levelNext - 1;
 473           }
 474           //Platform::DebugPrintf("Line[%04d] Prev[%20s] Cur[%20s] Level[%x]\n", lineCurrent+1, prevWord, s, levelCurrent);
 475           strcpy(prevWord, s);
 476         }
 477       }
 478     }
 479     if (atEOL) {
 480       int levelUse = levelCurrent;
 481
 482       if (foldAtElse && (levelMinCurrentElse < levelUse)) {
 483         levelUse = levelMinCurrentElse;
 484       }
 485       if (foldAtBegin && (levelMinCurrentBegin < levelUse)) {
 486         levelUse = levelMinCurrentBegin;
 487       }
 488       int lev = levelUse | levelNext << 16;
 489       if (visibleChars == 0 && foldCompact)
 490         lev |= SC_FOLDLEVELWHITEFLAG;
 491
 492       if (levelUse < levelNext)
 493         lev |= SC_FOLDLEVELHEADERFLAG;
 494       if (lev != styler.LevelAt(lineCurrent)) {
 495         styler.SetLevel(lineCurrent, lev);
 496       }
 497       //Platform::DebugPrintf("Line[%04d] ---------------------------------------------------- Level[%x]\n", lineCurrent+1, levelCurrent);
 498       lineCurrent++;
 499       levelCurrent = levelNext;
 500       //levelMinCurrent = levelCurrent;
 501       levelMinCurrentElse = levelCurrent;
 502       levelMinCurrentBegin = levelCurrent;
 503       visibleChars = 0;
 504     }
 505     /***************************************/
 506     if (!isspacechar(ch)) visibleChars++;
 507   }
 508
 509   /***************************************/
 510 //  Platform::DebugPrintf("Line[%04d] ---------------------------------------------------- Level[%x]\n", lineCurrent+1, levelCurrent);
 511 }
 512
 513 //=============================================================================
 514 static void FoldVHDLDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, WordList *[],
 515                        Accessor &styler) {
 516   FoldNoBoxVHDLDoc(startPos, length, initStyle, styler);
 517 }
 518
 519 //=============================================================================
 520 static const char * const VHDLWordLists[] = {
 521             "Keywords",
 522             "Operators",
 523             "Attributes",
 524             "Standard Functions",
 525             "Standard Packages",
 526             "Standard Types",
 527             "User Words",
 528             0,
 529         };
 530
 531
 532 LexerModule lmVHDL(SCLEX_VHDL, ColouriseVHDLDoc, "vhdl", FoldVHDLDoc, VHDLWordLists);
 533
 534
 535 // Keyword:
 536 //    access after alias all architecture array assert attribute begin block body buffer bus case component
 537 //    configuration constant disconnect downto else elsif end entity exit file for function generate generic
 538 //    group guarded if impure in inertial inout is label library linkage literal loop map new next null of
 539 //    on open others out package port postponed procedure process pure range record register reject report
 540 //    return select severity shared signal subtype then to transport type unaffected units until use variable
 541 //    wait when while with
 542 //
 543 // Operators:
 544 //    abs and mod nand nor not or rem rol ror sla sll sra srl xnor xor
 545 //
 546 // Attributes:
 547 //    left right low high ascending image value pos val succ pred leftof rightof base range reverse_range
 548 //    length delayed stable quiet transaction event active last_event last_active last_value driving
 549 //    driving_value simple_name path_name instance_name
 550 //
 551 // Std Functions:
 552 //    now readline read writeline write endfile resolved to_bit to_bitvector to_stdulogic to_stdlogicvector
 553 //    to_stdulogicvector to_x01 to_x01z to_UX01 rising_edge falling_edge is_x shift_left shift_right rotate_left
 554 //    rotate_right resize to_integer to_unsigned to_signed std_match to_01
 555 //
 556 // Std Packages:
 557 //    std ieee work standard textio std_logic_1164 std_logic_arith std_logic_misc std_logic_signed
 558 //    std_logic_textio std_logic_unsigned numeric_bit numeric_std math_complex math_real vital_primitives
 559 //    vital_timing
 560 //
 561 // Std Types:
 562 //    boolean bit character severity_level integer real time delay_length natural positive string bit_vector
 563 //    file_open_kind file_open_status line text side width std_ulogic std_ulogic_vector std_logic
 564 //    std_logic_vector X01 X01Z UX01 UX01Z unsigned signed
 565 //
 566