contrib/src/stc/scintilla/src/LexPerl.cxx

   1 // SciTE - Scintilla based Text Editor
   2 // LexPerl.cxx - lexer for subset of Perl
   3 // Copyright 1998-2000 by Neil Hodgson <neilh@scintilla.org>
   4 // The License.txt file describes the conditions under which this software may be distributed.
   5
   6 #include <stdlib.h>
   7 #include <string.h>
   8 #include <ctype.h>
   9 #include <stdio.h>
  10 #include <stdarg.h>
  11
  12 #include "Platform.h"
  13
  14 #include "PropSet.h"
  15 #include "Accessor.h"
  16 #include "KeyWords.h"
  17 #include "Scintilla.h"
  18 #include "SciLexer.h"
  19
  20 inline bool isPerlOperator(char ch) {
  21         if (isalnum(ch))
  22                 return false;
  23         // '.' left out as it is used to make up numbers
  24         if (ch == '%' || ch == '^' || ch == '&' || ch == '*' || ch == '\\' ||
  25                 ch == '(' || ch == ')' || ch == '-' || ch == '+' ||
  26                 ch == '=' || ch == '|' || ch == '{' || ch == '}' ||
  27                 ch == '[' || ch == ']' || ch == ':' || ch == ';' ||
  28                 ch == '<' || ch == '>' || ch == ',' || ch == '/' ||
  29                 ch == '?' || ch == '!' || ch == '.' || ch == '~')
  30                 return true;
  31         return false;
  32 }
  33
  34 static int classifyWordPerl(unsigned int start, unsigned int end, WordList &keywords, StylingContext &styler) {
  35         char s[100];
  36         bool wordIsNumber = isdigit(styler[start]) || (styler[start] == '.');
  37         for (unsigned int i = 0; i < end - start + 1 && i < 30; i++) {
  38                 s[i] = styler[start + i];
  39                 s[i + 1] = '\0';
  40         }
  41         char chAttr = SCE_PL_IDENTIFIER;
  42         if (wordIsNumber)
  43                 chAttr = SCE_PL_NUMBER;
  44         else {
  45                 if (keywords.InList(s))
  46                         chAttr = SCE_PL_WORD;
  47         }
  48         styler.ColourTo(end, chAttr);
  49         return chAttr;
  50 }
  51
  52 static bool isEndVar(char ch) {
  53         return !isalnum(ch) && ch != '#' && ch != '$' &&
  54                ch != '_' && ch != '\'';
  55 }
  56
  57 static bool isMatch(StylingContext &styler, int lengthDoc, int pos, const char *val) {
  58         if ((pos + static_cast<int>(strlen(val))) >= lengthDoc) {
  59                 return false;
  60         }
  61         while (*val) {
  62                 if (*val != styler[pos++]) {
  63                         return false;
  64                 }
  65                 val++;
  66         }
  67         return true;
  68 }
  69
  70 static bool isOKQuote(char ch) {
  71         if (isalnum(ch))
  72                 return false;
  73         if (isspace(ch))
  74                 return false;
  75         if (iscntrl(ch))
  76                 return false;
  77         return true;
  78 }
  79
  80 static char opposite(char ch) {
  81         if (ch == '(')
  82                 return ')';
  83         if (ch == '[')
  84                 return ']';
  85         if (ch == '{')
  86                 return '}';
  87         if (ch == '<')
  88                 return '>';
  89         return ch;
  90 }
  91
  92 static void ColourisePerlDoc(unsigned int startPos, int length, int initStyle,
  93                              WordList *keywordlists[], StylingContext &styler) {
  94
  95         // Lexer for perl often has to backtrack to start of current style to determine
  96         // which characters are being used as quotes, how deeply nested is the
  97         // start position and what the termination string is for here documents
  98
  99         WordList &keywords = *keywordlists[0];
 100
 101         char sooked[100];
 102         int quotes = 0;
 103         char quoteDown = 'd';
 104         char quoteUp = 'd';
 105         int quoteRep = 1;
 106         int sookedpos = 0;
 107         bool preferRE = true;
 108         sooked[sookedpos] = '\0';
 109         int state = initStyle;
 110         int lengthDoc = startPos + length;
 111         // If in a long distance lexical state, seek to the beginning  to find quote characters
 112         if (state == SCE_PL_HERE || state == SCE_PL_REGEX ||
 113                 state == SCE_PL_REGSUBST || state == SCE_PL_LONGQUOTE) {
 114                 while ((startPos > 1) && (styler.StyleAt(startPos - 1) == state)) {
 115                         startPos--;
 116                 }
 117                 state = SCE_PL_DEFAULT;
 118         }
 119         styler.StartAt(startPos);
 120         char chPrev = ' ';
 121         char chNext = styler[startPos];
 122         styler.StartSegment(startPos);
 123         for (int i = startPos; i <= lengthDoc; i++) {
 124                 char ch = chNext;
 125                 chNext = styler.SafeGetCharAt(i + 1);
 126                 char chNext2 = styler.SafeGetCharAt(i + 2);
 127
 128                 if (styler.IsLeadByte(ch)) {
 129                         chNext = styler.SafeGetCharAt(i + 2);
 130                         chPrev = ' ';
 131                         i += 1;
 132                         continue;
 133                 }
 134
 135                 if (state == SCE_PL_DEFAULT) {
 136                         if (iswordstart(ch)) {
 137                                 styler.ColourTo(i - 1, state);
 138                                 if (ch == 's' && !isalnum(chNext)) {
 139                                         state = SCE_PL_REGSUBST;
 140                                         quotes = 0;
 141                                         quoteUp = '\0';
 142                                         quoteDown = '\0';
 143                                         quoteRep = 2;
 144                                 } else if (ch == 'm' && !isalnum(chNext)) {
 145                                         state = SCE_PL_REGEX;
 146                                         quotes = 0;
 147                                         quoteUp = '\0';
 148                                         quoteDown = '\0';
 149                                         quoteRep = 1;
 150                                 } else if (ch == 't' && chNext == 'r' && !isalnum(chNext2)) {
 151                                         state = SCE_PL_REGSUBST;
 152                                         quotes = 0;
 153                                         quoteUp = '\0';
 154                                         quoteDown = '\0';
 155                                         quoteRep = 2;
 156                                         i++;
 157                                         chNext = chNext2;
 158                                 } else if (ch == 'q' && (chNext == 'q' || chNext == 'r' || chNext == 'w' || chNext == 'x') && !isalnum(chNext2)) {
 159                                         state = SCE_PL_LONGQUOTE;
 160                                         i++;
 161                                         chNext = chNext2;
 162                                         quotes = 0;
 163                                         quoteUp = '\0';
 164                                         quoteDown = '\0';
 165                                         quoteRep = 1;
 166                                 } else {
 167                                         state = SCE_PL_WORD;
 168                                         preferRE = false;
 169                                 }
 170                         } else if (ch == '#') {
 171                                 styler.ColourTo(i - 1, state);
 172                                 state = SCE_PL_COMMENTLINE;
 173                         } else if (ch == '\"') {
 174                                 styler.ColourTo(i - 1, state);
 175                                 state = SCE_PL_STRING;
 176                         } else if (ch == '\'') {
 177                                 if (chPrev == '&') {
 178                                         // Archaic call
 179                                         styler.ColourTo(i, state);
 180                                 } else {
 181                                         styler.ColourTo(i - 1, state);
 182                                         state = SCE_PL_CHARACTER;
 183                                 }
 184                         } else if (ch == '`') {
 185                                 styler.ColourTo(i - 1, state);
 186                                 state = SCE_PL_BACKTICKS;
 187                         } else if (ch == '$') {
 188                                 preferRE = false;
 189                                 styler.ColourTo(i - 1, state);
 190                                 if (isalnum(chNext) || chNext == '#' || chNext == '$' || chNext == '_') {
 191                                         state = SCE_PL_SCALAR;
 192                                 } else if (chNext != '{' && chNext != '[') {
 193                                         styler.ColourTo(i, SCE_PL_SCALAR);
 194                                         i++;
 195                                         ch = ' ';
 196                                         chNext = ' ';
 197                                 } else {
 198                                         styler.ColourTo(i, SCE_PL_SCALAR);
 199                                 }
 200                         } else if (ch == '@') {
 201                                 preferRE = false;
 202                                 styler.ColourTo(i - 1, state);
 203                                 if (isalpha(chNext) || chNext == '#' || chNext == '$' || chNext == '_') {
 204                                         state = SCE_PL_ARRAY;
 205                                 } else if (chNext != '{' && chNext != '[') {
 206                                         styler.ColourTo(i, SCE_PL_ARRAY);
 207                                         i++;
 208                                         ch = ' ';
 209                                 } else {
 210                                         styler.ColourTo(i, SCE_PL_ARRAY);
 211                                 }
 212                         } else if (ch == '%') {
 213                                 preferRE = false;
 214                                 styler.ColourTo(i - 1, state);
 215                                 if (isalpha(chNext) || chNext == '#' || chNext == '$' || chNext == '_') {
 216                                         state = SCE_PL_HASH;
 217                                 } else if (chNext != '{' && chNext != '[') {
 218                                         styler.ColourTo(i, SCE_PL_HASH);
 219                                         i++;
 220                                         ch = ' ';
 221                                 } else {
 222                                         styler.ColourTo(i, SCE_PL_HASH);
 223                                 }
 224                         } else if (ch == '*') {
 225                                 styler.ColourTo(i - 1, state);
 226                                 state = SCE_PL_SYMBOLTABLE;
 227                         } else if (ch == '/' && preferRE) {
 228                                 styler.ColourTo(i - 1, state);
 229                                 state = SCE_PL_REGEX;
 230                                 quoteUp = '/';
 231                                 quoteDown = '/';
 232                                 quotes = 1;
 233                                 quoteRep = 1;
 234                         } else if (ch == '<' && chNext == '<') {
 235                                 styler.ColourTo(i - 1, state);
 236                                 state = SCE_PL_HERE;
 237                                 i++;
 238                                 ch = chNext;
 239                                 chNext = chNext2;
 240                                 quotes = 0;
 241                                 sookedpos = 0;
 242                                 sooked[sookedpos] = '\0';
 243                         } else if (ch == '=' && isalpha(chNext)) {
 244                                 styler.ColourTo(i - 1, state);
 245                                 state = SCE_PL_POD;
 246                                 quotes = 0;
 247                                 sookedpos = 0;
 248                                 sooked[sookedpos] = '\0';
 249                         } else if (isPerlOperator(ch)) {
 250                                 if (ch == ')' || ch == ']')
 251                                         preferRE = false;
 252                                 else
 253                                         preferRE = true;
 254                                 styler.ColourTo(i - 1, state);
 255                                 styler.ColourTo(i, SCE_PL_OPERATOR);
 256                         }
 257                 } else if (state == SCE_PL_WORD) {
 258                         if (!iswordchar(ch) && ch != '\'') {    // Archaic Perl has quotes inside names
 259                                 if (isMatch(styler, lengthDoc, styler.GetStartSegment(), "__DATA__")) {
 260                                         styler.ColourTo(i, SCE_PL_DATASECTION);
 261                                         state = SCE_PL_DATASECTION;
 262                                 } else if (isMatch(styler, lengthDoc, styler.GetStartSegment(), "__END__")) {
 263                                         styler.ColourTo(i, SCE_PL_DATASECTION);
 264                                         state = SCE_PL_DATASECTION;
 265                                 } else {
 266                                         if (classifyWordPerl(styler.GetStartSegment(), i - 1, keywords, styler) == SCE_PL_WORD)
 267                                                 preferRE = true;
 268                                         state = SCE_PL_DEFAULT;
 269                                         if (ch == '#') {
 270                                                 state = SCE_PL_COMMENTLINE;
 271                                         } else if (ch == '\"') {
 272                                                 state = SCE_PL_STRING;
 273                                         } else if (ch == '\'') {
 274                                                 state = SCE_PL_CHARACTER;
 275                                         } else if (ch == '<' && chNext == '<') {
 276                                                 state = SCE_PL_HERE;
 277                                                 quotes = 0;
 278                                                 sookedpos = 0;
 279                                                 sooked[sookedpos] = '\0';
 280                                         } else if (isPerlOperator(ch)) {
 281                                                 if (ch == ')' || ch == ']')
 282                                                         preferRE = false;
 283                                                 else
 284                                                         preferRE = true;
 285                                                 styler.ColourTo(i, SCE_PL_OPERATOR);
 286                                                 state = SCE_PL_DEFAULT;
 287                                         }
 288                                 }
 289                         }
 290                 } else {
 291                         if (state == SCE_PL_COMMENTLINE) {
 292                                 if (ch == '\r' || ch == '\n') {
 293                                         styler.ColourTo(i - 1, state);
 294                                         state = SCE_PL_DEFAULT;
 295                                 }
 296                         } else if (state == SCE_PL_HERE) {
 297                                 if (isalnum(ch) && quotes < 2) {
 298                                         sooked[sookedpos++] = ch;
 299                                         sooked[sookedpos] = '\0';
 300                                         if (quotes == 0)
 301                                                 quotes = 1;
 302                                 } else {
 303                                         quotes++;
 304                                 }
 305
 306                                 if (quotes > 1 && isMatch(styler, lengthDoc, i, sooked)) {
 307                                         styler.ColourTo(i + sookedpos - 1, SCE_PL_HERE);
 308                                         state = SCE_PL_DEFAULT;
 309                                         i += sookedpos;
 310                                         chNext = ' ';
 311                                 }
 312                         } else if (state == SCE_PL_STRING) {
 313                                 if (ch == '\\') {
 314                                         if (chNext == '\"' || chNext == '\'' || chNext == '\\') {
 315                                                 i++;
 316                                                 ch = chNext;
 317                                                 chNext = styler.SafeGetCharAt(i + 1);
 318                                         }
 319                                 } else if (ch == '\"') {
 320                                         styler.ColourTo(i, state);
 321                                         state = SCE_PL_DEFAULT;
 322                                         i++;
 323                                         ch = chNext;
 324                                         chNext = styler.SafeGetCharAt(i + 1);
 325                                 }
 326                         } else if (state == SCE_PL_CHARACTER) {
 327                                 if (ch == '\\') {
 328                                         if (chNext == '\"' || chNext == '\'' || chNext == '\\') {
 329                                                 i++;
 330                                                 ch = chNext;
 331                                                 chNext = styler.SafeGetCharAt(i + 1);
 332                                         }
 333                                 } else if (ch == '\'') {
 334                                         styler.ColourTo(i, state);
 335                                         state = SCE_PL_DEFAULT;
 336                                         i++;
 337                                         ch = chNext;
 338                                         chNext = styler.SafeGetCharAt(i + 1);
 339                                 }
 340                         } else if (state == SCE_PL_BACKTICKS) {
 341                                 if (ch == '`') {
 342                                         styler.ColourTo(i, state);
 343                                         state = SCE_PL_DEFAULT;
 344                                         i++;
 345                                         ch = chNext;
 346                                         chNext = styler.SafeGetCharAt(i + 1);
 347                                 }
 348                         } else if (state == SCE_PL_POD) {
 349                                 if (ch == '=') {
 350                                         if (isMatch(styler, lengthDoc, i, "=cut")) {
 351                                                 styler.ColourTo(i - 1 + 4, state);
 352                                                 i += 4;
 353                                                 state = SCE_PL_DEFAULT;
 354                                                 chNext = ' ';
 355                                                 ch = ' ';
 356                                         }
 357                                 }
 358                         } else if (state == SCE_PL_SCALAR) {
 359                                 if (isEndVar(ch)) {
 360                                         styler.ColourTo(i - 1, state);
 361                                         state = SCE_PL_DEFAULT;
 362                                 }
 363                         } else if (state == SCE_PL_ARRAY) {
 364                                 if (isEndVar(ch)) {
 365                                         styler.ColourTo(i - 1, state);
 366                                         state = SCE_PL_DEFAULT;
 367                                 }
 368                         } else if (state == SCE_PL_HASH) {
 369                                 if (isEndVar(ch)) {
 370                                         styler.ColourTo(i - 1, state);
 371                                         state = SCE_PL_DEFAULT;
 372                                 }
 373                         } else if (state == SCE_PL_SYMBOLTABLE) {
 374                                 if (isEndVar(ch)) {
 375                                         styler.ColourTo(i - 1, state);
 376                                         state = SCE_PL_DEFAULT;
 377                                 }
 378                         } else if (state == SCE_PL_REF) {
 379                                 if (isEndVar(ch)) {
 380                                         styler.ColourTo(i - 1, state);
 381                                         state = SCE_PL_DEFAULT;
 382                                 }
 383                         } else if (state == SCE_PL_REGEX) {
 384                                 if (!quoteUp && !isspace(ch)) {
 385                                         quoteUp = ch;
 386                                         quoteDown = opposite(ch);
 387                                         quotes++;
 388                                 } else {
 389                                         if (ch == quoteDown && chPrev != '\\') {
 390                                                 quotes--;
 391                                                 if (quotes == 0) {
 392                                                         quoteRep--;
 393                                                         if (quoteUp == quoteDown) {
 394                                                                 quotes++;
 395                                                         }
 396                                                 }
 397                                                 if (!isalpha(chNext)) {
 398                                                         if (quoteRep <= 0) {
 399                                                                 styler.ColourTo(i, state);
 400                                                                 state = SCE_PL_DEFAULT;
 401                                                                 ch = ' ';
 402                                                         }
 403                                                 }
 404                                         } else if (ch == quoteUp && chPrev != '\\') {
 405                                                 quotes++;
 406                                         } else if (!isalpha(chNext)) {
 407                                                 if (quoteRep <= 0) {
 408                                                         styler.ColourTo(i, state);
 409                                                         state = SCE_PL_DEFAULT;
 410                                                         ch = ' ';
 411                                                 }
 412                                         }
 413                                 }
 414                         } else if (state == SCE_PL_REGSUBST) {
 415                                 if (!quoteUp && !isspace(ch)) {
 416                                         quoteUp = ch;
 417                                         quoteDown = opposite(ch);
 418                                         quotes++;
 419                                 } else {
 420                                         if (ch == quoteDown && chPrev != '\\') {
 421                                                 quotes--;
 422                                                 if (quotes == 0) {
 423                                                         quoteRep--;
 424                                                 }
 425                                                 if (!isalpha(chNext)) {
 426                                                         if (quoteRep <= 0) {
 427                                                                 styler.ColourTo(i, state);
 428                                                                 state = SCE_PL_DEFAULT;
 429                                                                 ch = ' ';
 430                                                         }
 431                                                 }
 432                                                 if (quoteUp == quoteDown) {
 433                                                         quotes++;
 434                                                 }
 435                                         } else if (ch == quoteUp && chPrev != '\\') {
 436                                                 quotes++;
 437                                         } else if (!isalpha(chNext)) {
 438                                                 if (quoteRep <= 0) {
 439                                                         styler.ColourTo(i, state);
 440                                                         state = SCE_PL_DEFAULT;
 441                                                         ch = ' ';
 442                                                 }
 443                                         }
 444                                 }
 445                         } else if (state == SCE_PL_LONGQUOTE) {
 446                                 if (!quoteDown && !isspace(ch)) {
 447                                         quoteUp = ch;
 448                                         quoteDown = opposite(quoteUp);
 449                                         quotes++;
 450                                 } else if (ch == quoteDown) {
 451                                         quotes--;
 452                                         if (quotes == 0) {
 453                                                 quoteRep--;
 454                                                 if (quoteRep <= 0) {
 455                                                         styler.ColourTo(i, state);
 456                                                         state = SCE_PL_DEFAULT;
 457                                                         ch = ' ';
 458                                                 }
 459                                                 if (quoteUp == quoteDown) {
 460                                                         quotes++;
 461                                                 }
 462                                         }
 463                                 } else if (ch == quoteUp) {
 464                                         quotes++;
 465                                 }
 466                         }
 467
 468                         if (state == SCE_PL_DEFAULT) {    // One of the above succeeded
 469                                 if (ch == '#') {
 470                                         state = SCE_PL_COMMENTLINE;
 471                                 } else if (ch == '\"') {
 472                                         state = SCE_PL_STRING;
 473                                 } else if (ch == '\'') {
 474                                         state = SCE_PL_CHARACTER;
 475                                 } else if (iswordstart(ch)) {
 476                                         state = SCE_PL_WORD;
 477                                         preferRE = false;
 478                                 } else if (isoperator(ch)) {
 479                                         styler.ColourTo(i, SCE_PL_OPERATOR);
 480                                 }
 481                         }
 482                 }
 483                 chPrev = ch;
 484         }
 485         styler.ColourTo(lengthDoc, state);
 486 }
 487
 488 LexerModule lmPerl(SCLEX_PERL, ColourisePerlDoc);