scintilla/lexers/LexTxt2tags.cxx

   1 /******************************************************************
   2  *  LexTxt2tags.cxx
   3  *
   4  *  A simple Txt2tags lexer for scintilla.
   5  *
   6  *
   7  *  Adapted by Eric Forgeot
   8  *  Based on the LexMarkdown.cxx by Jon Strait - jstrait@moonloop.net
   9  *
  10  *  What could be improved:
  11  *   - Verbatim lines could be like for raw lines : when there is no space between the ``` and the following text, the first letter should be colored so the user would understand there must be a space for a valid tag.
  12  *   - marks such as bold, italic, strikeout, underline should begin to be highlighted only when they are closed and valid.
  13  *   - verbatim and raw area should be highlighted too.
  14  *
  15  *  The License.txt file describes the conditions under which this
  16  *  software may be distributed.
  17  *
  18  *****************************************************************/
  19
  20 #include <stdlib.h>
  21 #include <string.h>
  22 #include <stdio.h>
  23 #include <stdarg.h>
  24 #include <assert.h>
  25
  26 #include "ILexer.h"
  27 #include "Scintilla.h"
  28 #include "SciLexer.h"
  29
  30 #include "WordList.h"
  31 #include "LexAccessor.h"
  32 #include "Accessor.h"
  33 #include "StyleContext.h"
  34 #include "CharacterSet.h"
  35 #include "LexerModule.h"
  36
  37 #ifdef SCI_NAMESPACE
  38 using namespace Scintilla;
  39 #endif
  40
  41
  42
  43 static inline bool IsNewline(const int ch) {
  44     return (ch == '\n' || ch == '\r');
  45 }
  46
  47 // True if can follow ch down to the end with possibly trailing whitespace
  48 static bool FollowToLineEnd(const int ch, const int state, const Sci_PositionU endPos, StyleContext &sc) {
  49     Sci_PositionU i = 0;
  50     while (sc.GetRelative(++i) == ch)
  51         ;
  52     // Skip over whitespace
  53     while (IsASpaceOrTab(sc.GetRelative(i)) && sc.currentPos + i < endPos)
  54         ++i;
  55     if (IsNewline(sc.GetRelative(i)) || sc.currentPos + i == endPos) {
  56         sc.Forward(i);
  57         sc.ChangeState(state);
  58         sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
  59         return true;
  60     }
  61     else return false;
  62 }
  63
  64 // Does the previous line have more than spaces and tabs?
  65 static bool HasPrevLineContent(StyleContext &sc) {
  66     Sci_Position i = 0;
  67     // Go back to the previous newline
  68     while ((--i + sc.currentPos) && !IsNewline(sc.GetRelative(i)))
  69         ;
  70     while (--i + sc.currentPos) {
  71         if (IsNewline(sc.GetRelative(i)))
  72             break;
  73         if (!IsASpaceOrTab(sc.GetRelative(i)))
  74             return true;
  75     }
  76     return false;
  77 }
  78
  79 // Separator line
  80 static bool IsValidHrule(const Sci_PositionU endPos, StyleContext &sc) {
  81     int count = 1;
  82     Sci_PositionU i = 0;
  83     for (;;) {
  84         ++i;
  85         int c = sc.GetRelative(i);
  86         if (c == sc.ch)
  87             ++count;
  88         // hit a terminating character
  89         else if (!IsASpaceOrTab(c) || sc.currentPos + i == endPos) {
  90             // Are we a valid HRULE
  91             if ((IsNewline(c) || sc.currentPos + i == endPos) &&
  92                     count >= 20 && !HasPrevLineContent(sc)) {
  93                 sc.SetState(SCE_TXT2TAGS_HRULE);
  94                 sc.Forward(i);
  95                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
  96                 return true;
  97             }
  98             else {
  99                 sc.SetState(SCE_TXT2TAGS_DEFAULT);
 100                 return false;
 101             }
 102         }
 103     }
 104 }
 105
 106 static void ColorizeTxt2tagsDoc(Sci_PositionU startPos, Sci_Position length, int initStyle,
 107                                WordList **, Accessor &styler) {
 108     Sci_PositionU endPos = startPos + length;
 109     int precharCount = 0;
 110     // Don't advance on a new loop iteration and retry at the same position.
 111     // Useful in the corner case of having to start at the beginning file position
 112     // in the default state.
 113     bool freezeCursor = false;
 114
 115     StyleContext sc(startPos, length, initStyle, styler);
 116
 117     while (sc.More()) {
 118         // Skip past escaped characters
 119         if (sc.ch == '\\') {
 120             sc.Forward();
 121             continue;
 122         }
 123
 124         // A blockquotes resets the line semantics
 125         if (sc.state == SCE_TXT2TAGS_BLOCKQUOTE){
 126             sc.Forward(2);
 127             sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 128         }
 129         // An option colors the whole line
 130         if (sc.state == SCE_TXT2TAGS_OPTION){
 131             FollowToLineEnd('%', SCE_TXT2TAGS_OPTION, endPos, sc);
 132         }
 133         if (sc.state == SCE_TXT2TAGS_POSTPROC){
 134             FollowToLineEnd('%', SCE_TXT2TAGS_POSTPROC, endPos, sc);
 135         }
 136         if (sc.state == SCE_TXT2TAGS_PREPROC){
 137             FollowToLineEnd('%', SCE_TXT2TAGS_PREPROC, endPos, sc);
 138         }
 139         // A comment colors the whole line
 140         if (sc.state == SCE_TXT2TAGS_COMMENT){
 141             FollowToLineEnd('%', SCE_TXT2TAGS_COMMENT, endPos, sc);
 142         }
 143         // Conditional state-based actions
 144         if (sc.state == SCE_TXT2TAGS_CODE2) {
 145         if (IsNewline(sc.ch))
 146                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 147             if (sc.Match("``") && sc.GetRelative(-2) != ' ') {
 148                 sc.Forward(2);
 149                 sc.SetState(SCE_TXT2TAGS_DEFAULT);
 150             }
 151         }
 152         // Table
 153         else if (sc.state == SCE_TXT2TAGS_CODE) {
 154         if (IsNewline(sc.ch))
 155                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 156             if (sc.ch == '|' && sc.chPrev != ' ')
 157                 sc.ForwardSetState(SCE_TXT2TAGS_DEFAULT);
 158         }
 159         // Strong
 160         else if (sc.state == SCE_TXT2TAGS_STRONG1) {
 161         if (IsNewline(sc.ch))
 162                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 163             if (sc.Match("**") && sc.chPrev != ' ') {
 164                 sc.Forward(2);
 165                 sc.SetState(SCE_TXT2TAGS_DEFAULT);
 166             }
 167         }
 168         // Emphasis
 169         else if (sc.state == SCE_TXT2TAGS_EM1) {
 170         if (IsNewline(sc.ch))
 171                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 172             if (sc.Match("//") && sc.chPrev != ' ') {
 173                 sc.Forward(2);
 174                 sc.ForwardSetState(SCE_TXT2TAGS_DEFAULT);
 175            }
 176         }
 177         // Underline
 178         else if (sc.state == SCE_TXT2TAGS_EM2) {
 179         if (IsNewline(sc.ch))
 180                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 181             if (sc.Match("__") && sc.chPrev != ' ') {
 182                 sc.Forward(2);
 183                 sc.ForwardSetState(SCE_TXT2TAGS_DEFAULT);
 184            }
 185         }
 186         // codeblock
 187         else if (sc.state == SCE_TXT2TAGS_CODEBK) {
 188                 if (IsNewline(sc.ch))
 189                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 190             if (sc.atLineStart && sc.Match("```")) {
 191                 Sci_Position i = 1;
 192                 while (!IsNewline(sc.GetRelative(i)) && sc.currentPos + i < endPos)
 193                     i++;
 194                 sc.Forward(i);
 195                 sc.SetState(SCE_TXT2TAGS_DEFAULT);
 196             }
 197         }
 198         // strikeout
 199         else if (sc.state == SCE_TXT2TAGS_STRIKEOUT) {
 200         if (IsNewline(sc.ch))
 201                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 202             if (sc.Match("--") && sc.chPrev != ' ') {
 203                 sc.Forward(2);
 204                 sc.SetState(SCE_TXT2TAGS_DEFAULT);
 205             }
 206         }
 207         // Headers
 208         else if (sc.state == SCE_TXT2TAGS_LINE_BEGIN) {
 209             if (sc.Match("======"))
 210                 {
 211                 sc.SetState(SCE_TXT2TAGS_HEADER6);
 212                 sc.Forward();
 213                 }
 214             else if (sc.Match("====="))
 215                 {
 216                 sc.SetState(SCE_TXT2TAGS_HEADER5);
 217                 sc.Forward();
 218                 }
 219             else if (sc.Match("===="))
 220                 {
 221                 sc.SetState(SCE_TXT2TAGS_HEADER4);
 222                 sc.Forward();
 223                 }
 224             else if (sc.Match("==="))
 225                 {
 226                 sc.SetState(SCE_TXT2TAGS_HEADER3);
 227                 sc.Forward();
 228                 }
 229                 //SetStateAndZoom(SCE_TXT2TAGS_HEADER3, 3, '=', sc);
 230             else if (sc.Match("==")) {
 231                 sc.SetState(SCE_TXT2TAGS_HEADER2);
 232                 sc.Forward();
 233                 }
 234                 //SetStateAndZoom(SCE_TXT2TAGS_HEADER2, 2, '=', sc);
 235             else if (sc.Match("=")) {
 236                 // Catch the special case of an unordered list
 237                 if (sc.chNext == '.' && IsASpaceOrTab(sc.GetRelative(2))) {
 238                     precharCount = 0;
 239                     sc.SetState(SCE_TXT2TAGS_PRECHAR);
 240                 }
 241                 else
 242                     {
 243                     sc.SetState(SCE_TXT2TAGS_HEADER1);
 244                     sc.Forward();
 245                     }
 246                     //SetStateAndZoom(SCE_TXT2TAGS_HEADER1, 1, '=', sc);
 247             }
 248
 249             // Numbered title
 250             else if (sc.Match("++++++"))
 251                 {
 252                 sc.SetState(SCE_TXT2TAGS_HEADER6);
 253                 sc.Forward();
 254                 }
 255             else if (sc.Match("+++++"))
 256                 {
 257                 sc.SetState(SCE_TXT2TAGS_HEADER5);
 258                 sc.Forward();
 259                 }
 260             else if (sc.Match("++++"))
 261                 {
 262                 sc.SetState(SCE_TXT2TAGS_HEADER4);
 263                 sc.Forward();
 264                 }
 265             else if (sc.Match("+++"))
 266                 {
 267                 sc.SetState(SCE_TXT2TAGS_HEADER3);
 268                 sc.Forward();
 269                 }
 270                 //SetStateAndZoom(SCE_TXT2TAGS_HEADER3, 3, '+', sc);
 271             else if (sc.Match("++")) {
 272                 sc.SetState(SCE_TXT2TAGS_HEADER2);
 273                 sc.Forward();
 274                 }
 275                 //SetStateAndZoom(SCE_TXT2TAGS_HEADER2, 2, '+', sc);
 276             else if (sc.Match("+")) {
 277                 // Catch the special case of an unordered list
 278                 if (sc.chNext == ' ' && IsASpaceOrTab(sc.GetRelative(1))) {
 279                  //    if (IsNewline(sc.ch)) {
 280                         //precharCount = 0;
 281                 //              sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 282                                 //sc.SetState(SCE_TXT2TAGS_PRECHAR);
 283                                 //      }
 284                 //    else {
 285                 //    precharCount = 0;
 286                     sc.SetState(SCE_TXT2TAGS_OLIST_ITEM);
 287                     sc.Forward(2);
 288                     sc.SetState(SCE_TXT2TAGS_DEFAULT);
 289                //     sc.SetState(SCE_TXT2TAGS_PRECHAR);
 290                                 //      }
 291                 }
 292                 else
 293                     {
 294                     sc.SetState(SCE_TXT2TAGS_HEADER1);
 295                     sc.Forward();
 296                     }
 297             }
 298
 299
 300             // Codeblock
 301             else if (sc.Match("```")) {
 302                 if (!HasPrevLineContent(sc))
 303               //  if (!FollowToLineEnd(sc))
 304                     sc.SetState(SCE_TXT2TAGS_CODEBK);
 305                 else
 306                     sc.SetState(SCE_TXT2TAGS_DEFAULT);
 307             }
 308
 309             // Preproc
 310             else if (sc.Match("%!preproc")) {
 311                 sc.SetState(SCE_TXT2TAGS_PREPROC);
 312             }
 313             // Postproc
 314             else if (sc.Match("%!postproc")) {
 315                 sc.SetState(SCE_TXT2TAGS_POSTPROC);
 316             }
 317             // Option
 318             else if (sc.Match("%!")) {
 319                 sc.SetState(SCE_TXT2TAGS_OPTION);
 320             }
 321
 322              // Comment
 323             else if (sc.ch == '%') {
 324                 sc.SetState(SCE_TXT2TAGS_COMMENT);
 325             }
 326             // list
 327             else if (sc.ch == '-') {
 328                     precharCount = 0;
 329                     sc.SetState(SCE_TXT2TAGS_PRECHAR);
 330             }
 331             // def list
 332             else if (sc.ch == ':') {
 333                     precharCount = 0;
 334                    sc.SetState(SCE_TXT2TAGS_OLIST_ITEM);
 335                    sc.Forward(1);
 336                    sc.SetState(SCE_TXT2TAGS_PRECHAR);
 337             }
 338             else if (IsNewline(sc.ch))
 339                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 340             else {
 341                 precharCount = 0;
 342                 sc.SetState(SCE_TXT2TAGS_PRECHAR);
 343             }
 344         }
 345
 346         // The header lasts until the newline
 347         else if (sc.state == SCE_TXT2TAGS_HEADER1 || sc.state == SCE_TXT2TAGS_HEADER2 ||
 348                 sc.state == SCE_TXT2TAGS_HEADER3 || sc.state == SCE_TXT2TAGS_HEADER4 ||
 349                 sc.state == SCE_TXT2TAGS_HEADER5 || sc.state == SCE_TXT2TAGS_HEADER6) {
 350             if (IsNewline(sc.ch))
 351                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 352         }
 353
 354         // New state only within the initial whitespace
 355         if (sc.state == SCE_TXT2TAGS_PRECHAR) {
 356             // Blockquote
 357             if (sc.Match("\"\"\"") && precharCount < 5){
 358
 359                 sc.SetState(SCE_TXT2TAGS_BLOCKQUOTE);
 360                 sc.Forward(1);
 361                 }
 362             /*
 363             // Begin of code block
 364             else if (!HasPrevLineContent(sc) && (sc.chPrev == '\t' || precharCount >= 4))
 365                 sc.SetState(SCE_TXT2TAGS_CODEBK);
 366             */
 367             // HRule - Total of 20 or more hyphens, asterisks, or underscores
 368             // on a line by themselves
 369             else if ((sc.ch == '-' ) && IsValidHrule(endPos, sc))
 370                 ;
 371             // Unordered list
 372             else if ((sc.ch == '-') && IsASpaceOrTab(sc.chNext)) {
 373                 sc.SetState(SCE_TXT2TAGS_ULIST_ITEM);
 374                 sc.ForwardSetState(SCE_TXT2TAGS_DEFAULT);
 375             }
 376             // Ordered list
 377             else if (IsADigit(sc.ch)) {
 378                 Sci_Position digitCount = 0;
 379                 while (IsADigit(sc.GetRelative(++digitCount)))
 380                     ;
 381                 if (sc.GetRelative(digitCount) == '.' &&
 382                         IsASpaceOrTab(sc.GetRelative(digitCount + 1))) {
 383                     sc.SetState(SCE_TXT2TAGS_OLIST_ITEM);
 384                     sc.Forward(digitCount + 1);
 385                     sc.SetState(SCE_TXT2TAGS_DEFAULT);
 386                 }
 387             }
 388             // Alternate Ordered list
 389             else if (sc.ch == '+' && sc.chNext == ' ' && IsASpaceOrTab(sc.GetRelative(2))) {
 390             //    sc.SetState(SCE_TXT2TAGS_OLIST_ITEM);
 391             //    sc.Forward(2);
 392              //   sc.SetState(SCE_TXT2TAGS_DEFAULT);
 393             }
 394             else if (sc.ch != ' ' || precharCount > 2)
 395                 sc.SetState(SCE_TXT2TAGS_DEFAULT);
 396             else
 397                 ++precharCount;
 398         }
 399
 400         // New state anywhere in doc
 401         if (sc.state == SCE_TXT2TAGS_DEFAULT) {
 402          //   if (sc.atLineStart && sc.ch == '#') {
 403          //       sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 404          //       freezeCursor = true;
 405          //   }
 406             // Links and Images
 407             if (sc.Match("![") || sc.ch == '[') {
 408                 Sci_Position i = 0, j = 0, k = 0;
 409                 Sci_Position len = endPos - sc.currentPos;
 410                 while (i < len && (sc.GetRelative(++i) != ']' || sc.GetRelative(i - 1) == '\\'))
 411                     ;
 412                 if (sc.GetRelative(i) == ']') {
 413                     j = i;
 414                     if (sc.GetRelative(++i) == '(') {
 415                         while (i < len && (sc.GetRelative(++i) != '(' || sc.GetRelative(i - 1) == '\\'))
 416                             ;
 417                         if (sc.GetRelative(i) == '(')
 418                             k = i;
 419                     }
 420
 421                     else if (sc.GetRelative(i) == '[' || sc.GetRelative(++i) == '[') {
 422                         while (i < len && (sc.GetRelative(++i) != ']' || sc.GetRelative(i - 1) == '\\'))
 423                             ;
 424                         if (sc.GetRelative(i) == ']')
 425                             k = i;
 426                     }
 427                 }
 428                 // At least a link text
 429                 if (j) {
 430                     sc.SetState(SCE_TXT2TAGS_LINK);
 431                     sc.Forward(j);
 432                     // Also has a URL or reference portion
 433                     if (k)
 434                         sc.Forward(k - j);
 435                     sc.ForwardSetState(SCE_TXT2TAGS_DEFAULT);
 436                 }
 437             }
 438             // Code - also a special case for alternate inside spacing
 439             if (sc.Match("``") && sc.GetRelative(3) != ' ') {
 440                 sc.SetState(SCE_TXT2TAGS_CODE2);
 441                 sc.Forward();
 442             }
 443             else if (sc.ch == '|' && sc.GetRelative(3) != ' ') {
 444                 sc.SetState(SCE_TXT2TAGS_CODE);
 445             }
 446             // Strong
 447             else if (sc.Match("**") && sc.GetRelative(2) != ' ') {
 448                 sc.SetState(SCE_TXT2TAGS_STRONG1);
 449                 sc.Forward();
 450            }
 451             // Emphasis
 452             else if (sc.Match("//") && sc.GetRelative(2) != ' ') {
 453                 sc.SetState(SCE_TXT2TAGS_EM1);
 454                 sc.Forward();
 455             }
 456             else if (sc.Match("__") && sc.GetRelative(2) != ' ') {
 457                 sc.SetState(SCE_TXT2TAGS_EM2);
 458                 sc.Forward();
 459             }
 460             // Strikeout
 461             else if (sc.Match("--") && sc.GetRelative(2) != ' ') {
 462                 sc.SetState(SCE_TXT2TAGS_STRIKEOUT);
 463                 sc.Forward();
 464             }
 465
 466             // Beginning of line
 467             else if (IsNewline(sc.ch))
 468                 sc.SetState(SCE_TXT2TAGS_LINE_BEGIN);
 469         }
 470         // Advance if not holding back the cursor for this iteration.
 471         if (!freezeCursor)
 472             sc.Forward();
 473         freezeCursor = false;
 474     }
 475     sc.Complete();
 476 }
 477
 478 LexerModule lmTxt2tags(SCLEX_TXT2TAGS, ColorizeTxt2tagsDoc, "txt2tags");
 479
 480