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