scintilla/lexers/LexMarkdown.cxx

   1 /******************************************************************
   2  *  LexMarkdown.cxx
   3  *
   4  *  A simple Markdown lexer for scintilla.
   5  *
   6  *  Includes highlighting for some extra features from the
   7  *  Pandoc implementation; strikeout, using '#.' as a default
   8  *  ordered list item marker, and delimited code blocks.
   9  *
  10  *  Limitations:
  11  *
  12  *  Standard indented code blocks are not highlighted at all,
  13  *  as it would conflict with other indentation schemes. Use
  14  *  delimited code blocks for blanket highlighting of an
  15  *  entire code block.  Embedded HTML is not highlighted either.
  16  *  Blanket HTML highlighting has issues, because some Markdown
  17  *  implementations allow Markdown markup inside of the HTML. Also,
  18  *  there is a following blank line issue that can't be ignored,
  19  *  explained in the next paragraph. Embedded HTML and code
  20  *  blocks would be better supported with language specific
  21  *  highlighting.
  22  *
  23  *  The highlighting aims to accurately reflect correct syntax,
  24  *  but a few restrictions are relaxed. Delimited code blocks are
  25  *  highlighted, even if the line following the code block is not blank.
  26  *  Requiring a blank line after a block, breaks the highlighting
  27  *  in certain cases, because of the way Scintilla ends up calling
  28  *  the lexer.
  29  *
  30  *  Written by Jon Strait - jstrait@moonloop.net
  31  *
  32  *  The License.txt file describes the conditions under which this
  33  *  software may be distributed.
  34  *
  35  *****************************************************************/
  36
  37 #include <stdlib.h>
  38 #include <string.h>
  39 #include <stdio.h>
  40 #include <stdarg.h>
  41 #include <assert.h>
  42
  43 #include "ILexer.h"
  44 #include "Scintilla.h"
  45 #include "SciLexer.h"
  46
  47 #include "WordList.h"
  48 #include "LexAccessor.h"
  49 #include "Accessor.h"
  50 #include "StyleContext.h"
  51 #include "CharacterSet.h"
  52 #include "LexerModule.h"
  53
  54 #ifdef SCI_NAMESPACE
  55 using namespace Scintilla;
  56 #endif
  57
  58 static inline bool IsNewline(const int ch) {
  59     return (ch == '\n' || ch == '\r');
  60 }
  61
  62 // True if can follow ch down to the end with possibly trailing whitespace
  63 static bool FollowToLineEnd(const int ch, const int state, const unsigned int endPos, StyleContext &sc) {
  64     unsigned int i = 0;
  65     while (sc.GetRelative(++i) == ch)
  66         ;
  67     // Skip over whitespace
  68     while (IsASpaceOrTab(sc.GetRelative(i)) && sc.currentPos + i < endPos)
  69         ++i;
  70     if (IsNewline(sc.GetRelative(i)) || sc.currentPos + i == endPos) {
  71         sc.Forward(i);
  72         sc.ChangeState(state);
  73         sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
  74         return true;
  75     }
  76     else return false;
  77 }
  78
  79 // Set the state on text section from current to length characters,
  80 // then set the rest until the newline to default, except for any characters matching token
  81 static void SetStateAndZoom(const int state, const int length, const int token, StyleContext &sc) {
  82     sc.SetState(state);
  83     sc.Forward(length);
  84     sc.SetState(SCE_MARKDOWN_DEFAULT);
  85     sc.Forward();
  86     bool started = false;
  87     while (sc.More() && !IsNewline(sc.ch)) {
  88         if (sc.ch == token && !started) {
  89             sc.SetState(state);
  90             started = true;
  91         }
  92         else if (sc.ch != token) {
  93             sc.SetState(SCE_MARKDOWN_DEFAULT);
  94             started = false;
  95         }
  96         sc.Forward();
  97     }
  98     sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
  99 }
 100
 101 // Does the previous line have more than spaces and tabs?
 102 static bool HasPrevLineContent(StyleContext &sc) {
 103     int i = 0;
 104     // Go back to the previous newline
 105     while ((--i + (int)sc.currentPos) >= 0 && !IsNewline(sc.GetRelative(i)))
 106         ;
 107     while ((--i + (int)sc.currentPos) >= 0) {
 108         if (IsNewline(sc.GetRelative(i)))
 109             break;
 110         if (!IsASpaceOrTab(sc.GetRelative(i)))
 111             return true;
 112     }
 113     return false;
 114 }
 115
 116 static bool AtTermStart(StyleContext &sc) {
 117     return sc.currentPos == 0 || isspacechar(sc.chPrev);
 118 }
 119
 120 static bool IsValidHrule(const unsigned int endPos, StyleContext &sc) {
 121     int count = 1;
 122     unsigned int i = 0;
 123     for (;;) {
 124         ++i;
 125         int c = sc.GetRelative(i);
 126         if (c == sc.ch)
 127             ++count;
 128         // hit a terminating character
 129         else if (!IsASpaceOrTab(c) || sc.currentPos + i == endPos) {
 130             // Are we a valid HRULE
 131             if ((IsNewline(c) || sc.currentPos + i == endPos) &&
 132                     count >= 3 && !HasPrevLineContent(sc)) {
 133                 sc.SetState(SCE_MARKDOWN_HRULE);
 134                 sc.Forward(i);
 135                 sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 136                 return true;
 137             }
 138             else {
 139                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 140                 return false;
 141             }
 142         }
 143     }
 144 }
 145
 146 static void ColorizeMarkdownDoc(unsigned int startPos, int length, int initStyle,
 147                                WordList **, Accessor &styler) {
 148     unsigned int endPos = startPos + length;
 149     int precharCount = 0;
 150     // Don't advance on a new loop iteration and retry at the same position.
 151     // Useful in the corner case of having to start at the beginning file position
 152     // in the default state.
 153     bool freezeCursor = false;
 154
 155     StyleContext sc(startPos, length, initStyle, styler);
 156
 157     while (sc.More()) {
 158         // Skip past escaped characters
 159         if (sc.ch == '\\') {
 160             sc.Forward();
 161             continue;
 162         }
 163
 164         // A blockquotes resets the line semantics
 165         if (sc.state == SCE_MARKDOWN_BLOCKQUOTE)
 166             sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 167
 168         // Conditional state-based actions
 169         if (sc.state == SCE_MARKDOWN_CODE2) {
 170             if (sc.Match("``") && sc.GetRelative(-2) != ' ') {
 171                 sc.Forward(2);
 172                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 173             }
 174         }
 175         else if (sc.state == SCE_MARKDOWN_CODE) {
 176             if (sc.ch == '`' && sc.chPrev != ' ')
 177                 sc.ForwardSetState(SCE_MARKDOWN_DEFAULT);
 178         }
 179         /* De-activated because it gets in the way of other valid indentation
 180          * schemes, for example multiple paragraphs inside a list item.
 181         // Code block
 182         else if (sc.state == SCE_MARKDOWN_CODEBK) {
 183             bool d = true;
 184             if (IsNewline(sc.ch)) {
 185                 if (sc.chNext != '\t') {
 186                     for (int c = 1; c < 5; ++c) {
 187                         if (sc.GetRelative(c) != ' ')
 188                             d = false;
 189                     }
 190                 }
 191             }
 192             else if (sc.atLineStart) {
 193                 if (sc.ch != '\t' ) {
 194                     for (int i = 0; i < 4; ++i) {
 195                         if (sc.GetRelative(i) != ' ')
 196                             d = false;
 197                     }
 198                 }
 199             }
 200             if (!d)
 201                 sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 202         }
 203         */
 204         // Strong
 205         else if (sc.state == SCE_MARKDOWN_STRONG1) {
 206             if (sc.Match("**") && sc.chPrev != ' ') {
 207                 sc.Forward(2);
 208                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 209             }
 210         }
 211         else if (sc.state == SCE_MARKDOWN_STRONG2) {
 212             if (sc.Match("__") && sc.chPrev != ' ') {
 213                 sc.Forward(2);
 214                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 215             }
 216         }
 217         // Emphasis
 218         else if (sc.state == SCE_MARKDOWN_EM1) {
 219             if (sc.ch == '*' && sc.chPrev != ' ')
 220                 sc.ForwardSetState(SCE_MARKDOWN_DEFAULT);
 221         }
 222         else if (sc.state == SCE_MARKDOWN_EM2) {
 223             if (sc.ch == '_' && sc.chPrev != ' ')
 224                 sc.ForwardSetState(SCE_MARKDOWN_DEFAULT);
 225         }
 226         else if (sc.state == SCE_MARKDOWN_CODEBK) {
 227             if (sc.atLineStart && sc.Match("~~~")) {
 228                 int i = 1;
 229                 while (!IsNewline(sc.GetRelative(i)) && sc.currentPos + i < endPos)
 230                     i++;
 231                 sc.Forward(i);
 232                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 233             }
 234         }
 235         else if (sc.state == SCE_MARKDOWN_STRIKEOUT) {
 236             if (sc.Match("~~") && sc.chPrev != ' ') {
 237                 sc.Forward(2);
 238                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 239             }
 240         }
 241         else if (sc.state == SCE_MARKDOWN_LINE_BEGIN) {
 242             // Header
 243             if (sc.Match("######"))
 244                 SetStateAndZoom(SCE_MARKDOWN_HEADER6, 6, '#', sc);
 245             else if (sc.Match("#####"))
 246                 SetStateAndZoom(SCE_MARKDOWN_HEADER5, 5, '#', sc);
 247             else if (sc.Match("####"))
 248                 SetStateAndZoom(SCE_MARKDOWN_HEADER4, 4, '#', sc);
 249             else if (sc.Match("###"))
 250                 SetStateAndZoom(SCE_MARKDOWN_HEADER3, 3, '#', sc);
 251             else if (sc.Match("##"))
 252                 SetStateAndZoom(SCE_MARKDOWN_HEADER2, 2, '#', sc);
 253             else if (sc.Match("#")) {
 254                 // Catch the special case of an unordered list
 255                 if (sc.chNext == '.' && IsASpaceOrTab(sc.GetRelative(2))) {
 256                     precharCount = 0;
 257                     sc.SetState(SCE_MARKDOWN_PRECHAR);
 258                 }
 259                 else
 260                     SetStateAndZoom(SCE_MARKDOWN_HEADER1, 1, '#', sc);
 261             }
 262             // Code block
 263             else if (sc.Match("~~~")) {
 264                 if (!HasPrevLineContent(sc))
 265                     sc.SetState(SCE_MARKDOWN_CODEBK);
 266                 else
 267                     sc.SetState(SCE_MARKDOWN_DEFAULT);
 268             }
 269             else if (sc.ch == '=') {
 270                 if (HasPrevLineContent(sc) && FollowToLineEnd('=', SCE_MARKDOWN_HEADER1, endPos, sc))
 271                     ;
 272                 else
 273                     sc.SetState(SCE_MARKDOWN_DEFAULT);
 274             }
 275             else if (sc.ch == '-') {
 276                 if (HasPrevLineContent(sc) && FollowToLineEnd('-', SCE_MARKDOWN_HEADER2, endPos, sc))
 277                     ;
 278                 else {
 279                     precharCount = 0;
 280                     sc.SetState(SCE_MARKDOWN_PRECHAR);
 281                 }
 282             }
 283             else if (IsNewline(sc.ch))
 284                 sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 285             else {
 286                 precharCount = 0;
 287                 sc.SetState(SCE_MARKDOWN_PRECHAR);
 288             }
 289         }
 290
 291         // The header lasts until the newline
 292         else if (sc.state == SCE_MARKDOWN_HEADER1 || sc.state == SCE_MARKDOWN_HEADER2 ||
 293                 sc.state == SCE_MARKDOWN_HEADER3 || sc.state == SCE_MARKDOWN_HEADER4 ||
 294                 sc.state == SCE_MARKDOWN_HEADER5 || sc.state == SCE_MARKDOWN_HEADER6) {
 295             if (IsNewline(sc.ch))
 296                 sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 297         }
 298
 299         // New state only within the initial whitespace
 300         if (sc.state == SCE_MARKDOWN_PRECHAR) {
 301             // Blockquote
 302             if (sc.ch == '>' && precharCount < 5)
 303                 sc.SetState(SCE_MARKDOWN_BLOCKQUOTE);
 304             /*
 305             // Begin of code block
 306             else if (!HasPrevLineContent(sc) && (sc.chPrev == '\t' || precharCount >= 4))
 307                 sc.SetState(SCE_MARKDOWN_CODEBK);
 308             */
 309             // HRule - Total of three or more hyphens, asterisks, or underscores
 310             // on a line by themselves
 311             else if ((sc.ch == '-' || sc.ch == '*' || sc.ch == '_') && IsValidHrule(endPos, sc))
 312                 ;
 313             // Unordered list
 314             else if ((sc.ch == '-' || sc.ch == '*' || sc.ch == '+') && IsASpaceOrTab(sc.chNext)) {
 315                 sc.SetState(SCE_MARKDOWN_ULIST_ITEM);
 316                 sc.ForwardSetState(SCE_MARKDOWN_DEFAULT);
 317             }
 318             // Ordered list
 319             else if (IsADigit(sc.ch)) {
 320                 int digitCount = 0;
 321                 while (IsADigit(sc.GetRelative(++digitCount)))
 322                     ;
 323                 if (sc.GetRelative(digitCount) == '.' &&
 324                         IsASpaceOrTab(sc.GetRelative(digitCount + 1))) {
 325                     sc.SetState(SCE_MARKDOWN_OLIST_ITEM);
 326                     sc.Forward(digitCount + 1);
 327                     sc.SetState(SCE_MARKDOWN_DEFAULT);
 328                 }
 329             }
 330             // Alternate Ordered list
 331             else if (sc.ch == '#' && sc.chNext == '.' && IsASpaceOrTab(sc.GetRelative(2))) {
 332                 sc.SetState(SCE_MARKDOWN_OLIST_ITEM);
 333                 sc.Forward(2);
 334                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 335             }
 336             else if (sc.ch != ' ' || precharCount > 2)
 337                 sc.SetState(SCE_MARKDOWN_DEFAULT);
 338             else
 339                 ++precharCount;
 340         }
 341
 342         // New state anywhere in doc
 343         if (sc.state == SCE_MARKDOWN_DEFAULT) {
 344             if (sc.atLineStart && sc.ch == '#') {
 345                 sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 346                 freezeCursor = true;
 347             }
 348             // Links and Images
 349             if (sc.Match("![") || sc.ch == '[') {
 350                 int i = 0, j = 0, k = 0;
 351                 int len = endPos - sc.currentPos;
 352                 while (i < len && (sc.GetRelative(++i) != ']' || sc.GetRelative(i - 1) == '\\'))
 353                     ;
 354                 if (sc.GetRelative(i) == ']') {
 355                     j = i;
 356                     if (sc.GetRelative(++i) == '(') {
 357                         while (i < len && (sc.GetRelative(++i) != ')' || sc.GetRelative(i - 1) == '\\'))
 358                             ;
 359                         if (sc.GetRelative(i) == ')')
 360                             k = i;
 361                     }
 362                     else if (sc.GetRelative(i) == '[' || sc.GetRelative(++i) == '[') {
 363                         while (i < len && (sc.GetRelative(++i) != ']' || sc.GetRelative(i - 1) == '\\'))
 364                             ;
 365                         if (sc.GetRelative(i) == ']')
 366                             k = i;
 367                     }
 368                 }
 369                 // At least a link text
 370                 if (j) {
 371                     sc.SetState(SCE_MARKDOWN_LINK);
 372                     sc.Forward(j);
 373                     // Also has a URL or reference portion
 374                     if (k)
 375                         sc.Forward(k - j);
 376                     sc.ForwardSetState(SCE_MARKDOWN_DEFAULT);
 377                 }
 378             }
 379             // Code - also a special case for alternate inside spacing
 380             if (sc.Match("``") && sc.GetRelative(3) != ' ' && AtTermStart(sc)) {
 381                 sc.SetState(SCE_MARKDOWN_CODE2);
 382                 sc.Forward();
 383             }
 384             else if (sc.ch == '`' && sc.chNext != ' ' && AtTermStart(sc)) {
 385                 sc.SetState(SCE_MARKDOWN_CODE);
 386             }
 387             // Strong
 388             else if (sc.Match("**") && sc.GetRelative(2) != ' ' && AtTermStart(sc)) {
 389                 sc.SetState(SCE_MARKDOWN_STRONG1);
 390                 sc.Forward();
 391            }
 392             else if (sc.Match("__") && sc.GetRelative(2) != ' ' && AtTermStart(sc)) {
 393                 sc.SetState(SCE_MARKDOWN_STRONG2);
 394                 sc.Forward();
 395             }
 396             // Emphasis
 397             else if (sc.ch == '*' && sc.chNext != ' ' && AtTermStart(sc)) {
 398                 sc.SetState(SCE_MARKDOWN_EM1);
 399             }
 400             else if (sc.ch == '_' && sc.chNext != ' ' && AtTermStart(sc)) {
 401                 sc.SetState(SCE_MARKDOWN_EM2);
 402             }
 403             // Strikeout
 404             else if (sc.Match("~~") && sc.GetRelative(2) != ' ' && AtTermStart(sc)) {
 405                 sc.SetState(SCE_MARKDOWN_STRIKEOUT);
 406                 sc.Forward();
 407             }
 408             // Beginning of line
 409             else if (IsNewline(sc.ch)) {
 410                 sc.SetState(SCE_MARKDOWN_LINE_BEGIN);
 411             }
 412         }
 413         // Advance if not holding back the cursor for this iteration.
 414         if (!freezeCursor)
 415             sc.Forward();
 416         freezeCursor = false;
 417     }
 418     sc.Complete();
 419 }
 420
 421 LexerModule lmMarkdown(SCLEX_MARKDOWN, ColorizeMarkdownDoc, "markdown");