ext/scintilla/src/LexPython.cxx

   1 // Scintilla source code edit control
   2 /** @file LexPython.cxx
   3  ** Lexer for Python.
   4  **/
   5 // Copyright 1998-2002 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #include <stdlib.h>
   9 #include <string.h>
  10 #include <ctype.h>
  11 #include <stdio.h>
  12 #include <stdarg.h>
  13
  14 #include "Platform.h"
  15
  16 #include "PropSet.h"
  17 #include "Accessor.h"
  18 #include "StyleContext.h"
  19 #include "KeyWords.h"
  20 #include "Scintilla.h"
  21 #include "SciLexer.h"
  22
  23 #ifdef SCI_NAMESPACE
  24 using namespace Scintilla;
  25 #endif
  26
  27 enum kwType { kwOther, kwClass, kwDef, kwImport };
  28 static const int indicatorWhitespace = 1;
  29
  30 static bool IsPyComment(Accessor &styler, int pos, int len) {
  31         return len > 0 && styler[pos] == '#';
  32 }
  33
  34 static bool IsPyStringStart(int ch, int chNext, int chNext2) {
  35         if (ch == '\'' || ch == '"')
  36                 return true;
  37         if (ch == 'u' || ch == 'U') {
  38                 if (chNext == '"' || chNext == '\'')
  39                         return true;
  40                 if ((chNext == 'r' || chNext == 'R') && (chNext2 == '"' || chNext2 == '\''))
  41                         return true;
  42         }
  43         if ((ch == 'r' || ch == 'R') && (chNext == '"' || chNext == '\''))
  44                 return true;
  45
  46         return false;
  47 }
  48
  49 /* Return the state to use for the string starting at i; *nextIndex will be set to the first index following the quote(s) */
  50 static int GetPyStringState(Accessor &styler, int i, unsigned int *nextIndex) {
  51         char ch = styler.SafeGetCharAt(i);
  52         char chNext = styler.SafeGetCharAt(i + 1);
  53
  54         // Advance beyond r, u, or ur prefix, but bail if there are any unexpected chars
  55         if (ch == 'r' || ch == 'R') {
  56                 i++;
  57                 ch = styler.SafeGetCharAt(i);
  58                 chNext = styler.SafeGetCharAt(i + 1);
  59         } else if (ch == 'u' || ch == 'U') {
  60                 if (chNext == 'r' || chNext == 'R')
  61                         i += 2;
  62                 else
  63                         i += 1;
  64                 ch = styler.SafeGetCharAt(i);
  65                 chNext = styler.SafeGetCharAt(i + 1);
  66         }
  67
  68         if (ch != '"' && ch != '\'') {
  69                 *nextIndex = i + 1;
  70                 return SCE_P_DEFAULT;
  71         }
  72
  73         if (ch == chNext && ch == styler.SafeGetCharAt(i + 2)) {
  74                 *nextIndex = i + 3;
  75
  76                 if (ch == '"')
  77                         return SCE_P_TRIPLEDOUBLE;
  78                 else
  79                         return SCE_P_TRIPLE;
  80         } else {
  81                 *nextIndex = i + 1;
  82
  83                 if (ch == '"')
  84                         return SCE_P_STRING;
  85                 else
  86                         return SCE_P_CHARACTER;
  87         }
  88 }
  89
  90 static inline bool IsAWordChar(int ch) {
  91         return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_');
  92 }
  93
  94 static inline bool IsAWordStart(int ch) {
  95         return (ch < 0x80) && (isalnum(ch) || ch == '_');
  96 }
  97
  98 static void ColourisePyDoc(unsigned int startPos, int length, int initStyle,
  99                            WordList *keywordlists[], Accessor &styler) {
 100
 101         int endPos = startPos + length;
 102
 103         // Backtrack to previous line in case need to fix its tab whinging
 104         int lineCurrent = styler.GetLine(startPos);
 105         if (startPos > 0) {
 106                 if (lineCurrent > 0) {
 107                         lineCurrent--;
 108                         startPos = styler.LineStart(lineCurrent);
 109                         if (startPos == 0)
 110                                 initStyle = SCE_P_DEFAULT;
 111                         else
 112                                 initStyle = styler.StyleAt(startPos - 1);
 113                 }
 114         }
 115
 116         WordList &keywords = *keywordlists[0];
 117         WordList &keywords2 = *keywordlists[1];
 118
 119         const int whingeLevel = styler.GetPropertyInt("tab.timmy.whinge.level");
 120
 121         initStyle = initStyle & 31;
 122         if (initStyle == SCE_P_STRINGEOL) {
 123                 initStyle = SCE_P_DEFAULT;
 124         }
 125
 126         kwType kwLast = kwOther;
 127         int spaceFlags = 0;
 128         styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
 129         bool hexadecimal = false;
 130
 131         StyleContext sc(startPos, endPos - startPos, initStyle, styler);
 132
 133         bool indentGood = true;
 134         int startIndicator = sc.currentPos;
 135
 136         for (; sc.More(); sc.Forward()) {
 137
 138                 if (sc.atLineStart) {
 139                         styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
 140                         indentGood = true;
 141                         if (whingeLevel == 1) {
 142                                 indentGood = (spaceFlags & wsInconsistent) == 0;
 143                         } else if (whingeLevel == 2) {
 144                                 indentGood = (spaceFlags & wsSpaceTab) == 0;
 145                         } else if (whingeLevel == 3) {
 146                                 indentGood = (spaceFlags & wsSpace) == 0;
 147                         } else if (whingeLevel == 4) {
 148                                 indentGood = (spaceFlags & wsTab) == 0;
 149                         }
 150                         if (!indentGood) {
 151                                 styler.IndicatorFill(startIndicator, sc.currentPos, indicatorWhitespace, 0);
 152                                 startIndicator = sc.currentPos;
 153                         }
 154                 }
 155
 156                 if (sc.atLineEnd) {
 157                         if ((sc.state == SCE_P_DEFAULT) ||
 158                                 (sc.state == SCE_P_TRIPLE) ||
 159                                 (sc.state == SCE_P_TRIPLEDOUBLE)) {
 160                                 // Perform colourisation of white space and triple quoted strings at end of each line to allow
 161                                 // tab marking to work inside white space and triple quoted strings
 162                                 sc.SetState(sc.state);
 163                         }
 164                         lineCurrent++;
 165                         if ((sc.state == SCE_P_STRING) || (sc.state == SCE_P_CHARACTER)) {
 166                                 sc.ChangeState(SCE_P_STRINGEOL);
 167                                 sc.ForwardSetState(SCE_P_DEFAULT);
 168                         }
 169                         if (!sc.More())
 170                                 break;
 171                 }
 172
 173                 bool needEOLCheck = false;
 174
 175                 // Check for a state end
 176                 if (sc.state == SCE_P_OPERATOR) {
 177                         kwLast = kwOther;
 178                         sc.SetState(SCE_P_DEFAULT);
 179                 } else if (sc.state == SCE_P_NUMBER) {
 180                         if (!IsAWordChar(sc.ch) &&
 181                                 !(!hexadecimal && ((sc.ch == '+' || sc.ch == '-') && (sc.chPrev == 'e' || sc.chPrev == 'E')))) {
 182                                 sc.SetState(SCE_P_DEFAULT);
 183                         }
 184                 } else if (sc.state == SCE_P_IDENTIFIER) {
 185                         if ((sc.ch == '.') || (!IsAWordChar(sc.ch))) {
 186                                 char s[100];
 187                                 sc.GetCurrent(s, sizeof(s));
 188                                 int style = SCE_P_IDENTIFIER;
 189                                 if ((kwLast == kwImport) && (strcmp(s, "as") == 0)) {
 190                                         style = SCE_P_WORD;
 191                                 } else if (keywords.InList(s)) {
 192                                         style = SCE_P_WORD;
 193                                 } else if (kwLast == kwClass) {
 194                                         style = SCE_P_CLASSNAME;
 195                                 } else if (kwLast == kwDef) {
 196                                         style = SCE_P_DEFNAME;
 197                                 } else if (keywords2.InList(s)) {
 198                                         style = SCE_P_WORD2;
 199                                 }
 200                                 sc.ChangeState(style);
 201                                 sc.SetState(SCE_P_DEFAULT);
 202                                 if (style == SCE_P_WORD) {
 203                                         if (0 == strcmp(s, "class"))
 204                                                 kwLast = kwClass;
 205                                         else if (0 == strcmp(s, "def"))
 206                                                 kwLast = kwDef;
 207                                         else if (0 == strcmp(s, "import"))
 208                                                 kwLast = kwImport;
 209                                         else
 210                                                 kwLast = kwOther;
 211                                 } else {
 212                                         kwLast = kwOther;
 213                                 }
 214                         }
 215                 } else if ((sc.state == SCE_P_COMMENTLINE) || (sc.state == SCE_P_COMMENTBLOCK)) {
 216                         if (sc.ch == '\r' || sc.ch == '\n') {
 217                                 sc.SetState(SCE_P_DEFAULT);
 218                         }
 219                 } else if (sc.state == SCE_P_DECORATOR) {
 220                         if (!IsAWordChar(sc.ch)) {
 221                                 sc.SetState(SCE_P_DEFAULT);
 222                         }
 223                 } else if ((sc.state == SCE_P_STRING) || (sc.state == SCE_P_CHARACTER)) {
 224                         if (sc.ch == '\\') {
 225                                 if ((sc.chNext == '\r') && (sc.GetRelative(2) == '\n')) {
 226                                         sc.Forward();
 227                                 }
 228                                 sc.Forward();
 229                         } else if ((sc.state == SCE_P_STRING) && (sc.ch == '\"')) {
 230                                 sc.ForwardSetState(SCE_P_DEFAULT);
 231                                 needEOLCheck = true;
 232                         } else if ((sc.state == SCE_P_CHARACTER) && (sc.ch == '\'')) {
 233                                 sc.ForwardSetState(SCE_P_DEFAULT);
 234                                 needEOLCheck = true;
 235                         }
 236                 } else if (sc.state == SCE_P_TRIPLE) {
 237                         if (sc.ch == '\\') {
 238                                 sc.Forward();
 239                         } else if (sc.Match("\'\'\'")) {
 240                                 sc.Forward();
 241                                 sc.Forward();
 242                                 sc.ForwardSetState(SCE_P_DEFAULT);
 243                                 needEOLCheck = true;
 244                         }
 245                 } else if (sc.state == SCE_P_TRIPLEDOUBLE) {
 246                         if (sc.ch == '\\') {
 247                                 sc.Forward();
 248                         } else if (sc.Match("\"\"\"")) {
 249                                 sc.Forward();
 250                                 sc.Forward();
 251                                 sc.ForwardSetState(SCE_P_DEFAULT);
 252                                 needEOLCheck = true;
 253                         }
 254                 }
 255
 256                 if (!indentGood && !IsASpaceOrTab(sc.ch)) {
 257                         styler.IndicatorFill(startIndicator, sc.currentPos, indicatorWhitespace, 1);
 258                         startIndicator = sc.currentPos;
 259                         indentGood = true;
 260                 }
 261
 262                 // State exit code may have moved on to end of line
 263                 if (needEOLCheck && sc.atLineEnd) {
 264                         lineCurrent++;
 265                         styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
 266                         if (!sc.More())
 267                                 break;
 268                 }
 269
 270                 // Check for a new state starting character
 271                 if (sc.state == SCE_P_DEFAULT) {
 272                         if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 273                                 if (sc.ch == '0' && (sc.chNext == 'x' || sc.chNext == 'X')) {
 274                                         hexadecimal = true;
 275                                 } else {
 276                                         hexadecimal = false;
 277                                 }
 278                                 sc.SetState(SCE_P_NUMBER);
 279                         } else if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch)) || sc.ch == '`') {
 280                                 sc.SetState(SCE_P_OPERATOR);
 281                         } else if (sc.ch == '#') {
 282                                 sc.SetState(sc.chNext == '#' ? SCE_P_COMMENTBLOCK : SCE_P_COMMENTLINE);
 283                         } else if (sc.ch == '@') {
 284                                 sc.SetState(SCE_P_DECORATOR);
 285                         } else if (IsPyStringStart(sc.ch, sc.chNext, sc.GetRelative(2))) {
 286                                 unsigned int nextIndex = 0;
 287                                 sc.SetState(GetPyStringState(styler, sc.currentPos, &nextIndex));
 288                                 while (nextIndex > (sc.currentPos + 1) && sc.More()) {
 289                                         sc.Forward();
 290                                 }
 291                         } else if (IsAWordStart(sc.ch)) {
 292                                 sc.SetState(SCE_P_IDENTIFIER);
 293                         }
 294                 }
 295         }
 296         styler.IndicatorFill(startIndicator, sc.currentPos, indicatorWhitespace, 0);
 297         sc.Complete();
 298 }
 299
 300 static bool IsCommentLine(int line, Accessor &styler) {
 301         int pos = styler.LineStart(line);
 302         int eol_pos = styler.LineStart(line + 1) - 1;
 303         for (int i = pos; i < eol_pos; i++) {
 304                 char ch = styler[i];
 305                 if (ch == '#')
 306                         return true;
 307                 else if (ch != ' ' && ch != '\t')
 308                         return false;
 309         }
 310         return false;
 311 }
 312
 313 static bool IsQuoteLine(int line, Accessor &styler) {
 314         int style = styler.StyleAt(styler.LineStart(line)) & 31;
 315         return ((style == SCE_P_TRIPLE) || (style == SCE_P_TRIPLEDOUBLE));
 316 }
 317
 318
 319 static void FoldPyDoc(unsigned int startPos, int length, int /*initStyle - unused*/,
 320                       WordList *[], Accessor &styler) {
 321         const int maxPos = startPos + length;
 322         const int maxLines = styler.GetLine(maxPos - 1);             // Requested last line
 323         const int docLines = styler.GetLine(styler.Length() - 1);  // Available last line
 324         const bool foldComment = styler.GetPropertyInt("fold.comment.python") != 0;
 325         const bool foldQuotes = styler.GetPropertyInt("fold.quotes.python") != 0;
 326
 327         // Backtrack to previous non-blank line so we can determine indent level
 328         // for any white space lines (needed esp. within triple quoted strings)
 329         // and so we can fix any preceding fold level (which is why we go back
 330         // at least one line in all cases)
 331         int spaceFlags = 0;
 332         int lineCurrent = styler.GetLine(startPos);
 333         int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
 334         while (lineCurrent > 0) {
 335                 lineCurrent--;
 336                 indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
 337                 if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG) &&
 338                         (!IsCommentLine(lineCurrent, styler)) &&
 339                         (!IsQuoteLine(lineCurrent, styler)))
 340                         break;
 341         }
 342         int indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
 343
 344         // Set up initial loop state
 345         startPos = styler.LineStart(lineCurrent);
 346         int prev_state = SCE_P_DEFAULT & 31;
 347         if (lineCurrent >= 1)
 348                 prev_state = styler.StyleAt(startPos - 1) & 31;
 349         int prevQuote = foldQuotes && ((prev_state == SCE_P_TRIPLE) || (prev_state == SCE_P_TRIPLEDOUBLE));
 350         int prevComment = 0;
 351         if (lineCurrent >= 1)
 352                 prevComment = foldComment && IsCommentLine(lineCurrent - 1, styler);
 353
 354         // Process all characters to end of requested range or end of any triple quote
 355         // or comment that hangs over the end of the range.  Cap processing in all cases
 356         // to end of document (in case of unclosed quote or comment at end).
 357         while ((lineCurrent <= docLines) && ((lineCurrent <= maxLines) || prevQuote || prevComment)) {
 358
 359                 // Gather info
 360                 int lev = indentCurrent;
 361                 int lineNext = lineCurrent + 1;
 362                 int indentNext = indentCurrent;
 363                 int quote = false;
 364                 if (lineNext <= docLines) {
 365                         // Information about next line is only available if not at end of document
 366                         indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
 367                         int style = styler.StyleAt(styler.LineStart(lineNext)) & 31;
 368                         quote = foldQuotes && ((style == SCE_P_TRIPLE) || (style == SCE_P_TRIPLEDOUBLE));
 369                 }
 370                 const int quote_start = (quote && !prevQuote);
 371                 const int quote_continue = (quote && prevQuote);
 372                 const int comment = foldComment && IsCommentLine(lineCurrent, styler);
 373                 const int comment_start = (comment && !prevComment && (lineNext <= docLines) &&
 374                                            IsCommentLine(lineNext, styler) && (lev > SC_FOLDLEVELBASE));
 375                 const int comment_continue = (comment && prevComment);
 376                 if ((!quote || !prevQuote) && !comment)
 377                         indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
 378                 if (quote)
 379                         indentNext = indentCurrentLevel;
 380                 if (indentNext & SC_FOLDLEVELWHITEFLAG)
 381                         indentNext = SC_FOLDLEVELWHITEFLAG | indentCurrentLevel;
 382
 383                 if (quote_start) {
 384                         // Place fold point at start of triple quoted string
 385                         lev |= SC_FOLDLEVELHEADERFLAG;
 386                 } else if (quote_continue || prevQuote) {
 387                         // Add level to rest of lines in the string
 388                         lev = lev + 1;
 389                 } else if (comment_start) {
 390                         // Place fold point at start of a block of comments
 391                         lev |= SC_FOLDLEVELHEADERFLAG;
 392                 } else if (comment_continue) {
 393                         // Add level to rest of lines in the block
 394                         lev = lev + 1;
 395                 }
 396
 397                 // Skip past any blank lines for next indent level info; we skip also
 398                 // comments (all comments, not just those starting in column 0)
 399                 // which effectively folds them into surrounding code rather
 400                 // than screwing up folding.
 401
 402                 while (!quote &&
 403                         (lineNext < docLines) &&
 404                         ((indentNext & SC_FOLDLEVELWHITEFLAG) ||
 405                          (lineNext <= docLines && IsCommentLine(lineNext, styler)))) {
 406
 407                         lineNext++;
 408                         indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
 409                 }
 410
 411                 const int levelAfterComments = indentNext & SC_FOLDLEVELNUMBERMASK;
 412                 const int levelBeforeComments = Platform::Maximum(indentCurrentLevel,levelAfterComments);
 413
 414                 // Now set all the indent levels on the lines we skipped
 415                 // Do this from end to start.  Once we encounter one line
 416                 // which is indented more than the line after the end of
 417                 // the comment-block, use the level of the block before
 418
 419                 int skipLine = lineNext;
 420                 int skipLevel = levelAfterComments;
 421
 422                 while (--skipLine > lineCurrent) {
 423                         int skipLineIndent = styler.IndentAmount(skipLine, &spaceFlags, NULL);
 424
 425                         if ((skipLineIndent & SC_FOLDLEVELNUMBERMASK) > levelAfterComments)
 426                                 skipLevel = levelBeforeComments;
 427
 428                         int whiteFlag = skipLineIndent & SC_FOLDLEVELWHITEFLAG;
 429
 430                         styler.SetLevel(skipLine, skipLevel | whiteFlag);
 431                 }
 432
 433                 // Set fold header on non-quote/non-comment line
 434                 if (!quote && !comment && !(indentCurrent & SC_FOLDLEVELWHITEFLAG) ) {
 435                         if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK))
 436                                 lev |= SC_FOLDLEVELHEADERFLAG;
 437                 }
 438
 439                 // Keep track of triple quote and block comment state of previous line
 440                 prevQuote = quote;
 441                 prevComment = comment_start || comment_continue;
 442
 443                 // Set fold level for this line and move to next line
 444                 styler.SetLevel(lineCurrent, lev);
 445                 indentCurrent = indentNext;
 446                 lineCurrent = lineNext;
 447         }
 448
 449         // NOTE: Cannot set level of last line here because indentCurrent doesn't have
 450         // header flag set; the loop above is crafted to take care of this case!
 451         //styler.SetLevel(lineCurrent, indentCurrent);
 452 }
 453
 454 static const char * const pythonWordListDesc[] = {
 455         "Keywords",
 456         "Highlighted identifiers",
 457         0
 458 };
 459
 460 LexerModule lmPython(SCLEX_PYTHON, ColourisePyDoc, "python", FoldPyDoc,
 461                                          pythonWordListDesc);