contrib/src/stc/scintilla/src/LexPython.cxx

   1 // Scintilla source code edit control
   2 /** @file LexPython.cxx
   3  ** Lexer for Python.
   4  **/
   5 // Copyright 1998-2001 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 enum kwType { kwOther, kwClass, kwDef, kwImport };
  24
  25 static bool IsPyComment(Accessor &styler, int pos, int len) {
  26         return len>0 && styler[pos]=='#';
  27 }
  28
  29 static bool IsPyStringStart(int ch, int chNext, int chNext2) {
  30         if (ch == '\'' || ch == '"')
  31                 return true;
  32         if (ch == 'u' || ch == 'U') {
  33                 if (chNext == '"' || chNext == '\'')
  34                         return true;
  35                 if ((chNext == 'r' || chNext == 'R') && (chNext2 == '"' || chNext2 == '\''))
  36                         return true;
  37         }
  38         if ((ch == 'r' || ch == 'R') && (chNext == '"' || chNext == '\''))
  39                 return true;
  40
  41         return false;
  42 }
  43
  44 /* Return the state to use for the string starting at i; *nextIndex will be set to the first index following the quote(s) */
  45 static int GetPyStringState(Accessor &styler, int i, int *nextIndex) {
  46         char ch = styler.SafeGetCharAt(i);
  47         char chNext = styler.SafeGetCharAt(i + 1);
  48
  49         // Advance beyond r, u, or ur prefix, but bail if there are any unexpected chars
  50         if (ch == 'r' || ch == 'R') {
  51                 i++;
  52                 ch = styler.SafeGetCharAt(i);
  53                 chNext = styler.SafeGetCharAt(i + 1);
  54         }
  55         else if (ch == 'u' || ch == 'U') {
  56                 if (chNext == 'r' || chNext == 'R')
  57                         i += 2;
  58                 else
  59                         i += 1;
  60                 ch = styler.SafeGetCharAt(i);
  61                 chNext = styler.SafeGetCharAt(i + 1);
  62         }
  63
  64         if (ch != '"' && ch != '\'') {
  65                 *nextIndex = i + 1;
  66                 return SCE_P_DEFAULT;
  67         }
  68
  69         if (ch == chNext && ch == styler.SafeGetCharAt(i + 2)) {
  70                 *nextIndex = i + 3;
  71
  72                 if (ch == '"')
  73                         return SCE_P_TRIPLEDOUBLE;
  74                 else
  75                         return SCE_P_TRIPLE;
  76         } else {
  77                 *nextIndex = i + 1;
  78
  79                 if (ch == '"')
  80                         return SCE_P_STRING;
  81                 else
  82                         return SCE_P_CHARACTER;
  83         }
  84 }
  85
  86 inline bool IsAWordChar(int  ch) {
  87         return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_');
  88 }
  89
  90 inline bool IsAWordStart(int ch) {
  91         return (ch < 0x80) && (isalnum(ch) || ch == '_');
  92 }
  93
  94 static void ColourisePyDoc(unsigned int startPos, int length, int initStyle,
  95                                                    WordList *keywordlists[], Accessor &styler) {
  96
  97         int endPos = startPos + length;
  98
  99         // Backtrack to previous line in case need to fix its tab whinging
 100         int lineCurrent = styler.GetLine(startPos);
 101         if (startPos > 0) {
 102                 if (lineCurrent > 0) {
 103                         startPos = styler.LineStart(lineCurrent-1);
 104                         if (startPos == 0)
 105                                 initStyle = SCE_P_DEFAULT;
 106                         else
 107                                 initStyle = styler.StyleAt(startPos-1);
 108                 }
 109         }
 110
 111         WordList &keywords = *keywordlists[0];
 112
 113         const int whingeLevel = styler.GetPropertyInt("tab.timmy.whinge.level");
 114
 115         initStyle = initStyle & 31;
 116         if (initStyle == SCE_P_STRINGEOL) {
 117                 initStyle = SCE_P_DEFAULT;
 118         }
 119
 120         kwType kwLast = kwOther;
 121         int spaceFlags = 0;
 122         styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
 123
 124         // Python uses a different mask because bad indentation is marked by oring with 32
 125         StyleContext sc(startPos, endPos-startPos, initStyle, styler, 0x7f);
 126
 127         for (; sc.More(); sc.Forward()) {
 128
 129                 if (sc.atLineStart) {
 130                         const char chBad = static_cast<char>(64);
 131                         const char chGood = static_cast<char>(0);
 132                         char chFlags = chGood;
 133                         if (whingeLevel == 1) {
 134                                 chFlags = (spaceFlags & wsInconsistent) ? chBad : chGood;
 135                         } else if (whingeLevel == 2) {
 136                                 chFlags = (spaceFlags & wsSpaceTab) ? chBad : chGood;
 137                         } else if (whingeLevel == 3) {
 138                                 chFlags = (spaceFlags & wsSpace) ? chBad : chGood;
 139                         } else if (whingeLevel == 4) {
 140                                 chFlags = (spaceFlags & wsTab) ? chBad : chGood;
 141                         }
 142                         styler.SetFlags(chFlags, static_cast<char>(sc.state));
 143                 }
 144
 145                 if (sc.atLineEnd) {
 146                         if ((sc.state == SCE_P_DEFAULT) ||
 147                                 (sc.state == SCE_P_TRIPLE) ||
 148                                 (sc.state == SCE_P_TRIPLEDOUBLE)) {
 149                                 // Perform colourisation of white space and triple quoted strings at end of each line to allow
 150                                 // tab marking to work inside white space and triple quoted strings
 151                                 sc.ForwardSetState(sc.state);
 152                         }
 153                         lineCurrent++;
 154                         styler.IndentAmount(lineCurrent, &spaceFlags, IsPyComment);
 155                         if ((sc.state == SCE_P_STRING) || (sc.state == SCE_P_CHARACTER)) {
 156                                 sc.ChangeState(SCE_P_STRINGEOL);
 157                                 sc.ForwardSetState(SCE_P_DEFAULT);
 158                         }
 159                 }
 160
 161                 // Check for a state end
 162                 if (sc.state == SCE_P_OPERATOR) {
 163                         kwLast = kwOther;
 164                         sc.SetState(SCE_C_DEFAULT);
 165                 } else if (sc.state == SCE_P_NUMBER) {
 166                         if (!IsAWordChar(sc.ch)) {
 167                                 sc.SetState(SCE_P_DEFAULT);
 168                         }
 169                 } else if (sc.state == SCE_P_WORD) {
 170                         if ((sc.ch == '.') || (!IsAWordChar(sc.ch))) {
 171                                 char s[100];
 172                                 sc.GetCurrent(s, sizeof(s));
 173                                 int style = SCE_P_IDENTIFIER;
 174                                 if ((kwLast == kwImport) && (strcmp(s, "as") == 0)) {
 175                                         style = SCE_P_WORD;
 176                                 } else if (keywords.InList(s)) {
 177                                         style = SCE_P_WORD;
 178                                 } else if (kwLast == kwClass) {
 179                                         style = SCE_P_CLASSNAME;
 180                                 } else if (kwLast == kwDef) {
 181                                         style = SCE_P_DEFNAME;
 182                                 }
 183                                 sc.ChangeState(style);
 184                                 sc.SetState(SCE_P_DEFAULT);
 185                                 if (style == SCE_P_WORD) {
 186                                         if (0 == strcmp(s, "class"))
 187                                                 kwLast = kwClass;
 188                                         else if (0 == strcmp(s, "def"))
 189                                                 kwLast = kwDef;
 190                                         else if (0 == strcmp(s, "import"))
 191                                                 kwLast = kwImport;
 192                                         else
 193                                                 kwLast = kwOther;
 194                                 } else if (style == SCE_P_CLASSNAME) {
 195                                                 kwLast = kwOther;
 196                                 } else if (style == SCE_P_DEFNAME) {
 197                                                 kwLast = kwOther;
 198                                 }
 199                         }
 200                 } else if ((sc.state == SCE_P_COMMENTLINE) || (sc.state == SCE_P_COMMENTBLOCK)) {
 201                         if (sc.ch == '\r' || sc.ch == '\n') {
 202                                 sc.SetState(SCE_P_DEFAULT);
 203                         }
 204                 } else if ((sc.state == SCE_P_STRING) || (sc.state == SCE_P_CHARACTER)) {
 205                         if (sc.ch == '\\') {
 206                                 if ((sc.chNext == '\r') && (sc.GetRelative(2) == '\n')) {
 207                                         sc.Forward();
 208                                 }
 209                                 sc.Forward();
 210                         } else if ((sc.state == SCE_P_STRING) && (sc.ch == '\"')) {
 211                                 sc.ForwardSetState(SCE_P_DEFAULT);
 212                         } else if ((sc.state == SCE_P_CHARACTER) && (sc.ch == '\'')) {
 213                                 sc.ForwardSetState(SCE_P_DEFAULT);
 214                         }
 215                 } else if (sc.state == SCE_P_TRIPLE) {
 216                         if (sc.ch == '\\') {
 217                                 sc.Forward();
 218                         } else if (sc.Match("\'\'\'")) {
 219                                 sc.Forward();
 220                                 sc.Forward();
 221                                 sc.ForwardSetState(SCE_P_DEFAULT);
 222                         }
 223                 } else if (sc.state == SCE_P_TRIPLEDOUBLE) {
 224                         if (sc.ch == '\\') {
 225                                 sc.Forward();
 226                         } else if (sc.Match("\"\"\"")) {
 227                                 sc.Forward();
 228                                 sc.Forward();
 229                                 sc.ForwardSetState(SCE_P_DEFAULT);
 230                         }
 231                 }
 232
 233                 // Check for a new state starting character
 234                 if (sc.state == SCE_P_DEFAULT) {
 235                         if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch)) || sc.ch == '`') {
 236                                 sc.SetState(SCE_P_OPERATOR);
 237                         } else if (sc.ch == '#') {
 238                                 sc.SetState(sc.chNext == '#' ? SCE_P_COMMENTBLOCK : SCE_P_COMMENTLINE);
 239                         } else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 240                                 sc.SetState(SCE_P_NUMBER);
 241                         } else if (IsPyStringStart(sc.ch, sc.chNext, sc.GetRelative(2))) {
 242                                 int nextIndex = 0;
 243                                 sc.SetState(GetPyStringState(styler, sc.currentPos, &nextIndex));
 244                                 while (nextIndex > (sc.currentPos+1)) {
 245                                         sc.Forward();
 246                                 }
 247                         } else if (IsAWordStart(sc.ch)) {
 248                                 sc.SetState(SCE_P_WORD);
 249                         }
 250                 }
 251         }
 252         sc.Complete();
 253 }
 254
 255 static bool IsCommentLine(int line, Accessor &styler) {
 256         int pos = styler.LineStart(line);
 257         int eol_pos = styler.LineStart(line+1) - 1;
 258         for (int i = pos; i < eol_pos; i++) {
 259                 char ch = styler[i];
 260                 if (ch == '#')
 261                         return true;
 262                 else if (ch != ' ' && ch != '\t')
 263                         return false;
 264         }
 265         return false;
 266 }
 267
 268 static bool IsQuoteLine(int line, Accessor &styler) {
 269         int style = styler.StyleAt(styler.LineStart(line)) & 31;
 270         return ((style == SCE_P_TRIPLE) || (style== SCE_P_TRIPLEDOUBLE));
 271 }
 272
 273
 274 static void FoldPyDoc(unsigned int startPos, int length, int /*initStyle - unused*/,
 275                                         WordList *[], Accessor &styler) {
 276         const int maxPos = startPos + length;
 277         const int maxLines = styler.GetLine(maxPos-1);             // Requested last line
 278         const int docLines = styler.GetLine(styler.Length() - 1);  // Available last line
 279         const bool foldComment = styler.GetPropertyInt("fold.comment.python");
 280         const bool foldQuotes = styler.GetPropertyInt("fold.quotes.python");
 281
 282         // Backtrack to previous non-blank line so we can determine indent level
 283         // for any white space lines (needed esp. within triple quoted strings)
 284         // and so we can fix any preceding fold level (which is why we go back
 285         // at least one line in all cases)
 286         int spaceFlags = 0;
 287         int lineCurrent = styler.GetLine(startPos);
 288         int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
 289         while (lineCurrent > 0) {
 290                 lineCurrent--;
 291                 indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, NULL);
 292                 if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG) &&
 293                         (!IsCommentLine(lineCurrent, styler)) &&
 294                         (!IsQuoteLine(lineCurrent, styler)))
 295                         break;
 296         }
 297         int indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
 298
 299         // Set up initial loop state
 300         startPos = styler.LineStart(lineCurrent);
 301         int prev_state = SCE_P_DEFAULT & 31;
 302         if (lineCurrent >= 1)
 303                 prev_state = styler.StyleAt(startPos-1) & 31;
 304         int prevQuote = foldQuotes && ((prev_state == SCE_P_TRIPLE) || (prev_state == SCE_P_TRIPLEDOUBLE));
 305         int prevComment = 0;
 306         if (lineCurrent >= 1)
 307                 prevComment = foldComment && IsCommentLine(lineCurrent - 1, styler);
 308
 309         // Process all characters to end of requested range or end of any triple quote
 310         // or comment that hangs over the end of the range.  Cap processing in all cases
 311         // to end of document (in case of unclosed quote or comment at end).
 312         while ((lineCurrent <= docLines) && ((lineCurrent <= maxLines) || prevQuote || prevComment)) {
 313
 314                 // Gather info
 315                 int lev = indentCurrent;
 316                 int lineNext = lineCurrent + 1;
 317                 int indentNext = indentCurrent;
 318                 int quote = false;
 319                 if (lineNext <= docLines) {
 320                         // Information about next line is only available if not at end of document
 321                         indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
 322                         int style = styler.StyleAt(styler.LineStart(lineNext)) & 31;
 323                         quote = foldQuotes && ((style == SCE_P_TRIPLE) || (style== SCE_P_TRIPLEDOUBLE));
 324                 }
 325                 const int quote_start = (quote && !prevQuote);
 326                 const int quote_continue = (quote && prevQuote);
 327                 const int comment = foldComment && IsCommentLine(lineCurrent, styler);
 328                 const int comment_start = (comment && !prevComment && (lineNext <= docLines) &&
 329                         IsCommentLine(lineNext, styler) && (lev > SC_FOLDLEVELBASE));
 330                 const int comment_continue = (comment && prevComment);
 331                 if ((!quote || !prevQuote) && !comment)
 332                         indentCurrentLevel = indentCurrent & SC_FOLDLEVELNUMBERMASK;
 333                 if (quote)
 334                         indentNext = indentCurrentLevel;
 335                 if (indentNext & SC_FOLDLEVELWHITEFLAG)
 336                         indentNext = SC_FOLDLEVELWHITEFLAG | indentCurrentLevel;
 337
 338                 if (quote_start) {
 339                         // Place fold point at start of triple quoted string
 340                         lev |= SC_FOLDLEVELHEADERFLAG;
 341                 } else if (quote_continue || prevQuote) {
 342                         // Add level to rest of lines in the string
 343                         lev = lev + 1;
 344                 } else if (comment_start) {
 345                         // Place fold point at start of a block of comments
 346                         lev |= SC_FOLDLEVELHEADERFLAG;
 347                 } else if (comment_continue) {
 348                         // Add level to rest of lines in the block
 349                         lev = lev + 1;
 350                 }
 351
 352                 // Skip past any blank lines for next indent level info; we skip also comments
 353                 // starting in column 0 which effectively folds them into surrounding code rather
 354                 // than screwing up folding.
 355                 const int saveIndentNext = indentNext;
 356                 while (!quote &&
 357                        (lineNext < docLines) &&
 358                        ((indentNext & SC_FOLDLEVELWHITEFLAG) ||
 359                         (lineNext <= docLines && styler[styler.LineStart(lineNext)] == '#'))) {
 360
 361                         lineNext++;
 362                         indentNext = styler.IndentAmount(lineNext, &spaceFlags, NULL);
 363                 }
 364
 365                 // Next compute max indent level of current line and next non-blank line.
 366                 // This is the level to which we set all the intervening blank or comment lines.
 367                 const int skip_level = Platform::Maximum(indentCurrentLevel,
 368                                                    indentNext & SC_FOLDLEVELNUMBERMASK);
 369
 370                 // Now set all the indent levels on the lines we skipped
 371                 int skipLine = lineCurrent + 1;
 372                 int skipIndentNext = saveIndentNext;
 373                 while (skipLine < lineNext) {
 374                         int skipLineLevel = skip_level;
 375                         if (skipIndentNext & SC_FOLDLEVELWHITEFLAG)
 376                                 skipLineLevel = SC_FOLDLEVELWHITEFLAG | skipLineLevel;
 377                         styler.SetLevel(skipLine, skipLineLevel);
 378                         skipLine++;
 379                         skipIndentNext = styler.IndentAmount(skipLine, &spaceFlags, NULL);
 380                 }
 381
 382                 // Set fold header on non-quote/non-comment line
 383                 if (!quote && !comment && !(indentCurrent & SC_FOLDLEVELWHITEFLAG) ) {
 384                         if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK))
 385                                 lev |= SC_FOLDLEVELHEADERFLAG;
 386                 }
 387
 388                 // Keep track of triple quote and block comment state of previous line
 389                 prevQuote = quote;
 390                 prevComment = comment_start || comment_continue;
 391
 392                 // Set fold level for this line and move to next line
 393                 styler.SetLevel(lineCurrent, lev);
 394                 indentCurrent = indentNext;
 395                 lineCurrent = lineNext;
 396         }
 397
 398         // NOTE: Cannot set level of last line here because indentCurrent doesn't have
 399         // header flag set; the loop above is crafted to take care of this case!
 400         //styler.SetLevel(lineCurrent, indentCurrent);
 401 }
 402
 403 LexerModule lmPython(SCLEX_PYTHON, ColourisePyDoc, "python", FoldPyDoc);