contrib/src/stc/scintilla/src/Document.cxx

   1 // Scintilla source code edit control
   2 /** @file Document.cxx
   3  ** Text document that handles notifications, DBCS, styling, words and end of line.
   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 <stdio.h>
  11 #include <ctype.h>
  12
  13 #include "Platform.h"
  14
  15 #include "Scintilla.h"
  16 #include "SVector.h"
  17 #include "CellBuffer.h"
  18 #include "Document.h"
  19 #include "RESearch.h"
  20
  21 // This is ASCII specific but is safe with chars >= 0x80
  22 static inline bool isspacechar(unsigned char ch) {
  23         return (ch == ' ') || ((ch >= 0x09) && (ch <= 0x0d));
  24 }
  25
  26 Document::Document() {
  27         refCount = 0;
  28 #ifdef unix
  29         eolMode = SC_EOL_LF;
  30 #else
  31         eolMode = SC_EOL_CRLF;
  32 #endif
  33         dbcsCodePage = 0;
  34         stylingBits = 5;
  35         stylingBitsMask = 0x1F;
  36         stylingMask = 0;
  37         SetWordChars(0);
  38         endStyled = 0;
  39         styleClock = 0;
  40         enteredCount = 0;
  41         enteredReadOnlyCount = 0;
  42         tabInChars = 8;
  43         indentInChars = 0;
  44         useTabs = true;
  45         tabIndents = true;
  46         backspaceUnindents = false;
  47         watchers = 0;
  48         lenWatchers = 0;
  49
  50         matchesValid = false;
  51         pre = 0;
  52         substituted = 0;
  53 }
  54
  55 Document::~Document() {
  56         for (int i = 0; i < lenWatchers; i++) {
  57                 watchers[i].watcher->NotifyDeleted(this, watchers[i].userData);
  58         }
  59         delete []watchers;
  60         watchers = 0;
  61         lenWatchers = 0;
  62         delete pre;
  63         pre = 0;
  64         delete []substituted;
  65         substituted = 0;
  66 }
  67
  68 // Increase reference count and return its previous value.
  69 int Document::AddRef() {
  70         return refCount++;
  71 }
  72
  73 // Decrease reference count and return its previous value.
  74 // Delete the document if reference count reaches zero.
  75 int Document::Release() {
  76         int curRefCount = --refCount;
  77         if (curRefCount == 0)
  78                 delete this;
  79         return curRefCount;
  80 }
  81
  82 void Document::SetSavePoint() {
  83         cb.SetSavePoint();
  84         NotifySavePoint(true);
  85 }
  86
  87 int Document::AddMark(int line, int markerNum) {
  88         int prev = cb.AddMark(line, markerNum);
  89         DocModification mh(SC_MOD_CHANGEMARKER, LineStart(line), 0, 0, 0);
  90         NotifyModified(mh);
  91         return prev;
  92 }
  93
  94 void Document::DeleteMark(int line, int markerNum) {
  95         cb.DeleteMark(line, markerNum);
  96         DocModification mh(SC_MOD_CHANGEMARKER, LineStart(line), 0, 0, 0);
  97         NotifyModified(mh);
  98 }
  99
 100 void Document::DeleteMarkFromHandle(int markerHandle) {
 101         cb.DeleteMarkFromHandle(markerHandle);
 102         DocModification mh(SC_MOD_CHANGEMARKER, 0, 0, 0, 0);
 103         NotifyModified(mh);
 104 }
 105
 106 void Document::DeleteAllMarks(int markerNum) {
 107         cb.DeleteAllMarks(markerNum);
 108         DocModification mh(SC_MOD_CHANGEMARKER, 0, 0, 0, 0);
 109         NotifyModified(mh);
 110 }
 111
 112 int Document::LineStart(int line) {
 113         return cb.LineStart(line);
 114 }
 115
 116 int Document::LineEnd(int line) {
 117         if (line == LinesTotal() - 1) {
 118                 return LineStart(line + 1);
 119         } else {
 120                 int position = LineStart(line + 1) - 1;
 121                 // When line terminator is CR+LF, may need to go back one more
 122                 if ((position > LineStart(line)) && (cb.CharAt(position - 1) == '\r')) {
 123                         position--;
 124                 }
 125                 return position;
 126         }
 127 }
 128
 129 int Document::LineFromPosition(int pos) {
 130         return cb.LineFromPosition(pos);
 131 }
 132
 133 int Document::LineEndPosition(int position) {
 134         return LineEnd(LineFromPosition(position));
 135 }
 136
 137 int Document::VCHomePosition(int position) {
 138         int line = LineFromPosition(position);
 139         int startPosition = LineStart(line);
 140         int endLine = LineStart(line + 1) - 1;
 141         int startText = startPosition;
 142         while (startText < endLine && (cb.CharAt(startText) == ' ' || cb.CharAt(startText) == '\t' ) )
 143                 startText++;
 144         if (position == startText)
 145                 return startPosition;
 146         else
 147                 return startText;
 148 }
 149
 150 int Document::SetLevel(int line, int level) {
 151         int prev = cb.SetLevel(line, level);
 152         if (prev != level) {
 153                 DocModification mh(SC_MOD_CHANGEFOLD | SC_MOD_CHANGEMARKER,
 154                                    LineStart(line), 0, 0, 0);
 155                 mh.line = line;
 156                 mh.foldLevelNow = level;
 157                 mh.foldLevelPrev = prev;
 158                 NotifyModified(mh);
 159         }
 160         return prev;
 161 }
 162
 163 static bool IsSubordinate(int levelStart, int levelTry) {
 164         if (levelTry & SC_FOLDLEVELWHITEFLAG)
 165                 return true;
 166         else
 167                 return (levelStart & SC_FOLDLEVELNUMBERMASK) < (levelTry & SC_FOLDLEVELNUMBERMASK);
 168 }
 169
 170 int Document::GetLastChild(int lineParent, int level) {
 171         if (level == -1)
 172                 level = GetLevel(lineParent) & SC_FOLDLEVELNUMBERMASK;
 173         int maxLine = LinesTotal();
 174         int lineMaxSubord = lineParent;
 175         while (lineMaxSubord < maxLine - 1) {
 176                 EnsureStyledTo(LineStart(lineMaxSubord + 2));
 177                 if (!IsSubordinate(level, GetLevel(lineMaxSubord + 1)))
 178                         break;
 179                 lineMaxSubord++;
 180         }
 181         if (lineMaxSubord > lineParent) {
 182                 if (level > (GetLevel(lineMaxSubord + 1) & SC_FOLDLEVELNUMBERMASK)) {
 183                         // Have chewed up some whitespace that belongs to a parent so seek back
 184                         if (GetLevel(lineMaxSubord) & SC_FOLDLEVELWHITEFLAG) {
 185                                 lineMaxSubord--;
 186                         }
 187                 }
 188         }
 189         return lineMaxSubord;
 190 }
 191
 192 int Document::GetFoldParent(int line) {
 193         int level = GetLevel(line);
 194         int lineLook = line - 1;
 195         while ((lineLook > 0) && (
 196                     (!(GetLevel(lineLook) & SC_FOLDLEVELHEADERFLAG)) ||
 197                     ((GetLevel(lineLook) & SC_FOLDLEVELNUMBERMASK) >= level))
 198               ) {
 199                 lineLook--;
 200         }
 201         if ((GetLevel(lineLook) & SC_FOLDLEVELHEADERFLAG) &&
 202                 ((GetLevel(lineLook) & SC_FOLDLEVELNUMBERMASK) < level)) {
 203                 return lineLook;
 204         } else {
 205                 return -1;
 206         }
 207 }
 208
 209 int Document::ClampPositionIntoDocument(int pos) {
 210         return Platform::Clamp(pos, 0, Length());
 211 }
 212
 213 bool Document::IsCrLf(int pos) {
 214         if (pos < 0)
 215                 return false;
 216         if (pos >= (Length() - 1))
 217                 return false;
 218         return (cb.CharAt(pos) == '\r') && (cb.CharAt(pos + 1) == '\n');
 219 }
 220
 221 bool Document::IsDBCS(int pos) {
 222         if (dbcsCodePage) {
 223                 if (SC_CP_UTF8 == dbcsCodePage) {
 224                         unsigned char ch = static_cast<unsigned char>(cb.CharAt(pos));
 225                         return ch >= 0x80;
 226                 } else {
 227                         // Anchor DBCS calculations at start of line because start of line can
 228                         // not be a DBCS trail byte.
 229                         int startLine = pos;
 230                         while (startLine > 0 && cb.CharAt(startLine) != '\r' && cb.CharAt(startLine) != '\n')
 231                                 startLine--;
 232                         while (startLine <= pos) {
 233                                 if (Platform::IsDBCSLeadByte(dbcsCodePage, cb.CharAt(startLine))) {
 234                                         startLine++;
 235                                         if (startLine >= pos)
 236                                                 return true;
 237                                 }
 238                                 startLine++;
 239                         }
 240                 }
 241         }
 242         return false;
 243 }
 244
 245 int Document::LenChar(int pos) {
 246         if (IsCrLf(pos)) {
 247                 return 2;
 248         } else if (SC_CP_UTF8 == dbcsCodePage) {
 249                 unsigned char ch = static_cast<unsigned char>(cb.CharAt(pos));
 250                 if (ch < 0x80)
 251                         return 1;
 252                 int len = 2;
 253                 if (ch >= (0x80 + 0x40 + 0x20))
 254                         len = 3;
 255                 int lengthDoc = Length();
 256                 if ((pos + len) > lengthDoc)
 257                         return lengthDoc -pos;
 258                 else
 259                         return len;
 260         } else if (IsDBCS(pos)) {
 261                 return 2;
 262         } else {
 263                 return 1;
 264         }
 265 }
 266
 267 // Normalise a position so that it is not halfway through a two byte character.
 268 // This can occur in two situations -
 269 // When lines are terminated with \r\n pairs which should be treated as one character.
 270 // When displaying DBCS text such as Japanese.
 271 // If moving, move the position in the indicated direction.
 272 int Document::MovePositionOutsideChar(int pos, int moveDir, bool checkLineEnd) {
 273         //Platform::DebugPrintf("NoCRLF %d %d\n", pos, moveDir);
 274         // If out of range, just return value - should be fixed up after
 275         if (pos < 0)
 276                 return pos;
 277         if (pos > Length())
 278                 return pos;
 279
 280         // Position 0 and Length() can not be between any two characters
 281         if (pos == 0)
 282                 return pos;
 283         if (pos == Length())
 284                 return pos;
 285
 286         // assert pos > 0 && pos < Length()
 287         if (checkLineEnd && IsCrLf(pos - 1)) {
 288                 if (moveDir > 0)
 289                         return pos + 1;
 290                 else
 291                         return pos - 1;
 292         }
 293
 294         // Not between CR and LF
 295
 296         if (dbcsCodePage) {
 297                 if (SC_CP_UTF8 == dbcsCodePage) {
 298                         unsigned char ch = static_cast<unsigned char>(cb.CharAt(pos));
 299                         while ((pos > 0) && (pos < Length()) && (ch >= 0x80) && (ch < (0x80 + 0x40))) {
 300                                 // ch is a trail byte
 301                                 if (moveDir > 0)
 302                                         pos++;
 303                                 else
 304                                         pos--;
 305                                 ch = static_cast<unsigned char>(cb.CharAt(pos));
 306                         }
 307                 } else {
 308                         // Anchor DBCS calculations at start of line because start of line can
 309                         // not be a DBCS trail byte.
 310                         int startLine = pos;
 311                         while (startLine > 0 && cb.CharAt(startLine) != '\r' && cb.CharAt(startLine) != '\n')
 312                                 startLine--;
 313                         bool atLeadByte = false;
 314                         while (startLine < pos) {
 315                                 if (atLeadByte)
 316                                         atLeadByte = false;
 317                                 else if (Platform::IsDBCSLeadByte(dbcsCodePage, cb.CharAt(startLine)))
 318                                         atLeadByte = true;
 319                                 else
 320                                         atLeadByte = false;
 321                                 startLine++;
 322                         }
 323
 324
 325                         if (atLeadByte) {
 326                                 // Position is between a lead byte and a trail byte
 327                                 if (moveDir > 0)
 328                                         return pos + 1;
 329                                 else
 330                                         return pos - 1;
 331                         }
 332                 }
 333         }
 334
 335         return pos;
 336 }
 337
 338 void Document::ModifiedAt(int pos) {
 339         if (endStyled > pos)
 340                 endStyled = pos;
 341 }
 342
 343 // Document only modified by gateways DeleteChars, InsertStyledString, Undo, Redo, and SetStyleAt.
 344 // SetStyleAt does not change the persistent state of a document
 345
 346 // Unlike Undo, Redo, and InsertStyledString, the pos argument is a cell number not a char number
 347 void Document::DeleteChars(int pos, int len) {
 348         if (len == 0)
 349                 return;
 350         if ((pos + len) > Length())
 351                 return;
 352         if (cb.IsReadOnly() && enteredReadOnlyCount == 0) {
 353                 enteredReadOnlyCount++;
 354                 NotifyModifyAttempt();
 355                 enteredReadOnlyCount--;
 356         }
 357         if (enteredCount == 0) {
 358                 enteredCount++;
 359                 if (!cb.IsReadOnly()) {
 360                         NotifyModified(
 361                             DocModification(
 362                                 SC_MOD_BEFOREDELETE | SC_PERFORMED_USER,
 363                                 pos, len,
 364                                 0, 0));
 365                         int prevLinesTotal = LinesTotal();
 366                         bool startSavePoint = cb.IsSavePoint();
 367                         const char *text = cb.DeleteChars(pos * 2, len * 2);
 368                         if (startSavePoint && cb.IsCollectingUndo())
 369                                 NotifySavePoint(!startSavePoint);
 370                         if ((pos < Length()) || (pos == 0))
 371                                 ModifiedAt(pos);
 372                         else
 373                                 ModifiedAt(pos-1);
 374                         NotifyModified(
 375                             DocModification(
 376                                 SC_MOD_DELETETEXT | SC_PERFORMED_USER,
 377                                 pos, len,
 378                                 LinesTotal() - prevLinesTotal, text));
 379                 }
 380                 enteredCount--;
 381         }
 382 }
 383
 384 void Document::InsertStyledString(int position, char *s, int insertLength) {
 385         if (cb.IsReadOnly() && enteredReadOnlyCount == 0) {
 386                 enteredReadOnlyCount++;
 387                 NotifyModifyAttempt();
 388                 enteredReadOnlyCount--;
 389         }
 390         if (enteredCount == 0) {
 391                 enteredCount++;
 392                 if (!cb.IsReadOnly()) {
 393                         NotifyModified(
 394                             DocModification(
 395                                 SC_MOD_BEFOREINSERT | SC_PERFORMED_USER,
 396                                 position / 2, insertLength / 2,
 397                                 0, 0));
 398                         int prevLinesTotal = LinesTotal();
 399                         bool startSavePoint = cb.IsSavePoint();
 400                         const char *text = cb.InsertString(position, s, insertLength);
 401                         if (startSavePoint && cb.IsCollectingUndo())
 402                                 NotifySavePoint(!startSavePoint);
 403                         ModifiedAt(position / 2);
 404                         NotifyModified(
 405                             DocModification(
 406                                 SC_MOD_INSERTTEXT | SC_PERFORMED_USER,
 407                                 position / 2, insertLength / 2,
 408                                 LinesTotal() - prevLinesTotal, text));
 409                 }
 410                 enteredCount--;
 411         }
 412 }
 413
 414 int Document::Undo() {
 415         int newPos = 0;
 416         if (enteredCount == 0) {
 417                 enteredCount++;
 418                 bool startSavePoint = cb.IsSavePoint();
 419                 int steps = cb.StartUndo();
 420                 //Platform::DebugPrintf("Steps=%d\n", steps);
 421                 for (int step = 0; step < steps; step++) {
 422                         int prevLinesTotal = LinesTotal();
 423                         const Action &action = cb.GetUndoStep();
 424                         if (action.at == removeAction) {
 425                                 NotifyModified(DocModification(
 426                                                    SC_MOD_BEFOREINSERT | SC_PERFORMED_UNDO, action));
 427                         } else {
 428                                 NotifyModified(DocModification(
 429                                                    SC_MOD_BEFOREDELETE | SC_PERFORMED_UNDO, action));
 430                         }
 431                         cb.PerformUndoStep();
 432                         int cellPosition = action.position / 2;
 433                         ModifiedAt(cellPosition);
 434                         newPos = cellPosition;
 435
 436                         int modFlags = SC_PERFORMED_UNDO;
 437                         // With undo, an insertion action becomes a deletion notification
 438                         if (action.at == removeAction) {
 439                                 newPos += action.lenData;
 440                                 modFlags |= SC_MOD_INSERTTEXT;
 441                         } else {
 442                                 modFlags |= SC_MOD_DELETETEXT;
 443                         }
 444                         if (step == steps - 1)
 445                                 modFlags |= SC_LASTSTEPINUNDOREDO;
 446                         NotifyModified(DocModification(modFlags, cellPosition, action.lenData,
 447                                                        LinesTotal() - prevLinesTotal, action.data));
 448                 }
 449
 450                 bool endSavePoint = cb.IsSavePoint();
 451                 if (startSavePoint != endSavePoint)
 452                         NotifySavePoint(endSavePoint);
 453                 enteredCount--;
 454         }
 455         return newPos;
 456 }
 457
 458 int Document::Redo() {
 459         int newPos = 0;
 460         if (enteredCount == 0) {
 461                 enteredCount++;
 462                 bool startSavePoint = cb.IsSavePoint();
 463                 int steps = cb.StartRedo();
 464                 for (int step = 0; step < steps; step++) {
 465                         int prevLinesTotal = LinesTotal();
 466                         const Action &action = cb.GetRedoStep();
 467                         if (action.at == insertAction) {
 468                                 NotifyModified(DocModification(
 469                                                    SC_MOD_BEFOREINSERT | SC_PERFORMED_REDO, action));
 470                         } else {
 471                                 NotifyModified(DocModification(
 472                                                    SC_MOD_BEFOREDELETE | SC_PERFORMED_REDO, action));
 473                         }
 474                         cb.PerformRedoStep();
 475                         ModifiedAt(action.position / 2);
 476                         newPos = action.position / 2;
 477
 478                         int modFlags = SC_PERFORMED_REDO;
 479                         if (action.at == insertAction) {
 480                                 newPos += action.lenData;
 481                                 modFlags |= SC_MOD_INSERTTEXT;
 482                         } else {
 483                                 modFlags |= SC_MOD_DELETETEXT;
 484                         }
 485                         if (step == steps - 1)
 486                                 modFlags |= SC_LASTSTEPINUNDOREDO;
 487                         NotifyModified(
 488                             DocModification(modFlags, action.position / 2, action.lenData,
 489                                             LinesTotal() - prevLinesTotal, action.data));
 490                 }
 491
 492                 bool endSavePoint = cb.IsSavePoint();
 493                 if (startSavePoint != endSavePoint)
 494                         NotifySavePoint(endSavePoint);
 495                 enteredCount--;
 496         }
 497         return newPos;
 498 }
 499
 500 void Document::InsertChar(int pos, char ch) {
 501         char chs[2];
 502         chs[0] = ch;
 503         chs[1] = 0;
 504         InsertStyledString(pos*2, chs, 2);
 505 }
 506
 507 // Insert a null terminated string
 508 void Document::InsertString(int position, const char *s) {
 509         InsertString(position, s, strlen(s));
 510 }
 511
 512 // Insert a string with a length
 513 void Document::InsertString(int position, const char *s, int insertLength) {
 514         char *sWithStyle = new char[insertLength * 2];
 515         if (sWithStyle) {
 516                 for (int i = 0; i < insertLength; i++) {
 517                         sWithStyle[i*2] = s[i];
 518                         sWithStyle[i*2 + 1] = 0;
 519                 }
 520                 InsertStyledString(position*2, sWithStyle, insertLength*2);
 521                 delete []sWithStyle;
 522         }
 523 }
 524
 525 void Document::ChangeChar(int pos, char ch) {
 526         DeleteChars(pos, 1);
 527         InsertChar(pos, ch);
 528 }
 529
 530 void Document::DelChar(int pos) {
 531         DeleteChars(pos, LenChar(pos));
 532 }
 533
 534 int Document::DelCharBack(int pos) {
 535         if (pos <= 0) {
 536                 return pos;
 537         } else if (IsCrLf(pos - 2)) {
 538                 DeleteChars(pos - 2, 2);
 539                 return pos - 2;
 540         } else if (SC_CP_UTF8 == dbcsCodePage) {
 541                 int startChar = MovePositionOutsideChar(pos - 1, -1, false);
 542                 DeleteChars(startChar, pos - startChar);
 543                 return startChar;
 544         } else if (IsDBCS(pos - 1)) {
 545                 DeleteChars(pos - 2, 2);
 546                 return pos - 2;
 547         } else {
 548                 DeleteChars(pos - 1, 1);
 549                 return pos - 1;
 550         }
 551 }
 552
 553 static bool isindentchar(char ch) {
 554         return (ch == ' ') || (ch == '\t');
 555 }
 556
 557 static int NextTab(int pos, int tabSize) {
 558         return ((pos / tabSize) + 1) * tabSize;
 559 }
 560
 561 static void CreateIndentation(char *linebuf, int length, int indent, int tabSize, bool insertSpaces) {
 562         length--;       // ensure space for \0
 563         if (!insertSpaces) {
 564                 while ((indent >= tabSize) && (length > 0)) {
 565                         *linebuf++ = '\t';
 566                         indent -= tabSize;
 567                         length--;
 568                 }
 569         }
 570         while ((indent > 0) && (length > 0)) {
 571                 *linebuf++ = ' ';
 572                 indent--;
 573                 length--;
 574         }
 575         *linebuf = '\0';
 576 }
 577
 578 int Document::GetLineIndentation(int line) {
 579         int indent = 0;
 580         if ((line >= 0) && (line < LinesTotal())) {
 581                 int lineStart = LineStart(line);
 582                 int length = Length();
 583                 for (int i = lineStart;i < length;i++) {
 584                         char ch = cb.CharAt(i);
 585                         if (ch == ' ')
 586                                 indent++;
 587                         else if (ch == '\t')
 588                                 indent = NextTab(indent, tabInChars);
 589                         else
 590                                 return indent;
 591                 }
 592         }
 593         return indent;
 594 }
 595
 596 void Document::SetLineIndentation(int line, int indent) {
 597         int indentOfLine = GetLineIndentation(line);
 598         if (indent < 0)
 599                 indent = 0;
 600         if (indent != indentOfLine) {
 601                 char linebuf[1000];
 602                 CreateIndentation(linebuf, sizeof(linebuf), indent, tabInChars, !useTabs);
 603                 int thisLineStart = LineStart(line);
 604                 int indentPos = GetLineIndentPosition(line);
 605                 DeleteChars(thisLineStart, indentPos - thisLineStart);
 606                 InsertString(thisLineStart, linebuf);
 607         }
 608 }
 609
 610 int Document::GetLineIndentPosition(int line) {
 611         if (line < 0)
 612                 return 0;
 613         int pos = LineStart(line);
 614         int length = Length();
 615         while ((pos < length) && isindentchar(cb.CharAt(pos))) {
 616                 pos++;
 617         }
 618         return pos;
 619 }
 620
 621 int Document::GetColumn(int pos) {
 622         int column = 0;
 623         int line = LineFromPosition(pos);
 624         if ((line >= 0) && (line < LinesTotal())) {
 625                 for (int i = LineStart(line);i < pos;) {
 626                         char ch = cb.CharAt(i);
 627                         if (ch == '\t') {
 628                                 column = NextTab(column, tabInChars);
 629                                 i++;
 630                         } else if (ch == '\r') {
 631                                 return column;
 632                         } else if (ch == '\n') {
 633                                 return column;
 634                         } else {
 635                                 column++;
 636                                 i = MovePositionOutsideChar(i + 1, 1);
 637                         }
 638                 }
 639         }
 640         return column;
 641 }
 642
 643 int Document::FindColumn(int line, int column) {
 644         int position = LineStart(line);
 645         int columnCurrent = 0;
 646         if ((line >= 0) && (line < LinesTotal())) {
 647                 while (columnCurrent < column) {
 648                         char ch = cb.CharAt(position);
 649                         if (ch == '\t') {
 650                                 columnCurrent = NextTab(columnCurrent, tabInChars);
 651                                 position++;
 652                         } else if (ch == '\r') {
 653                                 return position;
 654                         } else if (ch == '\n') {
 655                                 return position;
 656                         } else {
 657                                 columnCurrent++;
 658                                 position = MovePositionOutsideChar(position + 1, 1);
 659                         }
 660                 }
 661         }
 662         return position;
 663 }
 664
 665 void Document::Indent(bool forwards, int lineBottom, int lineTop) {
 666         // Dedent - suck white space off the front of the line to dedent by equivalent of a tab
 667         for (int line = lineBottom; line >= lineTop; line--) {
 668                 int indentOfLine = GetLineIndentation(line);
 669                 if (forwards)
 670                         SetLineIndentation(line, indentOfLine + IndentSize());
 671                 else
 672                         SetLineIndentation(line, indentOfLine - IndentSize());
 673         }
 674 }
 675
 676 void Document::ConvertLineEnds(int eolModeSet) {
 677         BeginUndoAction();
 678         for (int pos = 0; pos < Length(); pos++) {
 679                 if (cb.CharAt(pos) == '\r') {
 680                         if (cb.CharAt(pos + 1) == '\n') {
 681                                 if (eolModeSet != SC_EOL_CRLF) {
 682                                         DeleteChars(pos, 2);
 683                                         if (eolModeSet == SC_EOL_CR)
 684                                                 InsertString(pos, "\r", 1);
 685                                         else
 686                                                 InsertString(pos, "\n", 1);
 687                                 } else {
 688                                         pos++;
 689                                 }
 690                         } else {
 691                                 if (eolModeSet != SC_EOL_CR) {
 692                                         DeleteChars(pos, 1);
 693                                         if (eolModeSet == SC_EOL_CRLF) {
 694                                                 InsertString(pos, "\r\n", 2);
 695                                                 pos++;
 696                                         } else {
 697                                                 InsertString(pos, "\n", 1);
 698                                         }
 699                                 }
 700                         }
 701                 } else if (cb.CharAt(pos) == '\n') {
 702                         if (eolModeSet != SC_EOL_LF) {
 703                                 DeleteChars(pos, 1);
 704                                 if (eolModeSet == SC_EOL_CRLF) {
 705                                         InsertString(pos, "\r\n", 2);
 706                                         pos++;
 707                                 } else {
 708                                         InsertString(pos, "\r", 1);
 709                                 }
 710                         }
 711                 }
 712         }
 713         EndUndoAction();
 714 }
 715
 716 Document::charClassification Document::WordCharClass(unsigned char ch) {
 717         if ((SC_CP_UTF8 == dbcsCodePage) && (ch >= 0x80))
 718                 return ccWord;
 719         return charClass[ch];
 720 }
 721
 722 /**
 723  * Used by commmands that want to select whole words.
 724  * Finds the start of word at pos when delta < 0 or the end of the word when delta >= 0.
 725  */
 726 int Document::ExtendWordSelect(int pos, int delta, bool onlyWordCharacters) {
 727         charClassification ccStart = ccWord;
 728         if (delta < 0) {
 729                 if (!onlyWordCharacters)
 730                         ccStart = WordCharClass(cb.CharAt(pos-1));
 731                 while (pos > 0 && (WordCharClass(cb.CharAt(pos - 1)) == ccStart))
 732                         pos--;
 733         } else {
 734                 if (!onlyWordCharacters)
 735                         ccStart = WordCharClass(cb.CharAt(pos));
 736                 while (pos < (Length()) && (WordCharClass(cb.CharAt(pos)) == ccStart))
 737                         pos++;
 738         }
 739         return pos;
 740 }
 741
 742 /**
 743  * Find the start of the next word in either a forward (delta >= 0) or backwards direction
 744  * (delta < 0).
 745  * This is looking for a transition between character classes although there is also some
 746  * additional movement to transit white space.
 747  * Used by cursor movement by word commands.
 748  */
 749 int Document::NextWordStart(int pos, int delta) {
 750         if (delta < 0) {
 751                 while (pos > 0 && (WordCharClass(cb.CharAt(pos - 1)) == ccSpace))
 752                         pos--;
 753                 if (pos > 0) {
 754                         charClassification ccStart = WordCharClass(cb.CharAt(pos-1));
 755                         while (pos > 0 && (WordCharClass(cb.CharAt(pos - 1)) == ccStart)) {
 756                                 pos--;
 757                         }
 758                 }
 759         } else {
 760                 charClassification ccStart = WordCharClass(cb.CharAt(pos));
 761                 while (pos < (Length()) && (WordCharClass(cb.CharAt(pos)) == ccStart))
 762                         pos++;
 763                 while (pos < (Length()) && (WordCharClass(cb.CharAt(pos)) == ccSpace))
 764                         pos++;
 765         }
 766         return pos;
 767 }
 768
 769 /**
 770  * Check that the character at the given position is a word or punctuation character and that
 771  * the previous character is of a different character class.
 772  */
 773 bool Document::IsWordStartAt(int pos) {
 774         if (pos > 0) {
 775                 charClassification ccPos = WordCharClass(CharAt(pos));
 776                 return (ccPos == ccWord || ccPos == ccPunctuation) &&
 777                         (ccPos != WordCharClass(CharAt(pos - 1)));
 778         }
 779         return true;
 780 }
 781
 782 /**
 783  * Check that the character at the given position is a word or punctuation character and that
 784  * the next character is of a different character class.
 785  */
 786 bool Document::IsWordEndAt(int pos) {
 787         if (pos < Length() - 1) {
 788                 charClassification ccPrev = WordCharClass(CharAt(pos-1));
 789                 return (ccPrev == ccWord || ccPrev == ccPunctuation) &&
 790                         (ccPrev != WordCharClass(CharAt(pos)));
 791         }
 792         return true;
 793 }
 794
 795 /**
 796  * Check that the given range is has transitions between character classes at both
 797  * ends and where the characters on the inside are word or punctuation characters.
 798  */
 799 bool Document::IsWordAt(int start, int end) {
 800         return IsWordStartAt(start) && IsWordEndAt(end);
 801 }
 802
 803 // The comparison and case changing functions here assume ASCII
 804 // or extended ASCII such as the normal Windows code page.
 805
 806 static inline char MakeUpperCase(char ch) {
 807         if (ch < 'a' || ch > 'z')
 808                 return ch;
 809         else
 810                 return static_cast<char>(ch - 'a' + 'A');
 811 }
 812
 813 static inline char MakeLowerCase(char ch) {
 814         if (ch < 'A' || ch > 'Z')
 815                 return ch;
 816         else
 817                 return static_cast<char>(ch - 'A' + 'a');
 818 }
 819
 820 // Define a way for the Regular Expression code to access the document
 821 class DocumentIndexer : public CharacterIndexer {
 822         Document *pdoc;
 823         int end;
 824 public:
 825 DocumentIndexer(Document *pdoc_, int end_) :
 826         pdoc(pdoc_), end(end_) {}
 827
 828         virtual char CharAt(int index) {
 829                 if (index < 0 || index >= end)
 830                         return 0;
 831                 else
 832                         return pdoc->CharAt(index);
 833         }
 834 };
 835
 836 /**
 837  * Find text in document, supporting both forward and backward
 838  * searches (just pass minPos > maxPos to do a backward search)
 839  * Has not been tested with backwards DBCS searches yet.
 840  */
 841 long Document::FindText(int minPos, int maxPos, const char *s,
 842                         bool caseSensitive, bool word, bool wordStart, bool regExp,
 843                         int *length) {
 844         if (regExp) {
 845                 if (!pre)
 846                         pre = new RESearch();
 847                 if (!pre)
 848                         return -1;
 849
 850                 int startPos;
 851                 int endPos;
 852
 853                 if (minPos <= maxPos) {
 854                         startPos = minPos;
 855                         endPos = maxPos;
 856                 } else {
 857                         startPos = maxPos;
 858                         endPos = minPos;
 859                 }
 860
 861                 // Range endpoints should not be inside DBCS characters, but just in case, move them.
 862                 startPos = MovePositionOutsideChar(startPos, 1, false);
 863                 endPos = MovePositionOutsideChar(endPos, 1, false);
 864
 865                 const char *errmsg = pre->Compile(s, *length, caseSensitive);
 866                 if (errmsg) {
 867                         return -1;
 868                 }
 869                 // Find a variable in a property file: \$(\([A-Za-z0-9_.]+\))
 870                 // Replace first '.' with '-' in each property file variable reference:
 871                 //     Search: \$(\([A-Za-z0-9_-]+\)\.\([A-Za-z0-9_.]+\))
 872                 //     Replace: $(\1-\2)
 873                 int lineRangeStart = LineFromPosition(startPos);
 874                 int lineRangeEnd = LineFromPosition(endPos);
 875                 if ((startPos >= LineEnd(lineRangeStart)) && (lineRangeStart < lineRangeEnd)) {
 876                         // the start position is at end of line or between line end characters.
 877                         lineRangeStart++;
 878                         startPos = LineStart(lineRangeStart);
 879                 }
 880                 int pos = -1;
 881                 int lenRet = 0;
 882                 char searchEnd = s[*length - 1];
 883                 if (*length == 1) {
 884                         // These produce empty selections so nudge them on if needed
 885                         if (s[0] == '^') {
 886                                 if (startPos == LineStart(lineRangeStart))
 887                                         startPos++;
 888                         } else if (s[0] == '$') {
 889                                 if ((startPos == LineEnd(lineRangeStart)) && (lineRangeStart < lineRangeEnd))
 890                                         startPos = LineStart(lineRangeStart + 1);
 891                         }
 892                         lineRangeStart = LineFromPosition(startPos);
 893                         lineRangeEnd = LineFromPosition(endPos);
 894                 }
 895                 for (int line = lineRangeStart; line <= lineRangeEnd; line++) {
 896                         int startOfLine = LineStart(line);
 897                         int endOfLine = LineEnd(line);
 898                         if (line == lineRangeStart) {
 899                                 if ((startPos != startOfLine) && (s[0] == '^'))
 900                                         continue;       // Can't match start of line if start position after start of line
 901                                 startOfLine = startPos;
 902                         }
 903                         if (line == lineRangeEnd) {
 904                                 if ((endPos != endOfLine) && (searchEnd == '$'))
 905                                         continue;       // Can't match end of line if end position before end of line
 906                                 endOfLine = endPos;
 907                         }
 908                         DocumentIndexer di(this, endOfLine);
 909                         int success = pre->Execute(di, startOfLine, endOfLine);
 910                         if (success) {
 911                                 pos = pre->bopat[0];
 912                                 lenRet = pre->eopat[0] - pre->bopat[0];
 913                                 break;
 914                         }
 915                 }
 916                 *length = lenRet;
 917                 return pos;
 918
 919         } else {
 920
 921                 bool forward = minPos <= maxPos;
 922                 int increment = forward ? 1 : -1;
 923
 924                 // Range endpoints should not be inside DBCS characters, but just in case, move them.
 925                 int startPos = MovePositionOutsideChar(minPos, increment, false);
 926                 int endPos = MovePositionOutsideChar(maxPos, increment, false);
 927
 928                 // Compute actual search ranges needed
 929                 int lengthFind = *length;
 930                 if (lengthFind == -1)
 931                         lengthFind = strlen(s);
 932                 int endSearch = endPos;
 933                 if (startPos <= endPos) {
 934                         endSearch = endPos - lengthFind + 1;
 935                 }
 936                 //Platform::DebugPrintf("Find %d %d %s %d\n", startPos, endPos, ft->lpstrText, lengthFind);
 937                 char firstChar = s[0];
 938                 if (!caseSensitive)
 939                         firstChar = static_cast<char>(MakeUpperCase(firstChar));
 940                 int pos = startPos;
 941                 while (forward ? (pos < endSearch) : (pos >= endSearch)) {
 942                         char ch = CharAt(pos);
 943                         if (caseSensitive) {
 944                                 if (ch == firstChar) {
 945                                         bool found = true;
 946                                         for (int posMatch = 1; posMatch < lengthFind && found; posMatch++) {
 947                                                 ch = CharAt(pos + posMatch);
 948                                                 if (ch != s[posMatch])
 949                                                         found = false;
 950                                         }
 951                                         if (found) {
 952                                                 if ((!word && !wordStart) ||
 953                                                         word && IsWordAt(pos, pos + lengthFind) ||
 954                                                         wordStart && IsWordStartAt(pos))
 955                                                         return pos;
 956                                         }
 957                                 }
 958                         } else {
 959                                 if (MakeUpperCase(ch) == firstChar) {
 960                                         bool found = true;
 961                                         for (int posMatch = 1; posMatch < lengthFind && found; posMatch++) {
 962                                                 ch = CharAt(pos + posMatch);
 963                                                 if (MakeUpperCase(ch) != MakeUpperCase(s[posMatch]))
 964                                                         found = false;
 965                                         }
 966                                         if (found) {
 967                                                 if ((!word && !wordStart) ||
 968                                                         word && IsWordAt(pos, pos + lengthFind) ||
 969                                                         wordStart && IsWordStartAt(pos))
 970                                                         return pos;
 971                                         }
 972                                 }
 973                         }
 974                         pos += increment;
 975                         if (dbcsCodePage) {
 976                                 // Ensure trying to match from start of character
 977                                 pos = MovePositionOutsideChar(pos, increment, false);
 978                         }
 979                 }
 980         }
 981         //Platform::DebugPrintf("Not found\n");
 982         return -1;
 983 }
 984
 985 const char *Document::SubstituteByPosition(const char *text, int *length) {
 986         if (!pre)
 987                 return 0;
 988         delete []substituted;
 989         substituted = 0;
 990         DocumentIndexer di(this, Length());
 991         if (!pre->GrabMatches(di))
 992                 return 0;
 993         unsigned int lenResult = 0;
 994         for (int i = 0; i < *length; i++) {
 995                 if ((text[i] == '\\') && (text[i + 1] >= '1' && text[i + 1] <= '9')) {
 996                         unsigned int patNum = text[i + 1] - '0';
 997                         lenResult += pre->eopat[patNum] - pre->bopat[patNum];
 998                         i++;
 999                 } else {
1000                         lenResult++;
1001                 }
1002         }
1003         substituted = new char[lenResult + 1];
1004         if (!substituted)
1005                 return 0;
1006         char *o = substituted;
1007         for (int j = 0; j < *length; j++) {
1008                 if ((text[j] == '\\') && (text[j + 1] >= '1' && text[j + 1] <= '9')) {
1009                         unsigned int patNum = text[j + 1] - '0';
1010                         unsigned int len = pre->eopat[patNum] - pre->bopat[patNum];
1011                         if (pre->pat[patNum])   // Will be null if try for a match that did not occur
1012                                 memcpy(o, pre->pat[patNum], len);
1013                         o += len;
1014                         j++;
1015                 } else {
1016                         *o++ = text[j];
1017                 }
1018         }
1019         *o = '\0';
1020         *length = lenResult;
1021         return substituted;
1022 }
1023
1024 int Document::LinesTotal() {
1025         return cb.Lines();
1026 }
1027
1028 void Document::ChangeCase(Range r, bool makeUpperCase) {
1029         for (int pos = r.start; pos < r.end; pos++) {
1030                 char ch = CharAt(pos);
1031                 if (dbcsCodePage && IsDBCS(pos)) {
1032                         pos += LenChar(pos);
1033                 } else {
1034                         if (makeUpperCase) {
1035                                 if (islower(ch)) {
1036                                         ChangeChar(pos, static_cast<char>(MakeUpperCase(ch)));
1037                                 }
1038                         } else {
1039                                 if (isupper(ch)) {
1040                                         ChangeChar(pos, static_cast<char>(MakeLowerCase(ch)));
1041                                 }
1042                         }
1043                 }
1044         }
1045 }
1046
1047 void Document::SetWordChars(unsigned char *chars) {
1048         int ch;
1049         for (ch = 0; ch < 256; ch++) {
1050                 if (ch == '\r' || ch == '\n')
1051                         charClass[ch] = ccNewLine;
1052                 else if (ch < 0x20 || ch == ' ')
1053                         charClass[ch] = ccSpace;
1054                 else
1055                         charClass[ch] = ccPunctuation;
1056         }
1057         if (chars) {
1058                 while (*chars) {
1059                         charClass[*chars] = ccWord;
1060                         chars++;
1061                 }
1062         } else {
1063                 for (ch = 0; ch < 256; ch++) {
1064                         if (ch >= 0x80 || isalnum(ch) || ch == '_')
1065                                 charClass[ch] = ccWord;
1066                 }
1067         }
1068 }
1069
1070 void Document::SetStylingBits(int bits) {
1071         stylingBits = bits;
1072         stylingBitsMask = 0;
1073         for (int bit = 0; bit < stylingBits; bit++) {
1074                 stylingBitsMask <<= 1;
1075                 stylingBitsMask |= 1;
1076         }
1077 }
1078
1079 void Document::StartStyling(int position, char mask) {
1080         stylingMask = mask;
1081         endStyled = position;
1082 }
1083
1084 void Document::SetStyleFor(int length, char style) {
1085         if (enteredCount == 0) {
1086                 enteredCount++;
1087                 int prevEndStyled = endStyled;
1088                 if (cb.SetStyleFor(endStyled, length, style, stylingMask)) {
1089                         DocModification mh(SC_MOD_CHANGESTYLE | SC_PERFORMED_USER,
1090                                            prevEndStyled, length);
1091                         NotifyModified(mh);
1092                 }
1093                 endStyled += length;
1094                 enteredCount--;
1095         }
1096 }
1097
1098 void Document::SetStyles(int length, char *styles) {
1099         if (enteredCount == 0) {
1100                 enteredCount++;
1101                 int prevEndStyled = endStyled;
1102                 bool didChange = false;
1103                 for (int iPos = 0; iPos < length; iPos++, endStyled++) {
1104                         PLATFORM_ASSERT(endStyled < Length());
1105                         if (cb.SetStyleAt(endStyled, styles[iPos], stylingMask)) {
1106                                 didChange = true;
1107                         }
1108                 }
1109                 if (didChange) {
1110                         DocModification mh(SC_MOD_CHANGESTYLE | SC_PERFORMED_USER,
1111                                            prevEndStyled, endStyled - prevEndStyled);
1112                         NotifyModified(mh);
1113                 }
1114                 enteredCount--;
1115         }
1116 }
1117
1118 bool Document::EnsureStyledTo(int pos) {
1119         if (pos > GetEndStyled()) {
1120                 styleClock++;
1121                 if (styleClock > 0x100000) {
1122                         styleClock = 0;
1123                 }
1124         }
1125         // Ask the watchers to style, and stop as soon as one responds.
1126         for (int i = 0; pos > GetEndStyled() && i < lenWatchers; i++)
1127                 watchers[i].watcher->NotifyStyleNeeded(this, watchers[i].userData, pos);
1128         return pos <= GetEndStyled();
1129 }
1130
1131 bool Document::AddWatcher(DocWatcher *watcher, void *userData) {
1132         for (int i = 0; i < lenWatchers; i++) {
1133                 if ((watchers[i].watcher == watcher) &&
1134                         (watchers[i].userData == userData))
1135                         return false;
1136         }
1137         WatcherWithUserData *pwNew = new WatcherWithUserData[lenWatchers + 1];
1138         if (!pwNew)
1139                 return false;
1140         for (int j = 0; j < lenWatchers; j++)
1141                 pwNew[j] = watchers[j];
1142         pwNew[lenWatchers].watcher = watcher;
1143         pwNew[lenWatchers].userData = userData;
1144         delete []watchers;
1145         watchers = pwNew;
1146         lenWatchers++;
1147         return true;
1148 }
1149
1150 bool Document::RemoveWatcher(DocWatcher *watcher, void *userData) {
1151         for (int i = 0; i < lenWatchers; i++) {
1152                 if ((watchers[i].watcher == watcher) &&
1153                         (watchers[i].userData == userData)) {
1154                         if (lenWatchers == 1) {
1155                                 delete []watchers;
1156                                 watchers = 0;
1157                                 lenWatchers = 0;
1158                         } else {
1159                                 WatcherWithUserData *pwNew = new WatcherWithUserData[lenWatchers];
1160                                 if (!pwNew)
1161                                         return false;
1162                                 for (int j = 0; j < lenWatchers - 1; j++) {
1163                                         pwNew[j] = (j < i) ? watchers[j] : watchers[j + 1];
1164                                 }
1165                                 delete []watchers;
1166                                 watchers = pwNew;
1167                                 lenWatchers--;
1168                         }
1169                         return true;
1170                 }
1171         }
1172         return false;
1173 }
1174
1175 void Document::NotifyModifyAttempt() {
1176         for (int i = 0; i < lenWatchers; i++) {
1177                 watchers[i].watcher->NotifyModifyAttempt(this, watchers[i].userData);
1178         }
1179 }
1180
1181 void Document::NotifySavePoint(bool atSavePoint) {
1182         for (int i = 0; i < lenWatchers; i++) {
1183                 watchers[i].watcher->NotifySavePoint(this, watchers[i].userData, atSavePoint);
1184         }
1185 }
1186
1187 void Document::NotifyModified(DocModification mh) {
1188         for (int i = 0; i < lenWatchers; i++) {
1189                 watchers[i].watcher->NotifyModified(this, mh, watchers[i].userData);
1190         }
1191 }
1192
1193 bool Document::IsWordPartSeparator(char ch) {
1194         return ispunct(ch) && (WordCharClass(ch) == ccWord);
1195 }
1196
1197 int Document::WordPartLeft(int pos) {
1198         if (pos > 0) {
1199                 --pos;
1200                 char startChar = cb.CharAt(pos);
1201                 if (IsWordPartSeparator(startChar)) {
1202                         while (pos > 0 && IsWordPartSeparator(cb.CharAt(pos))) {
1203                                 --pos;
1204                         }
1205                 }
1206                 if (pos > 0) {
1207                         startChar = cb.CharAt(pos);
1208                         --pos;
1209                         if (islower(startChar)) {
1210                                 while (pos > 0 && islower(cb.CharAt(pos)))
1211                                         --pos;
1212                                 if (!isupper(cb.CharAt(pos)) && !islower(cb.CharAt(pos)))
1213                                         ++pos;
1214                         } else if (isupper(startChar)) {
1215                                 while (pos > 0 && isupper(cb.CharAt(pos)))
1216                                         --pos;
1217                                 if (!isupper(cb.CharAt(pos)))
1218                                         ++pos;
1219                         } else if (isdigit(startChar)) {
1220                                 while (pos > 0 && isdigit(cb.CharAt(pos)))
1221                                         --pos;
1222                                 if (!isdigit(cb.CharAt(pos)))
1223                                         ++pos;
1224                         } else if (ispunct(startChar)) {
1225                                 while (pos > 0 && ispunct(cb.CharAt(pos)))
1226                                         --pos;
1227                                 if (!ispunct(cb.CharAt(pos)))
1228                                         ++pos;
1229                         } else if (isspacechar(startChar)) {
1230                                 while (pos > 0 && isspacechar(cb.CharAt(pos)))
1231                                         --pos;
1232                                 if (!isspacechar(cb.CharAt(pos)))
1233                                         ++pos;
1234                         }
1235                 }
1236         }
1237         return pos;
1238 }
1239
1240 int Document::WordPartRight(int pos) {
1241         char startChar = cb.CharAt(pos);
1242         int length = Length();
1243         if (IsWordPartSeparator(startChar)) {
1244                 while (pos < length && IsWordPartSeparator(cb.CharAt(pos)))
1245                         ++pos;
1246                 startChar = cb.CharAt(pos);
1247         }
1248         if (islower(startChar)) {
1249                 while (pos < length && islower(cb.CharAt(pos)))
1250                         ++pos;
1251         } else if (isupper(startChar)) {
1252                 if (islower(cb.CharAt(pos + 1))) {
1253                         ++pos;
1254                         while (pos < length && islower(cb.CharAt(pos)))
1255                                 ++pos;
1256                 } else {
1257                         while (pos < length && isupper(cb.CharAt(pos)))
1258                                 ++pos;
1259                 }
1260                 if (islower(cb.CharAt(pos)) && isupper(cb.CharAt(pos - 1)))
1261                         --pos;
1262         } else if (isdigit(startChar)) {
1263                 while (pos < length && isdigit(cb.CharAt(pos)))
1264                         ++pos;
1265         } else if (ispunct(startChar)) {
1266                 while (pos < length && ispunct(cb.CharAt(pos)))
1267                         ++pos;
1268         } else if (isspacechar(startChar)) {
1269                 while (pos < length && isspacechar(cb.CharAt(pos)))
1270                         ++pos;
1271         }
1272         return pos;
1273 }