document.cxx

porting scintilla to qt

// Scintilla source code edit control
/** @file Document.cxx
 ** Text document that handles notifications, DBCS, styling, words and end of line.
 **/
// Copyright 1998-2003 by Neil Hodgson <[e-mail]>
// The License.txt file describes the conditions under which this software may be distributed.

#include <stdlib.h>
#include <string.h>
#include <stdio.h>
#include <ctype.h>

#include "Platform.h"

#include "Scintilla.h"
#include "SplitVector.h"
#include "Partitioning.h"
#include "RunStyles.h"
#include "CellBuffer.h"
#include "CharClassify.h"
#include "Decoration.h"
#include "Document.h"
#include "RESearch.h"

#ifdef SCI_NAMESPACE
using namespace Scintilla;
#endif

// This is ASCII specific but is safe with chars >= 0x80
static inline bool isspacechar(unsigned char ch) {
	return (ch == ' ') || ((ch >= 0x09) && (ch <= 0x0d));
}

static inline bool IsPunctuation(char ch) {
	return isascii(ch) && ispunct(ch);
}

static inline bool IsADigit(char ch) {
	return isascii(ch) && isdigit(ch);
}

static inline bool IsLowerCase(char ch) {
	return isascii(ch) && islower(ch);
}

static inline bool IsUpperCase(char ch) {
	return isascii(ch) && isupper(ch);
}

Document::Document() {
	refCount = 0;
#ifdef unix
	eolMode = SC_EOL_LF;
#else
	eolMode = SC_EOL_CRLF;
#endif
	dbcsCodePage = 0;
	stylingBits = 5;
	stylingBitsMask = 0x1F;
	stylingMask = 0;
	endStyled = 0;
	styleClock = 0;
	enteredModification = 0;
	enteredStyling = 0;
	enteredReadOnlyCount = 0;
	tabInChars = 8;
	indentInChars = 0;
	actualIndentInChars = 8;
	useTabs = true;
	tabIndents = true;
	backspaceUnindents = false;
	watchers = 0;
	lenWatchers = 0;

	matchesValid = false;
	pre = 0;
	substituted = 0;
}

Document::~Document() {
	for (int i = 0; i < lenWatchers; i++) {
		watchers[i].watcher->NotifyDeleted(this, watchers[i].userData);
	}
	delete []watchers;
	watchers = 0;
	lenWatchers = 0;
	delete pre;
	pre = 0;
	delete []substituted;
	substituted = 0;
}

// Increase reference count and return its previous value.
int Document::AddRef() {
	return refCount++;
}

// Decrease reference count and return its previous value.
// Delete the document if reference count reaches zero.
int Document::Release() {
	int curRefCount = --refCount;
	if (curRefCount == 0)
		delete this;
	return curRefCount;
}

void Document::SetSavePoint() {
	cb.SetSavePoint();
	NotifySavePoint(true);
}

int Document::AddMark(int line, int markerNum) {
	int prev = cb.AddMark(line, markerNum);
	DocModification mh(SC_MOD_CHANGEMARKER, LineStart(line), 0, 0, 0, line);
	NotifyModified(mh);
	return prev;
}

void Document::AddMarkSet(int line, int valueSet) {
	unsigned int m = valueSet;
	for (int i = 0; m; i++, m >>= 1)
		if (m & 1)
			cb.AddMark(line, i);
	DocModification mh(SC_MOD_CHANGEMARKER, LineStart(line), 0, 0, 0, line);
	NotifyModified(mh);
}

void Document::DeleteMark(int line, int markerNum) {
	cb.DeleteMark(line, markerNum);
	DocModification mh(SC_MOD_CHANGEMARKER, LineStart(line), 0, 0, 0, line);
	NotifyModified(mh);
}

void Document::DeleteMarkFromHandle(int markerHandle) {
	cb.DeleteMarkFromHandle(markerHandle);
	DocModification mh(SC_MOD_CHANGEMARKER, 0, 0, 0, 0);
	mh.line = -1;
	NotifyModified(mh);
}

void Document::DeleteAllMarks(int markerNum) {
	cb.DeleteAllMarks(markerNum);
	DocModification mh(SC_MOD_CHANGEMARKER, 0, 0, 0, 0);
	mh.line = -1;
	NotifyModified(mh);
}

int Document::LineStart(int line) const {
	return cb.LineStart(line);
}

int Document::LineEnd(int line) const {
	if (line == LinesTotal() - 1) {
		return LineStart(line + 1);
	} else {
		int position = LineStart(line + 1) - 1;
		// When line terminator is CR+LF, may need to go back one more
		if ((position > LineStart(line)) && (cb.CharAt(position - 1) == '\r')) {
			position--;
		}
		return position;
	}
}

int Document::LineFromPosition(int pos) {
	return cb.LineFromPosition(pos);
}

int Document::LineEndPosition(int position) {
	return LineEnd(LineFromPosition(position));
}

int Document::VCHomePosition(int position) {
	int line = LineFromPosition(position);
	int startPosition = LineStart(line);
	int endLine = LineStart(line + 1) - 1;
	int startText = startPosition;
	while (startText < endLine && (cb.CharAt(startText) == ' ' || cb.CharAt(startText) == '\t' ) )
		startText++;
	if (position == startText)
		return startPosition;
	else
		return startText;
}

int Document::SetLevel(int line, int level) {
	int prev = cb.SetLevel(line, level);
	if (prev != level) {
		DocModification mh(SC_MOD_CHANGEFOLD | SC_MOD_CHANGEMARKER,
		                   LineStart(line), 0, 0, 0, line);
		mh.foldLevelNow = level;
		mh.foldLevelPrev = prev;
		NotifyModified(mh);
	}
	return prev;
}

static bool IsSubordinate(int levelStart, int levelTry) {
	if (levelTry & SC_FOLDLEVELWHITEFLAG)
		return true;
	else
		return (levelStart & SC_FOLDLEVELNUMBERMASK) < (levelTry & SC_FOLDLEVELNUMBERMASK);
}

int Document::GetLastChild(int lineParent, int level) {
	if (level == -1)
		level = GetLevel(lineParent) & SC_FOLDLEVELNUMBERMASK;
	int maxLine = LinesTotal();
	int lineMaxSubord = lineParent;
	while (lineMaxSubord < maxLine - 1) {
		EnsureStyledTo(LineStart(lineMaxSubord + 2));
		if (!IsSubordinate(level, GetLevel(lineMaxSubord + 1)))
			break;
		lineMaxSubord++;
	}
	if (lineMaxSubord > lineParent) {
		if (level > (GetLevel(lineMaxSubord + 1) & SC_FOLDLEVELNUMBERMASK)) {
			// Have chewed up some whitespace that belongs to a parent so seek back
			if (GetLevel(lineMaxSubord) & SC_FOLDLEVELWHITEFLAG) {
				lineMaxSubord--;
			}
		}
	}
	return lineMaxSubord;
}

int Document::GetFoldParent(int line) {
	int level = GetLevel(line) & SC_FOLDLEVELNUMBERMASK;
	int lineLook = line - 1;
	while ((lineLook > 0) && (
	            (!(GetLevel(lineLook) & SC_FOLDLEVELHEADERFLAG)) ||
	            ((GetLevel(lineLook) & SC_FOLDLEVELNUMBERMASK) >= level))
	      ) {
		lineLook--;
	}
	if ((GetLevel(lineLook) & SC_FOLDLEVELHEADERFLAG) &&
	        ((GetLevel(lineLook) & SC_FOLDLEVELNUMBERMASK) < level)) {
		return lineLook;
	} else {
		return -1;
	}
}

int Document::ClampPositionIntoDocument(int pos) {
	return Platform::Clamp(pos, 0, Length());
}

bool Document::IsCrLf(int pos) {
	if (pos < 0)
		return false;
	if (pos >= (Length() - 1))
		return false;
	return (cb.CharAt(pos) == '\r') && (cb.CharAt(pos + 1) == '\n');
}

static const int maxBytesInDBCSCharacter=5;

int Document::LenChar(int pos) {
	if (pos < 0) {
		return 1;
	} else if (IsCrLf(pos)) {
		return 2;
	} else if (SC_CP_UTF8 == dbcsCodePage) {
		unsigned char ch = static_cast<unsigned char>(cb.CharAt(pos));
		if (ch < 0x80)
			return 1;
		int len = 2;
		if (ch >= (0x80 + 0x40 + 0x20 + 0x10))
			len = 4;
		else if (ch >= (0x80 + 0x40 + 0x20))
			len = 3;
		int lengthDoc = Length();
		if ((pos + len) > lengthDoc)
			return lengthDoc -pos;
		else
			return len;
	} else if (dbcsCodePage) {
		char mbstr[maxBytesInDBCSCharacter+1];
		int i;
		for (i=0; i<Platform::DBCSCharMaxLength(); i++) {
			mbstr[i] = cb.CharAt(pos+i);
		}
		mbstr[i] = '\0';
		return Platform::DBCSCharLength(dbcsCodePage, mbstr);
	} else {
		return 1;
	}
}

static bool IsTrailByte(int ch) {
	return (ch >= 0x80) && (ch < (0x80 + 0x40));
}

static int BytesFromLead(int leadByte) {
	if (leadByte > 0xF4) {
		// Characters longer than 4 bytes not possible in current UTF-8
		return 0;
	} else if (leadByte >= 0xF0) {
		return 4;
	} else if (leadByte >= 0xE0) {
		return 3;
	} else if (leadByte >= 0xC2) {
		return 2;
	}
	return 0;
}

bool Document::InGoodUTF8(int pos, int &start, int &end) {
	int lead = pos;
	while ((lead>0) && (pos-lead < 4) && IsTrailByte(static_cast<unsigned char>(cb.CharAt(lead-1))))
		lead--;
	start = 0;
	if (lead > 0) {
		start = lead-1;
	}
	int leadByte = static_cast<unsigned char>(cb.CharAt(start));
	int bytes = BytesFromLead(leadByte);
	if (bytes == 0) {
		return false;
	} else {
		int trailBytes = bytes - 1;
		int len = pos - lead + 1;
		if (len > trailBytes)
			// pos too far from lead
			return false;
		// Check that there are enough trails for this lead
		int trail = pos + 1;
		while ((trail-lead<trailBytes) && (trail < Length())) {
			if (!IsTrailByte(static_cast<unsigned char>(cb.CharAt(trail)))) {
				return false;
			}
			trail++;
		}
		end = start + bytes;
		return true;
	}
}

// Normalise a position so that it is not halfway through a two byte character.
// This can occur in two situations -
// When lines are terminated with \r\n pairs which should be treated as one character.
// When displaying DBCS text such as Japanese.
// If moving, move the position in the indicated direction.
int Document::MovePositionOutsideChar(int pos, int moveDir, bool checkLineEnd) {
	//Platform::DebugPrintf("NoCRLF %d %d\n", pos, moveDir);
	// If out of range, just return minimum/maximum value.
	if (pos <= 0)
		return 0;
	if (pos >= Length())
		return Length();

	// PLATFORM_ASSERT(pos > 0 && pos < Length());
	if (checkLineEnd && IsCrLf(pos - 1)) {
		if (moveDir > 0)
			return pos + 1;
		else
			return pos - 1;
	}

	// Not between CR and LF

	if (dbcsCodePage) {
		if (SC_CP_UTF8 == dbcsCodePage) {
			unsigned char ch = static_cast<unsigned char>(cb.CharAt(pos));
			int startUTF = pos;
			int endUTF = pos;
			if (IsTrailByte(ch) && InGoodUTF8(pos, startUTF, endUTF)) {
				// ch is a trail byte within a UTF-8 character
				if (moveDir > 0)
					pos = endUTF;
				else
					pos = startUTF;
			}
		} else {
			// Anchor DBCS calculations at start of line because start of line can
			// not be a DBCS trail byte.
			int posCheck = LineStart(LineFromPosition(pos));
			while (posCheck < pos) {
				char mbstr[maxBytesInDBCSCharacter+1];
				int i;
				for(i=0;i<Platform::DBCSCharMaxLength();i++) {
					mbstr[i] = cb.CharAt(posCheck+i);
				}
				mbstr[i] = '\0';

				int mbsize = Platform::DBCSCharLength(dbcsCodePage, mbstr);
				if (posCheck + mbsize == pos) {
					return pos;
				} else if (posCheck + mbsize > pos) {
					if (moveDir > 0) {
						return posCheck + mbsize;
					} else {
						return posCheck;
					}
				}
				posCheck += mbsize;
			}
		}
	}

	return pos;
}

void Document::ModifiedAt(int pos) {
	if (endStyled > pos)
		endStyled = pos;
}

void Document::CheckReadOnly() {
	if (cb.IsReadOnly() && enteredReadOnlyCount == 0) {
		enteredReadOnlyCount++;
		NotifyModifyAttempt();
		enteredReadOnlyCount--;
	}
}

// Document only modified by gateways DeleteChars, InsertString, Undo, Redo, and SetStyleAt.
// SetStyleAt does not change the persistent state of a document

bool Document::DeleteChars(int pos, int len) {
	if (len == 0)
		return false;
	if ((pos + len) > Length())
		return false;
	CheckReadOnly();
	if (enteredModification != 0) {
		return false;
	} else {
		enteredModification++;
		if (!cb.IsReadOnly()) {
			NotifyModified(
			    DocModification(
			        SC_MOD_BEFOREDELETE | SC_PERFORMED_USER,
			        pos, len,
			        0, 0));
			int prevLinesTotal = LinesTotal();
			bool startSavePoint = cb.IsSavePoint();
			bool startSequence = false;
			const char *text = cb.DeleteChars(pos, len, startSequence);
			if (startSavePoint && cb.IsCollectingUndo())
				NotifySavePoint(!startSavePoint);
			if ((pos < Length()) || (pos == 0))
				ModifiedAt(pos);
			else
				ModifiedAt(pos-1);
			NotifyModified(
			    DocModification(
			        SC_MOD_DELETETEXT | SC_PERFORMED_USER | (startSequence?SC_STARTACTION:0),
			        pos, len,
			        LinesTotal() - prevLinesTotal, text));
		}
		enteredModification--;
	}
	return !cb.IsReadOnly();
}

/**
 * Insert a string with a length.
 */
bool Document::InsertString(int position, const char *s, int insertLength) {
	if (insertLength <= 0) {
		return false;
	}
	CheckReadOnly();
	if (enteredModification != 0) {
		return false;
	} else {
		enteredModification++;
		if (!cb.IsReadOnly()) {
			NotifyModified(
			    DocModification(
			        SC_MOD_BEFOREINSERT | SC_PERFORMED_USER,
			        position, insertLength,
			        0, s));
			int prevLinesTotal = LinesTotal();
			bool startSavePoint = cb.IsSavePoint();
			bool startSequence = false;
			const char *text = cb.InsertString(position, s, insertLength, startSequence);
			if (startSavePoint && cb.IsCollectingUndo())
				NotifySavePoint(!startSavePoint);
			ModifiedAt(position);
			NotifyModified(
			    DocModification(
			        SC_MOD_INSERTTEXT | SC_PERFORMED_USER | (startSequence?SC_STARTACTION:0),
			        position, insertLength,
			        LinesTotal() - prevLinesTotal, text));
		}
		enteredModification--;
	}
	return !cb.IsReadOnly();
}

int Document::Undo() {
	int newPos = -1;
	CheckReadOnly();
	if (enteredModification == 0) {
		enteredModification++;
		if (!cb.IsReadOnly()) {
			bool startSavePoint = cb.IsSavePoint();
			bool multiLine = false;
			int steps = cb.StartUndo();
			//Platform::DebugPrintf("Steps=%d\n", steps);
			for (int step = 0; step < steps; step++) {
				const int prevLinesTotal = LinesTotal();
				const Action &action = cb.GetUndoStep();
				if (action.at == removeAction) {
					NotifyModified(DocModification(
									SC_MOD_BEFOREINSERT | SC_PERFORMED_UNDO, action));
				} else {
					NotifyModified(DocModification(
									SC_MOD_BEFOREDELETE | SC_PERFORMED_UNDO, action));
				}
				cb.PerformUndoStep();
				int cellPosition = action.position;
				ModifiedAt(cellPosition);
				newPos = cellPosition;

				int modFlags = SC_PERFORMED_UNDO;
				// With undo, an insertion action becomes a deletion notification
				if (action.at == removeAction) {
					newPos += action.lenData;
					modFlags |= SC_MOD_INSERTTEXT;
				} else {
					modFlags |= SC_MOD_DELETETEXT;
				}
				if (steps > 1)
					modFlags |= SC_MULTISTEPUNDOREDO;
				const int linesAdded = LinesTotal() - prevLinesTotal;
				if (linesAdded != 0)
					multiLine = true;
				if (step == steps - 1) {
					modFlags |= SC_LASTSTEPINUNDOREDO;
					if (multiLine)
						modFlags |= SC_MULTILINEUNDOREDO;
				}
				NotifyModified(DocModification(modFlags, cellPosition, action.lenData,
											   linesAdded, action.data));
			}

			bool endSavePoint = cb.IsSavePoint();
			if (startSavePoint != endSavePoint)
				NotifySavePoint(endSavePoint);
		}
		enteredModification--;
	}