lexothers.cxx.svn-base

Notepad++ is a generic source code editor (it tries to be anyway) and Notepad replacement written in

// Scintilla source code edit control
/** @file LexOthers.cxx
 ** Lexers for batch files, diff results, properties files, make files and error lists.
 ** Also lexer for LaTeX documents.
 **/
// Copyright 1998-2001 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 <ctype.h>
#include <stdio.h>
#include <stdarg.h>

#include "Platform.h"

#include "PropSet.h"
#include "Accessor.h"
#include "KeyWords.h"
#include "Scintilla.h"
#include "SciLexer.h"

static bool Is0To9(char ch) {
	return (ch >= '0') && (ch <= '9');
}

static bool Is1To9(char ch) {
	return (ch >= '1') && (ch <= '9');
}

static inline bool AtEOL(Accessor &styler, unsigned int i) {
	return (styler[i] == '\n') ||
	       ((styler[i] == '\r') && (styler.SafeGetCharAt(i + 1) != '\n'));
}

// Tests for BATCH Operators
static bool IsBOperator(char ch) {
	return (ch == '=') || (ch == '+') || (ch == '>') || (ch == '<') ||
		(ch == '|') || (ch == '?') || (ch == '*');
}

// Tests for BATCH Separators
static bool IsBSeparator(char ch) {
	return (ch == '\\') || (ch == '.') || (ch == ';') ||
		(ch == '\"') || (ch == '\'') || (ch == '/') || (ch == ')');
}

static void ColouriseBatchLine(
    char *lineBuffer,
    unsigned int lengthLine,
    unsigned int startLine,
    unsigned int endPos,
    WordList *keywordlists[],
    Accessor &styler) {

	unsigned int offset = 0;	// Line Buffer Offset
	unsigned int enVarEnd;		// Environment Variable End point
	unsigned int cmdLoc;		// External Command / Program Location
	char wordBuffer[81];		// Word Buffer - large to catch long paths
	unsigned int wbl;		// Word Buffer Length
	unsigned int wbo;		// Word Buffer Offset - also Special Keyword Buffer Length
	WordList &keywords = *keywordlists[0];      // Internal Commands
	WordList &keywords2 = *keywordlists[1];     // External Commands (optional)

	// CHOICE, ECHO, GOTO, PROMPT and SET have Default Text that may contain Regular Keywords
	//   Toggling Regular Keyword Checking off improves readability
	// Other Regular Keywords and External Commands / Programs might also benefit from toggling
	//   Need a more robust algorithm to properly toggle Regular Keyword Checking
	bool continueProcessing = true;	// Used to toggle Regular Keyword Checking
	// Special Keywords are those that allow certain characters without whitespace after the command
	// Examples are: cd. cd\ md. rd. dir| dir> echo: echo. path=
	// Special Keyword Buffer used to determine if the first n characters is a Keyword
	char sKeywordBuffer[10];	// Special Keyword Buffer
	bool sKeywordFound;		// Exit Special Keyword for-loop if found

	// Skip initial spaces
	while ((offset < lengthLine) && (isspacechar(lineBuffer[offset]))) {
		offset++;
	}
	// Colorize Default Text
	styler.ColourTo(startLine + offset - 1, SCE_BAT_DEFAULT);
	// Set External Command / Program Location
	cmdLoc = offset;

	// Check for Fake Label (Comment) or Real Label - return if found
	if (lineBuffer[offset] == ':') {
		if (lineBuffer[offset + 1] == ':') {
			// Colorize Fake Label (Comment) - :: is similar to REM, see http://content.techweb.com/winmag/columns/explorer/2000/21.htm
			styler.ColourTo(endPos, SCE_BAT_COMMENT);
		} else {
			// Colorize Real Label
			styler.ColourTo(endPos, SCE_BAT_LABEL);
		}
		return;
	// Check for Drive Change (Drive Change is internal command) - return if found
	} else if ((isalpha(lineBuffer[offset])) &&
		(lineBuffer[offset + 1] == ':') &&
		((isspacechar(lineBuffer[offset + 2])) ||
		(((lineBuffer[offset + 2] == '\\')) &&
		(isspacechar(lineBuffer[offset + 3]))))) {
		// Colorize Regular Keyword
		styler.ColourTo(endPos, SCE_BAT_WORD);
		return;
	}

	// Check for Hide Command (@ECHO OFF/ON)
	if (lineBuffer[offset] == '@') {
		styler.ColourTo(startLine + offset, SCE_BAT_HIDE);
		offset++;
	// Check for Argument (%n) or Environment Variable (%x...%)
	} else if (lineBuffer[offset] == '%') {
		enVarEnd = offset + 1;
		// Search end of word for second % (can be a long path)
		while ((enVarEnd < lengthLine) &&
			(!isspacechar(lineBuffer[enVarEnd])) &&
			(lineBuffer[enVarEnd] != '%') &&
			(!IsBOperator(lineBuffer[enVarEnd])) &&
			(!IsBSeparator(lineBuffer[enVarEnd]))) {
			enVarEnd++;
		}
		// Check for Argument (%n)
		if ((Is0To9(lineBuffer[offset + 1])) &&
			(lineBuffer[enVarEnd] != '%')) {
			// Colorize Argument
			styler.ColourTo(startLine + offset + 1, SCE_BAT_IDENTIFIER);
			offset += 2;
			// Check for External Command / Program
			if (offset < lengthLine && !isspacechar(lineBuffer[offset])) {
				cmdLoc = offset;
			}
		// Check for Environment Variable (%x...%)
		} else if ((lineBuffer[offset + 1] != '%') &&
			(lineBuffer[enVarEnd] == '%')) {
			offset = enVarEnd;
			// Colorize Environment Variable
			styler.ColourTo(startLine + offset, SCE_BAT_IDENTIFIER);
			offset++;
			// Check for External Command / Program
			if (offset < lengthLine && !isspacechar(lineBuffer[offset])) {
				cmdLoc = offset;
			}
		}
	}
	// Skip next spaces
	while ((offset < lengthLine) && (isspacechar(lineBuffer[offset]))) {
		offset++;
	}

	// Read remainder of line word-at-a-time or remainder-of-word-at-a-time
	while (offset < lengthLine) {
		if (offset > startLine) {
			// Colorize Default Text
			styler.ColourTo(startLine + offset - 1, SCE_BAT_DEFAULT);
		}
		// Copy word from Line Buffer into Word Buffer
		wbl = 0;
		for (; offset < lengthLine && wbl < 80 &&
		        !isspacechar(lineBuffer[offset]); wbl++, offset++) {
			wordBuffer[wbl] = static_cast<char>(tolower(lineBuffer[offset]));
		}
		wordBuffer[wbl] = '\0';
		wbo = 0;

		// Check for Comment - return if found
		if (CompareCaseInsensitive(wordBuffer, "rem") == 0) {
			styler.ColourTo(endPos, SCE_BAT_COMMENT);
			return;
		}
		// Check for Separator
		if (IsBSeparator(wordBuffer[0])) {
			// Check for External Command / Program
			if ((cmdLoc == offset - wbl) &&
				((wordBuffer[0] == ':') ||
				(wordBuffer[0] == '\\') ||
				(wordBuffer[0] == '.'))) {
				// Reset Offset to re-process remainder of word
				offset -= (wbl - 1);
				// Colorize External Command / Program
				if (!keywords2) {
					styler.ColourTo(startLine + offset - 1, SCE_BAT_COMMAND);
				} else if (keywords2.InList(wordBuffer)) {
					styler.ColourTo(startLine + offset - 1, SCE_BAT_COMMAND);
				} else {
					styler.ColourTo(startLine + offset - 1, SCE_BAT_DEFAULT);
				}
				// Reset External Command / Program Location
				cmdLoc = offset;
			} else {
				// Reset Offset to re-process remainder of word
				offset -= (wbl - 1);
				// Colorize Default Text
				styler.ColourTo(startLine + offset - 1, SCE_BAT_DEFAULT);
			}
		// Check for Regular Keyword in list
		} else if ((keywords.InList(wordBuffer)) &&
			(continueProcessing)) {
			// ECHO, GOTO, PROMPT and SET require no further Regular Keyword Checking
			if ((CompareCaseInsensitive(wordBuffer, "echo") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "goto") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "prompt") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "set") == 0)) {
				continueProcessing = false;
			}
			// Identify External Command / Program Location for ERRORLEVEL, and EXIST
			if ((CompareCaseInsensitive(wordBuffer, "errorlevel") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "exist") == 0)) {
				// Reset External Command / Program Location
				cmdLoc = offset;
				// Skip next spaces
				while ((cmdLoc < lengthLine) &&
					(isspacechar(lineBuffer[cmdLoc]))) {
					cmdLoc++;
				}
				// Skip comparison
				while ((cmdLoc < lengthLine) &&
					(!isspacechar(lineBuffer[cmdLoc]))) {
					cmdLoc++;
				}
				// Skip next spaces
				while ((cmdLoc < lengthLine) &&
					(isspacechar(lineBuffer[cmdLoc]))) {
					cmdLoc++;
				}
			// Identify External Command / Program Location for CALL, DO, LOADHIGH and LH
			} else if ((CompareCaseInsensitive(wordBuffer, "call") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "do") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "loadhigh") == 0) ||
				(CompareCaseInsensitive(wordBuffer, "lh") == 0)) {
				// Reset External Command / Program Location
				cmdLoc = offset;
				// Skip next spaces
				while ((cmdLoc < lengthLine) &&
					(isspacechar(lineBuffer[cmdLoc]))) {
					cmdLoc++;
				}
			}
			// Colorize Regular keyword
			styler.ColourTo(startLine + offset - 1, SCE_BAT_WORD);
			// No need to Reset Offset
		// Check for Special Keyword in list, External Command / Program, or Default Text
		} else if ((wordBuffer[0] != '%') &&
			(!IsBOperator(wordBuffer[0])) &&
			(continueProcessing)) {
			// Check for Special Keyword
			//     Affected Commands are in Length range 2-6
			//     Good that ERRORLEVEL, EXIST, CALL, DO, LOADHIGH, and LH are unaffected
			sKeywordFound = false;
			for (unsigned int keywordLength = 2; keywordLength < wbl && keywordLength < 7 && !sKeywordFound; keywordLength++) {
				wbo = 0;
				// Copy Keyword Length from Word Buffer into Special Keyword Buffer
				for (; wbo < keywordLength; wbo++) {
					sKeywordBuffer[wbo] = static_cast<char>(wordBuffer[wbo]);
				}
				sKeywordBuffer[wbo] = '\0';
				// Check for Special Keyword in list
				if ((keywords.InList(sKeywordBuffer)) &&
					((IsBOperator(wordBuffer[wbo])) ||
					(IsBSeparator(wordBuffer[wbo])))) {
					sKeywordFound = true;
					// ECHO requires no further Regular Keyword Checking
					if (CompareCaseInsensitive(sKeywordBuffer, "echo") == 0) {
						continueProcessing = false;
					}
					// Colorize Special Keyword as Regular Keyword
					styler.ColourTo(startLine + offset - 1 - (wbl - wbo), SCE_BAT_WORD);
					// Reset Offset to re-process remainder of word
					offset -= (wbl - wbo);
				}
			}
			// Check for External Command / Program or Default Text
			if (!sKeywordFound) {
				wbo = 0;
				// Check for External Command / Program
				if (cmdLoc == offset - wbl) {
					// Read up to %, Operator or Separator
					while ((wbo < wbl) &&
						(wordBuffer[wbo] != '%') &&
						(!IsBOperator(wordBuffer[wbo])) &&
						(!IsBSeparator(wordBuffer[wbo]))) {
						wbo++;
					}
					// Reset External Command / Program Location
					cmdLoc = offset - (wbl - wbo);
					// Reset Offset to re-process remainder of word
					offset -= (wbl - wbo);
					// CHOICE requires no further Regular Keyword Checking
					if (CompareCaseInsensitive(wordBuffer, "choice") == 0) {
						continueProcessing = false;
					}
					// Check for START (and its switches) - What follows is External Command \ Program
					if (CompareCaseInsensitive(wordBuffer, "start") == 0) {
						// Reset External Command / Program Location
						cmdLoc = offset;
						// Skip next spaces
						while ((cmdLoc < lengthLine) &&
							(isspacechar(lineBuffer[cmdLoc]))) {
							cmdLoc++;
						}
						// Reset External Command / Program Location if command switch detected
						if (lineBuffer[cmdLoc] == '/') {
							// Skip command switch
							while ((cmdLoc < lengthLine) &&
								(!isspacechar(lineBuffer[cmdLoc]))) {
								cmdLoc++;
							}
							// Skip next spaces
							while ((cmdLoc < lengthLine) &&
								(isspacechar(lineBuffer[cmdLoc]))) {
								cmdLoc++;
							}
						}
					}
					// Colorize External Command / Program
					if (!keywords2) {
						styler.ColourTo(startLine + offset - 1, SCE_BAT_COMMAND);
					} else if (keywords2.InList(wordBuffer)) {
						styler.ColourTo(startLine + offset - 1, SCE_BAT_COMMAND);
					} else {
						styler.ColourTo(startLine + offset - 1, SCE_BAT_DEFAULT);
					}
					// No need to Reset Offset
				// Check for Default Text
				} else {
					// Read up to %, Operator or Separator
					while ((wbo < wbl) &&
						(wordBuffer[wbo] != '%') &&
						(!IsBOperator(wordBuffer[wbo])) &&
						(!IsBSeparator(wordBuffer[wbo]))) {
						wbo++;
					}
					// Colorize Default Text
					styler.ColourTo(startLine + offset - 1 - (wbl - wbo), SCE_BAT_DEFAULT);
					// Reset Offset to re-process remainder of word
					offset -= (wbl - wbo);
				}
			}
		// Check for Argument  (%n), Environment Variable (%x...%) or Local Variable (%%a)
		} else if (wordBuffer[0] == '%') {
			// Colorize Default Text
			styler.ColourTo(startLine + offset - 1 - wbl, SCE_BAT_DEFAULT);
			wbo++;
			// Search to end of word for second % (can be a long path)
			while ((wbo < wbl) &&
				(wordBuffer[wbo] != '%') &&
				(!IsBOperator(wordBuffer[wbo])) &&
				(!IsBSeparator(wordBuffer[wbo]))) {
				wbo++;
			}
			// Check for Argument (%n)
			if ((Is0To9(wordBuffer[1])) &&
				(wordBuffer[wbo] != '%')) {
				// Check for External Command / Program
				if (cmdLoc == offset - wbl) {
					cmdLoc = offset - (wbl - 2);
				}
				// Colorize Argument
				styler.ColourTo(startLine + offset - 1 - (wbl - 2), SCE_BAT_IDENTIFIER);
				// Reset Offset to re-process remainder of word
				offset -= (wbl - 2);
			// Check for Environment Variable (%x...%)
			} else if ((wordBuffer[1] != '%') &&
				(wordBuffer[wbo] == '%')) {
				wbo++;
				// Check for External Command / Program
				if (cmdLoc == offset - wbl) {
					cmdLoc = offset - (wbl - wbo);
				}
				// Colorize Environment Variable
				styler.ColourTo(startLine + offset - 1 - (wbl - wbo), SCE_BAT_IDENTIFIER);
				// Reset Offset to re-process remainder of word
				offset -= (wbl - wbo);
			// Check for Local Variable (%%a)
			} else if (
				(wbl > 2) &&
				(wordBuffer[1] == '%') &&
				(wordBuffer[2] != '%') &&
				(!IsBOperator(wordBuffer[2])) &&
				(!IsBSeparator(wordBuffer[2]))) {
				// Check for External Command / Program
				if (cmdLoc == offset - wbl) {
					cmdLoc = offset - (wbl - 3);
				}
				// Colorize Local Variable
				styler.ColourTo(startLine + offset - 1 - (wbl - 3), SCE_BAT_IDENTIFIER);
				// Reset Offset to re-process remainder of word
				offset -= (wbl - 3);
			}