lexperl.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 LexPerl.cxx
 ** Lexer for subset of Perl.
 **/
// Copyright 1998-2005 by Neil Hodgson <[e-mail]>
// Lexical analysis fixes by Kein-Hong Man <[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"

#define PERLNUM_BINARY 1    // order is significant: 1-4 cannot have a dot
#define PERLNUM_HEX 2
#define PERLNUM_OCTAL 3
#define PERLNUM_FLOAT 4     // actually exponent part
#define PERLNUM_DECIMAL 5   // 1-5 are numbers; 6-7 are strings
#define PERLNUM_VECTOR 6
#define PERLNUM_V_VECTOR 7
#define PERLNUM_BAD 8

#define BACK_NONE 0         // lookback state for bareword disambiguation:
#define BACK_OPERATOR 1     // whitespace/comments are insignificant
#define BACK_KEYWORD 2      // operators/keywords are needed for disambiguation

#define HERE_DELIM_MAX 256

static inline bool isEOLChar(char ch) {
	return (ch == '\r') || (ch == '\n');
}

static bool isSingleCharOp(char ch) {
	char strCharSet[2];
	strCharSet[0] = ch;
	strCharSet[1] = '\0';
	return (NULL != strstr("rwxoRWXOezsfdlpSbctugkTBMAC", strCharSet));
}

static inline bool isPerlOperator(char ch) {
	if (ch == '^' || ch == '&' || ch == '\\' ||
	        ch == '(' || ch == ')' || ch == '-' || ch == '+' ||
	        ch == '=' || ch == '|' || ch == '{' || ch == '}' ||
	        ch == '[' || ch == ']' || ch == ':' || ch == ';' ||
	        ch == '>' || ch == ',' ||
	        ch == '?' || ch == '!' || ch == '.' || ch == '~')
		return true;
	// these chars are already tested before this call
	// ch == '%' || ch == '*' || ch == '<' || ch == '/' ||
	return false;
}

static bool isPerlKeyword(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler) {
	char s[100];
    unsigned int i, len = end - start;
    if (len > 30) { len = 30; }
	for (i = 0; i < len; i++, start++) s[i] = styler[start];
    s[i] = '\0';
	return keywords.InList(s);
}

// Note: as lexer uses chars, UTF-8 bytes are considered as <0 values
// Note: iswordchar() was used in only one place in LexPerl, it is
// unnecessary as '.' is processed as the concatenation operator, so
// only isWordStart() is used in LexPerl

static inline bool isWordStart(char ch) {
	return !isascii(ch) || isalnum(ch) || ch == '_';
}

static inline bool isEndVar(char ch) {
	return isascii(ch) && !isalnum(ch) && ch != '#' && ch != '$' &&
	       ch != '_' && ch != '\'';
}

static inline bool isNonQuote(char ch) {
	return !isascii(ch) || isalnum(ch) || ch == '_';
}

static inline char actualNumStyle(int numberStyle) {
    if (numberStyle == PERLNUM_VECTOR || numberStyle == PERLNUM_V_VECTOR) {
        return SCE_PL_STRING;
    } else if (numberStyle == PERLNUM_BAD) {
        return SCE_PL_ERROR;
    }
    return SCE_PL_NUMBER;
}

static bool isMatch(Accessor &styler, int lengthDoc, int pos, const char *val) {
	if ((pos + static_cast<int>(strlen(val))) >= lengthDoc) {
		return false;
	}
	while (*val) {
		if (*val != styler[pos++]) {
			return false;
		}
		val++;
	}
	return true;
}

static char opposite(char ch) {
	if (ch == '(')
		return ')';
	if (ch == '[')
		return ']';
	if (ch == '{')
		return '}';
	if (ch == '<')
		return '>';
	return ch;
}

static void ColourisePerlDoc(unsigned int startPos, int length, int initStyle,
                             WordList *keywordlists[], Accessor &styler) {

	// Lexer for perl often has to backtrack to start of current style to determine
	// which characters are being used as quotes, how deeply nested is the
	// start position and what the termination string is for here documents

	WordList &keywords = *keywordlists[0];

	class HereDocCls {
	public:
		int State;		// 0: '<<' encountered
		// 1: collect the delimiter
		// 2: here doc text (lines after the delimiter)
		char Quote;		// the char after '<<'
		bool Quoted;		// true if Quote in ('\'','"','`')
		int DelimiterLength;	// strlen(Delimiter)
		char *Delimiter;	// the Delimiter, 256: sizeof PL_tokenbuf
		HereDocCls() {
			State = 0;
            Quote = 0;
            Quoted = false;
			DelimiterLength = 0;
			Delimiter = new char[HERE_DELIM_MAX];
			Delimiter[0] = '\0';
		}
		~HereDocCls() {
			delete []Delimiter;
		}
	};
	HereDocCls HereDoc;	// TODO: FIFO for stacked here-docs

	class QuoteCls {
		public:
		int  Rep;
		int  Count;
		char Up;
		char Down;
		QuoteCls() {
			this->New(1);
		}
		void New(int r) {
			Rep   = r;
			Count = 0;
			Up    = '\0';
			Down  = '\0';
		}
		void Open(char u) {
			Count++;
			Up    = u;
			Down  = opposite(Up);
		}
	};
	QuoteCls Quote;

	int state = initStyle;
	char numState = PERLNUM_DECIMAL;
	int dotCount = 0;
	unsigned int lengthDoc = startPos + length;
	//int sookedpos = 0; // these have no apparent use, see POD state
	//char sooked[100];
	//sooked[sookedpos] = '\0';

	// If in a long distance lexical state, seek to the beginning to find quote characters
	// Perl strings can be multi-line with embedded newlines, so backtrack.
	// Perl numbers have additional state during lexing, so backtrack too.
	if (state == SCE_PL_HERE_Q || state == SCE_PL_HERE_QQ || state == SCE_PL_HERE_QX) {
		while ((startPos > 1) && (styler.StyleAt(startPos) != SCE_PL_HERE_DELIM)) {
			startPos--;
		}
		startPos = styler.LineStart(styler.GetLine(startPos));
		state = styler.StyleAt(startPos - 1);
	}
	if ( state == SCE_PL_STRING_Q
	|| state == SCE_PL_STRING_QQ
	|| state == SCE_PL_STRING_QX
	|| state == SCE_PL_STRING_QR
	|| state == SCE_PL_STRING_QW
	|| state == SCE_PL_REGEX
	|| state == SCE_PL_REGSUBST
	|| state == SCE_PL_STRING
	|| state == SCE_PL_BACKTICKS
	|| state == SCE_PL_CHARACTER
	|| state == SCE_PL_NUMBER
	|| state == SCE_PL_IDENTIFIER
    || state == SCE_PL_ERROR
	) {
		while ((startPos > 1) && (styler.StyleAt(startPos - 1) == state)) {
			startPos--;
		}
		state = SCE_PL_DEFAULT;
	}

    // lookback at start of lexing to set proper state for backflag
    // after this, they are updated when elements are lexed
    int backflag = BACK_NONE;
    unsigned int backPos = startPos;
    if (backPos > 0) {
        backPos--;
        int sty = SCE_PL_DEFAULT;
        while ((backPos > 0) && (sty = styler.StyleAt(backPos),
               sty == SCE_PL_DEFAULT || sty == SCE_PL_COMMENTLINE))
            backPos--;
        if (sty == SCE_PL_OPERATOR)
            backflag = BACK_OPERATOR;
        else if (sty == SCE_PL_WORD)
            backflag = BACK_KEYWORD;
    }

	styler.StartAt(startPos);
	char chPrev = styler.SafeGetCharAt(startPos - 1);
	if (startPos == 0)
		chPrev = '\n';
	char chNext = styler[startPos];
	styler.StartSegment(startPos);

	for (unsigned int i = startPos; i < lengthDoc; i++) {
		char ch = chNext;
		// if the current character is not consumed due to the completion of an
		// earlier style, lexing can be restarted via a simple goto
	restartLexer:
		chNext = styler.SafeGetCharAt(i + 1);
		char chNext2 = styler.SafeGetCharAt(i + 2);

		if (styler.IsLeadByte(ch)) {
			chNext = styler.SafeGetCharAt(i + 2);
			chPrev = ' ';
			i += 1;
			continue;
		}
		if ((chPrev == '\r' && ch == '\n')) {	// skip on DOS/Windows
			styler.ColourTo(i, state);
			chPrev = ch;
			continue;
		}

		if (HereDoc.State == 1 && isEOLChar(ch)) {
			// Begin of here-doc (the line after the here-doc delimiter):
			// Lexically, the here-doc starts from the next line after the >>, but the
			// first line of here-doc seem to follow the style of the last EOL sequence
			HereDoc.State = 2;
			if (HereDoc.Quoted) {
				if (state == SCE_PL_HERE_DELIM) {
					// Missing quote at end of string! We are stricter than perl.
					// Colour here-doc anyway while marking this bit as an error.
					state = SCE_PL_ERROR;
				}
				styler.ColourTo(i - 1, state);
				switch (HereDoc.Quote) {
				case '\'':
					state = SCE_PL_HERE_Q ;
					break;
				case '"':
					state = SCE_PL_HERE_QQ;
					break;
				case '`':
					state = SCE_PL_HERE_QX;
					break;
				}
			} else {
				styler.ColourTo(i - 1, state);
				switch (HereDoc.Quote) {
				case '\\':
					state = SCE_PL_HERE_Q ;
					break;
				default :
					state = SCE_PL_HERE_QQ;
				}
			}
		}

		if (state == SCE_PL_DEFAULT) {
			if ((isascii(ch) && isdigit(ch)) || (isascii(chNext) && isdigit(chNext) &&
				(ch == '.' || ch == 'v'))) {
				state = SCE_PL_NUMBER;
                backflag = BACK_NONE;
				numState = PERLNUM_DECIMAL;
				dotCount = 0;
				if (ch == '0') {	// hex,bin,octal
					if (chNext == 'x') {
						numState = PERLNUM_HEX;
					} else if (chNext == 'b') {
                        numState = PERLNUM_BINARY;
                    } else if (isascii(chNext) && isdigit(chNext)) {
                        numState = PERLNUM_OCTAL;
                    }
                    if (numState != PERLNUM_DECIMAL) {
						i++;
						ch = chNext;
						chNext = chNext2;
                    }
				} else if (ch == 'v') {	// vector
					numState = PERLNUM_V_VECTOR;
				}
			} else if (isWordStart(ch)) {
                // if immediately prefixed by '::', always a bareword
                state = SCE_PL_WORD;
                if (chPrev == ':' && styler.SafeGetCharAt(i - 2) == ':') {
                    state = SCE_PL_IDENTIFIER;
                }
                unsigned int kw = i + 1;
                // first check for possible quote-like delimiter
				if (ch == 's' && !isNonQuote(chNext)) {
					state = SCE_PL_REGSUBST;
					Quote.New(2);
				} else if (ch == 'm' && !isNonQuote(chNext)) {
					state = SCE_PL_REGEX;
					Quote.New(1);
				} else if (ch == 'q' && !isNonQuote(chNext)) {
					state = SCE_PL_STRING_Q;
					Quote.New(1);
				} else if (ch == 'y' && !isNonQuote(chNext)) {
					state = SCE_PL_REGSUBST;
					Quote.New(2);
				} else if (ch == 't' && chNext == 'r' && !isNonQuote(chNext2)) {
					state = SCE_PL_REGSUBST;
					Quote.New(2);
                    kw++;
				} else if (ch == 'q' && (chNext == 'q' || chNext == 'r' || chNext == 'w' || chNext == 'x') && !isNonQuote(chNext2)) {
					if      (chNext == 'q') state = SCE_PL_STRING_QQ;
					else if (chNext == 'x') state = SCE_PL_STRING_QX;
					else if (chNext == 'r') state = SCE_PL_STRING_QR;
					else if (chNext == 'w') state = SCE_PL_STRING_QW;
					Quote.New(1);
                    kw++;
				} else if (ch == 'x' && (chNext == '=' ||	// repetition
                           !isWordStart(chNext) ||
                           (isdigit(chPrev) && isdigit(chNext)))) {
                    state = SCE_PL_OPERATOR;
                }
                // if potentially a keyword, scan forward and grab word, then check
                // if it's really one; if yes, disambiguation test is performed
                // otherwise it is always a bareword and we skip a lot of scanning
                // note: keywords assumed to be limited to [_a-zA-Z] only
                if (state == SCE_PL_WORD) {
                    while (isWordStart(styler.SafeGetCharAt(kw))) kw++;
                    if (!isPerlKeyword(styler.GetStartSegment(), kw, keywords, styler)) {
                        state = SCE_PL_IDENTIFIER;
                    }
                }
                // if already SCE_PL_IDENTIFIER, then no ambiguity, skip this
                // for quote-like delimiters/keywords, attempt to disambiguate
                // to select for bareword, change state -> SCE_PL_IDENTIFIER
                if (state != SCE_PL_IDENTIFIER && i > 0) {
                    unsigned int j = i;
                    bool moreback = false;      // true if passed newline/comments
                    bool brace = false;         // true if opening brace found
                    char ch2;
                    // first look backwards past whitespace/comments for EOLs
                    // if BACK_NONE, neither operator nor keyword, so skip test
                    if (backflag != BACK_NONE) {
                        while (--j > backPos) {
                            if (isEOLChar(styler.SafeGetCharAt(j)))
                                moreback = true;
                        }
                        ch2 = styler.SafeGetCharAt(j);
                        if (ch2 == '{' && !moreback) {
                            // {bareword: possible variable spec
                            brace = true;
                        } else if ((ch2 == '&' && styler.SafeGetCharAt(j - 1) != '&')
                                // &bareword: subroutine call
                                || (ch2 == '>' && styler.SafeGetCharAt(j - 1) == '-')
                                // ->bareword: part of variable spec
                                || (ch2 == 'b' && styler.Match(j - 2, "su"))) {
                                // sub bareword: subroutine declaration
                                // (implied BACK_KEYWORD, no keywords end in 'sub'!)
                            state = SCE_PL_IDENTIFIER;
                        }
                        // if status still ambiguous, look forward after word past
                        // tabs/spaces only; if ch2 isn't one of '[{(,' it can never
                        // match anything, so skip the whole thing
                        j = kw;
                        if (state != SCE_PL_IDENTIFIER
                            && (ch2 == '{' || ch2 == '(' || ch2 == '['|| ch2 == ',')
                            && kw < lengthDoc) {
                            while (ch2 = styler.SafeGetCharAt(j),
                                   (ch2 == ' ' || ch2 == '\t') && j < lengthDoc) {
                                j++;
                            }
                            if ((ch2 == '}' && brace)
                             // {bareword}: variable spec
                             || (ch2 == '=' && styler.SafeGetCharAt(j + 1) == '>')) {
                             // [{(, bareword=>: hash literal
                                state = SCE_PL_IDENTIFIER;
                            }
                        }
                    }
                }
                backflag = BACK_NONE;
                // an identifier or bareword
                if (state == SCE_PL_IDENTIFIER) {
                    if ((!isWordStart(chNext) && chNext != '\'')
                        || (chNext == '.' && chNext2 == '.')) {
                        // We need that if length of word == 1!
                        // This test is copied from the SCE_PL_WORD handler.
                        styler.ColourTo(i, SCE_PL_IDENTIFIER);
                        state = SCE_PL_DEFAULT;