lexruby.cxx

porting scintilla to qt

static int skipWhitespace(int startPos,
                           int endPos,
                           Accessor &styler) {
    for (int i = startPos; i < endPos; i++) {
        if (!iswhitespace(styler[i])) {
            return i;
        }
    }
    return endPos;
}
    
// This routine looks for false positives like
// undef foo, <<
// There aren't too many.
//
// iPrev points to the start of <<

static bool sureThisIsHeredoc(int iPrev,
                              Accessor &styler,
                              char *prevWord) {
                    
    // Not so fast, since Ruby's so dynamic.  Check the context
    // to make sure we're OK.
    int prevStyle;
    int lineStart = styler.GetLine(iPrev);
    int lineStartPosn = styler.LineStart(lineStart);
    styler.Flush();

    // Find the first word after some whitespace
    int firstWordPosn = skipWhitespace(lineStartPosn, iPrev, styler);
    if (firstWordPosn >= iPrev) {
        // Have something like {^     <<}
		//XXX Look at the first previous non-comment non-white line
		// to establish the context.  Not too likely though.
        return true;
    } else {
        switch (prevStyle = styler.StyleAt(firstWordPosn)) {
        case SCE_RB_WORD:
        case SCE_RB_WORD_DEMOTED:
        case SCE_RB_IDENTIFIER:
            break;
        default:
            return true;
        }
    }
    int firstWordEndPosn = firstWordPosn;
    char *dst = prevWord;
    for (;;) {
        if (firstWordEndPosn >= iPrev ||
            styler.StyleAt(firstWordEndPosn) != prevStyle) {
            *dst = 0;
            break;
        }
        *dst++ = styler[firstWordEndPosn];
        firstWordEndPosn += 1;
    }
    //XXX Write a style-aware thing to regex scintilla buffer objects
    if (!strcmp(prevWord, "undef")
        || !strcmp(prevWord, "def")
        || !strcmp(prevWord, "alias")) {
        // These keywords are what we were looking for
        return false;
    }
    return true;
}

// Routine that saves us from allocating a buffer for the here-doc target
// targetEndPos points one past the end of the current target
static bool haveTargetMatch(int currPos,
                            int lengthDoc,
                            int targetStartPos,
                            int targetEndPos,
                            Accessor &styler) {
    if (lengthDoc - currPos < targetEndPos - targetStartPos) {
        return false;
    }
    int i, j;
    for (i = targetStartPos, j = currPos;
         i < targetEndPos && j < lengthDoc;
         i++, j++) {
        if (styler[i] != styler[j]) {
            return false;
        }
    }
    return true;
}

// We need a check because the form
// [identifier] <<[target]
// is ambiguous.  The Ruby lexer/parser resolves it by
// looking to see if [identifier] names a variable or a
// function.  If it's the first, it's the start of a here-doc.
// If it's a var, it's an operator.  This lexer doesn't
// maintain a symbol table, so it looks ahead to see what's
// going on, in cases where we have
// ^[white-space]*[identifier([.|::]identifier)*][white-space]*<<[target]
//
// If there's no occurrence of [target] on a line, assume we don't.

// return true == yes, we have no heredocs

static bool sureThisIsNotHeredoc(int lt2StartPos,
                                 Accessor &styler) {
    int prevStyle;
     // Use full document, not just part we're styling
    int lengthDoc = styler.Length();
    int lineStart = styler.GetLine(lt2StartPos);
    int lineStartPosn = styler.LineStart(lineStart);
    styler.Flush();
    const bool definitely_not_a_here_doc = true;
    const bool looks_like_a_here_doc = false;
    
    // Find the first word after some whitespace
    int firstWordPosn = skipWhitespace(lineStartPosn, lt2StartPos, styler);
    if (firstWordPosn >= lt2StartPos) {
        return definitely_not_a_here_doc;
    }
    prevStyle = styler.StyleAt(firstWordPosn);
    // If we have '<<' following a keyword, it's not a heredoc
    if (prevStyle != SCE_RB_IDENTIFIER) {
        return definitely_not_a_here_doc;
    }
    int newStyle = prevStyle;
    // Some compilers incorrectly warn about uninit newStyle
    for (firstWordPosn += 1; firstWordPosn <= lt2StartPos; firstWordPosn += 1) {
        // Inner loop looks at the name
        for (; firstWordPosn <= lt2StartPos; firstWordPosn += 1) {
            newStyle = styler.StyleAt(firstWordPosn);
            if (newStyle != prevStyle) {
                break;
            }
        }
        // Do we have '::' or '.'?
        if (firstWordPosn < lt2StartPos && newStyle == SCE_RB_OPERATOR) {
            char ch = styler[firstWordPosn];
            if (ch == '.') {
                // yes
            } else if (ch == ':') {
                if (styler.StyleAt(++firstWordPosn) != SCE_RB_OPERATOR) {
                    return definitely_not_a_here_doc;
                } else if (styler[firstWordPosn] != ':') {
                    return definitely_not_a_here_doc;
                }
            } else {
                break;
            }
        } else {
            break;
        }
    }
    // Skip next batch of white-space
    firstWordPosn = skipWhitespace(firstWordPosn, lt2StartPos, styler);
    if (firstWordPosn != lt2StartPos) {
        // Have [[^ws[identifier]ws[*something_else*]ws<<
        return definitely_not_a_here_doc;
    }
    // OK, now 'j' will point to the current spot moving ahead
	int j = firstWordPosn + 1;
    if (styler.StyleAt(j) != SCE_RB_OPERATOR || styler[j] != '<') {
        // This shouldn't happen
        return definitely_not_a_here_doc;
    }
    int nextLineStartPosn = styler.LineStart(lineStart + 1);
    if (nextLineStartPosn >= lengthDoc) {
        return definitely_not_a_here_doc;
    }
    j = skipWhitespace(j + 1, nextLineStartPosn, styler);
    if (j >= lengthDoc) {
        return definitely_not_a_here_doc;
    }
    bool allow_indent;
    int target_start, target_end;
    // From this point on no more styling, since we're looking ahead
    if (styler[j] == '-') {
        allow_indent = true;
        j++;
    } else {
        allow_indent = false;
    }

    // Allow for quoted targets.
    char target_quote = 0;
    switch (styler[j]) {
    case '\'':
    case '"':
    case '`':
        target_quote = styler[j];
        j += 1;
    }
    
    if (isSafeAlnum(styler[j])) {
        // Init target_end because some compilers think it won't
        // be initialized by the time it's used
        target_start = target_end = j;
        j++;
    } else {
        return definitely_not_a_here_doc;
    }
    for (; j < lengthDoc; j++) {
        if (!isSafeAlnum(styler[j])) {
            if (target_quote && styler[j] != target_quote) {
                // unquoted end
                return definitely_not_a_here_doc;
            }

            // And for now make sure that it's a newline
            // don't handle arbitrary expressions yet
            
            target_end = j;
			if (target_quote) {
				// Now we can move to the character after the string delimiter.
				j += 1;
			}
            j = skipWhitespace(j, lengthDoc, styler);
            if (j >= lengthDoc) {
                return definitely_not_a_here_doc;
            } else {
                char ch = styler[j];
                if (ch == '#' || isEOLChar(ch)) {
                    // This is OK, so break and continue;
                    break;
                } else {
                    return definitely_not_a_here_doc;
                }
            }
        }
    }

    // Just look at the start of each line
    int last_line = styler.GetLine(lengthDoc - 1);
    // But don't go too far
    if (last_line > lineStart + 50) {
        last_line = lineStart + 50;
    }
    for (int line_num = lineStart + 1; line_num <= last_line; line_num++) {
        if (allow_indent) {
            j = skipWhitespace(styler.LineStart(line_num), lengthDoc, styler);
        } else {
            j = styler.LineStart(line_num);
        }
        // target_end is one past the end
        if (haveTargetMatch(j, lengthDoc, target_start, target_end, styler)) {
            // We got it
            return looks_like_a_here_doc;
        }
    }
    return definitely_not_a_here_doc;
}

//todo: if we aren't looking at a stdio character,
// move to the start of the first line that is not in a 
// multi-line construct

static void synchronizeDocStart(unsigned int& startPos,
                                int &length,
                                int &initStyle,
                                Accessor &styler,
                                bool skipWhiteSpace=false) {

    styler.Flush();
    int style = actual_style(styler.StyleAt(startPos));
    switch (style) {
        case SCE_RB_STDIN:
        case SCE_RB_STDOUT:
        case SCE_RB_STDERR:
            // Don't do anything else with these.
            return;
    }
    
    int pos = startPos;
    // Quick way to characterize each line
    int lineStart;
    for (lineStart = styler.GetLine(pos); lineStart > 0; lineStart--) {
        // Now look at the style before the previous line's EOL
        pos = styler.LineStart(lineStart) - 1;
        if (pos <= 10) {
            lineStart = 0;
            break;
        }
        char ch = styler.SafeGetCharAt(pos);
        char chPrev = styler.SafeGetCharAt(pos - 1);
        if (ch == '\n' && chPrev == '\r') {
            pos--;
        }
        if (styler.SafeGetCharAt(pos - 1) == '\\') {
            // Continuation line -- keep going
        } else if (actual_style(styler.StyleAt(pos)) != SCE_RB_DEFAULT) {
            // Part of multi-line construct -- keep going
        } else if (currLineContainsHereDelims(pos, styler)) {
            // Keep going, with pos and length now pointing
            // at the end of the here-doc delimiter
        } else if (skipWhiteSpace && isEmptyLine(pos, styler)) {
            // Keep going
        } else {
            break;
        }
    }
    pos = styler.LineStart(lineStart);
    length += (startPos - pos);
    startPos = pos;
    initStyle = SCE_RB_DEFAULT;
}

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

	// Lexer for Ruby 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;
        // States
        // 0: '<<' encountered
		// 1: collect the delimiter
        // 1b: text between the end of the delimiter and the EOL
		// 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[256];	// the Delimiter, limit of 256: from Perl
        bool CanBeIndented;
		HereDocCls() {
			State = 0;
			DelimiterLength = 0;
			Delimiter[0] = '\0';
            CanBeIndented = false;
		}
	};
	HereDocCls HereDoc;	

	QuoteCls Quote;

    int numDots = 0;  // For numbers --
                      // Don't start lexing in the middle of a num

    synchronizeDocStart(startPos, length, initStyle, styler, // ref args
                        false);