lexperl.cxx.svn-base

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

                    }
                // a keyword
                } else if (state == SCE_PL_WORD) {
                    i = kw - 1;
                    if (ch == '_' && chNext == '_' &&
                        (isMatch(styler, lengthDoc, styler.GetStartSegment(), "__DATA__")
                      || isMatch(styler, lengthDoc, styler.GetStartSegment(), "__END__"))) {
                        styler.ColourTo(i, SCE_PL_DATASECTION);
                        state = SCE_PL_DATASECTION;
                    } else {
                        styler.ColourTo(i, SCE_PL_WORD);
                        state = SCE_PL_DEFAULT;
                        backflag = BACK_KEYWORD;
                        backPos = i;
                    }
                    ch = styler.SafeGetCharAt(i);
                    chNext = styler.SafeGetCharAt(i + 1);
                // a repetition operator 'x'
                } else if (state == SCE_PL_OPERATOR) {
                    styler.ColourTo(i, SCE_PL_OPERATOR);
                    state = SCE_PL_DEFAULT;
                // quote-like delimiter, skip one char if double-char delimiter
                } else {
                    i = kw - 1;
                    chNext = styler.SafeGetCharAt(i + 1);
                }
			} else if (ch == '#') {
				state = SCE_PL_COMMENTLINE;
			} else if (ch == '\"') {
				state = SCE_PL_STRING;
				Quote.New(1);
				Quote.Open(ch);
                backflag = BACK_NONE;
			} else if (ch == '\'') {
				if (chPrev == '&') {
					// Archaic call
					styler.ColourTo(i, state);
				} else {
					state = SCE_PL_CHARACTER;
					Quote.New(1);
					Quote.Open(ch);
				}
                backflag = BACK_NONE;
			} else if (ch == '`') {
				state = SCE_PL_BACKTICKS;
				Quote.New(1);
				Quote.Open(ch);
                backflag = BACK_NONE;
			} else if (ch == '$') {
				if ((chNext == '{') || isspacechar(chNext)) {
					styler.ColourTo(i, SCE_PL_SCALAR);
				} else {
					state = SCE_PL_SCALAR;
					if ((chNext == '`' && chNext2 == '`')
                     || (chNext == ':' && chNext2 == ':')) {
						i += 2;
						ch = styler.SafeGetCharAt(i);
						chNext = styler.SafeGetCharAt(i + 1);
					} else {
						i++;
						ch = chNext;
						chNext = chNext2;
					}
				}
                backflag = BACK_NONE;
			} else if (ch == '@') {
				if (!isascii(chNext) || isalpha(chNext) || chNext == '#' || chNext == '$'
					|| chNext == '_' || chNext == '+' || chNext == '-') {
					state = SCE_PL_ARRAY;
                } else if (chNext == ':' && chNext2 == ':') {
                    state = SCE_PL_ARRAY;
                    i += 2;
                    ch = styler.SafeGetCharAt(i);
                    chNext = styler.SafeGetCharAt(i + 1);
				} else if (chNext != '{' && chNext != '[') {
					styler.ColourTo(i, SCE_PL_ARRAY);
				} else {
					styler.ColourTo(i, SCE_PL_ARRAY);
				}
                backflag = BACK_NONE;
			} else if (ch == '%') {
				if (!isascii(chNext) || isalpha(chNext) || chNext == '#' || chNext == '$'
                    || chNext == '_' || chNext == '!' || chNext == '^') {
					state = SCE_PL_HASH;
                    i++;
                    ch = chNext;
                    chNext = chNext2;
                } else if (chNext == ':' && chNext2 == ':') {
                    state = SCE_PL_HASH;
                    i += 2;
                    ch = styler.SafeGetCharAt(i);
                    chNext = styler.SafeGetCharAt(i + 1);
				} else if (chNext == '{') {
					styler.ColourTo(i, SCE_PL_HASH);
				} else {
					styler.ColourTo(i, SCE_PL_OPERATOR);
				}
                backflag = BACK_NONE;
			} else if (ch == '*') {
                char strch[2];
                strch[0] = chNext;
                strch[1] = '\0';
                if (chNext == ':' && chNext2 == ':') {
                    state = SCE_PL_SYMBOLTABLE;
                    i += 2;
                    ch = styler.SafeGetCharAt(i);
                    chNext = styler.SafeGetCharAt(i + 1);
				} else if (!isascii(chNext) || isalpha(chNext) || chNext == '_'
                        || NULL != strstr("^/|,\\\";#%^:?<>)[]", strch)) {
					state = SCE_PL_SYMBOLTABLE;
                    i++;
                    ch = chNext;
                    chNext = chNext2;
				} else if (chNext == '{') {
					styler.ColourTo(i, SCE_PL_SYMBOLTABLE);
				} else {
					if (chNext == '*') {	// exponentiation
						i++;
						ch = chNext;
						chNext = chNext2;
					}
					styler.ColourTo(i, SCE_PL_OPERATOR);
				}
                backflag = BACK_NONE;
			} else if (ch == '/' || (ch == '<' && chNext == '<')) {
				// Explicit backward peeking to set a consistent preferRE for
				// any slash found, so no longer need to track preferRE state.
				// Find first previous significant lexed element and interpret.
                // Test for HERE doc start '<<' shares this code, helps to
                // determine if it should be an operator.
				bool preferRE = false;
                bool isHereDoc = (ch == '<');
                bool hereDocSpace = false;      // these are for corner case:
                bool hereDocScalar = false;     // SCALAR [whitespace] '<<'
				unsigned int bk = (i > 0)? i - 1: 0;
				char bkch;
				styler.Flush();
                if (styler.StyleAt(bk) == SCE_PL_DEFAULT)
                    hereDocSpace = true;
				while ((bk > 0) && (styler.StyleAt(bk) == SCE_PL_DEFAULT ||
					styler.StyleAt(bk) == SCE_PL_COMMENTLINE)) {
					bk--;
				}
				if (bk == 0) {
					// position 0 won't really be checked; rarely happens
					// hard to fix due to an unsigned index i
					preferRE = true;
				} else {
					int bkstyle = styler.StyleAt(bk);
					bkch = styler.SafeGetCharAt(bk);
					switch(bkstyle) {
					case SCE_PL_OPERATOR:
						preferRE = true;
						if (bkch == ')' || bkch == ']') {
							preferRE = false;
						} else if (bkch == '}') {
							// backtrack further, count balanced brace pairs
							// if a brace pair found, see if it's a variable
							int braceCount = 1;
							while (--bk > 0) {
								bkstyle = styler.StyleAt(bk);
								if (bkstyle == SCE_PL_OPERATOR) {
									bkch = styler.SafeGetCharAt(bk);
									if (bkch == ';') {	// early out
										break;
									} else if (bkch == '}') {
										braceCount++;
									} else if (bkch == '{') {
										if (--braceCount == 0)
											break;
									}
								}
							}
							if (bk == 0) {
								// at beginning, true
							} else if (braceCount == 0) {
								// balanced { found, bk>0, skip more whitespace
								if (styler.StyleAt(--bk) == SCE_PL_DEFAULT) {
									while (bk > 0) {
										bkstyle = styler.StyleAt(--bk);
										if (bkstyle != SCE_PL_DEFAULT)
											break;
									}
								}
								bkstyle = styler.StyleAt(bk);
								if (bkstyle == SCE_PL_SCALAR
								 || bkstyle == SCE_PL_ARRAY
								 || bkstyle == SCE_PL_HASH
								 || bkstyle == SCE_PL_SYMBOLTABLE
								 || bkstyle == SCE_PL_OPERATOR) {
									preferRE = false;
								}
							}
						}
						break;
					case SCE_PL_IDENTIFIER:
						preferRE = true;
						if (bkch == '>') {	// inputsymbol
							preferRE = false;
							break;
						}
						// backtrack to find "->" or "::" before identifier
						while (bk > 0 && styler.StyleAt(bk) == SCE_PL_IDENTIFIER) {
							bk--;
						}
						while (bk > 0) {
							bkstyle = styler.StyleAt(bk);
							if (bkstyle == SCE_PL_DEFAULT ||
							    bkstyle == SCE_PL_COMMENTLINE) {
							} else if (bkstyle == SCE_PL_OPERATOR) {
								// gcc 3.2.3 bloats if more compact form used
								bkch = styler.SafeGetCharAt(bk);
								if (bkch == '>') { // "->"
									if (styler.SafeGetCharAt(bk - 1) == '-') {
										preferRE = false;
										break;
									}
								} else if (bkch == ':') { // "::"
									if (styler.SafeGetCharAt(bk - 1) == ':') {
										preferRE = false;
										break;
									}
								}
							} else {// bare identifier, usually a function call but Perl
								// optimizes them as pseudo-constants, then the next
								// '/' will be a divide; favour divide over regex
								// if there is a whitespace after the '/'
								if (isspacechar(chNext)) {
									preferRE = false;
								}
								break;
							}
							bk--;
						}
						break;
                    case SCE_PL_SCALAR:     // for $var<< case
                        hereDocScalar = true;
                        break;
					// other styles uses the default, preferRE=false
					case SCE_PL_WORD:
					case SCE_PL_POD:
					case SCE_PL_POD_VERB:
					case SCE_PL_HERE_Q:
					case SCE_PL_HERE_QQ:
					case SCE_PL_HERE_QX:
						preferRE = true;
						break;
					}
				}
                if (isHereDoc) {    // handle HERE doc
                    // if SCALAR whitespace '<<', *always* a HERE doc
                    if (preferRE || (hereDocSpace && hereDocScalar)) {
                        state = SCE_PL_HERE_DELIM;
                        HereDoc.State = 0;
                    } else {        // << operator
						i++;
						ch = chNext;
						chNext = chNext2;
                        styler.ColourTo(i, SCE_PL_OPERATOR);
                    }
                } else {            // handle regexp
                    if (preferRE) {
                        state = SCE_PL_REGEX;
                        Quote.New(1);
                        Quote.Open(ch);
                    } else {        // / operator
                        styler.ColourTo(i, SCE_PL_OPERATOR);
                    }
                }
                backflag = BACK_NONE;
			} else if (ch == '<') {
				// looks forward for matching > on same line
				unsigned int fw = i + 1;
				while (fw < lengthDoc) {
					char fwch = styler.SafeGetCharAt(fw);
					if (fwch == ' ') {
						if (styler.SafeGetCharAt(fw-1) != '\\' ||
						    styler.SafeGetCharAt(fw-2) != '\\')
						break;
					} else if (isEOLChar(fwch) || isspacechar(fwch)) {
						break;
					} else if (fwch == '>') {
						if ((fw - i) == 2 &&	// '<=>' case
						    styler.SafeGetCharAt(fw-1) == '=') {
							styler.ColourTo(fw, SCE_PL_OPERATOR);
						} else {
							styler.ColourTo(fw, SCE_PL_IDENTIFIER);
						}
						i = fw;
						ch = fwch;
						chNext = styler.SafeGetCharAt(i+1);
					}
					fw++;
				}
				styler.ColourTo(i, SCE_PL_OPERATOR);
                backflag = BACK_NONE;
			} else if (ch == '='	// POD
			           && isalpha(chNext)
			           && (isEOLChar(chPrev))) {
				state = SCE_PL_POD;
                backflag = BACK_NONE;
				//sookedpos = 0;
				//sooked[sookedpos] = '\0';
			} else if (ch == '-'	// file test operators
			           && isSingleCharOp(chNext)
			           && !isalnum((chNext2 = styler.SafeGetCharAt(i+2)))) {
				styler.ColourTo(i + 1, SCE_PL_WORD);
				state = SCE_PL_DEFAULT;
				i++;
				ch = chNext;
				chNext = chNext2;
                backflag = BACK_NONE;
			} else if (isPerlOperator(ch)) {
				if (ch == '.' && chNext == '.') { // .. and ...
					i++;
					if (chNext2 == '.') { i++; }
					state = SCE_PL_DEFAULT;
					ch = styler.SafeGetCharAt(i);
					chNext = styler.SafeGetCharAt(i + 1);
				}
				styler.ColourTo(i, SCE_PL_OPERATOR);
                backflag = BACK_OPERATOR;
                backPos = i;
			} else {
				// keep colouring defaults to make restart easier
				styler.ColourTo(i, SCE_PL_DEFAULT);
			}
		} else if (state == SCE_PL_NUMBER) {
			if (ch == '.') {
				if (chNext == '.') {
					// double dot is always an operator
					goto numAtEnd;
				} else if (numState <= PERLNUM_FLOAT) {
					// non-decimal number or float exponent, consume next dot
					styler.ColourTo(i - 1, SCE_PL_NUMBER);
					styler.ColourTo(i, SCE_PL_OPERATOR);
					state = SCE_PL_DEFAULT;
				} else { // decimal or vectors allows dots
					dotCount++;
					if (numState == PERLNUM_DECIMAL) {
						if (dotCount > 1) {
							if (isdigit(chNext)) { // really a vector
								numState = PERLNUM_VECTOR;
							} else	// number then dot
								goto numAtEnd;
						}
					} else { // vectors
						if (!isdigit(chNext))	// vector then dot
							goto numAtEnd;
					}
				}
			} else if (ch == '_' && numState == PERLNUM_DECIMAL) {
				if (!isdigit(chNext)) {
					goto numAtEnd;
				}
			} else if (!isascii(ch) || isalnum(ch)) {
				if (numState == PERLNUM_VECTOR || numState == PERLNUM_V_VECTOR) {
					if (!isascii(ch) || isalpha(ch)) {
						if (dotCount == 0) { // change to word
							state = SCE_PL_IDENTIFIER;
						} else { // vector then word
							goto numAtEnd;
						}
					}
				} else if (numState == PERLNUM_DECIMAL) {
					if (ch == 'E' || ch == 'e') { // exponent
						numState = PERLNUM_FLOAT;
						if (chNext == '+' || chNext == '-') {
							i++;
							ch = chNext;
							chNext = chNext2;
						}
					} else if (!isascii(ch) || !isdigit(ch)) { // number then word
						goto numAtEnd;
					}
				} else if (numState == PERLNUM_FLOAT) {
					if (!isdigit(ch)) { // float then word
						goto numAtEnd;
					}
				} else if (numState == PERLNUM_OCTAL) {
                    if (!isdigit(ch))
                        goto numAtEnd;
                    else if (ch > '7')
                        numState = PERLNUM_BAD;
                } else if (numState == PERLNUM_BINARY) {
                    if (!isdigit(ch))
                        goto numAtEnd;
                    else if (ch > '1')
                        numState = PERLNUM_BAD;
                } else if (numState == PERLNUM_HEX) {
                    int ch2 = toupper(ch);
                    if (!isdigit(ch) && !(ch2 >= 'A' && ch2 <= 'F'))
                        goto numAtEnd;
				} else {//(numState == PERLNUM_BAD) {
                    if (!isdigit(ch))
                        goto numAtEnd;
                }
			} else {
				// complete current number or vector
			numAtEnd:
				styler.ColourTo(i - 1, actualNumStyle(numState));
				state = SCE_PL_DEFAULT;
				goto restartLexer;
			}
		} else if (state == SCE_PL_IDENTIFIER) {
			if (!isWordStart(chNext) && chNext != '\'') {
				styler.ColourTo(i, SCE_PL_IDENTIFIER);
				state = SCE_PL_DEFAULT;
				ch = ' ';
			}
		} else {
			if (state == SCE_PL_COMMENTLINE) {
				if (isEOLChar(ch)) {
					styler.ColourTo(i - 1, state);
					state = SCE_PL_DEFAULT;
					goto restartLexer;
				} else if (isEOLChar(chNext)) {
					styler.ColourTo(i, state);
					state = SCE_PL_DEFAULT;