| 1 | 1 | simandl | /* |
| 2 | | | ParseMaster, version 1.0 (pre-release) (2005/05/12) x6 |
| 3 | | | Copyright 2005, Dean Edwards |
| 4 | | | Web: http://dean.edwards.name/ |
| 5 | | | |
| 6 | | | This software is licensed under the CC-GNU LGPL |
| 7 | | | Web: http://creativecommons.org/licenses/LGPL/2.1/ |
| 8 | | | */ |
| 9 | | | |
| 10 | | | /* a multi-pattern parser */ |
| 11 | | | |
| 12 | | | /* --- (include) http://dean.edwards.name/my/fix-ie5.js --- */ |
| 13 | | | /* --- (require) http://dean.edwards.name/common/common.js --- */ |
| 14 | | | |
| 15 | | | function ParseMaster() { |
| 16 | | | // constants |
| 17 | | | var $EXPRESSION = 0, $REPLACEMENT = 1, $LENGTH = 2; |
| 18 | | | // used to determine nesting levels |
| 19 | | | var $GROUPS = /\(/g, $SUB_REPLACE = /\$\d/, $INDEXED = /^\$\d+$/, |
| 20 | | | $TRIM = /(['"])\1\+(.*)\+\1\1$/, $$ESCAPE = /\\./g, $QUOTE = /'/, |
| 21 | | | $$DELETED = /\001[^\001]*\001/g; |
| 22 | | | function $DELETE($match, $offset){return "\001" + $match[$offset] + "\001"}; |
| 23 | | | // public |
| 24 | | | this.add = function($expression, $replacement) { |
| 25 | | | if (!$replacement) $replacement = $DELETE; |
| 26 | | | // count the number of sub-expressions |
| 27 | | | // - add one because each pattern is itself a sub-expression |
| 28 | | | var $length = (_internalEscape(String($expression)).match($GROUPS) || "").length + 1; |
| 29 | | | // does the pattern deal with sub-expressions? |
| 30 | | | if (typeof $replacement == "string" && $SUB_REPLACE.test($replacement)) { |
| 31 | | | // a simple lookup? (e.g. "$2") |
| 32 | | | if ($INDEXED.test($replacement)) { |
| 33 | | | // store the index (used for fast retrieval of matched strings) |
| 34 | | | $replacement = parseInt($replacement.slice(1)) - 1; |
| 35 | | | } else { // a complicated lookup (e.g. "Hello $2 $1") |
| 36 | | | // build a function to do the lookup |
| 37 | | | var i = $length; |
| 38 | | | var $quote = $QUOTE.test(_internalEscape($replacement)) ? '"' : "'"; |
| 39 | | | while (i) $replacement = $replacement.split("$" + i--).join($quote + "+a[o+" + i + "]+" + $quote); |
| 40 | | | $replacement = new Function("a,o", "return" + $quote + $replacement.replace($TRIM, "$1") + $quote); |
| 41 | | | } |
| 42 | | | } |
| 43 | | | // pass the modified arguments |
| 44 | | | _add($expression || "/^$/", $replacement, $length); |
| 45 | | | }; |
| 46 | | | // execute the global replacement |
| 47 | | | this.exec = function($string) { |
| 48 | | | return _unescape(_escape($string, this.escapeChar).replace( |
| 49 | | | new RegExp(_patterns, this.ignoreCase ? "gi" : "g"), _replacement), this.escapeChar).replace($$DELETED, ""); |
| 50 | | | }; |
| 51 | | | // clear the patterns collections so that this object may be re-used |
| 52 | | | this.reset = function() { |
| 53 | | | _patterns.length = 0; |
| 54 | | | }; |
| 55 | | | |
| 56 | | | // private |
| 57 | | | var _patterns = []; // patterns stored by index |
| 58 | | | var _toString = function(){return "(" + String(this[$EXPRESSION]).slice(1, -1) + ")"}; |
| 59 | | | _patterns.toString = function(){return this.join("|")}; |
| 60 | | | // create and add a new pattern to the patterns collection |
| 61 | | | function _add() { |
| 62 | | | arguments.toString = _toString; |
| 63 | | | // store the pattern - as an arguments object (i think this is quicker..?) |
| 64 | | | _patterns[_patterns.length] = arguments; |
| 65 | | | } |
| 66 | | | // this is the global replace function (it's quite complicated) |
| 67 | | | function _replacement() { |
| 68 | | | if (!arguments[0]) return ""; |
| 69 | | | var i = 1, j = 0, $pattern; |
| 70 | | | // loop through the patterns |
| 71 | | | while ($pattern = _patterns[j++]) { |
| 72 | | | // do we have a result? |
| 73 | | | if (arguments[i]) { |
| 74 | | | var $replacement = $pattern[$REPLACEMENT]; |
| 75 | | | switch (typeof $replacement) { |
| 76 | | | case "function": return $replacement(arguments, i); |
| 77 | | | case "number": return arguments[$replacement + i]; |
| 78 | | | default: return $replacement; |
| 79 | | | } |
| 80 | | | // skip over references to sub-expressions |
| 81 | | | } else i += $pattern[$LENGTH]; |
| 82 | | | } |
| 83 | | | }; |
| 84 | | | // encode escaped characters |
| 85 | | | var _escaped = []; |
| 86 | | | function _escape($string, $escapeChar) { |
| 87 | | | return $escapeChar ? $string.replace(new RegExp("\\" + $escapeChar + "(.)", "g"), function($match, $char) { |
| 88 | | | _escaped[_escaped.length] = $char; |
| 89 | | | return $escapeChar; |
| 90 | | | }) : $string; |
| 91 | | | }; |
| 92 | | | // decode escaped characters |
| 93 | | | function _unescape($string, $escapeChar) { |
| 94 | | | var i = 0; |
| 95 | | | return $escapeChar ? $string.replace(new RegExp("\\" + $escapeChar, "g"), function() { |
| 96 | | | return $escapeChar + (_escaped[i++] || ""); |
| 97 | | | }) : $string; |
| 98 | | | }; |
| 99 | | | function _internalEscape($string) { |
| 100 | | | return $string.replace($$ESCAPE, ""); |
| 101 | | | }; |
| 102 | | | }; |
| 103 | | | Common.specialize({ |
| 104 | | | constructor: ParseMaster, |
| 105 | | | ignoreCase: false, |
| 106 | | | escapeChar: "" |
| 107 | | | }); |
