diff options
Diffstat (limited to 'node_modules/entities/lib')
-rw-r--r-- | node_modules/entities/lib/decode.js | 72 | ||||
-rw-r--r-- | node_modules/entities/lib/decode_codepoint.js | 26 | ||||
-rw-r--r-- | node_modules/entities/lib/encode.js | 73 |
3 files changed, 171 insertions, 0 deletions
diff --git a/node_modules/entities/lib/decode.js b/node_modules/entities/lib/decode.js new file mode 100644 index 000000000..5e48bdb68 --- /dev/null +++ b/node_modules/entities/lib/decode.js @@ -0,0 +1,72 @@ +var entityMap = require("../maps/entities.json"), + legacyMap = require("../maps/legacy.json"), + xmlMap = require("../maps/xml.json"), + decodeCodePoint = require("./decode_codepoint.js"); + +var decodeXMLStrict = getStrictDecoder(xmlMap), + decodeHTMLStrict = getStrictDecoder(entityMap); + +function getStrictDecoder(map){ + var keys = Object.keys(map).join("|"), + replace = getReplacer(map); + + keys += "|#[xX][\\da-fA-F]+|#\\d+"; + + var re = new RegExp("&(?:" + keys + ");", "g"); + + return function(str){ + return String(str).replace(re, replace); + }; +} + +var decodeHTML = (function(){ + var legacy = Object.keys(legacyMap) + .sort(sorter); + + var keys = Object.keys(entityMap) + .sort(sorter); + + for(var i = 0, j = 0; i < keys.length; i++){ + if(legacy[j] === keys[i]){ + keys[i] += ";?"; + j++; + } else { + keys[i] += ";"; + } + } + + var re = new RegExp("&(?:" + keys.join("|") + "|#[xX][\\da-fA-F]+;?|#\\d+;?)", "g"), + replace = getReplacer(entityMap); + + function replacer(str){ + if(str.substr(-1) !== ";") str += ";"; + return replace(str); + } + + //TODO consider creating a merged map + return function(str){ + return String(str).replace(re, replacer); + }; +}()); + +function sorter(a, b){ + return a < b ? 1 : -1; +} + +function getReplacer(map){ + return function replace(str){ + if(str.charAt(1) === "#"){ + if(str.charAt(2) === "X" || str.charAt(2) === "x"){ + return decodeCodePoint(parseInt(str.substr(3), 16)); + } + return decodeCodePoint(parseInt(str.substr(2), 10)); + } + return map[str.slice(1, -1)]; + }; +} + +module.exports = { + XML: decodeXMLStrict, + HTML: decodeHTML, + HTMLStrict: decodeHTMLStrict +};
\ No newline at end of file diff --git a/node_modules/entities/lib/decode_codepoint.js b/node_modules/entities/lib/decode_codepoint.js new file mode 100644 index 000000000..730d5bfa2 --- /dev/null +++ b/node_modules/entities/lib/decode_codepoint.js @@ -0,0 +1,26 @@ +var decodeMap = require("../maps/decode.json"); + +module.exports = decodeCodePoint; + +// modified version of https://github.com/mathiasbynens/he/blob/master/src/he.js#L94-L119 +function decodeCodePoint(codePoint){ + + if((codePoint >= 0xD800 && codePoint <= 0xDFFF) || codePoint > 0x10FFFF){ + return "\uFFFD"; + } + + if(codePoint in decodeMap){ + codePoint = decodeMap[codePoint]; + } + + var output = ""; + + if(codePoint > 0xFFFF){ + codePoint -= 0x10000; + output += String.fromCharCode(codePoint >>> 10 & 0x3FF | 0xD800); + codePoint = 0xDC00 | codePoint & 0x3FF; + } + + output += String.fromCharCode(codePoint); + return output; +} diff --git a/node_modules/entities/lib/encode.js b/node_modules/entities/lib/encode.js new file mode 100644 index 000000000..10c0c36e2 --- /dev/null +++ b/node_modules/entities/lib/encode.js @@ -0,0 +1,73 @@ +var inverseXML = getInverseObj(require("../maps/xml.json")), + xmlReplacer = getInverseReplacer(inverseXML); + +exports.XML = getInverse(inverseXML, xmlReplacer); + +var inverseHTML = getInverseObj(require("../maps/entities.json")), + htmlReplacer = getInverseReplacer(inverseHTML); + +exports.HTML = getInverse(inverseHTML, htmlReplacer); + +function getInverseObj(obj){ + return Object.keys(obj).sort().reduce(function(inverse, name){ + inverse[obj[name]] = "&" + name + ";"; + return inverse; + }, {}); +} + +function getInverseReplacer(inverse){ + var single = [], + multiple = []; + + Object.keys(inverse).forEach(function(k){ + if(k.length === 1){ + single.push("\\" + k); + } else { + multiple.push(k); + } + }); + + //TODO add ranges + multiple.unshift("[" + single.join("") + "]"); + + return new RegExp(multiple.join("|"), "g"); +} + +var re_nonASCII = /[^\0-\x7F]/g, + re_astralSymbols = /[\uD800-\uDBFF][\uDC00-\uDFFF]/g; + +function singleCharReplacer(c){ + return "&#x" + c.charCodeAt(0).toString(16).toUpperCase() + ";"; +} + +function astralReplacer(c){ + // http://mathiasbynens.be/notes/javascript-encoding#surrogate-formulae + var high = c.charCodeAt(0); + var low = c.charCodeAt(1); + var codePoint = (high - 0xD800) * 0x400 + low - 0xDC00 + 0x10000; + return "&#x" + codePoint.toString(16).toUpperCase() + ";"; +} + +function getInverse(inverse, re){ + function func(name){ + return inverse[name]; + } + + return function(data){ + return data + .replace(re, func) + .replace(re_astralSymbols, astralReplacer) + .replace(re_nonASCII, singleCharReplacer); + }; +} + +var re_xmlChars = getInverseReplacer(inverseXML); + +function escapeXML(data){ + return data + .replace(re_xmlChars, singleCharReplacer) + .replace(re_astralSymbols, astralReplacer) + .replace(re_nonASCII, singleCharReplacer); +} + +exports.escape = escapeXML; |