diff options
author | Florian Dold <florian.dold@gmail.com> | 2017-05-03 15:35:00 +0200 |
---|---|---|
committer | Florian Dold <florian.dold@gmail.com> | 2017-05-03 15:35:00 +0200 |
commit | de98e0b232509d5f40c135d540a70e415272ff85 (patch) | |
tree | a79222a5b58484ab3b80d18efcaaa7ccc4769b33 /node_modules/entities/lib/decode.js | |
parent | e0c9d480a73fa629c1e4a47d3e721f1d2d345406 (diff) |
node_modules
Diffstat (limited to 'node_modules/entities/lib/decode.js')
-rw-r--r-- | node_modules/entities/lib/decode.js | 72 |
1 files changed, 72 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 |