77 lines
2.9 KiB
JavaScript
77 lines
2.9 KiB
JavaScript
|
"use strict";
|
||
|
|
var __importDefault = (this && this.__importDefault) || function (mod) {
|
||
|
|
return (mod && mod.__esModule) ? mod : { "default": mod };
|
||
|
|
};
|
||
|
|
Object.defineProperty(exports, "__esModule", { value: true });
|
||
|
|
exports.encodeNonAsciiHTML = exports.encodeHTML = void 0;
|
||
|
|
var encode_html_js_1 = __importDefault(require("./generated/encode-html.js"));
|
||
|
|
var escape_js_1 = require("./escape.js");
|
||
|
|
var htmlReplacer = /[\t\n!-,./:-@[-`\f{-}$\x80-\uFFFF]/g;
|
||
|
|
/**
|
||
|
|
* Encodes all characters in the input using HTML entities. This includes
|
||
|
|
* characters that are valid ASCII characters in HTML documents, such as `#`.
|
||
|
|
*
|
||
|
|
* To get a more compact output, consider using the `encodeNonAsciiHTML`
|
||
|
|
* function, which will only encode characters that are not valid in HTML
|
||
|
|
* documents, as well as non-ASCII characters.
|
||
|
|
*
|
||
|
|
* If a character has no equivalent entity, a numeric hexadecimal reference
|
||
|
|
* (eg. `ü`) will be used.
|
||
|
|
*/
|
||
|
|
function encodeHTML(data) {
|
||
|
|
return encodeHTMLTrieRe(htmlReplacer, data);
|
||
|
|
}
|
||
|
|
exports.encodeHTML = encodeHTML;
|
||
|
|
/**
|
||
|
|
* Encodes all non-ASCII characters, as well as characters not valid in HTML
|
||
|
|
* documents using HTML entities. This function will not encode characters that
|
||
|
|
* are valid in HTML documents, such as `#`.
|
||
|
|
*
|
||
|
|
* If a character has no equivalent entity, a numeric hexadecimal reference
|
||
|
|
* (eg. `ü`) will be used.
|
||
|
|
*/
|
||
|
|
function encodeNonAsciiHTML(data) {
|
||
|
|
return encodeHTMLTrieRe(escape_js_1.xmlReplacer, data);
|
||
|
|
}
|
||
|
|
exports.encodeNonAsciiHTML = encodeNonAsciiHTML;
|
||
|
|
function encodeHTMLTrieRe(regExp, str) {
|
||
|
|
var ret = "";
|
||
|
|
var lastIdx = 0;
|
||
|
|
var match;
|
||
|
|
while ((match = regExp.exec(str)) !== null) {
|
||
|
|
var i = match.index;
|
||
|
|
ret += str.substring(lastIdx, i);
|
||
|
|
var char = str.charCodeAt(i);
|
||
|
|
var next = encode_html_js_1.default.get(char);
|
||
|
|
if (typeof next === "object") {
|
||
|
|
// We are in a branch. Try to match the next char.
|
||
|
|
if (i + 1 < str.length) {
|
||
|
|
var nextChar = str.charCodeAt(i + 1);
|
||
|
|
var value = typeof next.n === "number"
|
||
|
|
? next.n === nextChar
|
||
|
|
? next.o
|
||
|
|
: undefined
|
||
|
|
: next.n.get(nextChar);
|
||
|
|
if (value !== undefined) {
|
||
|
|
ret += value;
|
||
|
|
lastIdx = regExp.lastIndex += 1;
|
||
|
|
continue;
|
||
|
|
}
|
||
|
|
}
|
||
|
|
next = next.v;
|
||
|
|
}
|
||
|
|
// We might have a tree node without a value; skip and use a numeric entitiy.
|
||
|
|
if (next !== undefined) {
|
||
|
|
ret += next;
|
||
|
|
lastIdx = i + 1;
|
||
|
|
}
|
||
|
|
else {
|
||
|
|
var cp = (0, escape_js_1.getCodePoint)(str, i);
|
||
|
|
ret += "&#x".concat(cp.toString(16), ";");
|
||
|
|
// Increase by 1 if we have a surrogate pair
|
||
|
|
lastIdx = regExp.lastIndex += Number(cp !== char);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
return ret + str.substr(lastIdx);
|
||
|
|
}
|
||
|
|
//# sourceMappingURL=encode.js.map
|