entities 2.2.0 → 4.0.0

package/lib/decode.d.ts

@@ -1,5 +1,14 @@
-export declare const decodeXML: (str: string) => string;
-export declare const decodeHTMLStrict: (str: string) => string;
-export declare type MapType = Record<string, string>;
-export declare const decodeHTML: (str: string) => string;
+import htmlDecodeTree from "./generated/decode-data-html.js";
+import xmlDecodeTree from "./generated/decode-data-xml.js";
+import decodeCodePoint from "./decode_codepoint.js";
+export { htmlDecodeTree, xmlDecodeTree, decodeCodePoint };
+export declare enum BinTrieFlags {
+    VALUE_LENGTH = 49152,
+    BRANCH_LENGTH = 16256,
+    JUMP_TABLE = 127
+}
+export declare function determineBranch(decodeTree: Uint16Array, current: number, nodeIdx: number, char: number): number;
+export declare function decodeHTML(str: string): string;
+export declare function decodeHTMLStrict(str: string): string;
+export declare function decodeXML(str: string): string;
 //# sourceMappingURL=decode.d.ts.map

package/lib/decode.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"decode.d.ts","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":"AAOA,eAAO,MAAM,SAAS,QAOL,MAAM,WAP0B,CAAC;AAClD,eAAO,MAAM,gBAAgB,QAMZ,MAAM,WANoC,CAAC;AAE5D,oBAAY,OAAO,GAAG,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;AAS7C,eAAO,MAAM,UAAU,QAyBN,MAAM,WACnB,CAAC"}
+{"version":3,"file":"decode.d.ts","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":"AAAA,OAAO,cAAc,MAAM,iCAAiC,CAAC;AAC7D,OAAO,aAAa,MAAM,gCAAgC,CAAC;AAC3D,OAAO,eAAe,MAAM,uBAAuB,CAAC;AAGpD,OAAO,EAAE,cAAc,EAAE,aAAa,EAAE,eAAe,EAAE,CAAC;AAc1D,oBAAY,YAAY;IACpB,YAAY,QAAwB;IACpC,aAAa,QAAwB;IACrC,UAAU,MAAwB;CACrC;AA8GD,wBAAgB,eAAe,CAC3B,UAAU,EAAE,WAAW,EACvB,OAAO,EAAE,MAAM,EACf,OAAO,EAAE,MAAM,EACf,IAAI,EAAE,MAAM,GACb,MAAM,CAsCR;AAKD,wBAAgB,UAAU,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAE9C;AAED,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAEpD;AAED,wBAAgB,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAE7C"}

package/lib/decode.js

@@ -3,51 +3,144 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.decodeHTML = exports.decodeHTMLStrict = exports.decodeXML = void 0;
-var entities_json_1 = __importDefault(require("./maps/entities.json"));
-var legacy_json_1 = __importDefault(require("./maps/legacy.json"));
-var xml_json_1 = __importDefault(require("./maps/xml.json"));
-var decode_codepoint_1 = __importDefault(require("./decode_codepoint"));
-var strictEntityRe = /&(?:[a-zA-Z0-9]+|#[xX][\da-fA-F]+|#\d+);/g;
-exports.decodeXML = getStrictDecoder(xml_json_1.default);
-exports.decodeHTMLStrict = getStrictDecoder(entities_json_1.default);
-function getStrictDecoder(map) {
-    var replace = getReplacer(map);
-    return function (str) { return String(str).replace(strictEntityRe, replace); };
+exports.decodeXML = exports.decodeHTMLStrict = exports.decodeHTML = exports.determineBranch = exports.BinTrieFlags = exports.decodeCodePoint = exports.xmlDecodeTree = exports.htmlDecodeTree = void 0;
+var decode_data_html_js_1 = __importDefault(require("./generated/decode-data-html.js"));
+exports.htmlDecodeTree = decode_data_html_js_1.default;
+var decode_data_xml_js_1 = __importDefault(require("./generated/decode-data-xml.js"));
+exports.xmlDecodeTree = decode_data_xml_js_1.default;
+var decode_codepoint_js_1 = __importDefault(require("./decode_codepoint.js"));
+exports.decodeCodePoint = decode_codepoint_js_1.default;
+var BinTrieFlags;
+(function (BinTrieFlags) {
+    BinTrieFlags[BinTrieFlags["VALUE_LENGTH"] = 49152] = "VALUE_LENGTH";
+    BinTrieFlags[BinTrieFlags["BRANCH_LENGTH"] = 16256] = "BRANCH_LENGTH";
+    BinTrieFlags[BinTrieFlags["JUMP_TABLE"] = 127] = "JUMP_TABLE";
+})(BinTrieFlags = exports.BinTrieFlags || (exports.BinTrieFlags = {}));
+function getDecoder(decodeTree) {
+    return function decodeHTMLBinary(str, strict) {
+        var ret = "";
+        var lastIdx = 0;
+        var strIdx = 0;
+        while ((strIdx = str.indexOf("&", strIdx)) >= 0) {
+            ret += str.slice(lastIdx, strIdx);
+            lastIdx = strIdx;
+            // Skip the "&"
+            strIdx += 1;
+            // If we have a numeric entity, handle this separately.
+            if (str.charCodeAt(strIdx) === 35 /* NUM */) {
+                // Skip the leading "&#". For hex entities, also skip the leading "x".
+                var start = strIdx + 1;
+                var base = 10;
+                var cp = str.charCodeAt(start);
+                if ((cp | 32 /* To_LOWER_BIT */) === 120 /* LOWER_X */) {
+                    base = 16;
+                    strIdx += 1;
+                    start += 1;
+                }
+                do
+                    cp = str.charCodeAt(++strIdx);
+                while ((cp >= 48 /* ZERO */ && cp <= 57 /* NINE */) ||
+                    (base === 16 &&
+                        (cp | 32 /* To_LOWER_BIT */) >= 97 /* LOWER_A */ &&
+                        (cp | 32 /* To_LOWER_BIT */) <= 102 /* LOWER_F */));
+                if (start !== strIdx) {
+                    var entity = str.substring(start, strIdx);
+                    var parsed = parseInt(entity, base);
+                    if (str.charCodeAt(strIdx) === 59 /* SEMI */) {
+                        strIdx += 1;
+                    }
+                    else if (strict) {
+                        continue;
+                    }
+                    ret += (0, decode_codepoint_js_1.default)(parsed);
+                    lastIdx = strIdx;
+                }
+                continue;
+            }
+            var resultIdx = 0;
+            var excess = 1;
+            var treeIdx = 0;
+            var current = decodeTree[treeIdx];
+            for (; strIdx < str.length; strIdx++, excess++) {
+                treeIdx = determineBranch(decodeTree, current, treeIdx + 1, str.charCodeAt(strIdx));
+                if (treeIdx < 0)
+                    break;
+                current = decodeTree[treeIdx];
+                var masked = current & BinTrieFlags.VALUE_LENGTH;
+                // If the branch is a value, store it and continue
+                if (masked) {
+                    // If we have a legacy entity while parsing strictly, just skip the number of bytes
+                    if (!strict || str.charCodeAt(strIdx) === 59 /* SEMI */) {
+                        resultIdx = treeIdx;
+                        excess = 0;
+                    }
+                    // The mask is the number of bytes of the value, including the current byte.
+                    var valueLength = (masked >> 14) - 1;
+                    if (valueLength === 0)
+                        break;
+                    treeIdx += valueLength;
+                }
+            }
+            if (resultIdx !== 0) {
+                var valueLength = (decodeTree[resultIdx] & BinTrieFlags.VALUE_LENGTH) >> 14;
+                ret +=
+                    valueLength === 1
+                        ? String.fromCharCode(decodeTree[resultIdx] & ~BinTrieFlags.VALUE_LENGTH)
+                        : valueLength === 2
+                            ? String.fromCharCode(decodeTree[resultIdx + 1])
+                            : String.fromCharCode(decodeTree[resultIdx + 1], decodeTree[resultIdx + 2]);
+                lastIdx = strIdx - excess + 1;
+            }
+        }
+        return ret + str.slice(lastIdx);
+    };
 }
-var sorter = function (a, b) { return (a < b ? 1 : -1); };
-exports.decodeHTML = (function () {
-    var legacy = Object.keys(legacy_json_1.default).sort(sorter);
-    var keys = Object.keys(entities_json_1.default).sort(sorter);
-    for (var i = 0, j = 0; i < keys.length; i++) {
-        if (legacy[j] === keys[i]) {
-            keys[i] += ";?";
-            j++;
+function determineBranch(decodeTree, current, nodeIdx, char) {
+    var branchCount = (current & BinTrieFlags.BRANCH_LENGTH) >> 7;
+    var jumpOffset = current & BinTrieFlags.JUMP_TABLE;
+    // Case 1: Single branch encoded in jump offset
+    if (branchCount === 0) {
+        return jumpOffset !== 0 && char === jumpOffset ? nodeIdx : -1;
+    }
+    // Case 2: Multiple branches encoded in jump table
+    if (jumpOffset) {
+        var value = char - jumpOffset;
+        return value < 0 || value > branchCount
+            ? -1
+            : decodeTree[nodeIdx + value] - 1;
+    }
+    // Case 3: Multiple branches encoded in dictionary
+    // Binary search for the character.
+    var lo = nodeIdx;
+    var hi = lo + branchCount - 1;
+    while (lo <= hi) {
+        var mid = (lo + hi) >>> 1;
+        var midVal = decodeTree[mid];
+        if (midVal < char) {
+            lo = mid + 1;
+        }
+        else if (midVal > char) {
+            hi = mid - 1;
         }
         else {
-            keys[i] += ";";
+            return decodeTree[mid + branchCount];
         }
     }
-    var re = new RegExp("&(?:" + keys.join("|") + "|#[xX][\\da-fA-F]+;?|#\\d+;?)", "g");
-    var replace = getReplacer(entities_json_1.default);
-    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 getReplacer(map) {
-    return function replace(str) {
-        if (str.charAt(1) === "#") {
-            var secondChar = str.charAt(2);
-            if (secondChar === "X" || secondChar === "x") {
-                return decode_codepoint_1.default(parseInt(str.substr(3), 16));
-            }
-            return decode_codepoint_1.default(parseInt(str.substr(2), 10));
-        }
-        // eslint-disable-next-line @typescript-eslint/prefer-nullish-coalescing
-        return map[str.slice(1, -1)] || str;
-    };
+    return -1;
+}
+exports.determineBranch = determineBranch;
+var htmlDecoder = getDecoder(decode_data_html_js_1.default);
+var xmlDecoder = getDecoder(decode_data_xml_js_1.default);
+function decodeHTML(str) {
+    return htmlDecoder(str, false);
+}
+exports.decodeHTML = decodeHTML;
+function decodeHTMLStrict(str) {
+    return htmlDecoder(str, true);
+}
+exports.decodeHTMLStrict = decodeHTMLStrict;
+function decodeXML(str) {
+    return xmlDecoder(str, true);
 }
+exports.decodeXML = decodeXML;
+//# sourceMappingURL=decode.js.map

package/lib/decode.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode.js","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":";;;;;;AAAA,wFAA6D;AAKpD,yBALF,6BAAc,CAKE;AAJvB,sFAA2D;AAIlC,wBAJlB,4BAAa,CAIkB;AAHtC,8EAAoD;AAGZ,0BAHjC,6BAAe,CAGiC;AAcvD,IAAY,YAIX;AAJD,WAAY,YAAY;IACpB,mEAAoC,CAAA;IACpC,qEAAqC,CAAA;IACrC,6DAAkC,CAAA;AACtC,CAAC,EAJW,YAAY,GAAZ,oBAAY,KAAZ,oBAAY,QAIvB;AAED,SAAS,UAAU,CAAC,UAAuB;IACvC,OAAO,SAAS,gBAAgB,CAAC,GAAW,EAAE,MAAe;QACzD,IAAI,GAAG,GAAG,EAAE,CAAC;QACb,IAAI,OAAO,GAAG,CAAC,CAAC;QAChB,IAAI,MAAM,GAAG,CAAC,CAAC;QAEf,OAAO,CAAC,MAAM,GAAG,GAAG,CAAC,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,EAAE;YAC7C,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;YAClC,OAAO,GAAG,MAAM,CAAC;YACjB,eAAe;YACf,MAAM,IAAI,CAAC,CAAC;YAEZ,uDAAuD;YACvD,IAAI,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,iBAAkB,EAAE;gBAC1C,sEAAsE;gBACtE,IAAI,KAAK,GAAG,MAAM,GAAG,CAAC,CAAC;gBACvB,IAAI,IAAI,GAAG,EAAE,CAAC;gBAEd,IAAI,EAAE,GAAG,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;gBAC/B,IAAI,CAAC,EAAE,wBAAyB,CAAC,sBAAsB,EAAE;oBACrD,IAAI,GAAG,EAAE,CAAC;oBACV,MAAM,IAAI,CAAC,CAAC;oBACZ,KAAK,IAAI,CAAC,CAAC;iBACd;gBAED;oBAAG,EAAE,GAAG,GAAG,CAAC,UAAU,CAAC,EAAE,MAAM,CAAC,CAAC;uBAE7B,CAAC,EAAE,iBAAkB,IAAI,EAAE,iBAAkB,CAAC;oBAC9C,CAAC,IAAI,KAAK,EAAE;wBACR,CAAC,EAAE,wBAAyB,CAAC,oBAAqB;wBAClD,CAAC,EAAE,wBAAyB,CAAC,qBAAqB,CAAC,EACzD;gBAEF,IAAI,KAAK,KAAK,MAAM,EAAE;oBAClB,IAAM,MAAM,GAAG,GAAG,CAAC,SAAS,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;oBAC5C,IAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;oBAEtC,IAAI,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,kBAAmB,EAAE;wBAC3C,MAAM,IAAI,CAAC,CAAC;qBACf;yBAAM,IAAI,MAAM,EAAE;wBACf,SAAS;qBACZ;oBAED,GAAG,IAAI,IAAA,6BAAe,EAAC,MAAM,CAAC,CAAC;oBAC/B,OAAO,GAAG,MAAM,CAAC;iBACpB;gBAED,SAAS;aACZ;YAED,IAAI,SAAS,GAAG,CAAC,CAAC;YAClB,IAAI,MAAM,GAAG,CAAC,CAAC;YACf,IAAI,OAAO,GAAG,CAAC,CAAC;YAChB,IAAI,OAAO,GAAG,UAAU,CAAC,OAAO,CAAC,CAAC;YAElC,OAAO,MAAM,GAAG,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,MAAM,EAAE,EAAE;gBAC5C,OAAO,GAAG,eAAe,CACrB,UAAU,EACV,OAAO,EACP,OAAO,GAAG,CAAC,EACX,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,CACzB,CAAC;gBAEF,IAAI,OAAO,GAAG,CAAC;oBAAE,MAAM;gBAEvB,OAAO,GAAG,UAAU,CAAC,OAAO,CAAC,CAAC;gBAE9B,IAAM,MAAM,GAAG,OAAO,GAAG,YAAY,CAAC,YAAY,CAAC;gBAEnD,kDAAkD;gBAClD,IAAI,MAAM,EAAE;oBACR,mFAAmF;oBACnF,IAAI,CAAC,MAAM,IAAI,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,kBAAmB,EAAE;wBACtD,SAAS,GAAG,OAAO,CAAC;wBACpB,MAAM,GAAG,CAAC,CAAC;qBACd;oBAED,4EAA4E;oBAC5E,IAAM,WAAW,GAAG,CAAC,MAAM,IAAI,EAAE,CAAC,GAAG,CAAC,CAAC;oBAEvC,IAAI,WAAW,KAAK,CAAC;wBAAE,MAAM;oBAE7B,OAAO,IAAI,WAAW,CAAC;iBAC1B;aACJ;YAED,IAAI,SAAS,KAAK,CAAC,EAAE;gBACjB,IAAM,WAAW,GACb,CAAC,UAAU,CAAC,SAAS,CAAC,GAAG,YAAY,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;gBAC9D,GAAG;oBACC,WAAW,KAAK,CAAC;wBACb,CAAC,CAAC,MAAM,CAAC,YAAY,CACf,UAAU,CAAC,SAAS,CAAC,GAAG,CAAC,YAAY,CAAC,YAAY,CACrD;wBACH,CAAC,CAAC,WAAW,KAAK,CAAC;4BACnB,CAAC,CAAC,MAAM,CAAC,YAAY,CAAC,UAAU,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC;4BAChD,CAAC,CAAC,MAAM,CAAC,YAAY,CACf,UAAU,CAAC,SAAS,GAAG,CAAC,CAAC,EACzB,UAAU,CAAC,SAAS,GAAG,CAAC,CAAC,CAC5B,CAAC;gBACZ,OAAO,GAAG,MAAM,GAAG,MAAM,GAAG,CAAC,CAAC;aACjC;SACJ;QAED,OAAO,GAAG,GAAG,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IACpC,CAAC,CAAC;AACN,CAAC;AAED,SAAgB,eAAe,CAC3B,UAAuB,EACvB,OAAe,EACf,OAAe,EACf,IAAY;IAEZ,IAAM,WAAW,GAAG,CAAC,OAAO,GAAG,YAAY,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;IAChE,IAAM,UAAU,GAAG,OAAO,GAAG,YAAY,CAAC,UAAU,CAAC;IAErD,+CAA+C;IAC/C,IAAI,WAAW,KAAK,CAAC,EAAE;QACnB,OAAO,UAAU,KAAK,CAAC,IAAI,IAAI,KAAK,UAAU,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KACjE;IAED,kDAAkD;IAClD,IAAI,UAAU,EAAE;QACZ,IAAM,KAAK,GAAG,IAAI,GAAG,UAAU,CAAC;QAEhC,OAAO,KAAK,GAAG,CAAC,IAAI,KAAK,GAAG,WAAW;YACnC,CAAC,CAAC,CAAC,CAAC;YACJ,CAAC,CAAC,UAAU,CAAC,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC;KACzC;IAED,kDAAkD;IAElD,mCAAmC;IACnC,IAAI,EAAE,GAAG,OAAO,CAAC;IACjB,IAAI,EAAE,GAAG,EAAE,GAAG,WAAW,GAAG,CAAC,CAAC;IAE9B,OAAO,EAAE,IAAI,EAAE,EAAE;QACb,IAAM,GAAG,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;QAC5B,IAAM,MAAM,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC;QAE/B,IAAI,MAAM,GAAG,IAAI,EAAE;YACf,EAAE,GAAG,GAAG,GAAG,CAAC,CAAC;SAChB;aAAM,IAAI,MAAM,GAAG,IAAI,EAAE;YACtB,EAAE,GAAG,GAAG,GAAG,CAAC,CAAC;SAChB;aAAM;YACH,OAAO,UAAU,CAAC,GAAG,GAAG,WAAW,CAAC,CAAC;SACxC;KACJ;IAED,OAAO,CAAC,CAAC,CAAC;AACd,CAAC;AA3CD,0CA2CC;AAED,IAAM,WAAW,GAAG,UAAU,CAAC,6BAAc,CAAC,CAAC;AAC/C,IAAM,UAAU,GAAG,UAAU,CAAC,4BAAa,CAAC,CAAC;AAE7C,SAAgB,UAAU,CAAC,GAAW;IAClC,OAAO,WAAW,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;AACnC,CAAC;AAFD,gCAEC;AAED,SAAgB,gBAAgB,CAAC,GAAW;IACxC,OAAO,WAAW,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;AAClC,CAAC;AAFD,4CAEC;AAED,SAAgB,SAAS,CAAC,GAAW;IACjC,OAAO,UAAU,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;AACjC,CAAC;AAFD,8BAEC"}

package/lib/decode_codepoint.d.ts

@@ -1,2 +1,3 @@
+export declare function replaceCodePoint(codePoint: number): number;
 export default function decodeCodePoint(codePoint: number): string;
 //# sourceMappingURL=decode_codepoint.d.ts.map

package/lib/decode_codepoint.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"decode_codepoint.d.ts","sourceRoot":"","sources":["../src/decode_codepoint.ts"],"names":[],"mappings":"AAsBA,MAAM,CAAC,OAAO,UAAU,eAAe,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAUjE"}
+{"version":3,"file":"decode_codepoint.d.ts","sourceRoot":"","sources":["../src/decode_codepoint.ts"],"names":[],"mappings":"AAmDA,wBAAgB,gBAAgB,CAAC,SAAS,EAAE,MAAM,UAMjD;AAED,MAAM,CAAC,OAAO,UAAU,eAAe,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAEjE"}

package/lib/decode_codepoint.js

@@ -1,12 +1,39 @@
 "use strict";
-var __importDefault = (this && this.__importDefault) || function (mod) {
-    return (mod && mod.__esModule) ? mod : { "default": mod };
-};
+// Adapted from https://github.com/mathiasbynens/he/blob/36afe179392226cf1b6ccdb16ebbb7a5a844d93a/src/he.js#L106-L134
 Object.defineProperty(exports, "__esModule", { value: true });
-var decode_json_1 = __importDefault(require("./maps/decode.json"));
-// Adapted from https://github.com/mathiasbynens/he/blob/master/src/he.js#L94-L119
+exports.replaceCodePoint = void 0;
+var decodeMap = new Map([
+    [0, 65533],
+    [128, 8364],
+    [130, 8218],
+    [131, 402],
+    [132, 8222],
+    [133, 8230],
+    [134, 8224],
+    [135, 8225],
+    [136, 710],
+    [137, 8240],
+    [138, 352],
+    [139, 8249],
+    [140, 338],
+    [142, 381],
+    [145, 8216],
+    [146, 8217],
+    [147, 8220],
+    [148, 8221],
+    [149, 8226],
+    [150, 8211],
+    [151, 8212],
+    [152, 732],
+    [153, 8482],
+    [154, 353],
+    [155, 8250],
+    [156, 339],
+    [158, 382],
+    [159, 376],
+]);
 var fromCodePoint = 
-// eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
+// eslint-disable-next-line @typescript-eslint/no-unnecessary-condition, node/no-unsupported-features/es-builtins
 String.fromCodePoint ||
     function (codePoint) {
         var output = "";
@@ -18,13 +45,16 @@ String.fromCodePoint ||
         output += String.fromCharCode(codePoint);
         return output;
     };
-function decodeCodePoint(codePoint) {
+function replaceCodePoint(codePoint) {
+    var _a;
     if ((codePoint >= 0xd800 && codePoint <= 0xdfff) || codePoint > 0x10ffff) {
-        return "\uFFFD";
-    }
-    if (codePoint in decode_json_1.default) {
-        codePoint = decode_json_1.default[codePoint];
+        return 0xfffd;
     }
-    return fromCodePoint(codePoint);
+    return (_a = decodeMap.get(codePoint)) !== null && _a !== void 0 ? _a : codePoint;
+}
+exports.replaceCodePoint = replaceCodePoint;
+function decodeCodePoint(codePoint) {
+    return fromCodePoint(replaceCodePoint(codePoint));
 }
 exports.default = decodeCodePoint;
+//# sourceMappingURL=decode_codepoint.js.map

package/lib/decode_codepoint.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode_codepoint.js","sourceRoot":"","sources":["../src/decode_codepoint.ts"],"names":[],"mappings":";AAAA,qHAAqH;;;AAErH,IAAM,SAAS,GAAG,IAAI,GAAG,CAAC;IACtB,CAAC,CAAC,EAAE,KAAK,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,IAAI,CAAC;IACX,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;IACV,CAAC,GAAG,EAAE,GAAG,CAAC;CACb,CAAC,CAAC;AAEH,IAAM,aAAa;AACf,iHAAiH;AACjH,MAAM,CAAC,aAAa;IACpB,UAAU,SAAiB;QACvB,IAAI,MAAM,GAAG,EAAE,CAAC;QAEhB,IAAI,SAAS,GAAG,MAAM,EAAE;YACpB,SAAS,IAAI,OAAO,CAAC;YACrB,MAAM,IAAI,MAAM,CAAC,YAAY,CACzB,CAAC,CAAC,SAAS,KAAK,EAAE,CAAC,GAAG,KAAK,CAAC,GAAG,MAAM,CACxC,CAAC;YACF,SAAS,GAAG,MAAM,GAAG,CAAC,SAAS,GAAG,KAAK,CAAC,CAAC;SAC5C;QAED,MAAM,IAAI,MAAM,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC;QACzC,OAAO,MAAM,CAAC;IAClB,CAAC,CAAC;AAEN,SAAgB,gBAAgB,CAAC,SAAiB;;IAC9C,IAAI,CAAC,SAAS,IAAI,MAAM,IAAI,SAAS,IAAI,MAAM,CAAC,IAAI,SAAS,GAAG,QAAQ,EAAE;QACtE,OAAO,MAAM,CAAC;KACjB;IAED,OAAO,MAAA,SAAS,CAAC,GAAG,CAAC,SAAS,CAAC,mCAAI,SAAS,CAAC;AACjD,CAAC;AAND,4CAMC;AAED,SAAwB,eAAe,CAAC,SAAiB;IACrD,OAAO,aAAa,CAAC,gBAAgB,CAAC,SAAS,CAAC,CAAC,CAAC;AACtD,CAAC;AAFD,kCAEC"}

package/lib/encode-trie.d.ts

@@ -0,0 +1,8 @@
+export declare const getCodePoint: (str: string, index: number) => number;
+export declare function encodeHTMLTrieRe(regExp: RegExp, str: string): string;
+export interface TrieNode {
+    value?: string;
+    next?: Map<number, TrieNode>;
+}
+export declare function getTrie(map: Record<string, string>): Map<number, TrieNode>;
+//# sourceMappingURL=encode-trie.d.ts.map

package/lib/encode-trie.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode-trie.d.ts","sourceRoot":"","sources":["../src/encode-trie.ts"],"names":[],"mappings":"AAYA,eAAO,MAAM,YAAY,QAGT,MAAM,SAAS,MAAM,KAAG,MAQD,CAAC;AAIxC,wBAAgB,gBAAgB,CAAC,MAAM,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,MAAM,CAkCpE;AAED,MAAM,WAAW,QAAQ;IACrB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,IAAI,CAAC,EAAE,GAAG,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAC;CAChC;AAED,wBAAgB,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,QAAQ,CAAC,CAmB1E"}

package/lib/encode-trie.js

@@ -0,0 +1,78 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.getTrie = exports.encodeHTMLTrieRe = exports.getCodePoint = void 0;
+var entities_encode_json_1 = __importDefault(require("./maps/entities-encode.json"));
+function isHighSurrugate(c) {
+    return (c & 64512 /* Mask */) === 55296 /* High */;
+}
+// For compatibility with node < 4, we wrap `codePointAt`
+exports.getCodePoint = 
+// eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
+String.prototype.codePointAt != null
+    ? function (str, index) { return str.codePointAt(index); }
+    : // http://mathiasbynens.be/notes/javascript-encoding#surrogate-formulae
+        function (c, index) {
+            return isHighSurrugate(c.charCodeAt(index))
+                ? (c.charCodeAt(index) - 55296 /* High */) * 0x400 +
+                    c.charCodeAt(index + 1) -
+                    0xdc00 +
+                    0x10000
+                : c.charCodeAt(index);
+        };
+var htmlTrie = getTrie(entities_encode_json_1.default);
+function encodeHTMLTrieRe(regExp, str) {
+    var _a;
+    var ret = "";
+    var lastIdx = 0;
+    var match;
+    while ((match = regExp.exec(str)) !== null) {
+        var i = match.index;
+        var char = str.charCodeAt(i);
+        var next = htmlTrie.get(char);
+        if (next) {
+            if (next.next != null && i + 1 < str.length) {
+                var value = (_a = next.next.get(str.charCodeAt(i + 1))) === null || _a === void 0 ? void 0 : _a.value;
+                if (value != null) {
+                    ret += str.substring(lastIdx, i) + value;
+                    regExp.lastIndex += 1;
+                    lastIdx = i + 2;
+                    continue;
+                }
+            }
+            ret += str.substring(lastIdx, i) + next.value;
+            lastIdx = i + 1;
+        }
+        else {
+            ret += "".concat(str.substring(lastIdx, i), "&#x").concat((0, exports.getCodePoint)(str, i).toString(16), ";");
+            // Increase by 1 if we have a surrogate pair
+            lastIdx = regExp.lastIndex += Number(isHighSurrugate(char));
+        }
+    }
+    return ret + str.substr(lastIdx);
+}
+exports.encodeHTMLTrieRe = encodeHTMLTrieRe;
+function getTrie(map) {
+    var _a, _b, _c, _d;
+    var trie = new Map();
+    for (var _i = 0, _e = Object.keys(map); _i < _e.length; _i++) {
+        var decoded = _e[_i];
+        var entity = map[decoded];
+        // Resolve the key
+        var lastMap = trie;
+        for (var i = 0; i < decoded.length - 1; i++) {
+            var char = decoded.charCodeAt(i);
+            var next = (_a = lastMap.get(char)) !== null && _a !== void 0 ? _a : {};
+            lastMap.set(char, next);
+            lastMap = (_b = next.next) !== null && _b !== void 0 ? _b : (next.next = new Map());
+        }
+        var val = (_c = lastMap.get(decoded.charCodeAt(decoded.length - 1))) !== null && _c !== void 0 ? _c : {};
+        (_d = val.value) !== null && _d !== void 0 ? _d : (val.value = "&".concat(entity, ";"));
+        lastMap.set(decoded.charCodeAt(decoded.length - 1), val);
+    }
+    return trie;
+}
+exports.getTrie = getTrie;
+//# sourceMappingURL=encode-trie.js.map

package/lib/encode-trie.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode-trie.js","sourceRoot":"","sources":["../src/encode-trie.ts"],"names":[],"mappings":";;;;;;AAAA,qFAAkD;AAOlD,SAAS,eAAe,CAAC,CAAS;IAC9B,OAAO,CAAC,CAAC,mBAAiB,CAAC,qBAAmB,CAAC;AACnD,CAAC;AAED,yDAAyD;AAC5C,QAAA,YAAY;AACrB,uEAAuE;AACvE,MAAM,CAAC,SAAS,CAAC,WAAW,IAAI,IAAI;IAChC,CAAC,CAAC,UAAC,GAAW,EAAE,KAAa,IAAa,OAAA,GAAG,CAAC,WAAW,CAAC,KAAK,CAAE,EAAvB,CAAuB;IACjE,CAAC,CAAC,uEAAuE;QACvE,UAAC,CAAS,EAAE,KAAa;YACrB,OAAA,eAAe,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;gBAChC,CAAC,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,mBAAiB,CAAC,GAAG,KAAK;oBAC9C,CAAC,CAAC,UAAU,CAAC,KAAK,GAAG,CAAC,CAAC;oBACvB,MAAM;oBACN,OAAO;gBACT,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC;QALzB,CAKyB,CAAC;AAExC,IAAM,QAAQ,GAAG,OAAO,CAAC,8BAAO,CAAC,CAAC;AAElC,SAAgB,gBAAgB,CAAC,MAAc,EAAE,GAAW;;IACxD,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,KAAK,CAAC;IAEV,OAAO,CAAC,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE;QACxC,IAAM,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC;QACtB,IAAM,IAAI,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;QAC/B,IAAM,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAEhC,IAAI,IAAI,EAAE;YACN,IAAI,IAAI,CAAC,IAAI,IAAI,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE;gBACzC,IAAM,KAAK,GAAG,MAAA,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,0CAAE,KAAK,CAAC;gBAC1D,IAAI,KAAK,IAAI,IAAI,EAAE;oBACf,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,GAAG,KAAK,CAAC;oBACzC,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC;oBACtB,OAAO,GAAG,CAAC,GAAG,CAAC,CAAC;oBAChB,SAAS;iBACZ;aACJ;YAED,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC;YAC9C,OAAO,GAAG,CAAC,GAAG,CAAC,CAAC;SACnB;aAAM;YACH,GAAG,IAAI,UAAG,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,gBAAM,IAAA,oBAAY,EACjD,GAAG,EACH,CAAC,CACJ,CAAC,QAAQ,CAAC,EAAE,CAAC,MAAG,CAAC;YAClB,4CAA4C;YAC5C,OAAO,GAAG,MAAM,CAAC,SAAS,IAAI,MAAM,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC;SAC/D;KACJ;IAED,OAAO,GAAG,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;AACrC,CAAC;AAlCD,4CAkCC;AAOD,SAAgB,OAAO,CAAC,GAA2B;;IAC/C,IAAM,IAAI,GAAG,IAAI,GAAG,EAAoB,CAAC;IAEzC,KAAsB,UAAgB,EAAhB,KAAA,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,EAAhB,cAAgB,EAAhB,IAAgB,EAAE;QAAnC,IAAM,OAAO,SAAA;QACd,IAAM,MAAM,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;QAC5B,kBAAkB;QAClB,IAAI,OAAO,GAAG,IAAI,CAAC;QACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;YACzC,IAAM,IAAI,GAAG,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;YACnC,IAAM,IAAI,GAAG,MAAA,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,mCAAI,EAAE,CAAC;YACrC,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;YACxB,OAAO,SAAG,IAAI,CAAC,IAAI,oCAAT,IAAI,CAAC,IAAI,GAAK,IAAI,GAAG,EAAE,CAAA,CAAC;SACrC;QACD,IAAM,GAAG,GAAG,MAAA,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,mCAAI,EAAE,CAAC;QACtE,MAAA,GAAG,CAAC,KAAK,oCAAT,GAAG,CAAC,KAAK,GAAK,WAAI,MAAM,MAAG,EAAC;QAC5B,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;KAC5D;IAED,OAAO,IAAI,CAAC;AAChB,CAAC;AAnBD,0BAmBC"}

package/lib/encode.d.ts

@@ -5,7 +5,7 @@
  * If a character has no equivalent entity, a
  * numeric hexadecimal reference (eg. `ü`) will be used.
  */
-export declare const encodeXML: (data: string) => string;
+export declare function encodeXML(str: string): string;
 /**
  * Encodes all entities and non-ASCII characters in the input.
  *
@@ -16,7 +16,7 @@ export declare const encodeXML: (data: string) => string;
  * If a character has no equivalent entity, a
  * numeric hexadecimal reference (eg. `ü`) will be used.
  */
-export declare const encodeHTML: (data: string) => string;
+export declare function encodeHTML(data: string): string;
 /**
  * Encodes all non-ASCII characters, as well as characters not valid in HTML
  * documents using HTML entities.
@@ -24,7 +24,7 @@ export declare const encodeHTML: (data: string) => string;
  * If a character has no equivalent entity, a
  * numeric hexadecimal reference (eg. `ü`) will be used.
  */
-export declare const encodeNonAsciiHTML: (data: string) => string;
+export declare function encodeNonAsciiHTML(data: string): string;
 /**
  * Encodes all non-ASCII characters, as well as characters not valid in XML
  * documents using numeric hexadecimal reference (eg. `ü`).
@@ -34,14 +34,27 @@ export declare const encodeNonAsciiHTML: (data: string) => string;
  *
  * @param data String to escape.
  */
-export declare function escape(data: string): string;
+export declare const escape: typeof encodeXML;
 /**
- * Encodes all characters not valid in XML documents using numeric hexadecimal
- * reference (eg. `ü`).
+ * Encodes all characters not valid in XML documents using XML entities.
  *
  * Note that the output will be character-set dependent.
  *
  * @param data String to escape.
  */
-export declare function escapeUTF8(data: string): string;
+export declare const escapeUTF8: (data: string) => string;
+/**
+ * Encodes all characters that have to be escaped in HTML attributes,
+ * following {@link https://html.spec.whatwg.org/multipage/parsing.html#escapingString}.
+ *
+ * @param data String to escape.
+ */
+export declare const escapeAttribute: (data: string) => string;
+/**
+ * Encodes all characters that have to be escaped in HTML text,
+ * following {@link https://html.spec.whatwg.org/multipage/parsing.html#escapingString}.
+ *
+ * @param data String to escape.
+ */
+export declare const escapeText: (data: string) => string;
 //# sourceMappingURL=encode.d.ts.map

package/lib/encode.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"encode.d.ts","sourceRoot":"","sources":["../src/encode.ts"],"names":[],"mappings":"AAKA;;;;;;GAMG;AACH,eAAO,MAAM,SAAS,SAwIJ,MAAM,WAxI4B,CAAC;AAOrD;;;;;;;;;GASG;AACH,eAAO,MAAM,UAAU,SAkFL,MAAM,WAlFuC,CAAC;AAChE;;;;;;GAMG;AACH,eAAO,MAAM,kBAAkB,SA+Gb,MAAM,WA/GsC,CAAC;AAqF/D;;;;;;;;GAQG;AACH,wBAAgB,MAAM,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAE3C;AAED;;;;;;;GAOG;AACH,wBAAgB,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAE/C"}
+{"version":3,"file":"encode.d.ts","sourceRoot":"","sources":["../src/encode.ts"],"names":[],"mappings":"AAaA;;;;;;GAMG;AACH,wBAAgB,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CA0B7C;AAED;;;;;;;;;GASG;AACH,wBAAgB,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAE/C;AACD;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAEvD;AAED;;;;;;;;GAQG;AACH,eAAO,MAAM,MAAM,kBAAY,CAAC;AA2BhC;;;;;;GAMG;AACH,eAAO,MAAM,UAAU,SA7Bb,MAAM,KAAK,MA6BuC,CAAC;AAE7D;;;;;GAKG;AACH,eAAO,MAAM,eAAe,SArClB,MAAM,KAAK,MA4CpB,CAAC;AAEF;;;;;GAKG;AACH,eAAO,MAAM,UAAU,SApDb,MAAM,KAAK,MA4DpB,CAAC"}