entities 3.0.0 → 4.1.0

package/lib/esm/decode.js

@@ -0,0 +1,135 @@
+import htmlDecodeTree from "./generated/decode-data-html.js";
+import xmlDecodeTree from "./generated/decode-data-xml.js";
+import decodeCodePoint from "./decode_codepoint.js";
+// Re-export for use by eg. htmlparser2
+export { htmlDecodeTree, xmlDecodeTree, decodeCodePoint };
+export 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 || (BinTrieFlags = {}));
+function getDecoder(decodeTree) {
+    return function decodeHTMLBinary(str, strict) {
+        let ret = "";
+        let lastIdx = 0;
+        let 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".
+                let start = strIdx + 1;
+                let base = 10;
+                let 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) {
+                    const entity = str.substring(start, strIdx);
+                    const parsed = parseInt(entity, base);
+                    if (str.charCodeAt(strIdx) === 59 /* SEMI */) {
+                        strIdx += 1;
+                    }
+                    else if (strict) {
+                        continue;
+                    }
+                    ret += decodeCodePoint(parsed);
+                    lastIdx = strIdx;
+                }
+                continue;
+            }
+            let resultIdx = 0;
+            let excess = 1;
+            let treeIdx = 0;
+            let 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];
+                const 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.
+                    const valueLength = (masked >> 14) - 1;
+                    if (valueLength === 0)
+                        break;
+                    treeIdx += valueLength;
+                }
+            }
+            if (resultIdx !== 0) {
+                const 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);
+    };
+}
+export function determineBranch(decodeTree, current, nodeIdx, char) {
+    const branchCount = (current & BinTrieFlags.BRANCH_LENGTH) >> 7;
+    const 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) {
+        const 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.
+    let lo = nodeIdx;
+    let hi = lo + branchCount - 1;
+    while (lo <= hi) {
+        const mid = (lo + hi) >>> 1;
+        const midVal = decodeTree[mid];
+        if (midVal < char) {
+            lo = mid + 1;
+        }
+        else if (midVal > char) {
+            hi = mid - 1;
+        }
+        else {
+            return decodeTree[mid + branchCount];
+        }
+    }
+    return -1;
+}
+const htmlDecoder = getDecoder(htmlDecodeTree);
+const xmlDecoder = getDecoder(xmlDecodeTree);
+export function decodeHTML(str) {
+    return htmlDecoder(str, false);
+}
+export function decodeHTMLStrict(str) {
+    return htmlDecoder(str, true);
+}
+export function decodeXML(str) {
+    return xmlDecoder(str, true);
+}
+//# sourceMappingURL=decode.js.map

package/lib/esm/decode.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode.js","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;AAEpD,uCAAuC;AACvC,OAAO,EAAE,cAAc,EAAE,aAAa,EAAE,eAAe,EAAE,CAAC;AAc1D,MAAM,CAAN,IAAY,YAIX;AAJD,WAAY,YAAY;IACpB,mEAAoC,CAAA;IACpC,qEAAqC,CAAA;IACrC,6DAAkC,CAAA;AACtC,CAAC,EAJW,YAAY,KAAZ,YAAY,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,MAAM,MAAM,GAAG,GAAG,CAAC,SAAS,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;oBAC5C,MAAM,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,eAAe,CAAC,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,MAAM,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,MAAM,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,MAAM,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,MAAM,UAAU,eAAe,CAC3B,UAAuB,EACvB,OAAe,EACf,OAAe,EACf,IAAY;IAEZ,MAAM,WAAW,GAAG,CAAC,OAAO,GAAG,YAAY,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;IAChE,MAAM,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,MAAM,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,MAAM,GAAG,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;QAC5B,MAAM,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;AAED,MAAM,WAAW,GAAG,UAAU,CAAC,cAAc,CAAC,CAAC;AAC/C,MAAM,UAAU,GAAG,UAAU,CAAC,aAAa,CAAC,CAAC;AAE7C,MAAM,UAAU,UAAU,CAAC,GAAW;IAClC,OAAO,WAAW,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;AACnC,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,GAAW;IACxC,OAAO,WAAW,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;AAClC,CAAC;AAED,MAAM,UAAU,SAAS,CAAC,GAAW;IACjC,OAAO,UAAU,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;AACjC,CAAC"}

package/lib/esm/decode_codepoint.d.ts

@@ -0,0 +1,3 @@
+export declare function replaceCodePoint(codePoint: number): number;
+export default function decodeCodePoint(codePoint: number): string;
+//# sourceMappingURL=decode_codepoint.d.ts.map

package/lib/esm/decode_codepoint.d.ts.map

@@ -0,0 +1 @@
+{"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/esm/decode_codepoint.js

@@ -0,0 +1,55 @@
+// Adapted from https://github.com/mathiasbynens/he/blob/36afe179392226cf1b6ccdb16ebbb7a5a844d93a/src/he.js#L106-L134
+const 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],
+]);
+const fromCodePoint = 
+// eslint-disable-next-line @typescript-eslint/no-unnecessary-condition, node/no-unsupported-features/es-builtins
+String.fromCodePoint ||
+    function (codePoint) {
+        let output = "";
+        if (codePoint > 0xffff) {
+            codePoint -= 0x10000;
+            output += String.fromCharCode(((codePoint >>> 10) & 0x3ff) | 0xd800);
+            codePoint = 0xdc00 | (codePoint & 0x3ff);
+        }
+        output += String.fromCharCode(codePoint);
+        return output;
+    };
+export function replaceCodePoint(codePoint) {
+    var _a;
+    if ((codePoint >= 0xd800 && codePoint <= 0xdfff) || codePoint > 0x10ffff) {
+        return 0xfffd;
+    }
+    return (_a = decodeMap.get(codePoint)) !== null && _a !== void 0 ? _a : codePoint;
+}
+export default function decodeCodePoint(codePoint) {
+    return fromCodePoint(replaceCodePoint(codePoint));
+}
+//# sourceMappingURL=decode_codepoint.js.map

package/lib/esm/decode_codepoint.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode_codepoint.js","sourceRoot":"","sources":["../../src/decode_codepoint.ts"],"names":[],"mappings":"AAAA,qHAAqH;AAErH,MAAM,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,MAAM,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,MAAM,UAAU,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;AAED,MAAM,CAAC,OAAO,UAAU,eAAe,CAAC,SAAiB;IACrD,OAAO,aAAa,CAAC,gBAAgB,CAAC,SAAS,CAAC,CAAC,CAAC;AACtD,CAAC"}

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

@@ -0,0 +1,3 @@
+export declare const getCodePoint: (str: string, index: number) => number;
+export declare function encodeHTMLTrieRe(regExp: RegExp, str: string): string;
+//# sourceMappingURL=encode-trie.d.ts.map

package/lib/esm/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;AAExC,wBAAgB,gBAAgB,CAAC,MAAM,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,MAAM,CA6CpE"}

package/lib/esm/encode-trie.js

@@ -0,0 +1,55 @@
+import htmlTrie from "./generated/encode-html";
+function isHighSurrugate(c) {
+    return (c & 64512 /* Mask */) === 55296 /* High */;
+}
+// For compatibility with node < 4, we wrap `codePointAt`
+export const getCodePoint = 
+// eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
+String.prototype.codePointAt != null
+    ? (str, index) => str.codePointAt(index)
+    : // http://mathiasbynens.be/notes/javascript-encoding#surrogate-formulae
+        (c, index) => isHighSurrugate(c.charCodeAt(index))
+            ? (c.charCodeAt(index) - 55296 /* High */) * 0x400 +
+                c.charCodeAt(index + 1) -
+                0xdc00 +
+                0x10000
+            : c.charCodeAt(index);
+export function encodeHTMLTrieRe(regExp, str) {
+    var _a;
+    let ret = "";
+    let lastIdx = 0;
+    let match;
+    while ((match = regExp.exec(str)) !== null) {
+        const i = match.index;
+        const char = str.charCodeAt(i);
+        let next = htmlTrie.get(char);
+        if (next != null) {
+            if (typeof next !== "string") {
+                // We are in a branch. Try to match the next char.
+                if (i + 1 < str.length) {
+                    const value = typeof next.n === "number"
+                        ? next.n === str.charCodeAt(i + 1)
+                            ? next.o
+                            : null
+                        : next.n.get(str.charCodeAt(i + 1));
+                    if (value) {
+                        ret += str.substring(lastIdx, i) + value;
+                        lastIdx = regExp.lastIndex += 1;
+                        continue;
+                    }
+                }
+                // If we have a character without a value, use a numeric entitiy.
+                next = (_a = next.v) !== null && _a !== void 0 ? _a : `&#x${char.toString(16)};`;
+            }
+            ret += str.substring(lastIdx, i) + next;
+            lastIdx = i + 1;
+        }
+        else {
+            ret += `${str.substring(lastIdx, i)}&#x${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);
+}
+//# sourceMappingURL=encode-trie.js.map

package/lib/esm/encode-trie.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode-trie.js","sourceRoot":"","sources":["../../src/encode-trie.ts"],"names":[],"mappings":"AAAA,OAAO,QAAQ,MAAM,yBAAyB,CAAC;AAO/C,SAAS,eAAe,CAAC,CAAS;IAC9B,OAAO,CAAC,CAAC,mBAAiB,CAAC,qBAAmB,CAAC;AACnD,CAAC;AAED,yDAAyD;AACzD,MAAM,CAAC,MAAM,YAAY;AACrB,uEAAuE;AACvE,MAAM,CAAC,SAAS,CAAC,WAAW,IAAI,IAAI;IAChC,CAAC,CAAC,CAAC,GAAW,EAAE,KAAa,EAAU,EAAE,CAAC,GAAG,CAAC,WAAW,CAAC,KAAK,CAAE;IACjE,CAAC,CAAC,uEAAuE;QACvE,CAAC,CAAS,EAAE,KAAa,EAAU,EAAE,CACjC,eAAe,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;YAChC,CAAC,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,mBAAiB,CAAC,GAAG,KAAK;gBAC9C,CAAC,CAAC,UAAU,CAAC,KAAK,GAAG,CAAC,CAAC;gBACvB,MAAM;gBACN,OAAO;YACT,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;AAExC,MAAM,UAAU,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,MAAM,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC;QACtB,MAAM,IAAI,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;QAC/B,IAAI,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAE9B,IAAI,IAAI,IAAI,IAAI,EAAE;YACd,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE;gBAC1B,kDAAkD;gBAClD,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE;oBACpB,MAAM,KAAK,GACP,OAAO,IAAI,CAAC,CAAC,KAAK,QAAQ;wBACtB,CAAC,CAAC,IAAI,CAAC,CAAC,KAAK,GAAG,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC;4BAC9B,CAAC,CAAC,IAAI,CAAC,CAAC;4BACR,CAAC,CAAC,IAAI;wBACV,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;oBAE5C,IAAI,KAAK,EAAE;wBACP,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,GAAG,KAAK,CAAC;wBACzC,OAAO,GAAG,MAAM,CAAC,SAAS,IAAI,CAAC,CAAC;wBAChC,SAAS;qBACZ;iBACJ;gBAED,iEAAiE;gBACjE,IAAI,GAAG,MAAA,IAAI,CAAC,CAAC,mCAAI,MAAM,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC,GAAG,CAAC;aAC/C;YAED,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC;YACxC,OAAO,GAAG,CAAC,GAAG,CAAC,CAAC;SACnB;aAAM;YACH,GAAG,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,MAAM,YAAY,CACjD,GAAG,EACH,CAAC,CACJ,CAAC,QAAQ,CAAC,EAAE,CAAC,GAAG,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"}

package/lib/esm/encode.d.ts

@@ -0,0 +1,60 @@
+/**
+ * Encodes all non-ASCII characters, as well as characters not valid in XML
+ * documents using XML entities.
+ *
+ * If a character has no equivalent entity, a
+ * numeric hexadecimal reference (eg. `ü`) will be used.
+ */
+export declare function encodeXML(str: string): string;
+/**
+ * Encodes all entities and non-ASCII characters in the input.
+ *
+ * This includes characters that are valid ASCII characters in HTML documents.
+ * For example `#` will be encoded as `#`. To get a more compact output,
+ * consider using the `encodeNonAsciiHTML` function.
+ *
+ * If a character has no equivalent entity, a
+ * numeric hexadecimal reference (eg. `ü`) will be used.
+ */
+export declare function encodeHTML(data: string): string;
+/**
+ * Encodes all non-ASCII characters, as well as characters not valid in HTML
+ * documents using HTML entities.
+ *
+ * If a character has no equivalent entity, a
+ * numeric hexadecimal reference (eg. `ü`) will be used.
+ */
+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`).
+ *
+ * Have a look at `escapeUTF8` if you want a more concise output at the expense
+ * of reduced transportability.
+ *
+ * @param data String to escape.
+ */
+export declare const escape: typeof encodeXML;
+/**
+ * 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 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/esm/encode.d.ts.map

@@ -0,0 +1 @@
+{"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"}

package/lib/esm/encode.js

@@ -0,0 +1,119 @@
+import { encodeHTMLTrieRe, getCodePoint } from "./encode-trie.js";
+const htmlReplacer = /[\t\n!-,./:-@[-`\f{-}$\x80-\uFFFF]/g;
+const xmlReplacer = /["&'<>$\x80-\uFFFF]/g;
+const xmlCodeMap = new Map([
+    [34, "&quot;"],
+    [38, "&amp;"],
+    [39, "&apos;"],
+    [60, "&lt;"],
+    [62, "&gt;"],
+]);
+/**
+ * Encodes all non-ASCII characters, as well as characters not valid in XML
+ * documents using XML entities.
+ *
+ * If a character has no equivalent entity, a
+ * numeric hexadecimal reference (eg. `&#xfc;`) will be used.
+ */
+export function encodeXML(str) {
+    let ret = "";
+    let lastIdx = 0;
+    let match;
+    while ((match = xmlReplacer.exec(str)) !== null) {
+        const i = match.index;
+        const char = str.charCodeAt(i);
+        const next = xmlCodeMap.get(char);
+        if (next) {
+            ret += str.substring(lastIdx, i) + next;
+            lastIdx = i + 1;
+        }
+        else {
+            ret += `${str.substring(lastIdx, i)}&#x${getCodePoint(str, i).toString(16)};`;
+            // Increase by 1 if we have a surrogate pair
+            lastIdx = xmlReplacer.lastIndex += Number((char & 65408) === 0xd800);
+        }
+    }
+    return ret + str.substr(lastIdx);
+}
+/**
+ * Encodes all entities and non-ASCII characters in the input.
+ *
+ * This includes characters that are valid ASCII characters in HTML documents.
+ * For example `#` will be encoded as `&num;`. To get a more compact output,
+ * consider using the `encodeNonAsciiHTML` function.
+ *
+ * If a character has no equivalent entity, a
+ * numeric hexadecimal reference (eg. `&#xfc;`) will be used.
+ */
+export function encodeHTML(data) {
+    return encodeHTMLTrieRe(htmlReplacer, data);
+}
+/**
+ * Encodes all non-ASCII characters, as well as characters not valid in HTML
+ * documents using HTML entities.
+ *
+ * If a character has no equivalent entity, a
+ * numeric hexadecimal reference (eg. `&#xfc;`) will be used.
+ */
+export function encodeNonAsciiHTML(data) {
+    return encodeHTMLTrieRe(xmlReplacer, data);
+}
+/**
+ * Encodes all non-ASCII characters, as well as characters not valid in XML
+ * documents using numeric hexadecimal reference (eg. `&#xfc;`).
+ *
+ * Have a look at `escapeUTF8` if you want a more concise output at the expense
+ * of reduced transportability.
+ *
+ * @param data String to escape.
+ */
+export const escape = encodeXML;
+function getEscaper(regex, map) {
+    return function escape(data) {
+        let match;
+        let lastIdx = 0;
+        let result = "";
+        while ((match = regex.exec(data))) {
+            if (lastIdx !== match.index) {
+                result += data.substring(lastIdx, match.index);
+            }
+            // We know that this chararcter will be in the map.
+            result += map.get(match[0].charCodeAt(0));
+            // Every match will be of length 1
+            lastIdx = match.index + 1;
+        }
+        return result + data.substring(lastIdx);
+    };
+}
+/**
+ * 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 const escapeUTF8 = getEscaper(/[&<>'"]/g, xmlCodeMap);
+/**
+ * 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 const escapeAttribute = getEscaper(/["&\u00A0]/g, new Map([
+    [34, "&quot;"],
+    [38, "&amp;"],
+    [160, "&nbsp;"],
+]));
+/**
+ * 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 const escapeText = getEscaper(/[&<>\u00A0]/g, new Map([
+    [38, "&amp;"],
+    [60, "&lt;"],
+    [62, "&gt;"],
+    [160, "&nbsp;"],
+]));
+//# sourceMappingURL=encode.js.map

package/lib/esm/encode.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode.js","sourceRoot":"","sources":["../../src/encode.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,kBAAkB,CAAC;AAElE,MAAM,YAAY,GAAG,qCAAqC,CAAC;AAC3D,MAAM,WAAW,GAAG,sBAAsB,CAAC;AAE3C,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC;IACvB,CAAC,EAAE,EAAE,QAAQ,CAAC;IACd,CAAC,EAAE,EAAE,OAAO,CAAC;IACb,CAAC,EAAE,EAAE,QAAQ,CAAC;IACd,CAAC,EAAE,EAAE,MAAM,CAAC;IACZ,CAAC,EAAE,EAAE,MAAM,CAAC;CACf,CAAC,CAAC;AAEH;;;;;;GAMG;AACH,MAAM,UAAU,SAAS,CAAC,GAAW;IACjC,IAAI,GAAG,GAAG,EAAE,CAAC;IACb,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,KAAK,CAAC;IAEV,OAAO,CAAC,KAAK,GAAG,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,EAAE;QAC7C,MAAM,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC;QACtB,MAAM,IAAI,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;QAC/B,MAAM,IAAI,GAAG,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAElC,IAAI,IAAI,EAAE;YACN,GAAG,IAAI,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC;YACxC,OAAO,GAAG,CAAC,GAAG,CAAC,CAAC;SACnB;aAAM;YACH,GAAG,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC,CAAC,MAAM,YAAY,CACjD,GAAG,EACH,CAAC,CACJ,CAAC,QAAQ,CAAC,EAAE,CAAC,GAAG,CAAC;YAClB,4CAA4C;YAC5C,OAAO,GAAG,WAAW,CAAC,SAAS,IAAI,MAAM,CACrC,CAAC,IAAI,GAAG,KAAqB,CAAC,KAAK,MAAM,CAC5C,CAAC;SACL;KACJ;IAED,OAAO,GAAG,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;AACrC,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,UAAU,CAAC,IAAY;IACnC,OAAO,gBAAgB,CAAC,YAAY,EAAE,IAAI,CAAC,CAAC;AAChD,CAAC;AACD;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,IAAY;IAC3C,OAAO,gBAAgB,CAAC,WAAW,EAAE,IAAI,CAAC,CAAC;AAC/C,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,MAAM,GAAG,SAAS,CAAC;AAEhC,SAAS,UAAU,CACf,KAAa,EACb,GAAwB;IAExB,OAAO,SAAS,MAAM,CAAC,IAAY;QAC/B,IAAI,KAAK,CAAC;QACV,IAAI,OAAO,GAAG,CAAC,CAAC;QAChB,IAAI,MAAM,GAAG,EAAE,CAAC;QAEhB,OAAO,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE;YAC/B,IAAI,OAAO,KAAK,KAAK,CAAC,KAAK,EAAE;gBACzB,MAAM,IAAI,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC;aAClD;YAED,mDAAmD;YACnD,MAAM,IAAI,GAAG,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAE,CAAC;YAE3C,kCAAkC;YAClC,OAAO,GAAG,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC;SAC7B;QAED,OAAO,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;IAC5C,CAAC,CAAC;AACN,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,UAAU,GAAG,UAAU,CAAC,UAAU,EAAE,UAAU,CAAC,CAAC;AAE7D;;;;;GAKG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,UAAU,CACrC,aAAa,EACb,IAAI,GAAG,CAAC;IACJ,CAAC,EAAE,EAAE,QAAQ,CAAC;IACd,CAAC,EAAE,EAAE,OAAO,CAAC;IACb,CAAC,GAAG,EAAE,QAAQ,CAAC;CAClB,CAAC,CACL,CAAC;AAEF;;;;;GAKG;AACH,MAAM,CAAC,MAAM,UAAU,GAAG,UAAU,CAChC,cAAc,EACd,IAAI,GAAG,CAAC;IACJ,CAAC,EAAE,EAAE,OAAO,CAAC;IACb,CAAC,EAAE,EAAE,MAAM,CAAC;IACZ,CAAC,EAAE,EAAE,MAAM,CAAC;IACZ,CAAC,GAAG,EAAE,QAAQ,CAAC;CAClB,CAAC,CACL,CAAC"}

package/lib/esm/generated/decode-data-html.d.ts

@@ -0,0 +1,3 @@
+declare const _default: Uint16Array;
+export default _default;
+//# sourceMappingURL=decode-data-html.d.ts.map

package/lib/esm/generated/decode-data-html.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode-data-html.d.ts","sourceRoot":"","sources":["../../../src/generated/decode-data-html.ts"],"names":[],"mappings":";AAEA,wBAAivgE"}