entities 7.0.0 → 8.0.0

package/dist/decode-codepoint.d.ts

@@ -0,0 +1,8 @@
+/**
+ * Replace the given code point with a replacement character if it is a
+ * surrogate or is outside the valid range. Otherwise return the code
+ * point unchanged.
+ * @param codePoint Unicode code point to convert.
+ */
+export declare function replaceCodePoint(codePoint: number): number;
+//# sourceMappingURL=decode-codepoint.d.ts.map

package/dist/decode-codepoint.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode-codepoint.d.ts","sourceRoot":"","sources":["../src/decode-codepoint.ts"],"names":[],"mappings":"AAkCA;;;;;GAKG;AACH,wBAAgB,gBAAgB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAS1D"}

package/dist/decode-codepoint.js

@@ -0,0 +1,46 @@
+// Adapted from https://github.com/mathiasbynens/he/blob/36afe179392226cf1b6ccdb16ebbb7a5a844d93a/src/he.js#L106-L134
+const decodeMap = new Map([
+    [0, 65_533],
+    // C1 Unicode control character reference replacements
+    [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],
+]);
+/**
+ * Replace the given code point with a replacement character if it is a
+ * surrogate or is outside the valid range. Otherwise return the code
+ * point unchanged.
+ * @param codePoint Unicode code point to convert.
+ */
+export function replaceCodePoint(codePoint) {
+    if ((codePoint >= 0xd8_00 && codePoint <= 0xdf_ff) ||
+        codePoint > 0x10_ff_ff) {
+        return 0xff_fd;
+    }
+    return decodeMap.get(codePoint) ?? codePoint;
+}
+//# sourceMappingURL=decode-codepoint.js.map

package/dist/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,MAAM,CAAC;IACX,sDAAsD;IACtD,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;;;;;GAKG;AACH,MAAM,UAAU,gBAAgB,CAAC,SAAiB;IAC9C,IACI,CAAC,SAAS,IAAI,OAAO,IAAI,SAAS,IAAI,OAAO,CAAC;QAC9C,SAAS,GAAG,UAAU,EACxB,CAAC;QACC,OAAO,OAAO,CAAC;IACnB,CAAC;IAED,OAAO,SAAS,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,SAAS,CAAC;AACjD,CAAC"}

package/dist/{commonjs/decode.d.ts → decode.d.ts}

@@ -1,3 +1,6 @@
+/**
+ * Decoding mode for named entities.
+ */
 export declare enum DecodingMode {
     /** Entities in text nodes that can end with any character. */
     Legacy = 0,
@@ -25,7 +28,6 @@ export declare class EntityDecoder {
      *
      * For multi-byte named entities, this will be called multiple times,
      * with the second codepoint, and the same `consumed` value.
-     *
      * @param codepoint The decoded codepoint.
      * @param consumed The number of bytes consumed by the decoder.
      */
@@ -40,7 +42,6 @@ export declare class EntityDecoder {
      *
      * For multi-byte named entities, this will be called multiple times,
      * with the second codepoint, and the same `consumed` value.
-     *
      * @param codepoint The decoded codepoint.
      * @param consumed The number of bytes consumed by the decoder.
      */
@@ -64,7 +65,12 @@ export declare class EntityDecoder {
     private excess;
     /** The mode in which the decoder is operating. */
     private decodeMode;
-    /** Resets the instance to make it reusable. */
+    /** The number of characters that have been consumed in the current run. */
+    private runConsumed;
+    /**
+     * Resets the instance to make it reusable.
+     * @param decodeMode Entity decoding mode to use.
+     */
     startEntity(decodeMode: DecodingMode): void;
     /**
      * Write an entity to the decoder. This can be called multiple times with partial entities.
@@ -72,7 +78,6 @@ export declare class EntityDecoder {
      *
      * Mirrors the implementation of `getDecoder`, but with the ability to stop decoding if the
      * entity is incomplete, and resume when the next string is written.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The offset at which the entity begins. Should be 0 if this is not the first call.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -82,7 +87,6 @@ export declare class EntityDecoder {
      * Switches between the numeric decimal and hexadecimal states.
      *
      * Equivalent to the `Numeric character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -92,7 +96,6 @@ export declare class EntityDecoder {
      * Parses a hexadecimal numeric entity.
      *
      * Equivalent to the `Hexademical character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -102,7 +105,6 @@ export declare class EntityDecoder {
      * Parses a decimal numeric entity.
      *
      * Equivalent to the `Decimal character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -113,7 +115,6 @@ export declare class EntityDecoder {
      *
      * Implements the logic from the `Hexademical character reference start
      * state` and `Numeric character reference end state` in the HTML spec.
-     *
      * @param lastCp The last code point of the entity. Used to see if the
      *               entity was terminated with a semicolon.
      * @param expectedLength The minimum number of characters that should be
@@ -126,7 +127,6 @@ export declare class EntityDecoder {
      * Parses a named entity.
      *
      * Equivalent to the `Named character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -134,17 +134,14 @@ export declare class EntityDecoder {
     private stateNamedEntity;
     /**
      * Emit a named entity that was not terminated with a semicolon.
-     *
      * @returns The number of characters consumed.
      */
     private emitNotTerminatedNamedEntity;
     /**
      * Emit a named entity.
-     *
      * @param result The index of the entity in the decode tree.
      * @param valueLength The number of bytes in the entity.
      * @param consumed The number of characters consumed.
-     *
      * @returns The number of characters consumed.
      */
     private emitNamedEntityData;
@@ -152,7 +149,6 @@ export declare class EntityDecoder {
      * Signal to the parser that the end of the input was reached.
      *
      * Remaining data will be emitted and relevant errors will be produced.
-     *
      * @returns The number of characters consumed.
      */
     end(): number;
@@ -160,17 +156,15 @@ export declare class EntityDecoder {
 /**
  * Determines the branch of the current node that is taken given the current
  * character. This function is used to traverse the trie.
- *
  * @param decodeTree The trie.
  * @param current The current node.
- * @param nodeIdx The index right after the current node and its value.
+ * @param nodeIndex Index immediately after the current node header.
  * @param char The current character.
  * @returns The index of the next node, or -1 if no branch is taken.
  */
 export declare function determineBranch(decodeTree: Uint16Array, current: number, nodeIndex: number, char: number): number;
 /**
  * Decodes an HTML string.
- *
  * @param htmlString The string to decode.
  * @param mode The decoding mode.
  * @returns The decoded string.
@@ -178,26 +172,23 @@ export declare function determineBranch(decodeTree: Uint16Array, current: number
 export declare function decodeHTML(htmlString: string, mode?: DecodingMode): string;
 /**
  * Decodes an HTML string in an attribute.
- *
  * @param htmlAttribute The string to decode.
  * @returns The decoded string.
  */
 export declare function decodeHTMLAttribute(htmlAttribute: string): string;
 /**
  * Decodes an HTML string, requiring all entities to be terminated by a semicolon.
- *
  * @param htmlString The string to decode.
  * @returns The decoded string.
  */
 export declare function decodeHTMLStrict(htmlString: string): string;
 /**
  * Decodes an XML string, requiring all entities to be terminated by a semicolon.
- *
  * @param xmlString The string to decode.
  * @returns The decoded string.
  */
 export declare function decodeXML(xmlString: string): string;
-export { decodeCodePoint, fromCodePoint, replaceCodePoint, } from "./decode-codepoint.js";
+export { replaceCodePoint } from "./decode-codepoint.js";
 export { htmlDecodeTree } from "./generated/decode-data-html.js";
 export { xmlDecodeTree } from "./generated/decode-data-xml.js";
 //# sourceMappingURL=decode.d.ts.map

package/dist/decode.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode.d.ts","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":"AA6DA;;GAEG;AACH,oBAAY,YAAY;IACpB,8DAA8D;IAC9D,MAAM,IAAI;IACV,uDAAuD;IACvD,MAAM,IAAI;IACV,oEAAoE;IACpE,SAAS,IAAI;CAChB;AAED;;GAEG;AACH,MAAM,WAAW,mBAAmB;IAChC,uCAAuC,IAAI,IAAI,CAAC;IAChD,0CAA0C,CACtC,kBAAkB,EAAE,MAAM,GAC3B,IAAI,CAAC;IACR,iCAAiC,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;CACzD;AAED;;GAEG;AACH,qBAAa,aAAa;IAElB,wCAAwC;IAExC,OAAO,CAAC,QAAQ,CAAC,UAAU;IAC3B;;;;;;;OAOG;IACH,OAAO,CAAC,QAAQ,CAAC,aAAa;IAC9B,gDAAgD;IAChD,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC;;IAbxB,wCAAwC;IAEvB,UAAU,EAAE,WAAW;IACxC;;;;;;;OAOG;IACc,aAAa,EAAE,CAAC,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,KAAK,IAAI;IACtE,gDAAgD;IAC/B,MAAM,CAAC,EAAE,mBAAmB,GAAG,SAAS;IAG7D,wCAAwC;IACxC,OAAO,CAAC,KAAK,CAAkC;IAC/C,6DAA6D;IAC7D,OAAO,CAAC,QAAQ,CAAK;IACrB;;;;;OAKG;IACH,OAAO,CAAC,MAAM,CAAK;IAEnB,4CAA4C;IAC5C,OAAO,CAAC,SAAS,CAAK;IACtB,6DAA6D;IAC7D,OAAO,CAAC,MAAM,CAAK;IACnB,kDAAkD;IAClD,OAAO,CAAC,UAAU,CAAuB;IACzC,2EAA2E;IAC3E,OAAO,CAAC,WAAW,CAAK;IAExB;;;OAGG;IACH,WAAW,CAAC,UAAU,EAAE,YAAY,GAAG,IAAI;IAU3C;;;;;;;;;OASG;IACH,KAAK,CAAC,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM;IA8B5C;;;;;;;OAOG;IACH,OAAO,CAAC,iBAAiB;IAezB;;;;;;;OAOG;IACH,OAAO,CAAC,eAAe;IAmBvB;;;;;;;OAOG;IACH,OAAO,CAAC,mBAAmB;IAc3B;;;;;;;;;;;OAWG;IACH,OAAO,CAAC,iBAAiB;IA6BzB;;;;;;;OAOG;IACH,OAAO,CAAC,gBAAgB;IAoIxB;;;OAGG;IACH,OAAO,CAAC,4BAA4B;IAYpC;;;;;;OAMG;IACH,OAAO,CAAC,mBAAmB;IAsB3B;;;;;OAKG;IACH,GAAG,IAAI,MAAM;CA6BhB;AAmDD;;;;;;;;GAQG;AACH,wBAAgB,eAAe,CAC3B,UAAU,EAAE,WAAW,EACvB,OAAO,EAAE,MAAM,EACf,SAAS,EAAE,MAAM,EACjB,IAAI,EAAE,MAAM,GACb,MAAM,CA4CR;AAKD;;;;;GAKG;AACH,wBAAgB,UAAU,CACtB,UAAU,EAAE,MAAM,EAClB,IAAI,GAAE,YAAkC,GACzC,MAAM,CAER;AAED;;;;GAIG;AACH,wBAAgB,mBAAmB,CAAC,aAAa,EAAE,MAAM,GAAG,MAAM,CAEjE;AAED;;;;GAIG;AACH,wBAAgB,gBAAgB,CAAC,UAAU,EAAE,MAAM,GAAG,MAAM,CAE3D;AAED;;;;GAIG;AACH,wBAAgB,SAAS,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAEnD;AAED,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAEzD,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,aAAa,EAAE,MAAM,gCAAgC,CAAC"}

package/dist/{esm/decode.js → decode.js}

@@ -1,4 +1,4 @@
-import { fromCodePoint, replaceCodePoint } from "./decode-codepoint.js";
+import { replaceCodePoint } from "./decode-codepoint.js";
 import { htmlDecodeTree } from "./generated/decode-data-html.js";
 import { xmlDecodeTree } from "./generated/decode-data-xml.js";
 import { BinTrieFlags } from "./internal/bin-trie-flags.js";
@@ -18,7 +18,7 @@ var CharCodes;
     CharCodes[CharCodes["UPPER_Z"] = 90] = "UPPER_Z";
 })(CharCodes || (CharCodes = {}));
 /** Bit that needs to be set to convert an upper case ASCII character to lower case */
-const TO_LOWER_BIT = 32;
+const TO_LOWER_BIT = 0b10_0000;
 function isNumber(code) {
     return code >= CharCodes.ZERO && code <= CharCodes.NINE;
 }
@@ -36,6 +36,7 @@ function isAsciiAlphaNumeric(code) {
  *
  * Attribute values that aren't terminated properly aren't parsed, and shouldn't lead to a parser error.
  * See the example in https://html.spec.whatwg.org/multipage/parsing.html#named-character-reference-state
+ * @param code Code point to decode.
  */
 function isEntityInAttributeInvalidEnd(code) {
     return code === CharCodes.EQUALS || isAsciiAlphaNumeric(code);
@@ -48,6 +49,9 @@ var EntityDecoderState;
     EntityDecoderState[EntityDecoderState["NumericHex"] = 3] = "NumericHex";
     EntityDecoderState[EntityDecoderState["NamedEntity"] = 4] = "NamedEntity";
 })(EntityDecoderState || (EntityDecoderState = {}));
+/**
+ * Decoding mode for named entities.
+ */
 export var DecodingMode;
 (function (DecodingMode) {
     /** Entities in text nodes that can end with any character. */
@@ -61,6 +65,9 @@ export var DecodingMode;
  * Token decoder with support of writing partial entities.
  */
 export class EntityDecoder {
+    decodeTree;
+    emitCodePoint;
+    errors;
     constructor(
     /** The tree used to decode entities. */
     // biome-ignore lint/correctness/noUnusedPrivateClassMembers: False positive
@@ -70,7 +77,6 @@ export class EntityDecoder {
      *
      * For multi-byte named entities, this will be called multiple times,
      * with the second codepoint, and the same `consumed` value.
-     *
      * @param codepoint The decoded codepoint.
      * @param consumed The number of bytes consumed by the decoder.
      */
@@ -80,25 +86,30 @@ export class EntityDecoder {
         this.decodeTree = decodeTree;
         this.emitCodePoint = emitCodePoint;
         this.errors = errors;
-        /** The current state of the decoder. */
-        this.state = EntityDecoderState.EntityStart;
-        /** Characters that were consumed while parsing an entity. */
-        this.consumed = 1;
-        /**
-         * The result of the entity.
-         *
-         * Either the result index of a numeric entity, or the codepoint of a
-         * numeric entity.
-         */
-        this.result = 0;
-        /** The current index in the decode tree. */
-        this.treeIndex = 0;
-        /** The number of characters that were consumed in excess. */
-        this.excess = 1;
-        /** The mode in which the decoder is operating. */
-        this.decodeMode = DecodingMode.Strict;
     }
-    /** Resets the instance to make it reusable. */
+    /** The current state of the decoder. */
+    state = EntityDecoderState.EntityStart;
+    /** Characters that were consumed while parsing an entity. */
+    consumed = 1;
+    /**
+     * The result of the entity.
+     *
+     * Either the result index of a numeric entity, or the codepoint of a
+     * numeric entity.
+     */
+    result = 0;
+    /** The current index in the decode tree. */
+    treeIndex = 0;
+    /** The number of characters that were consumed in excess. */
+    excess = 1;
+    /** The mode in which the decoder is operating. */
+    decodeMode = DecodingMode.Strict;
+    /** The number of characters that have been consumed in the current run. */
+    runConsumed = 0;
+    /**
+     * Resets the instance to make it reusable.
+     * @param decodeMode Entity decoding mode to use.
+     */
     startEntity(decodeMode) {
         this.decodeMode = decodeMode;
         this.state = EntityDecoderState.EntityStart;
@@ -106,6 +117,7 @@ export class EntityDecoder {
         this.treeIndex = 0;
         this.excess = 1;
         this.consumed = 1;
+        this.runConsumed = 0;
     }
     /**
      * Write an entity to the decoder. This can be called multiple times with partial entities.
@@ -113,7 +125,6 @@ export class EntityDecoder {
      *
      * Mirrors the implementation of `getDecoder`, but with the ability to stop decoding if the
      * entity is incomplete, and resume when the next string is written.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The offset at which the entity begins. Should be 0 if this is not the first call.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -147,7 +158,6 @@ export class EntityDecoder {
      * Switches between the numeric decimal and hexadecimal states.
      *
      * Equivalent to the `Numeric character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -168,7 +178,6 @@ export class EntityDecoder {
      * Parses a hexadecimal numeric entity.
      *
      * Equivalent to the `Hexademical character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -195,7 +204,6 @@ export class EntityDecoder {
      * Parses a decimal numeric entity.
      *
      * Equivalent to the `Decimal character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -219,7 +227,6 @@ export class EntityDecoder {
      *
      * Implements the logic from the `Hexademical character reference start
      * state` and `Numeric character reference end state` in the HTML spec.
-     *
      * @param lastCp The last code point of the entity. Used to see if the
      *               entity was terminated with a semicolon.
      * @param expectedLength The minimum number of characters that should be
@@ -228,10 +235,9 @@ export class EntityDecoder {
      * @returns The number of characters that were consumed.
      */
     emitNumericEntity(lastCp, expectedLength) {
-        var _a;
         // Ensure we consumed at least one digit.
         if (this.consumed <= expectedLength) {
-            (_a = this.errors) === null || _a === void 0 ? void 0 : _a.absenceOfDigitsInNumericCharacterReference(this.consumed);
+            this.errors?.absenceOfDigitsInNumericCharacterReference(this.consumed);
             return 0;
         }
         // Figure out if this is a legit end of the entity
@@ -254,7 +260,6 @@ export class EntityDecoder {
      * Parses a named entity.
      *
      * Equivalent to the `Named character reference state` in the HTML spec.
-     *
      * @param input The string containing the entity (or a continuation of the entity).
      * @param offset The current offset.
      * @returns The number of characters that were consumed, or -1 if the entity is incomplete.
@@ -268,43 +273,40 @@ export class EntityDecoder {
             // Handle compact runs (possibly inline): valueLength == 0 and SEMI_REQUIRED bit set.
             if (valueLength === 0 && (current & BinTrieFlags.FLAG13) !== 0) {
                 const runLength = (current & BinTrieFlags.BRANCH_LENGTH) >> 7; /* 2..63 */
-                const firstChar = current & BinTrieFlags.JUMP_TABLE;
-                // Fast-fail if we don't have enough remaining input for the full run (incomplete entity)
-                if (offset + runLength > input.length)
-                    return -1;
-                // Verify first char
-                if (input.charCodeAt(offset) !== firstChar) {
-                    return this.result === 0
-                        ? 0
-                        : this.emitNotTerminatedNamedEntity();
-                }
-                offset++;
-                this.excess++;
-                // Remaining characters after the first
-                const remaining = runLength - 1;
-                // Iterate over packed 2-char words
-                for (let runPos = 1; runPos < runLength; runPos += 2) {
-                    const packedWord = decodeTree[this.treeIndex + 1 + ((runPos - 1) >> 1)];
-                    const low = packedWord & 0xff;
-                    if (input.charCodeAt(offset) !== low) {
+                // If we are starting a run, check the first char.
+                if (this.runConsumed === 0) {
+                    const firstChar = current & BinTrieFlags.JUMP_TABLE;
+                    if (input.charCodeAt(offset) !== firstChar) {
                         return this.result === 0
                             ? 0
                             : this.emitNotTerminatedNamedEntity();
                     }
                     offset++;
                     this.excess++;
-                    const high = (packedWord >> 8) & 0xff;
-                    if (runPos + 1 < runLength) {
-                        if (input.charCodeAt(offset) !== high) {
-                            return this.result === 0
-                                ? 0
-                                : this.emitNotTerminatedNamedEntity();
-                        }
-                        offset++;
-                        this.excess++;
+                    this.runConsumed++;
+                }
+                // Check remaining characters in the run.
+                while (this.runConsumed < runLength) {
+                    if (offset >= input.length) {
+                        return -1;
+                    }
+                    const charIndexInPacked = this.runConsumed - 1;
+                    const packedWord = decodeTree[this.treeIndex + 1 + (charIndexInPacked >> 1)];
+                    const expectedChar = charIndexInPacked % 2 === 0
+                        ? packedWord & 0xff
+                        : (packedWord >> 8) & 0xff;
+                    if (input.charCodeAt(offset) !== expectedChar) {
+                        this.runConsumed = 0;
+                        return this.result === 0
+                            ? 0
+                            : this.emitNotTerminatedNamedEntity();
                     }
+                    offset++;
+                    this.excess++;
+                    this.runConsumed++;
                 }
-                this.treeIndex += 1 + ((remaining + 1) >> 1);
+                this.runConsumed = 0;
+                this.treeIndex += 1 + (runLength >> 1);
                 current = decodeTree[this.treeIndex];
                 valueLength = (current & BinTrieFlags.VALUE_LENGTH) >> 14;
             }
@@ -359,24 +361,20 @@ export class EntityDecoder {
     }
     /**
      * Emit a named entity that was not terminated with a semicolon.
-     *
      * @returns The number of characters consumed.
      */
     emitNotTerminatedNamedEntity() {
-        var _a;
         const { result, decodeTree } = this;
         const valueLength = (decodeTree[result] & BinTrieFlags.VALUE_LENGTH) >> 14;
         this.emitNamedEntityData(result, valueLength, this.consumed);
-        (_a = this.errors) === null || _a === void 0 ? void 0 : _a.missingSemicolonAfterCharacterReference();
+        this.errors?.missingSemicolonAfterCharacterReference();
         return this.consumed;
     }
     /**
      * Emit a named entity.
-     *
      * @param result The index of the entity in the decode tree.
      * @param valueLength The number of bytes in the entity.
      * @param consumed The number of characters consumed.
-     *
      * @returns The number of characters consumed.
      */
     emitNamedEntityData(result, valueLength, consumed) {
@@ -395,11 +393,9 @@ export class EntityDecoder {
      * Signal to the parser that the end of the input was reached.
      *
      * Remaining data will be emitted and relevant errors will be produced.
-     *
      * @returns The number of characters consumed.
      */
     end() {
-        var _a;
         switch (this.state) {
             case EntityDecoderState.NamedEntity: {
                 // Emit a named entity if we have one.
@@ -417,7 +413,7 @@ export class EntityDecoder {
                 return this.emitNumericEntity(0, 3);
             }
             case EntityDecoderState.NumericStart: {
-                (_a = this.errors) === null || _a === void 0 ? void 0 : _a.absenceOfDigitsInNumericCharacterReference(this.consumed);
+                this.errors?.absenceOfDigitsInNumericCharacterReference(this.consumed);
                 return 0;
             }
             case EntityDecoderState.EntityStart: {
@@ -429,13 +425,12 @@ export class EntityDecoder {
 }
 /**
  * Creates a function that decodes entities in a string.
- *
  * @param decodeTree The decode tree.
  * @returns A function that decodes entities in a string.
  */
 function getDecoder(decodeTree) {
     let returnValue = "";
-    const decoder = new EntityDecoder(decodeTree, (data) => (returnValue += fromCodePoint(data)));
+    const decoder = new EntityDecoder(decodeTree, (data) => (returnValue += String.fromCodePoint(data)));
     return function decodeWithTrie(input, decodeMode) {
         let lastIndex = 0;
         let offset = 0;
@@ -462,10 +457,9 @@ function getDecoder(decodeTree) {
 /**
  * Determines the branch of the current node that is taken given the current
  * character. This function is used to traverse the trie.
- *
  * @param decodeTree The trie.
  * @param current The current node.
- * @param nodeIdx The index right after the current node and its value.
+ * @param nodeIndex Index immediately after the current node header.
  * @param char The current character.
  * @returns The index of the next node, or -1 if no branch is taken.
  */
@@ -512,7 +506,6 @@ const htmlDecoder = /* #__PURE__ */ getDecoder(htmlDecodeTree);
 const xmlDecoder = /* #__PURE__ */ getDecoder(xmlDecodeTree);
 /**
  * Decodes an HTML string.
- *
  * @param htmlString The string to decode.
  * @param mode The decoding mode.
  * @returns The decoded string.
@@ -522,7 +515,6 @@ export function decodeHTML(htmlString, mode = DecodingMode.Legacy) {
 }
 /**
  * Decodes an HTML string in an attribute.
- *
  * @param htmlAttribute The string to decode.
  * @returns The decoded string.
  */
@@ -531,7 +523,6 @@ export function decodeHTMLAttribute(htmlAttribute) {
 }
 /**
  * Decodes an HTML string, requiring all entities to be terminated by a semicolon.
- *
  * @param htmlString The string to decode.
  * @returns The decoded string.
  */
@@ -540,14 +531,13 @@ export function decodeHTMLStrict(htmlString) {
 }
 /**
  * Decodes an XML string, requiring all entities to be terminated by a semicolon.
- *
  * @param xmlString The string to decode.
  * @returns The decoded string.
  */
 export function decodeXML(xmlString) {
     return xmlDecoder(xmlString, DecodingMode.Strict);
 }
-export { decodeCodePoint, fromCodePoint, replaceCodePoint, } from "./decode-codepoint.js";
+export { replaceCodePoint } from "./decode-codepoint.js";
 // Re-export for use by eg. htmlparser2
 export { htmlDecodeTree } from "./generated/decode-data-html.js";
 export { xmlDecodeTree } from "./generated/decode-data-xml.js";

package/dist/decode.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"decode.js","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACzD,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,aAAa,EAAE,MAAM,gCAAgC,CAAC;AAC/D,OAAO,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAC;AAE5D,IAAW,SAaV;AAbD,WAAW,SAAS;IAChB,wCAAQ,CAAA;IACR,0CAAS,CAAA;IACT,8CAAW,CAAA;IACX,0CAAS,CAAA;IACT,0CAAS,CAAA;IACT,gDAAY,CAAA;IACZ,iDAAa,CAAA;IACb,iDAAa,CAAA;IACb,iDAAa,CAAA;IACb,gDAAY,CAAA;IACZ,gDAAY,CAAA;IACZ,gDAAY,CAAA;AAChB,CAAC,EAbU,SAAS,KAAT,SAAS,QAanB;AAED,sFAAsF;AACtF,MAAM,YAAY,GAAG,SAAS,CAAC;AAE/B,SAAS,QAAQ,CAAC,IAAY;IAC1B,OAAO,IAAI,IAAI,SAAS,CAAC,IAAI,IAAI,IAAI,IAAI,SAAS,CAAC,IAAI,CAAC;AAC5D,CAAC;AAED,SAAS,sBAAsB,CAAC,IAAY;IACxC,OAAO,CACH,CAAC,IAAI,IAAI,SAAS,CAAC,OAAO,IAAI,IAAI,IAAI,SAAS,CAAC,OAAO,CAAC;QACxD,CAAC,IAAI,IAAI,SAAS,CAAC,OAAO,IAAI,IAAI,IAAI,SAAS,CAAC,OAAO,CAAC,CAC3D,CAAC;AACN,CAAC;AAED,SAAS,mBAAmB,CAAC,IAAY;IACrC,OAAO,CACH,CAAC,IAAI,IAAI,SAAS,CAAC,OAAO,IAAI,IAAI,IAAI,SAAS,CAAC,OAAO,CAAC;QACxD,CAAC,IAAI,IAAI,SAAS,CAAC,OAAO,IAAI,IAAI,IAAI,SAAS,CAAC,OAAO,CAAC;QACxD,QAAQ,CAAC,IAAI,CAAC,CACjB,CAAC;AACN,CAAC;AAED;;;;;;GAMG;AACH,SAAS,6BAA6B,CAAC,IAAY;IAC/C,OAAO,IAAI,KAAK,SAAS,CAAC,MAAM,IAAI,mBAAmB,CAAC,IAAI,CAAC,CAAC;AAClE,CAAC;AAED,IAAW,kBAMV;AAND,WAAW,kBAAkB;IACzB,yEAAW,CAAA;IACX,2EAAY,CAAA;IACZ,+EAAc,CAAA;IACd,uEAAU,CAAA;IACV,yEAAW,CAAA;AACf,CAAC,EANU,kBAAkB,KAAlB,kBAAkB,QAM5B;AAED;;GAEG;AACH,MAAM,CAAN,IAAY,YAOX;AAPD,WAAY,YAAY;IACpB,8DAA8D;IAC9D,mDAAU,CAAA;IACV,uDAAuD;IACvD,mDAAU,CAAA;IACV,oEAAoE;IACpE,yDAAa,CAAA;AACjB,CAAC,EAPW,YAAY,KAAZ,YAAY,QAOvB;AAaD;;GAEG;AACH,MAAM,OAAO,aAAa;IAID;IASA;IAEA;IAdrB;IACI,wCAAwC;IACxC,4EAA4E;IAC3D,UAAuB;IACxC;;;;;;;OAOG;IACc,aAAqD;IACtE,gDAAgD;IAC/B,MAAwC;QAXxC,eAAU,GAAV,UAAU,CAAa;QASvB,kBAAa,GAAb,aAAa,CAAwC;QAErD,WAAM,GAAN,MAAM,CAAkC;IAC1D,CAAC;IAEJ,wCAAwC;IAChC,KAAK,GAAG,kBAAkB,CAAC,WAAW,CAAC;IAC/C,6DAA6D;IACrD,QAAQ,GAAG,CAAC,CAAC;IACrB;;;;;OAKG;IACK,MAAM,GAAG,CAAC,CAAC;IAEnB,4CAA4C;IACpC,SAAS,GAAG,CAAC,CAAC;IACtB,6DAA6D;IACrD,MAAM,GAAG,CAAC,CAAC;IACnB,kDAAkD;IAC1C,UAAU,GAAG,YAAY,CAAC,MAAM,CAAC;IACzC,2EAA2E;IACnE,WAAW,GAAG,CAAC,CAAC;IAExB;;;OAGG;IACH,WAAW,CAAC,UAAwB;QAChC,IAAI,CAAC,UAAU,GAAG,UAAU,CAAC;QAC7B,IAAI,CAAC,KAAK,GAAG,kBAAkB,CAAC,WAAW,CAAC;QAC5C,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;QAChB,IAAI,CAAC,SAAS,GAAG,CAAC,CAAC;QACnB,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;QAChB,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC;QAClB,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC;IACzB,CAAC;IAED;;;;;;;;;OASG;IACH,KAAK,CAAC,KAAa,EAAE,MAAc;QAC/B,QAAQ,IAAI,CAAC,KAAK,EAAE,CAAC;YACjB,KAAK,kBAAkB,CAAC,WAAW,CAAC,CAAC,CAAC;gBAClC,IAAI,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,KAAK,SAAS,CAAC,GAAG,EAAE,CAAC;oBAC7C,IAAI,CAAC,KAAK,GAAG,kBAAkB,CAAC,YAAY,CAAC;oBAC7C,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;oBACnB,OAAO,IAAI,CAAC,iBAAiB,CAAC,KAAK,EAAE,MAAM,GAAG,CAAC,CAAC,CAAC;gBACrD,CAAC;gBACD,IAAI,CAAC,KAAK,GAAG,kBAAkB,CAAC,WAAW,CAAC;gBAC5C,OAAO,IAAI,CAAC,gBAAgB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;YAChD,CAAC;YAED,KAAK,kBAAkB,CAAC,YAAY,CAAC,CAAC,CAAC;gBACnC,OAAO,IAAI,CAAC,iBAAiB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;YACjD,CAAC;YAED,KAAK,kBAAkB,CAAC,cAAc,CAAC,CAAC,CAAC;gBACrC,OAAO,IAAI,CAAC,mBAAmB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;YACnD,CAAC;YAED,KAAK,kBAAkB,CAAC,UAAU,CAAC,CAAC,CAAC;gBACjC,OAAO,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;YAC/C,CAAC;YAED,KAAK,kBAAkB,CAAC,WAAW,CAAC,CAAC,CAAC;gBAClC,OAAO,IAAI,CAAC,gBAAgB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;YAChD,CAAC;QACL,CAAC;IACL,CAAC;IAED;;;;;;;OAOG;IACK,iBAAiB,CAAC,KAAa,EAAE,MAAc;QACnD,IAAI,MAAM,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;YACzB,OAAO,CAAC,CAAC,CAAC;QACd,CAAC;QAED,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,GAAG,YAAY,CAAC,KAAK,SAAS,CAAC,OAAO,EAAE,CAAC;YAClE,IAAI,CAAC,KAAK,GAAG,kBAAkB,CAAC,UAAU,CAAC;YAC3C,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;YACnB,OAAO,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,MAAM,GAAG,CAAC,CAAC,CAAC;QACnD,CAAC;QAED,IAAI,CAAC,KAAK,GAAG,kBAAkB,CAAC,cAAc,CAAC;QAC/C,OAAO,IAAI,CAAC,mBAAmB,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC;IACnD,CAAC;IAED;;;;;;;OAOG;IACK,eAAe,CAAC,KAAa,EAAE,MAAc;QACjD,OAAO,MAAM,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC3B,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;YACtC,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,sBAAsB,CAAC,IAAI,CAAC,EAAE,CAAC;gBACjD,oDAAoD;gBACpD,MAAM,KAAK,GACP,IAAI,IAAI,SAAS,CAAC,IAAI;oBAClB,CAAC,CAAC,IAAI,GAAG,SAAS,CAAC,IAAI;oBACvB,CAAC,CAAC,CAAC,IAAI,GAAG,YAAY,CAAC,GAAG,SAAS,CAAC,OAAO,GAAG,EAAE,CAAC;gBACzD,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,GAAG,EAAE,GAAG,KAAK,CAAC;gBACvC,IAAI,CAAC,QAAQ,EAAE,CAAC;gBAChB,MAAM,EAAE,CAAC;YACb,CAAC;iBAAM,CAAC;gBACJ,OAAO,IAAI,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;YAC3C,CAAC;QACL,CAAC;QACD,OAAO,CAAC,CAAC,CAAC,CAAC,oBAAoB;IACnC,CAAC;IAED;;;;;;;OAOG;IACK,mBAAmB,CAAC,KAAa,EAAE,MAAc;QACrD,OAAO,MAAM,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC3B,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;YACtC,IAAI,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;gBACjB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,GAAG,EAAE,GAAG,CAAC,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;gBACzD,IAAI,CAAC,QAAQ,EAAE,CAAC;gBAChB,MAAM,EAAE,CAAC;YACb,CAAC;iBAAM,CAAC;gBACJ,OAAO,IAAI,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;YAC3C,CAAC;QACL,CAAC;QACD,OAAO,CAAC,CAAC,CAAC,CAAC,oBAAoB;IACnC,CAAC;IAED;;;;;;;;;;;OAWG;IACK,iBAAiB,CAAC,MAAc,EAAE,cAAsB;QAC5D,yCAAyC;QACzC,IAAI,IAAI,CAAC,QAAQ,IAAI,cAAc,EAAE,CAAC;YAClC,IAAI,CAAC,MAAM,EAAE,0CAA0C,CACnD,IAAI,CAAC,QAAQ,CAChB,CAAC;YACF,OAAO,CAAC,CAAC;QACb,CAAC;QAED,kDAAkD;QAClD,IAAI,MAAM,KAAK,SAAS,CAAC,IAAI,EAAE,CAAC;YAC5B,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;QACvB,CAAC;aAAM,IAAI,IAAI,CAAC,UAAU,KAAK,YAAY,CAAC,MAAM,EAAE,CAAC;YACjD,OAAO,CAAC,CAAC;QACb,CAAC;QAED,IAAI,CAAC,aAAa,CAAC,gBAAgB,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;QAEjE,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YACd,IAAI,MAAM,KAAK,SAAS,CAAC,IAAI,EAAE,CAAC;gBAC5B,IAAI,CAAC,MAAM,CAAC,uCAAuC,EAAE,CAAC;YAC1D,CAAC;YAED,IAAI,CAAC,MAAM,CAAC,iCAAiC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAC/D,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC;IACzB,CAAC;IAED;;;;;;;OAOG;IACK,gBAAgB,CAAC,KAAa,EAAE,MAAc;QAClD,MAAM,EAAE,UAAU,EAAE,GAAG,IAAI,CAAC;QAC5B,IAAI,OAAO,GAAG,UAAU,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;QACzC,8EAA8E;QAC9E,IAAI,WAAW,GAAG,CAAC,OAAO,GAAG,YAAY,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;QAE9D,OAAO,MAAM,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC3B,qFAAqF;YACrF,IAAI,WAAW,KAAK,CAAC,IAAI,CAAC,OAAO,GAAG,YAAY,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;gBAC7D,MAAM,SAAS,GACX,CAAC,OAAO,GAAG,YAAY,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW;gBAE5D,kDAAkD;gBAClD,IAAI,IAAI,CAAC,WAAW,KAAK,CAAC,EAAE,CAAC;oBACzB,MAAM,SAAS,GAAG,OAAO,GAAG,YAAY,CAAC,UAAU,CAAC;oBACpD,IAAI,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,KAAK,SAAS,EAAE,CAAC;wBACzC,OAAO,IAAI,CAAC,MAAM,KAAK,CAAC;4BACpB,CAAC,CAAC,CAAC;4BACH,CAAC,CAAC,IAAI,CAAC,4BAA4B,EAAE,CAAC;oBAC9C,CAAC;oBACD,MAAM,EAAE,CAAC;oBACT,IAAI,CAAC,MAAM,EAAE,CAAC;oBACd,IAAI,CAAC,WAAW,EAAE,CAAC;gBACvB,CAAC;gBAED,yCAAyC;gBACzC,OAAO,IAAI,CAAC,WAAW,GAAG,SAAS,EAAE,CAAC;oBAClC,IAAI,MAAM,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;wBACzB,OAAO,CAAC,CAAC,CAAC;oBACd,CAAC;oBAED,MAAM,iBAAiB,GAAG,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC;oBAC/C,MAAM,UAAU,GACZ,UAAU,CACN,IAAI,CAAC,SAAS,GAAG,CAAC,GAAG,CAAC,iBAAiB,IAAI,CAAC,CAAC,CAChD,CAAC;oBACN,MAAM,YAAY,GACd,iBAAiB,GAAG,CAAC,KAAK,CAAC;wBACvB,CAAC,CAAC,UAAU,GAAG,IAAI;wBACnB,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC;oBAEnC,IAAI,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,KAAK,YAAY,EAAE,CAAC;wBAC5C,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC;wBACrB,OAAO,IAAI,CAAC,MAAM,KAAK,CAAC;4BACpB,CAAC,CAAC,CAAC;4BACH,CAAC,CAAC,IAAI,CAAC,4BAA4B,EAAE,CAAC;oBAC9C,CAAC;oBACD,MAAM,EAAE,CAAC;oBACT,IAAI,CAAC,MAAM,EAAE,CAAC;oBACd,IAAI,CAAC,WAAW,EAAE,CAAC;gBACvB,CAAC;gBAED,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC;gBACrB,IAAI,CAAC,SAAS,IAAI,CAAC,GAAG,CAAC,SAAS,IAAI,CAAC,CAAC,CAAC;gBACvC,OAAO,GAAG,UAAU,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;gBACrC,WAAW,GAAG,CAAC,OAAO,GAAG,YAAY,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;YAC9D,CAAC;YAED,IAAI,MAAM,IAAI,KAAK,CAAC,MAAM;gBAAE,MAAM;YAElC,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;YAEtC;;;;;;eAMG;YACH,IACI,IAAI,KAAK,SAAS,CAAC,IAAI;gBACvB,WAAW,KAAK,CAAC;gBACjB,CAAC,OAAO,GAAG,YAAY,CAAC,MAAM,CAAC,KAAK,CAAC,EACvC,CAAC;gBACC,OAAO,IAAI,CAAC,mBAAmB,CAC3B,IAAI,CAAC,SAAS,EACd,WAAW,EACX,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,MAAM,CAC9B,CAAC;YACN,CAAC;YAED,IAAI,CAAC,SAAS,GAAG,eAAe,CAC5B,UAAU,EACV,OAAO,EACP,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,WAAW,CAAC,EACzC,IAAI,CACP,CAAC;YAEF,IAAI,IAAI,CAAC,SAAS,GAAG,CAAC,EAAE,CAAC;gBACrB,OAAO,IAAI,CAAC,MAAM,KAAK,CAAC;oBACpB,iCAAiC;oBACjC,CAAC,IAAI,CAAC,UAAU,KAAK,YAAY,CAAC,SAAS;wBACvC,8DAA8D;wBAC9D,CAAC,WAAW,KAAK,CAAC;4BACd,6CAA6C;4BAC7C,6BAA6B,CAAC,IAAI,CAAC,CAAC,CAAC;oBAC7C,CAAC,CAAC,CAAC;oBACH,CAAC,CAAC,IAAI,CAAC,4BAA4B,EAAE,CAAC;YAC9C,CAAC;YAED,OAAO,GAAG,UAAU,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;YACrC,WAAW,GAAG,CAAC,OAAO,GAAG,YAAY,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;YAE1D,kDAAkD;YAClD,IAAI,WAAW,KAAK,CAAC,EAAE,CAAC;gBACpB,2DAA2D;gBAC3D,IAAI,IAAI,KAAK,SAAS,CAAC,IAAI,EAAE,CAAC;oBAC1B,OAAO,IAAI,CAAC,mBAAmB,CAC3B,IAAI,CAAC,SAAS,EACd,WAAW,EACX,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,MAAM,CAC9B,CAAC;gBACN,CAAC;gBAED,2FAA2F;gBAC3F,IACI,IAAI,CAAC,UAAU,KAAK,YAAY,CAAC,MAAM;oBACvC,CAAC,OAAO,GAAG,YAAY,CAAC,MAAM,CAAC,KAAK,CAAC,EACvC,CAAC;oBACC,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,SAAS,CAAC;oBAC7B,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,MAAM,CAAC;oBAC7B,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;gBACpB,CAAC;YACL,CAAC;YACD,+CAA+C;YAC/C,MAAM,EAAE,CAAC;YACT,IAAI,CAAC,MAAM,EAAE,CAAC;QAClB,CAAC;QAED,OAAO,CAAC,CAAC,CAAC;IACd,CAAC;IAED;;;OAGG;IACK,4BAA4B;QAChC,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,GAAG,IAAI,CAAC;QAEpC,MAAM,WAAW,GACb,CAAC,UAAU,CAAC,MAAM,CAAC,GAAG,YAAY,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC;QAE3D,IAAI,CAAC,mBAAmB,CAAC,MAAM,EAAE,WAAW,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC7D,IAAI,CAAC,MAAM,EAAE,uCAAuC,EAAE,CAAC;QAEvD,OAAO,IAAI,CAAC,QAAQ,CAAC;IACzB,CAAC;IAED;;;;;;OAMG;IACK,mBAAmB,CACvB,MAAc,EACd,WAAmB,EACnB,QAAgB;QAEhB,MAAM,EAAE,UAAU,EAAE,GAAG,IAAI,CAAC;QAE5B,IAAI,CAAC,aAAa,CACd,WAAW,KAAK,CAAC;YACb,CAAC,CAAC,UAAU,CAAC,MAAM,CAAC;gBACd,CAAC,CAAC,YAAY,CAAC,YAAY,GAAG,YAAY,CAAC,MAAM,CAAC;YACxD,CAAC,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,EAC5B,QAAQ,CACX,CAAC;QACF,IAAI,WAAW,KAAK,CAAC,EAAE,CAAC;YACpB,0DAA0D;YAC1D,IAAI,CAAC,aAAa,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;QACzD,CAAC;QAED,OAAO,QAAQ,CAAC;IACpB,CAAC;IAED;;;;;OAKG;IACH,GAAG;QACC,QAAQ,IAAI,CAAC,KAAK,EAAE,CAAC;YACjB,KAAK,kBAAkB,CAAC,WAAW,CAAC,CAAC,CAAC;gBAClC,sCAAsC;gBACtC,OAAO,IAAI,CAAC,MAAM,KAAK,CAAC;oBACpB,CAAC,IAAI,CAAC,UAAU,KAAK,YAAY,CAAC,SAAS;wBACvC,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,SAAS,CAAC;oBACnC,CAAC,CAAC,IAAI,CAAC,4BAA4B,EAAE;oBACrC,CAAC,CAAC,CAAC,CAAC;YACZ,CAAC;YACD,mDAAmD;YACnD,KAAK,kBAAkB,CAAC,cAAc,CAAC,CAAC,CAAC;gBACrC,OAAO,IAAI,CAAC,iBAAiB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YACxC,CAAC;YACD,KAAK,kBAAkB,CAAC,UAAU,CAAC,CAAC,CAAC;gBACjC,OAAO,IAAI,CAAC,iBAAiB,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YACxC,CAAC;YACD,KAAK,kBAAkB,CAAC,YAAY,CAAC,CAAC,CAAC;gBACnC,IAAI,CAAC,MAAM,EAAE,0CAA0C,CACnD,IAAI,CAAC,QAAQ,CAChB,CAAC;gBACF,OAAO,CAAC,CAAC;YACb,CAAC;YACD,KAAK,kBAAkB,CAAC,WAAW,CAAC,CAAC,CAAC;gBAClC,iCAAiC;gBACjC,OAAO,CAAC,CAAC;YACb,CAAC;QACL,CAAC;IACL,CAAC;CACJ;AAED;;;;GAIG;AACH,SAAS,UAAU,CAAC,UAAuB;IACvC,IAAI,WAAW,GAAG,EAAE,CAAC;IACrB,MAAM,OAAO,GAAG,IAAI,aAAa,CAC7B,UAAU,EACV,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,WAAW,IAAI,MAAM,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,CACxD,CAAC;IAEF,OAAO,SAAS,cAAc,CAC1B,KAAa,EACb,UAAwB;QAExB,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,MAAM,GAAG,CAAC,CAAC;QAEf,OAAO,CAAC,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC;YAChD,WAAW,IAAI,KAAK,CAAC,KAAK,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC;YAE9C,OAAO,CAAC,WAAW,CAAC,UAAU,CAAC,CAAC;YAEhC,MAAM,MAAM,GAAG,OAAO,CAAC,KAAK,CACxB,KAAK;YACL,eAAe;YACf,MAAM,GAAG,CAAC,CACb,CAAC;YAEF,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC;gBACb,SAAS,GAAG,MAAM,GAAG,OAAO,CAAC,GAAG,EAAE,CAAC;gBACnC,MAAM;YACV,CAAC;YAED,SAAS,GAAG,MAAM,GAAG,MAAM,CAAC;YAC5B,uDAAuD;YACvD,MAAM,GAAG,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;QACtD,CAAC;QAED,MAAM,MAAM,GAAG,WAAW,GAAG,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC;QAEpD,2DAA2D;QAC3D,WAAW,GAAG,EAAE,CAAC;QAEjB,OAAO,MAAM,CAAC;IAClB,CAAC,CAAC;AACN,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,eAAe,CAC3B,UAAuB,EACvB,OAAe,EACf,SAAiB,EACjB,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,CAAC;QACpB,OAAO,UAAU,KAAK,CAAC,IAAI,IAAI,KAAK,UAAU,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACpE,CAAC;IAED,kDAAkD;IAClD,IAAI,UAAU,EAAE,CAAC;QACb,MAAM,KAAK,GAAG,IAAI,GAAG,UAAU,CAAC;QAEhC,OAAO,KAAK,GAAG,CAAC,IAAI,KAAK,IAAI,WAAW;YACpC,CAAC,CAAC,CAAC,CAAC;YACJ,CAAC,CAAC,UAAU,CAAC,SAAS,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC;IAC5C,CAAC;IAED,+EAA+E;IAC/E,MAAM,cAAc,GAAG,CAAC,WAAW,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;IAE9C;;;OAGG;IACH,IAAI,EAAE,GAAG,CAAC,CAAC;IACX,IAAI,EAAE,GAAG,WAAW,GAAG,CAAC,CAAC;IAEzB,OAAO,EAAE,IAAI,EAAE,EAAE,CAAC;QACd,MAAM,GAAG,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC;QAC5B,MAAM,IAAI,GAAG,GAAG,IAAI,CAAC,CAAC;QACtB,MAAM,MAAM,GAAG,UAAU,CAAC,SAAS,GAAG,IAAI,CAAC,CAAC;QAC5C,MAAM,MAAM,GAAG,CAAC,MAAM,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC;QAElD,IAAI,MAAM,GAAG,IAAI,EAAE,CAAC;YAChB,EAAE,GAAG,GAAG,GAAG,CAAC,CAAC;QACjB,CAAC;aAAM,IAAI,MAAM,GAAG,IAAI,EAAE,CAAC;YACvB,EAAE,GAAG,GAAG,GAAG,CAAC,CAAC;QACjB,CAAC;aAAM,CAAC;YACJ,OAAO,UAAU,CAAC,SAAS,GAAG,cAAc,GAAG,GAAG,CAAC,CAAC;QACxD,CAAC;IACL,CAAC;IAED,OAAO,CAAC,CAAC,CAAC;AACd,CAAC;AAED,MAAM,WAAW,GAAG,eAAe,CAAC,UAAU,CAAC,cAAc,CAAC,CAAC;AAC/D,MAAM,UAAU,GAAG,eAAe,CAAC,UAAU,CAAC,aAAa,CAAC,CAAC;AAE7D;;;;;GAKG;AACH,MAAM,UAAU,UAAU,CACtB,UAAkB,EAClB,OAAqB,YAAY,CAAC,MAAM;IAExC,OAAO,WAAW,CAAC,UAAU,EAAE,IAAI,CAAC,CAAC;AACzC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,mBAAmB,CAAC,aAAqB;IACrD,OAAO,WAAW,CAAC,aAAa,EAAE,YAAY,CAAC,SAAS,CAAC,CAAC;AAC9D,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,gBAAgB,CAAC,UAAkB;IAC/C,OAAO,WAAW,CAAC,UAAU,EAAE,YAAY,CAAC,MAAM,CAAC,CAAC;AACxD,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,SAAS,CAAC,SAAiB;IACvC,OAAO,UAAU,CAAC,SAAS,EAAE,YAAY,CAAC,MAAM,CAAC,CAAC;AACtD,CAAC;AAED,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACzD,uCAAuC;AACvC,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,aAAa,EAAE,MAAM,gCAAgC,CAAC"}

package/dist/{commonjs/encode.d.ts → encode.d.ts}

@@ -8,6 +8,7 @@
  *
  * If a character has no equivalent entity, a numeric hexadecimal reference
  * (eg. `ü`) will be used.
+ * @param input Input string to encode or decode.
  */
 export declare function encodeHTML(input: string): string;
 /**
@@ -17,6 +18,7 @@ export declare function encodeHTML(input: string): string;
  *
  * If a character has no equivalent entity, a numeric hexadecimal reference
  * (eg. `ü`) will be used.
+ * @param input Input string to encode or decode.
  */
 export declare function encodeNonAsciiHTML(input: string): string;
 //# sourceMappingURL=encode.d.ts.map

package/dist/encode.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode.d.ts","sourceRoot":"","sources":["../src/encode.ts"],"names":[],"mappings":"AAeA;;;;;;;;;;;GAWG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAEhD;AACD;;;;;;;;GAQG;AACH,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAExD"}

package/dist/{esm/encode.js → encode.js}

@@ -4,10 +4,10 @@ import { htmlTrie } from "./generated/encode-html.js";
  * We store the characters to consider as a compact bitset for fast lookups.
  */
 const HTML_BITSET = /* #__PURE__ */ new Uint32Array([
-    5632, // Bits for 09,0A,0C
-    4227923966, // 32..63 -> 21-2D (minus space), 2E,2F,3A-3F
-    4160749569, // 64..95 -> 40, 5B-5F
-    939524097, // 96..127-> 60, 7B-7D
+    0x16_00, // Bits for 09,0A,0C
+    0xfc_00_ff_fe, // 32..63 -> 21-2D (minus space), 2E,2F,3A-3F
+    0xf8_00_00_01, // 64..95 -> 40, 5B-5F
+    0x38_00_00_01, // 96..127-> 60, 7B-7D
 ]);
 const XML_BITSET = /* #__PURE__ */ new Uint32Array([0, XML_BITSET_VALUE, 0, 0]);
 /**
@@ -20,6 +20,7 @@ const XML_BITSET = /* #__PURE__ */ new Uint32Array([0, XML_BITSET_VALUE, 0, 0]);
  *
  * If a character has no equivalent entity, a numeric hexadecimal reference
  * (eg. `ü`) will be used.
+ * @param input Input string to encode or decode.
  */
 export function encodeHTML(input) {
     return encodeHTMLTrieRe(HTML_BITSET, input);
@@ -31,6 +32,7 @@ export function encodeHTML(input) {
  *
  * If a character has no equivalent entity, a numeric hexadecimal reference
  * (eg. `ü`) will be used.
+ * @param input Input string to encode or decode.
  */
 export function encodeNonAsciiHTML(input) {
     return encodeHTMLTrieRe(XML_BITSET, input);

package/dist/encode.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"encode.js","sourceRoot":"","sources":["../src/encode.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAC7D,OAAO,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAC;AAEtD;;GAEG;AACH,MAAM,WAAW,GAAG,eAAe,CAAC,IAAI,WAAW,CAAC;IAChD,OAAO,EAAE,oBAAoB;IAC7B,aAAa,EAAE,6CAA6C;IAC5D,aAAa,EAAE,sBAAsB;IACrC,aAAa,EAAE,sBAAsB;CACxC,CAAC,CAAC;AAEH,MAAM,UAAU,GAAG,eAAe,CAAC,IAAI,WAAW,CAAC,CAAC,CAAC,EAAE,gBAAgB,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;AAEhF;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACpC,OAAO,gBAAgB,CAAC,WAAW,EAAE,KAAK,CAAC,CAAC;AAChD,CAAC;AACD;;;;;;;;GAQG;AACH,MAAM,UAAU,kBAAkB,CAAC,KAAa;IAC5C,OAAO,gBAAgB,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC;AAC/C,CAAC;AAED,SAAS,gBAAgB,CAAC,MAAmB,EAAE,KAAa;IACxD,IAAI,GAAuB,CAAC;IAC5B,IAAI,IAAI,GAAG,CAAC,CAAC,CAAC,qCAAqC;IACnD,MAAM,EAAE,MAAM,EAAE,GAAG,KAAK,CAAC;IAEzB,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAC1C,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;QACrC,iDAAiD;QACjD,IAAI,IAAI,GAAG,IAAI,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,CAAC,KAAK,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;YACtD,SAAS;QACb,CAAC;QAED,IAAI,GAAG,KAAK,SAAS;YAAE,GAAG,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;aAClD,IAAI,IAAI,KAAK,KAAK;YAAE,GAAG,IAAI,KAAK,CAAC,SAAS,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QAE7D,IAAI,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAE9B,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC3B,IAAI,KAAK,GAAG,CAAC,GAAG,MAAM,EAAE,CAAC;gBACrB,MAAM,QAAQ,GAAG,KAAK,CAAC,UAAU,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;gBAC7C,MAAM,KAAK,GACP,OAAO,IAAI,CAAC,IAAI,KAAK,QAAQ;oBACzB,CAAC,CAAC,IAAI,CAAC,IAAI,KAAK,QAAQ;wBACpB,CAAC,CAAC,IAAI,CAAC,SAAS;wBAChB,CAAC,CAAC,SAAS;oBACf,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;gBAElC,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;oBACtB,GAAG,IAAI,KAAK,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,IAAI,GAAG,KAAK,GAAG,CAAC,CAAC;oBACjB,SAAS;gBACb,CAAC;YACL,CAAC;YACD,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC;QACtB,CAAC;QAED,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACrB,MAAM,EAAE,GAAG,YAAY,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;YACtC,GAAG,IAAI,MAAM,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,GAAG,CAAC;YAChC,IAAI,EAAE,KAAK,IAAI;gBAAE,KAAK,EAAE,CAAC;YACzB,IAAI,GAAG,KAAK,GAAG,CAAC,CAAC;QACrB,CAAC;aAAM,CAAC;YACJ,GAAG,IAAI,IAAI,CAAC;YACZ,IAAI,GAAG,KAAK,GAAG,CAAC,CAAC;QACrB,CAAC;IACL,CAAC;IAED,IAAI,GAAG,KAAK,SAAS;QAAE,OAAO,KAAK,CAAC;IACpC,IAAI,IAAI,GAAG,MAAM;QAAE,GAAG,IAAI,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;IAC7C,OAAO,GAAG,CAAC;AACf,CAAC"}