entities 6.0.1 → 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/{esm/decode.d.ts → decode.d.ts}

@@ -1,8 +1,6 @@
-export declare enum BinTrieFlags {
-    VALUE_LENGTH = 49152,
-    BRANCH_LENGTH = 16256,
-    JUMP_TABLE = 127
-}
+/**
+ * Decoding mode for named entities.
+ */
 export declare enum DecodingMode {
     /** Entities in text nodes that can end with any character. */
     Legacy = 0,
@@ -30,7 +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.
      */
@@ -45,7 +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.
      */
@@ -69,7 +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.
@@ -77,7 +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.
@@ -87,18 +87,15 @@ 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.
      */
     private stateNumericStart;
-    private addToNumericResult;
     /**
      * 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.
@@ -108,7 +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.
@@ -119,7 +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
@@ -132,7 +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.
@@ -140,17 +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;
@@ -158,7 +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;
@@ -166,17 +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.
@@ -184,26 +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 { replaceCodePoint } from "./decode-codepoint.js";
 export { htmlDecodeTree } from "./generated/decode-data-html.js";
 export { xmlDecodeTree } from "./generated/decode-data-xml.js";
-export { decodeCodePoint, replaceCodePoint, fromCodePoint, } from "./decode-codepoint.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"}