parse5 7.0.0 → 7.1.0

package/dist/cjs/index.d.ts

@@ -45,7 +45,7 @@ export declare function parse<T extends TreeAdapterTypeMap = DefaultTreeAdapterM
  * console.log(documentFragment.childNodes[0].tagName); //> 'table'
  *
  * // Parses the html fragment in the context of the parsed <table> element.
- * const trFragment = parser.parseFragment(documentFragment.childNodes[0], '<tr><td>Shake it, baby</td></tr>');
+ * const trFragment = parse5.parseFragment(documentFragment.childNodes[0], '<tr><td>Shake it, baby</td></tr>');
  *
  * console.log(trFragment.childNodes[0].childNodes[0].tagName); //> 'td'
  * ```

package/dist/cjs/parser/formatting-element-list.d.ts

@@ -1,5 +1,5 @@
 import type { TagToken } from '../common/token.js';
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
 export declare enum EntryType {
     Marker = 0,
     Element = 1
@@ -24,6 +24,11 @@ export declare class FormattingElementList<T extends TreeAdapterTypeMap> {
     pushElement(element: T['element'], token: TagToken): void;
     insertElementAfterBookmark(element: T['element'], token: TagToken): void;
     removeEntry(entry: Entry<T>): void;
+    /**
+     * Clears the list of formatting elements up to the last marker.
+     *
+     * @see https://html.spec.whatwg.org/multipage/parsing.html#clear-the-list-of-active-formatting-elements-up-to-the-last-marker
+     */
     clearToLastMarker(): void;
     getElementEntryInScopeWithTagName(tagName: string): ElementEntry<T> | null;
     getElementEntry(element: T['element']): ElementEntry<T> | undefined;

package/dist/cjs/parser/formatting-element-list.js

@@ -88,6 +88,11 @@ class FormattingElementList {
             this.entries.splice(entryIndex, 1);
         }
     }
+    /**
+     * Clears the list of formatting elements up to the last marker.
+     *
+     * @see https://html.spec.whatwg.org/multipage/parsing.html#clear-the-list-of-active-formatting-elements-up-to-the-last-marker
+     */
     clearToLastMarker() {
         const markerIdx = this.entries.indexOf(MARKER);
         if (markerIdx >= 0) {

package/dist/cjs/parser/index.js

@@ -1307,19 +1307,7 @@ function endTagInHead(p, token) {
             break;
         }
         case html_js_1.TAG_ID.TEMPLATE: {
-            if (p.openElements.tmplCount > 0) {
-                p.openElements.generateImpliedEndTagsThoroughly();
-                if (p.openElements.currentTagId !== html_js_1.TAG_ID.TEMPLATE) {
-                    p._err(token, error_codes_js_1.ERR.closingOfElementWithOpenChildElements);
-                }
-                p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.TEMPLATE);
-                p.activeFormattingElements.clearToLastMarker();
-                p.tmplInsertionModeStack.shift();
-                p._resetInsertionMode();
-            }
-            else {
-                p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
-            }
+            templateEndTagInHead(p, token);
             break;
         }
         default: {
@@ -1327,6 +1315,21 @@ function endTagInHead(p, token) {
         }
     }
 }
+function templateEndTagInHead(p, token) {
+    if (p.openElements.tmplCount > 0) {
+        p.openElements.generateImpliedEndTagsThoroughly();
+        if (p.openElements.currentTagId !== html_js_1.TAG_ID.TEMPLATE) {
+            p._err(token, error_codes_js_1.ERR.closingOfElementWithOpenChildElements);
+        }
+        p.openElements.popUntilTagNamePopped(html_js_1.TAG_ID.TEMPLATE);
+        p.activeFormattingElements.clearToLastMarker();
+        p.tmplInsertionModeStack.shift();
+        p._resetInsertionMode();
+    }
+    else {
+        p._err(token, error_codes_js_1.ERR.endTagWithoutMatchingOpenElement);
+    }
+}
 function tokenInHead(p, token) {
     p.openElements.pop();
     p.insertionMode = InsertionMode.AFTER_HEAD;
@@ -1435,7 +1438,7 @@ function endTagAfterHead(p, token) {
             break;
         }
         case html_js_1.TAG_ID.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         default: {
@@ -2095,6 +2098,7 @@ function endTagInBody(p, token) {
         case html_js_1.TAG_ID.MAIN:
         case html_js_1.TAG_ID.MENU:
         case html_js_1.TAG_ID.ASIDE:
+        case html_js_1.TAG_ID.BUTTON:
         case html_js_1.TAG_ID.CENTER:
         case html_js_1.TAG_ID.FIGURE:
         case html_js_1.TAG_ID.FOOTER:
@@ -2154,7 +2158,7 @@ function endTagInBody(p, token) {
             break;
         }
         case html_js_1.TAG_ID.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         default: {
@@ -2320,7 +2324,7 @@ function endTagInTable(p, token) {
             break;
         }
         case html_js_1.TAG_ID.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         case html_js_1.TAG_ID.BODY:
@@ -2457,7 +2461,7 @@ function endTagInColumnGroup(p, token) {
             break;
         }
         case html_js_1.TAG_ID.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         case html_js_1.TAG_ID.COL: {
@@ -2752,7 +2756,7 @@ function endTagInSelect(p, token) {
             break;
         }
         case html_js_1.TAG_ID.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         default:
@@ -2850,7 +2854,7 @@ function startTagInTemplate(p, token) {
 }
 function endTagInTemplate(p, token) {
     if (token.tagID === html_js_1.TAG_ID.TEMPLATE) {
-        endTagInHead(p, token);
+        templateEndTagInHead(p, token);
     }
 }
 function eofInTemplate(p, token) {

package/dist/cjs/parser/open-element-stack.d.ts

@@ -1,5 +1,5 @@
 import { TAG_ID as $ } from '../common/html.js';
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
 export interface StackHandler<T extends TreeAdapterTypeMap> {
     onItemPush: (node: T['parentNode'], tid: number, isTop: boolean) => void;
     onItemPop: (node: T['parentNode'], isTop: boolean) => void;

package/dist/cjs/serializer/index.d.ts

@@ -1,4 +1,4 @@
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
 import { type DefaultTreeAdapterMap } from '../tree-adapters/default.js';
 export interface SerializerOptions<T extends TreeAdapterTypeMap> {
     /**

package/dist/cjs/tokenizer/index.d.ts

@@ -78,10 +78,9 @@ declare const enum State {
     AMBIGUOUS_AMPERSAND = 73,
     NUMERIC_CHARACTER_REFERENCE = 74,
     HEXADEMICAL_CHARACTER_REFERENCE_START = 75,
-    DECIMAL_CHARACTER_REFERENCE_START = 76,
-    HEXADEMICAL_CHARACTER_REFERENCE = 77,
-    DECIMAL_CHARACTER_REFERENCE = 78,
-    NUMERIC_CHARACTER_REFERENCE_END = 79
+    HEXADEMICAL_CHARACTER_REFERENCE = 76,
+    DECIMAL_CHARACTER_REFERENCE = 77,
+    NUMERIC_CHARACTER_REFERENCE_END = 78
 }
 export declare const TokenizerMode: {
     readonly DATA: State.DATA;
@@ -241,7 +240,6 @@ export declare class Tokenizer {
     private _stateAmbiguousAmpersand;
     private _stateNumericCharacterReference;
     private _stateHexademicalCharacterReferenceStart;
-    private _stateDecimalCharacterReferenceStart;
     private _stateHexademicalCharacterReference;
     private _stateDecimalCharacterReference;
     private _stateNumericCharacterReferenceEnd;

package/dist/cjs/tokenizer/index.js

@@ -116,10 +116,9 @@ var State;
     State[State["AMBIGUOUS_AMPERSAND"] = 73] = "AMBIGUOUS_AMPERSAND";
     State[State["NUMERIC_CHARACTER_REFERENCE"] = 74] = "NUMERIC_CHARACTER_REFERENCE";
     State[State["HEXADEMICAL_CHARACTER_REFERENCE_START"] = 75] = "HEXADEMICAL_CHARACTER_REFERENCE_START";
-    State[State["DECIMAL_CHARACTER_REFERENCE_START"] = 76] = "DECIMAL_CHARACTER_REFERENCE_START";
-    State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 77] = "HEXADEMICAL_CHARACTER_REFERENCE";
-    State[State["DECIMAL_CHARACTER_REFERENCE"] = 78] = "DECIMAL_CHARACTER_REFERENCE";
-    State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 79] = "NUMERIC_CHARACTER_REFERENCE_END";
+    State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 76] = "HEXADEMICAL_CHARACTER_REFERENCE";
+    State[State["DECIMAL_CHARACTER_REFERENCE"] = 77] = "DECIMAL_CHARACTER_REFERENCE";
+    State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 78] = "NUMERIC_CHARACTER_REFERENCE_END";
 })(State || (State = {}));
 //Tokenizer initial states for different modes
 exports.TokenizerMode = {
@@ -277,9 +276,9 @@ class Tokenizer {
         this.consumedAfterSnapshot -= count;
         this.preprocessor.retreat(count);
     }
-    _reconsumeInState(state) {
+    _reconsumeInState(state, cp) {
         this.state = state;
-        this._unconsume(1);
+        this._callState(cp);
     }
     _advanceBy(count) {
         this.consumedAfterSnapshot += count;
@@ -475,13 +474,11 @@ class Tokenizer {
         this._createCharacterToken(type, ch);
     }
     _emitCodePoint(cp) {
-        let type = token_js_1.TokenType.CHARACTER;
-        if (isWhitespace(cp)) {
-            type = token_js_1.TokenType.WHITESPACE_CHARACTER;
-        }
-        else if (cp === unicode_js_1.CODE_POINTS.NULL) {
-            type = token_js_1.TokenType.NULL_CHARACTER;
-        }
+        const type = isWhitespace(cp)
+            ? token_js_1.TokenType.WHITESPACE_CHARACTER
+            : cp === unicode_js_1.CODE_POINTS.NULL
+                ? token_js_1.TokenType.NULL_CHARACTER
+                : token_js_1.TokenType.CHARACTER;
         this._appendCharToCurrentCharacterToken(type, String.fromCodePoint(cp));
     }
     //NOTE: used when we emit characters explicitly.
@@ -864,10 +861,6 @@ class Tokenizer {
                 this._stateHexademicalCharacterReferenceStart(cp);
                 break;
             }
-            case State.DECIMAL_CHARACTER_REFERENCE_START: {
-                this._stateDecimalCharacterReferenceStart(cp);
-                break;
-            }
             case State.HEXADEMICAL_CHARACTER_REFERENCE: {
                 this._stateHexademicalCharacterReference(cp);
                 break;
@@ -877,7 +870,7 @@ class Tokenizer {
                 break;
             }
             case State.NUMERIC_CHARACTER_REFERENCE_END: {
-                this._stateNumericCharacterReferenceEnd();
+                this._stateNumericCharacterReferenceEnd(cp);
                 break;
             }
             default: {
@@ -2775,7 +2768,7 @@ class Tokenizer {
         }
         else {
             this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
-            this._reconsumeInState(this.returnState);
+            this._reconsumeInState(this.returnState, cp);
         }
     }
     // Named character reference state
@@ -2808,7 +2801,7 @@ class Tokenizer {
             if (cp === unicode_js_1.CODE_POINTS.SEMICOLON) {
                 this._err(error_codes_js_1.ERR.unknownNamedCharacterReference);
             }
-            this._reconsumeInState(this.returnState);
+            this._reconsumeInState(this.returnState, cp);
         }
     }
     // Numeric character reference state
@@ -2818,9 +2811,16 @@ class Tokenizer {
         if (cp === unicode_js_1.CODE_POINTS.LATIN_SMALL_X || cp === unicode_js_1.CODE_POINTS.LATIN_CAPITAL_X) {
             this.state = State.HEXADEMICAL_CHARACTER_REFERENCE_START;
         }
+        // Inlined decimal character reference start state
+        else if (isAsciiDigit(cp)) {
+            this.state = State.DECIMAL_CHARACTER_REFERENCE;
+            this._stateDecimalCharacterReference(cp);
+        }
         else {
-            this.state = State.DECIMAL_CHARACTER_REFERENCE_START;
-            this._stateDecimalCharacterReferenceStart(cp);
+            this._err(error_codes_js_1.ERR.absenceOfDigitsInNumericCharacterReference);
+            this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
+            this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.NUMBER_SIGN);
+            this._reconsumeInState(this.returnState, cp);
         }
     }
     // Hexademical character reference start state
@@ -2838,20 +2838,6 @@ class Tokenizer {
             this.state = this.returnState;
         }
     }
-    // Decimal character reference start state
-    //------------------------------------------------------------------
-    _stateDecimalCharacterReferenceStart(cp) {
-        if (isAsciiDigit(cp)) {
-            this.state = State.DECIMAL_CHARACTER_REFERENCE;
-            this._stateDecimalCharacterReference(cp);
-        }
-        else {
-            this._err(error_codes_js_1.ERR.absenceOfDigitsInNumericCharacterReference);
-            this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.AMPERSAND);
-            this._flushCodePointConsumedAsCharacterReference(unicode_js_1.CODE_POINTS.NUMBER_SIGN);
-            this._reconsumeInState(this.returnState);
-        }
-    }
     // Hexademical character reference state
     //------------------------------------------------------------------
     _stateHexademicalCharacterReference(cp) {
@@ -2870,7 +2856,7 @@ class Tokenizer {
         else {
             this._err(error_codes_js_1.ERR.missingSemicolonAfterCharacterReference);
             this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
-            this._stateNumericCharacterReferenceEnd();
+            this._stateNumericCharacterReferenceEnd(cp);
         }
     }
     // Decimal character reference state
@@ -2885,12 +2871,12 @@ class Tokenizer {
         else {
             this._err(error_codes_js_1.ERR.missingSemicolonAfterCharacterReference);
             this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
-            this._stateNumericCharacterReferenceEnd();
+            this._stateNumericCharacterReferenceEnd(cp);
         }
     }
     // Numeric character reference end state
     //------------------------------------------------------------------
-    _stateNumericCharacterReferenceEnd() {
+    _stateNumericCharacterReferenceEnd(cp) {
         if (this.charRefCode === unicode_js_1.CODE_POINTS.NULL) {
             this._err(error_codes_js_1.ERR.nullCharacterReference);
             this.charRefCode = unicode_js_1.CODE_POINTS.REPLACEMENT_CHARACTER;
@@ -2914,7 +2900,7 @@ class Tokenizer {
             }
         }
         this._flushCodePointConsumedAsCharacterReference(this.charRefCode);
-        this._reconsumeInState(this.returnState);
+        this._reconsumeInState(this.returnState, cp);
     }
 }
 exports.Tokenizer = Tokenizer;

package/dist/cjs/tree-adapters/default.d.ts

@@ -1,16 +1,9 @@
 import { DOCUMENT_MODE, type NS } from '../common/html.js';
 import type { Attribute, Location, ElementLocation } from '../common/token.js';
 import type { TreeAdapter, TreeAdapterTypeMap } from './interface.js';
-export declare enum NodeType {
-    Document = "#document",
-    DocumentFragment = "#document-fragment",
-    Comment = "#comment",
-    Text = "#text",
-    DocumentType = "#documentType"
-}
 export interface Document {
     /** The name of the node. */
-    nodeName: NodeType.Document;
+    nodeName: '#document';
     /**
      * Document mode.
      *
@@ -23,7 +16,7 @@ export interface Document {
 }
 export interface DocumentFragment {
     /** The name of the node. */
-    nodeName: NodeType.DocumentFragment;
+    nodeName: '#document-fragment';
     /** The node's children. */
     childNodes: ChildNode[];
     /** Comment source code location info. Available if location info is enabled. */
@@ -47,7 +40,7 @@ export interface Element {
 }
 export interface CommentNode {
     /** The name of the node. */
-    nodeName: NodeType.Comment;
+    nodeName: '#comment';
     /** Parent node. */
     parentNode: ParentNode | null;
     /** Comment text. */
@@ -56,7 +49,7 @@ export interface CommentNode {
     sourceCodeLocation?: Location | null;
 }
 export interface TextNode {
-    nodeName: NodeType.Text;
+    nodeName: '#text';
     /** Parent node. */
     parentNode: ParentNode | null;
     /** Text content. */
@@ -72,7 +65,7 @@ export interface Template extends Element {
 }
 export interface DocumentType {
     /** The name of the node. */
-    nodeName: NodeType.DocumentType;
+    nodeName: '#documentType';
     /** Parent node. */
     parentNode: ParentNode | null;
     /** Document type name. */

package/dist/cjs/tree-adapters/default.js

@@ -1,18 +1,10 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.defaultTreeAdapter = exports.NodeType = void 0;
+exports.defaultTreeAdapter = void 0;
 const html_js_1 = require("../common/html.js");
-var NodeType;
-(function (NodeType) {
-    NodeType["Document"] = "#document";
-    NodeType["DocumentFragment"] = "#document-fragment";
-    NodeType["Comment"] = "#comment";
-    NodeType["Text"] = "#text";
-    NodeType["DocumentType"] = "#documentType";
-})(NodeType = exports.NodeType || (exports.NodeType = {}));
 function createTextNode(value) {
     return {
-        nodeName: NodeType.Text,
+        nodeName: '#text',
         value,
         parentNode: null,
     };
@@ -21,14 +13,14 @@ exports.defaultTreeAdapter = {
     //Node construction
     createDocument() {
         return {
-            nodeName: NodeType.Document,
+            nodeName: '#document',
             mode: html_js_1.DOCUMENT_MODE.NO_QUIRKS,
             childNodes: [],
         };
     },
     createDocumentFragment() {
         return {
-            nodeName: NodeType.DocumentFragment,
+            nodeName: '#document-fragment',
             childNodes: [],
         };
     },
@@ -44,7 +36,7 @@ exports.defaultTreeAdapter = {
     },
     createCommentNode(data) {
         return {
-            nodeName: NodeType.Comment,
+            nodeName: '#comment',
             data,
             parentNode: null,
         };
@@ -66,7 +58,7 @@ exports.defaultTreeAdapter = {
         return templateElement.content;
     },
     setDocumentType(document, name, publicId, systemId) {
-        const doctypeNode = document.childNodes.find((node) => node.nodeName === NodeType.DocumentType);
+        const doctypeNode = document.childNodes.find((node) => node.nodeName === '#documentType');
         if (doctypeNode) {
             doctypeNode.name = name;
             doctypeNode.publicId = publicId;
@@ -74,7 +66,7 @@ exports.defaultTreeAdapter = {
         }
         else {
             const node = {
-                nodeName: NodeType.DocumentType,
+                nodeName: '#documentType',
                 name,
                 publicId,
                 systemId,
@@ -166,7 +158,7 @@ exports.defaultTreeAdapter = {
         return node.nodeName === '#comment';
     },
     isDocumentTypeNode(node) {
-        return node.nodeName === NodeType.DocumentType;
+        return node.nodeName === '#documentType';
     },
     isElementNode(node) {
         return Object.prototype.hasOwnProperty.call(node, 'tagName');

package/dist/common/error-codes.js

@@ -60,5 +60,5 @@ export var ERR;
     ERR["misplacedStartTagForHeadElement"] = "misplaced-start-tag-for-head-element";
     ERR["nestedNoscriptInHead"] = "nested-noscript-in-head";
     ERR["eofInElementThatCanContainOnlyText"] = "eof-in-element-that-can-contain-only-text";
-})(ERR || (ERR = {}));
+})(ERR = ERR || (ERR = {}));
 //# sourceMappingURL=error-codes.js.map

package/dist/common/html.js

@@ -7,7 +7,7 @@ export var NS;
     NS["XLINK"] = "http://www.w3.org/1999/xlink";
     NS["XML"] = "http://www.w3.org/XML/1998/namespace";
     NS["XMLNS"] = "http://www.w3.org/2000/xmlns/";
-})(NS || (NS = {}));
+})(NS = NS || (NS = {}));
 export var ATTRS;
 (function (ATTRS) {
     ATTRS["TYPE"] = "type";
@@ -18,7 +18,7 @@ export var ATTRS;
     ATTRS["COLOR"] = "color";
     ATTRS["FACE"] = "face";
     ATTRS["SIZE"] = "size";
-})(ATTRS || (ATTRS = {}));
+})(ATTRS = ATTRS || (ATTRS = {}));
 /**
  * The mode of the document.
  *
@@ -29,7 +29,7 @@ export var DOCUMENT_MODE;
     DOCUMENT_MODE["NO_QUIRKS"] = "no-quirks";
     DOCUMENT_MODE["QUIRKS"] = "quirks";
     DOCUMENT_MODE["LIMITED_QUIRKS"] = "limited-quirks";
-})(DOCUMENT_MODE || (DOCUMENT_MODE = {}));
+})(DOCUMENT_MODE = DOCUMENT_MODE || (DOCUMENT_MODE = {}));
 export var TAG_NAMES;
 (function (TAG_NAMES) {
     TAG_NAMES["A"] = "a";
@@ -154,7 +154,7 @@ export var TAG_NAMES;
     TAG_NAMES["VAR"] = "var";
     TAG_NAMES["WBR"] = "wbr";
     TAG_NAMES["XMP"] = "xmp";
-})(TAG_NAMES || (TAG_NAMES = {}));
+})(TAG_NAMES = TAG_NAMES || (TAG_NAMES = {}));
 /**
  * Tag IDs are numeric IDs for known tag names.
  *
@@ -285,7 +285,7 @@ export var TAG_ID;
     TAG_ID[TAG_ID["VAR"] = 120] = "VAR";
     TAG_ID[TAG_ID["WBR"] = 121] = "WBR";
     TAG_ID[TAG_ID["XMP"] = 122] = "XMP";
-})(TAG_ID || (TAG_ID = {}));
+})(TAG_ID = TAG_ID || (TAG_ID = {}));
 const TAG_NAME_TO_ID = new Map([
     [TAG_NAMES.A, TAG_ID.A],
     [TAG_NAMES.ADDRESS, TAG_ID.ADDRESS],

package/dist/common/token.js

@@ -9,7 +9,7 @@ export var TokenType;
     TokenType[TokenType["DOCTYPE"] = 6] = "DOCTYPE";
     TokenType[TokenType["EOF"] = 7] = "EOF";
     TokenType[TokenType["HIBERNATION"] = 8] = "HIBERNATION";
-})(TokenType || (TokenType = {}));
+})(TokenType = TokenType || (TokenType = {}));
 export function getTokenAttr(token, attrName) {
     for (let i = token.attrs.length - 1; i >= 0; i--) {
         if (token.attrs[i].name === attrName) {

package/dist/common/unicode.js

@@ -39,7 +39,7 @@ export var CODE_POINTS;
     CODE_POINTS[CODE_POINTS["LATIN_SMALL_X"] = 120] = "LATIN_SMALL_X";
     CODE_POINTS[CODE_POINTS["LATIN_SMALL_Z"] = 122] = "LATIN_SMALL_Z";
     CODE_POINTS[CODE_POINTS["REPLACEMENT_CHARACTER"] = 65533] = "REPLACEMENT_CHARACTER";
-})(CODE_POINTS || (CODE_POINTS = {}));
+})(CODE_POINTS = CODE_POINTS || (CODE_POINTS = {}));
 export const SEQUENCES = {
     DASH_DASH: '--',
     CDATA_START: '[CDATA[',

package/dist/index.d.ts

@@ -45,7 +45,7 @@ export declare function parse<T extends TreeAdapterTypeMap = DefaultTreeAdapterM
  * console.log(documentFragment.childNodes[0].tagName); //> 'table'
  *
  * // Parses the html fragment in the context of the parsed <table> element.
- * const trFragment = parser.parseFragment(documentFragment.childNodes[0], '<tr><td>Shake it, baby</td></tr>');
+ * const trFragment = parse5.parseFragment(documentFragment.childNodes[0], '<tr><td>Shake it, baby</td></tr>');
  *
  * console.log(trFragment.childNodes[0].childNodes[0].tagName); //> 'td'
  * ```

package/dist/parser/formatting-element-list.d.ts

@@ -1,5 +1,5 @@
 import type { TagToken } from '../common/token.js';
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
 export declare enum EntryType {
     Marker = 0,
     Element = 1
@@ -24,6 +24,11 @@ export declare class FormattingElementList<T extends TreeAdapterTypeMap> {
     pushElement(element: T['element'], token: TagToken): void;
     insertElementAfterBookmark(element: T['element'], token: TagToken): void;
     removeEntry(entry: Entry<T>): void;
+    /**
+     * Clears the list of formatting elements up to the last marker.
+     *
+     * @see https://html.spec.whatwg.org/multipage/parsing.html#clear-the-list-of-active-formatting-elements-up-to-the-last-marker
+     */
     clearToLastMarker(): void;
     getElementEntryInScopeWithTagName(tagName: string): ElementEntry<T> | null;
     getElementEntry(element: T['element']): ElementEntry<T> | undefined;

package/dist/parser/formatting-element-list.js

@@ -4,7 +4,7 @@ export var EntryType;
 (function (EntryType) {
     EntryType[EntryType["Marker"] = 0] = "Marker";
     EntryType[EntryType["Element"] = 1] = "Element";
-})(EntryType || (EntryType = {}));
+})(EntryType = EntryType || (EntryType = {}));
 const MARKER = { type: EntryType.Marker };
 //List of formatting elements
 export class FormattingElementList {
@@ -85,6 +85,11 @@ export class FormattingElementList {
             this.entries.splice(entryIndex, 1);
         }
     }
+    /**
+     * Clears the list of formatting elements up to the last marker.
+     *
+     * @see https://html.spec.whatwg.org/multipage/parsing.html#clear-the-list-of-active-formatting-elements-up-to-the-last-marker
+     */
     clearToLastMarker() {
         const markerIdx = this.entries.indexOf(MARKER);
         if (markerIdx >= 0) {

package/dist/parser/index.js

@@ -1312,19 +1312,7 @@ function endTagInHead(p, token) {
             break;
         }
         case $.TEMPLATE: {
-            if (p.openElements.tmplCount > 0) {
-                p.openElements.generateImpliedEndTagsThoroughly();
-                if (p.openElements.currentTagId !== $.TEMPLATE) {
-                    p._err(token, ERR.closingOfElementWithOpenChildElements);
-                }
-                p.openElements.popUntilTagNamePopped($.TEMPLATE);
-                p.activeFormattingElements.clearToLastMarker();
-                p.tmplInsertionModeStack.shift();
-                p._resetInsertionMode();
-            }
-            else {
-                p._err(token, ERR.endTagWithoutMatchingOpenElement);
-            }
+            templateEndTagInHead(p, token);
             break;
         }
         default: {
@@ -1332,6 +1320,21 @@ function endTagInHead(p, token) {
         }
     }
 }
+function templateEndTagInHead(p, token) {
+    if (p.openElements.tmplCount > 0) {
+        p.openElements.generateImpliedEndTagsThoroughly();
+        if (p.openElements.currentTagId !== $.TEMPLATE) {
+            p._err(token, ERR.closingOfElementWithOpenChildElements);
+        }
+        p.openElements.popUntilTagNamePopped($.TEMPLATE);
+        p.activeFormattingElements.clearToLastMarker();
+        p.tmplInsertionModeStack.shift();
+        p._resetInsertionMode();
+    }
+    else {
+        p._err(token, ERR.endTagWithoutMatchingOpenElement);
+    }
+}
 function tokenInHead(p, token) {
     p.openElements.pop();
     p.insertionMode = InsertionMode.AFTER_HEAD;
@@ -1440,7 +1443,7 @@ function endTagAfterHead(p, token) {
             break;
         }
         case $.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         default: {
@@ -2100,6 +2103,7 @@ function endTagInBody(p, token) {
         case $.MAIN:
         case $.MENU:
         case $.ASIDE:
+        case $.BUTTON:
         case $.CENTER:
         case $.FIGURE:
         case $.FOOTER:
@@ -2159,7 +2163,7 @@ function endTagInBody(p, token) {
             break;
         }
         case $.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         default: {
@@ -2325,7 +2329,7 @@ function endTagInTable(p, token) {
             break;
         }
         case $.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         case $.BODY:
@@ -2462,7 +2466,7 @@ function endTagInColumnGroup(p, token) {
             break;
         }
         case $.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         case $.COL: {
@@ -2757,7 +2761,7 @@ function endTagInSelect(p, token) {
             break;
         }
         case $.TEMPLATE: {
-            endTagInHead(p, token);
+            templateEndTagInHead(p, token);
             break;
         }
         default:
@@ -2855,7 +2859,7 @@ function startTagInTemplate(p, token) {
 }
 function endTagInTemplate(p, token) {
     if (token.tagID === $.TEMPLATE) {
-        endTagInHead(p, token);
+        templateEndTagInHead(p, token);
     }
 }
 function eofInTemplate(p, token) {

package/dist/parser/open-element-stack.d.ts

@@ -1,5 +1,5 @@
 import { TAG_ID as $ } from '../common/html.js';
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
 export interface StackHandler<T extends TreeAdapterTypeMap> {
     onItemPush: (node: T['parentNode'], tid: number, isTop: boolean) => void;
     onItemPop: (node: T['parentNode'], isTop: boolean) => void;

package/dist/serializer/index.d.ts

@@ -1,4 +1,4 @@
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
 import { type DefaultTreeAdapterMap } from '../tree-adapters/default.js';
 export interface SerializerOptions<T extends TreeAdapterTypeMap> {
     /**

package/dist/tokenizer/index.d.ts

@@ -78,10 +78,9 @@ declare const enum State {
     AMBIGUOUS_AMPERSAND = 73,
     NUMERIC_CHARACTER_REFERENCE = 74,
     HEXADEMICAL_CHARACTER_REFERENCE_START = 75,
-    DECIMAL_CHARACTER_REFERENCE_START = 76,
-    HEXADEMICAL_CHARACTER_REFERENCE = 77,
-    DECIMAL_CHARACTER_REFERENCE = 78,
-    NUMERIC_CHARACTER_REFERENCE_END = 79
+    HEXADEMICAL_CHARACTER_REFERENCE = 76,
+    DECIMAL_CHARACTER_REFERENCE = 77,
+    NUMERIC_CHARACTER_REFERENCE_END = 78
 }
 export declare const TokenizerMode: {
     readonly DATA: State.DATA;
@@ -241,7 +240,6 @@ export declare class Tokenizer {
     private _stateAmbiguousAmpersand;
     private _stateNumericCharacterReference;
     private _stateHexademicalCharacterReferenceStart;
-    private _stateDecimalCharacterReferenceStart;
     private _stateHexademicalCharacterReference;
     private _stateDecimalCharacterReference;
     private _stateNumericCharacterReferenceEnd;

package/dist/tokenizer/index.js

@@ -113,10 +113,9 @@ var State;
     State[State["AMBIGUOUS_AMPERSAND"] = 73] = "AMBIGUOUS_AMPERSAND";
     State[State["NUMERIC_CHARACTER_REFERENCE"] = 74] = "NUMERIC_CHARACTER_REFERENCE";
     State[State["HEXADEMICAL_CHARACTER_REFERENCE_START"] = 75] = "HEXADEMICAL_CHARACTER_REFERENCE_START";
-    State[State["DECIMAL_CHARACTER_REFERENCE_START"] = 76] = "DECIMAL_CHARACTER_REFERENCE_START";
-    State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 77] = "HEXADEMICAL_CHARACTER_REFERENCE";
-    State[State["DECIMAL_CHARACTER_REFERENCE"] = 78] = "DECIMAL_CHARACTER_REFERENCE";
-    State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 79] = "NUMERIC_CHARACTER_REFERENCE_END";
+    State[State["HEXADEMICAL_CHARACTER_REFERENCE"] = 76] = "HEXADEMICAL_CHARACTER_REFERENCE";
+    State[State["DECIMAL_CHARACTER_REFERENCE"] = 77] = "DECIMAL_CHARACTER_REFERENCE";
+    State[State["NUMERIC_CHARACTER_REFERENCE_END"] = 78] = "NUMERIC_CHARACTER_REFERENCE_END";
 })(State || (State = {}));
 //Tokenizer initial states for different modes
 export const TokenizerMode = {
@@ -274,9 +273,9 @@ export class Tokenizer {
         this.consumedAfterSnapshot -= count;
         this.preprocessor.retreat(count);
     }
-    _reconsumeInState(state) {
+    _reconsumeInState(state, cp) {
         this.state = state;
-        this._unconsume(1);
+        this._callState(cp);
     }
     _advanceBy(count) {
         this.consumedAfterSnapshot += count;
@@ -472,13 +471,11 @@ export class Tokenizer {
         this._createCharacterToken(type, ch);
     }
     _emitCodePoint(cp) {
-        let type = TokenType.CHARACTER;
-        if (isWhitespace(cp)) {
-            type = TokenType.WHITESPACE_CHARACTER;
-        }
-        else if (cp === $.NULL) {
-            type = TokenType.NULL_CHARACTER;
-        }
+        const type = isWhitespace(cp)
+            ? TokenType.WHITESPACE_CHARACTER
+            : cp === $.NULL
+                ? TokenType.NULL_CHARACTER
+                : TokenType.CHARACTER;
         this._appendCharToCurrentCharacterToken(type, String.fromCodePoint(cp));
     }
     //NOTE: used when we emit characters explicitly.
@@ -861,10 +858,6 @@ export class Tokenizer {
                 this._stateHexademicalCharacterReferenceStart(cp);
                 break;
             }
-            case State.DECIMAL_CHARACTER_REFERENCE_START: {
-                this._stateDecimalCharacterReferenceStart(cp);
-                break;
-            }
             case State.HEXADEMICAL_CHARACTER_REFERENCE: {
                 this._stateHexademicalCharacterReference(cp);
                 break;
@@ -874,7 +867,7 @@ export class Tokenizer {
                 break;
             }
             case State.NUMERIC_CHARACTER_REFERENCE_END: {
-                this._stateNumericCharacterReferenceEnd();
+                this._stateNumericCharacterReferenceEnd(cp);
                 break;
             }
             default: {
@@ -2772,7 +2765,7 @@ export class Tokenizer {
         }
         else {
             this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
-            this._reconsumeInState(this.returnState);
+            this._reconsumeInState(this.returnState, cp);
         }
     }
     // Named character reference state
@@ -2805,7 +2798,7 @@ export class Tokenizer {
             if (cp === $.SEMICOLON) {
                 this._err(ERR.unknownNamedCharacterReference);
             }
-            this._reconsumeInState(this.returnState);
+            this._reconsumeInState(this.returnState, cp);
         }
     }
     // Numeric character reference state
@@ -2815,9 +2808,16 @@ export class Tokenizer {
         if (cp === $.LATIN_SMALL_X || cp === $.LATIN_CAPITAL_X) {
             this.state = State.HEXADEMICAL_CHARACTER_REFERENCE_START;
         }
+        // Inlined decimal character reference start state
+        else if (isAsciiDigit(cp)) {
+            this.state = State.DECIMAL_CHARACTER_REFERENCE;
+            this._stateDecimalCharacterReference(cp);
+        }
         else {
-            this.state = State.DECIMAL_CHARACTER_REFERENCE_START;
-            this._stateDecimalCharacterReferenceStart(cp);
+            this._err(ERR.absenceOfDigitsInNumericCharacterReference);
+            this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
+            this._flushCodePointConsumedAsCharacterReference($.NUMBER_SIGN);
+            this._reconsumeInState(this.returnState, cp);
         }
     }
     // Hexademical character reference start state
@@ -2835,20 +2835,6 @@ export class Tokenizer {
             this.state = this.returnState;
         }
     }
-    // Decimal character reference start state
-    //------------------------------------------------------------------
-    _stateDecimalCharacterReferenceStart(cp) {
-        if (isAsciiDigit(cp)) {
-            this.state = State.DECIMAL_CHARACTER_REFERENCE;
-            this._stateDecimalCharacterReference(cp);
-        }
-        else {
-            this._err(ERR.absenceOfDigitsInNumericCharacterReference);
-            this._flushCodePointConsumedAsCharacterReference($.AMPERSAND);
-            this._flushCodePointConsumedAsCharacterReference($.NUMBER_SIGN);
-            this._reconsumeInState(this.returnState);
-        }
-    }
     // Hexademical character reference state
     //------------------------------------------------------------------
     _stateHexademicalCharacterReference(cp) {
@@ -2867,7 +2853,7 @@ export class Tokenizer {
         else {
             this._err(ERR.missingSemicolonAfterCharacterReference);
             this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
-            this._stateNumericCharacterReferenceEnd();
+            this._stateNumericCharacterReferenceEnd(cp);
         }
     }
     // Decimal character reference state
@@ -2882,12 +2868,12 @@ export class Tokenizer {
         else {
             this._err(ERR.missingSemicolonAfterCharacterReference);
             this.state = State.NUMERIC_CHARACTER_REFERENCE_END;
-            this._stateNumericCharacterReferenceEnd();
+            this._stateNumericCharacterReferenceEnd(cp);
         }
     }
     // Numeric character reference end state
     //------------------------------------------------------------------
-    _stateNumericCharacterReferenceEnd() {
+    _stateNumericCharacterReferenceEnd(cp) {
         if (this.charRefCode === $.NULL) {
             this._err(ERR.nullCharacterReference);
             this.charRefCode = $.REPLACEMENT_CHARACTER;
@@ -2911,7 +2897,7 @@ export class Tokenizer {
             }
         }
         this._flushCodePointConsumedAsCharacterReference(this.charRefCode);
-        this._reconsumeInState(this.returnState);
+        this._reconsumeInState(this.returnState, cp);
     }
 }
 //# sourceMappingURL=index.js.map

package/dist/tree-adapters/default.d.ts

@@ -1,16 +1,9 @@
 import { DOCUMENT_MODE, type NS } from '../common/html.js';
 import type { Attribute, Location, ElementLocation } from '../common/token.js';
 import type { TreeAdapter, TreeAdapterTypeMap } from './interface.js';
-export declare enum NodeType {
-    Document = "#document",
-    DocumentFragment = "#document-fragment",
-    Comment = "#comment",
-    Text = "#text",
-    DocumentType = "#documentType"
-}
 export interface Document {
     /** The name of the node. */
-    nodeName: NodeType.Document;
+    nodeName: '#document';
     /**
      * Document mode.
      *
@@ -23,7 +16,7 @@ export interface Document {
 }
 export interface DocumentFragment {
     /** The name of the node. */
-    nodeName: NodeType.DocumentFragment;
+    nodeName: '#document-fragment';
     /** The node's children. */
     childNodes: ChildNode[];
     /** Comment source code location info. Available if location info is enabled. */
@@ -47,7 +40,7 @@ export interface Element {
 }
 export interface CommentNode {
     /** The name of the node. */
-    nodeName: NodeType.Comment;
+    nodeName: '#comment';
     /** Parent node. */
     parentNode: ParentNode | null;
     /** Comment text. */
@@ -56,7 +49,7 @@ export interface CommentNode {
     sourceCodeLocation?: Location | null;
 }
 export interface TextNode {
-    nodeName: NodeType.Text;
+    nodeName: '#text';
     /** Parent node. */
     parentNode: ParentNode | null;
     /** Text content. */
@@ -72,7 +65,7 @@ export interface Template extends Element {
 }
 export interface DocumentType {
     /** The name of the node. */
-    nodeName: NodeType.DocumentType;
+    nodeName: '#documentType';
     /** Parent node. */
     parentNode: ParentNode | null;
     /** Document type name. */

package/dist/tree-adapters/default.js

@@ -1,15 +1,7 @@
 import { DOCUMENT_MODE } from '../common/html.js';
-export var NodeType;
-(function (NodeType) {
-    NodeType["Document"] = "#document";
-    NodeType["DocumentFragment"] = "#document-fragment";
-    NodeType["Comment"] = "#comment";
-    NodeType["Text"] = "#text";
-    NodeType["DocumentType"] = "#documentType";
-})(NodeType || (NodeType = {}));
 function createTextNode(value) {
     return {
-        nodeName: NodeType.Text,
+        nodeName: '#text',
         value,
         parentNode: null,
     };
@@ -18,14 +10,14 @@ export const defaultTreeAdapter = {
     //Node construction
     createDocument() {
         return {
-            nodeName: NodeType.Document,
+            nodeName: '#document',
             mode: DOCUMENT_MODE.NO_QUIRKS,
             childNodes: [],
         };
     },
     createDocumentFragment() {
         return {
-            nodeName: NodeType.DocumentFragment,
+            nodeName: '#document-fragment',
             childNodes: [],
         };
     },
@@ -41,7 +33,7 @@ export const defaultTreeAdapter = {
     },
     createCommentNode(data) {
         return {
-            nodeName: NodeType.Comment,
+            nodeName: '#comment',
             data,
             parentNode: null,
         };
@@ -63,7 +55,7 @@ export const defaultTreeAdapter = {
         return templateElement.content;
     },
     setDocumentType(document, name, publicId, systemId) {
-        const doctypeNode = document.childNodes.find((node) => node.nodeName === NodeType.DocumentType);
+        const doctypeNode = document.childNodes.find((node) => node.nodeName === '#documentType');
         if (doctypeNode) {
             doctypeNode.name = name;
             doctypeNode.publicId = publicId;
@@ -71,7 +63,7 @@ export const defaultTreeAdapter = {
         }
         else {
             const node = {
-                nodeName: NodeType.DocumentType,
+                nodeName: '#documentType',
                 name,
                 publicId,
                 systemId,
@@ -163,7 +155,7 @@ export const defaultTreeAdapter = {
         return node.nodeName === '#comment';
     },
     isDocumentTypeNode(node) {
-        return node.nodeName === NodeType.DocumentType;
+        return node.nodeName === '#documentType';
     },
     isElementNode(node) {
         return Object.prototype.hasOwnProperty.call(node, 'tagName');

package/package.json

@@ -2,13 +2,13 @@
     "name": "parse5",
     "type": "module",
     "description": "HTML parser and serializer.",
-    "version": "7.0.0",
+    "version": "7.1.0",
     "author": "Ivan Nikulin <ifaaan@gmail.com> (https://github.com/inikulin)",
     "contributors": "https://github.com/inikulin/parse5/graphs/contributors",
     "homepage": "https://github.com/inikulin/parse5",
     "funding": "https://github.com/inikulin/parse5?sponsor=1",
     "dependencies": {
-        "entities": "^4.3.0"
+        "entities": "^4.4.0"
     },
     "keywords": [
         "html",
@@ -43,7 +43,7 @@
         "url": "git://github.com/inikulin/parse5.git"
     },
     "files": [
-        "dist",
-        "!*.map"
+        "dist/**/*.js",
+        "dist/**/*.d.ts"
     ]
 }

package/dist/cjs/package.json

@@ -1 +0,0 @@
-{"type":"commonjs"}