Socket
Socket
Sign inDemoInstall

htmlparser2

Package Overview
Dependencies
5
Maintainers
1
Versions
76
Alerts
File Explorer

Advanced tools

npm Scripts

Install Socket

Detect and block malicious and high-risk dependencies

Install

Comparing version 6.1.0 to 7.0.0

14

lib/FeedHandler.d.ts

@@ -1,2 +0,2 @@

import DomHandler, { DomHandlerOptions } from "domhandler";
import DomHandler, { DomHandlerOptions, Node } from "domhandler";
import { ParserOptions } from "./Parser";

@@ -49,2 +49,3 @@ declare enum FeedItemMediaMedium {

}
/** @deprecated Handler is no longer necessary; use `getFeed` or `parseFeed` instead. */
export declare class FeedHandler extends DomHandler {

@@ -61,9 +62,16 @@ feed?: Feed;

/**
* Get the feed object from the root of a DOM tree.
*
* @param dom - The DOM to to extract the feed from.
* @returns The feed.
*/
export declare function getFeed(dom: Node[]): Feed | null;
/**
* Parse a feed.
*
* @param feed The feed that should be parsed, as a string.
* @param options Optionally, options for parsing. When using this option, you should set `xmlMode` to `true`.
* @param options Optionally, options for parsing. When using this, you should set `xmlMode` to `true`.
*/
export declare function parseFeed(feed: string, options?: ParserOptions & DomHandlerOptions): Feed | undefined;
export declare function parseFeed(feed: string, options?: ParserOptions & DomHandlerOptions): Feed | null;
export {};
//# sourceMappingURL=FeedHandler.d.ts.map

162

lib/FeedHandler.js

@@ -40,3 +40,3 @@ "use strict";

Object.defineProperty(exports, "__esModule", { value: true });
exports.parseFeed = exports.FeedHandler = void 0;
exports.parseFeed = exports.getFeed = exports.FeedHandler = void 0;
var domhandler_1 = __importDefault(require("domhandler"));

@@ -59,3 +59,3 @@ var DomUtils = __importStar(require("domutils"));

})(FeedItemMediaExpression || (FeedItemMediaExpression = {}));
// TODO: Consume data as it is coming in
/** @deprecated Handler is no longer necessary; use `getFeed` or `parseFeed` instead. */
var FeedHandler = /** @class */ (function (_super) {

@@ -78,77 +78,91 @@ __extends(FeedHandler, _super);

FeedHandler.prototype.onend = function () {
var _a, _b;
var feedRoot = getOneElement(isValidFeed, this.dom);
if (!feedRoot) {
var feed = getFeed(this.dom);
if (feed) {
this.feed = feed;
this.handleCallback(null);
}
else {
this.handleCallback(new Error("couldn't find root of feed"));
return;
}
var feed = {};
if (feedRoot.name === "feed") {
var childs = feedRoot.children;
feed.type = "atom";
addConditionally(feed, "id", "id", childs);
addConditionally(feed, "title", "title", childs);
var href = getAttribute("href", getOneElement("link", childs));
};
return FeedHandler;
}(domhandler_1.default));
exports.FeedHandler = FeedHandler;
/**
* Get the feed object from the root of a DOM tree.
*
* @param dom - The DOM to to extract the feed from.
* @returns The feed.
*/
function getFeed(dom) {
var _a, _b;
var feedRoot = getOneElement(isValidFeed, dom);
if (!feedRoot)
return null;
var feed = {};
if (feedRoot.name === "feed") {
var childs = feedRoot.children;
feed.type = "atom";
addConditionally(feed, "id", "id", childs);
addConditionally(feed, "title", "title", childs);
var href = getAttribute("href", getOneElement("link", childs));
if (href) {
feed.link = href;
}
addConditionally(feed, "description", "subtitle", childs);
var updated = fetch("updated", childs);
if (updated) {
feed.updated = new Date(updated);
}
addConditionally(feed, "author", "email", childs, true);
feed.items = getElements("entry", childs).map(function (item) {
var entry = {};
var children = item.children;
addConditionally(entry, "id", "id", children);
addConditionally(entry, "title", "title", children);
var href = getAttribute("href", getOneElement("link", children));
if (href) {
feed.link = href;
entry.link = href;
}
addConditionally(feed, "description", "subtitle", childs);
var updated = fetch("updated", childs);
if (updated) {
feed.updated = new Date(updated);
var description = fetch("summary", children) || fetch("content", children);
if (description) {
entry.description = description;
}
addConditionally(feed, "author", "email", childs, true);
feed.items = getElements("entry", childs).map(function (item) {
var entry = {};
var children = item.children;
addConditionally(entry, "id", "id", children);
addConditionally(entry, "title", "title", children);
var href = getAttribute("href", getOneElement("link", children));
if (href) {
entry.link = href;
}
var description = fetch("summary", children) || fetch("content", children);
if (description) {
entry.description = description;
}
var pubDate = fetch("updated", children);
if (pubDate) {
entry.pubDate = new Date(pubDate);
}
entry.media = getMediaElements(children);
return entry;
});
}
else {
var childs = (_b = (_a = getOneElement("channel", feedRoot.children)) === null || _a === void 0 ? void 0 : _a.children) !== null && _b !== void 0 ? _b : [];
feed.type = feedRoot.name.substr(0, 3);
feed.id = "";
addConditionally(feed, "title", "title", childs);
addConditionally(feed, "link", "link", childs);
addConditionally(feed, "description", "description", childs);
var updated = fetch("lastBuildDate", childs);
if (updated) {
feed.updated = new Date(updated);
var pubDate = fetch("updated", children);
if (pubDate) {
entry.pubDate = new Date(pubDate);
}
addConditionally(feed, "author", "managingEditor", childs, true);
feed.items = getElements("item", feedRoot.children).map(function (item) {
var entry = {};
var children = item.children;
addConditionally(entry, "id", "guid", children);
addConditionally(entry, "title", "title", children);
addConditionally(entry, "link", "link", children);
addConditionally(entry, "description", "description", children);
var pubDate = fetch("pubDate", children);
if (pubDate)
entry.pubDate = new Date(pubDate);
entry.media = getMediaElements(children);
return entry;
});
entry.media = getMediaElements(children);
return entry;
});
}
else {
var childs = (_b = (_a = getOneElement("channel", feedRoot.children)) === null || _a === void 0 ? void 0 : _a.children) !== null && _b !== void 0 ? _b : [];
feed.type = feedRoot.name.substr(0, 3);
feed.id = "";
addConditionally(feed, "title", "title", childs);
addConditionally(feed, "link", "link", childs);
addConditionally(feed, "description", "description", childs);
var updated = fetch("lastBuildDate", childs);
if (updated) {
feed.updated = new Date(updated);
}
this.feed = feed;
this.handleCallback(null);
};
return FeedHandler;
}(domhandler_1.default));
exports.FeedHandler = FeedHandler;
addConditionally(feed, "author", "managingEditor", childs, true);
feed.items = getElements("item", feedRoot.children).map(function (item) {
var entry = {};
var children = item.children;
addConditionally(entry, "id", "guid", children);
addConditionally(entry, "title", "title", children);
addConditionally(entry, "link", "link", children);
addConditionally(entry, "description", "description", children);
var pubDate = fetch("pubDate", children);
if (pubDate)
entry.pubDate = new Date(pubDate);
entry.media = getMediaElements(children);
return entry;
});
}
return feed;
}
exports.getFeed = getFeed;
function getMediaElements(where) {

@@ -208,3 +222,3 @@ return getElements("media:content", where).map(function (elem) {

if (recurse === void 0) { recurse = false; }
return DomUtils.getText(DomUtils.getElementsByTagName(tagName, where, recurse, 1)).trim();
return DomUtils.textContent(DomUtils.getElementsByTagName(tagName, where, recurse, 1)).trim();
}

@@ -231,10 +245,10 @@ function getAttribute(name, elem) {

* @param feed The feed that should be parsed, as a string.
* @param options Optionally, options for parsing. When using this option, you should set `xmlMode` to `true`.
* @param options Optionally, options for parsing. When using this, you should set `xmlMode` to `true`.
*/
function parseFeed(feed, options) {
if (options === void 0) { options = { xmlMode: true }; }
var handler = new FeedHandler(options);
var handler = new domhandler_1.default(null, options);
new Parser_1.Parser(handler, options).end(feed);
return handler.feed;
return getFeed(handler.dom);
}
exports.parseFeed = parseFeed;

7

lib/Parser.d.ts

@@ -80,6 +80,7 @@ import Tokenizer from "./Tokenizer";

export declare class Parser {
private readonly options;
/** The start index of the last event. */
startIndex: number;
/** The end index of the last event. */
endIndex: number | null;
endIndex: number;
private tagname;

@@ -92,9 +93,9 @@ private attribname;

private readonly cbs;
private readonly options;
private readonly lowerCaseTagNames;
private readonly lowerCaseAttributeNames;
private readonly tokenizer;
constructor(cbs: Partial<Handler> | null, options?: ParserOptions);
constructor(cbs?: Partial<Handler> | null, options?: ParserOptions);
private updatePosition;
ontext(data: string): void;
protected isVoidElement(name: string): boolean;
onopentagname(name: string): void;

@@ -101,0 +102,0 @@ onopentagend(): void;

153

lib/Parser.js

@@ -18,51 +18,54 @@ "use strict";

var pTag = new Set(["p"]);
var openImpliesClose = {
tr: new Set(["tr", "th", "td"]),
th: new Set(["th"]),
td: new Set(["thead", "th", "td"]),
body: new Set(["head", "link", "script"]),
li: new Set(["li"]),
p: pTag,
h1: pTag,
h2: pTag,
h3: pTag,
h4: pTag,
h5: pTag,
h6: pTag,
select: formTags,
input: formTags,
output: formTags,
button: formTags,
datalist: formTags,
textarea: formTags,
option: new Set(["option"]),
optgroup: new Set(["optgroup", "option"]),
dd: new Set(["dt", "dd"]),
dt: new Set(["dt", "dd"]),
address: pTag,
article: pTag,
aside: pTag,
blockquote: pTag,
details: pTag,
div: pTag,
dl: pTag,
fieldset: pTag,
figcaption: pTag,
figure: pTag,
footer: pTag,
form: pTag,
header: pTag,
hr: pTag,
main: pTag,
nav: pTag,
ol: pTag,
pre: pTag,
section: pTag,
table: pTag,
ul: pTag,
rt: new Set(["rt", "rp"]),
rp: new Set(["rt", "rp"]),
tbody: new Set(["thead", "tbody"]),
tfoot: new Set(["thead", "tbody"]),
};
var tableSectionTags = new Set(["thead", "tbody"]);
var ddtTags = new Set(["dd", "dt"]);
var rtpTags = new Set(["rt", "rp"]);
var openImpliesClose = new Map([
["tr", new Set(["tr", "th", "td"])],
["th", new Set(["th"])],
["td", new Set(["thead", "th", "td"])],
["body", new Set(["head", "link", "script"])],
["li", new Set(["li"])],
["p", pTag],
["h1", pTag],
["h2", pTag],
["h3", pTag],
["h4", pTag],
["h5", pTag],
["h6", pTag],
["select", formTags],
["input", formTags],
["output", formTags],
["button", formTags],
["datalist", formTags],
["textarea", formTags],
["option", new Set(["option"])],
["optgroup", new Set(["optgroup", "option"])],
["dd", ddtTags],
["dt", ddtTags],
["address", pTag],
["article", pTag],
["aside", pTag],
["blockquote", pTag],
["details", pTag],
["div", pTag],
["dl", pTag],
["fieldset", pTag],
["figcaption", pTag],
["figure", pTag],
["footer", pTag],
["form", pTag],
["header", pTag],
["hr", pTag],
["main", pTag],
["nav", pTag],
["ol", pTag],
["pre", pTag],
["section", pTag],
["table", pTag],
["ul", pTag],
["rt", rtpTags],
["rp", rtpTags],
["tbody", tableSectionTags],
["tfoot", tableSectionTags],
]);
var voidElements = new Set([

@@ -106,6 +109,7 @@ "area",

var _a, _b, _c, _d, _e;
this.options = options;
/** The start index of the last event. */
this.startIndex = 0;
/** The end index of the last event. */
this.endIndex = null;
this.endIndex = 0;
this.tagname = "";

@@ -125,14 +129,4 @@ this.attribname = "";

}
Parser.prototype.updatePosition = function (initialOffset) {
if (this.endIndex === null) {
if (this.tokenizer.sectionStart <= initialOffset) {
this.startIndex = 0;
}
else {
this.startIndex = this.tokenizer.sectionStart - initialOffset;
}
}
else {
this.startIndex = this.endIndex + 1;
}
Parser.prototype.updatePosition = function (offset) {
this.startIndex = this.tokenizer.getAbsoluteSectionStart() - offset;
this.endIndex = this.tokenizer.getAbsoluteIndex();

@@ -143,8 +137,12 @@ };

var _a, _b;
this.updatePosition(1);
this.endIndex--;
this.startIndex = this.tokenizer.getAbsoluteSectionStart();
this.endIndex = this.tokenizer.getAbsoluteIndex() - 1;
(_b = (_a = this.cbs).ontext) === null || _b === void 0 ? void 0 : _b.call(_a, data);
};
Parser.prototype.isVoidElement = function (name) {
return !this.options.xmlMode && voidElements.has(name);
};
Parser.prototype.onopentagname = function (name) {
var _a, _b;
var _a, _b, _c, _d;
this.updatePosition(1);
if (this.lowerCaseTagNames) {

@@ -154,11 +152,11 @@ name = name.toLowerCase();

this.tagname = name;
if (!this.options.xmlMode &&
Object.prototype.hasOwnProperty.call(openImpliesClose, name)) {
var el = void 0;
var impliesClose = !this.options.xmlMode && openImpliesClose.get(name);
if (impliesClose) {
while (this.stack.length > 0 &&
openImpliesClose[name].has((el = this.stack[this.stack.length - 1]))) {
this.onclosetag(el);
impliesClose.has(this.stack[this.stack.length - 1])) {
var el = this.stack.pop();
(_b = (_a = this.cbs).onclosetag) === null || _b === void 0 ? void 0 : _b.call(_a, el);
}
}
if (this.options.xmlMode || !voidElements.has(name)) {
if (!this.isVoidElement(name)) {
this.stack.push(name);

@@ -172,3 +170,3 @@ if (foreignContextElements.has(name)) {

}
(_b = (_a = this.cbs).onopentagname) === null || _b === void 0 ? void 0 : _b.call(_a, name);
(_d = (_c = this.cbs).onopentagname) === null || _d === void 0 ? void 0 : _d.call(_c, name);
if (this.cbs.onopentag)

@@ -179,3 +177,3 @@ this.attribs = {};

var _a, _b;
this.updatePosition(1);
this.endIndex = this.tokenizer.getAbsoluteIndex();
if (this.attribs) {

@@ -185,5 +183,3 @@ (_b = (_a = this.cbs).onopentag) === null || _b === void 0 ? void 0 : _b.call(_a, this.tagname, this.attribs);

}
if (!this.options.xmlMode &&
this.cbs.onclosetag &&
voidElements.has(this.tagname)) {
if (this.cbs.onclosetag && this.isVoidElement(this.tagname)) {
this.cbs.onclosetag(this.tagname);

@@ -202,4 +198,3 @@ }

}
if (this.stack.length &&
(this.options.xmlMode || !voidElements.has(name))) {
if (this.stack.length && !this.isVoidElement(name)) {
var pos = this.stack.lastIndexOf(name);

@@ -279,2 +274,3 @@ if (pos !== -1) {

if (this.cbs.onprocessinginstruction) {
this.updatePosition(2);
var name_1 = this.getInstructionName(value);

@@ -286,2 +282,3 @@ this.cbs.onprocessinginstruction("!" + name_1, "!" + value);

if (this.cbs.onprocessinginstruction) {
this.updatePosition(2);
var name_2 = this.getInstructionName(value);

@@ -288,0 +285,0 @@ this.cbs.onprocessinginstruction("?" + name_2, "?" + value);

17

lib/Tokenizer.d.ts

@@ -87,2 +87,3 @@ /** All the states the tokenizer can be in. */

export default class Tokenizer {
private readonly cbs;
/** The current state the tokenizer is in. */

@@ -109,5 +110,5 @@ _state: State;

private ended;
private readonly cbs;
private readonly xmlMode;
private readonly decodeEntities;
private readonly entityTrie;
constructor(options: {

@@ -123,2 +124,6 @@ xmlMode?: boolean;

/**
* The start of the current section.
*/
getAbsoluteSectionStart(): number;
/**
* The current index within all of the written data.

@@ -165,8 +170,13 @@ */

private stateAfterSpecialLast;
private parseFixedEntity;
private parseLegacyEntity;
private trieIndex;
private trieCurrent;
private trieResult;
private trieExcess;
private stateBeforeEntity;
private stateInNamedEntity;
private emitNamedEntity;
private decodeNumericEntity;
private stateInNumericEntity;
private stateInHexEntity;
private allowLegacyEntity;
private cleanup;

@@ -182,3 +192,2 @@ /**

private getSection;
private emitToken;
private emitPartial;

@@ -185,0 +194,0 @@ }

372

lib/Tokenizer.js

@@ -7,26 +7,19 @@ "use strict";

var decode_codepoint_1 = __importDefault(require("entities/lib/decode_codepoint"));
var entities_json_1 = __importDefault(require("entities/lib/maps/entities.json"));
var legacy_json_1 = __importDefault(require("entities/lib/maps/legacy.json"));
var xml_json_1 = __importDefault(require("entities/lib/maps/xml.json"));
var decode_1 = require("entities/lib/decode");
function whitespace(c) {
return c === " " || c === "\n" || c === "\t" || c === "\f" || c === "\r";
return (c === 32 /* Space */ ||
c === 10 /* NewLine */ ||
c === 9 /* Tab */ ||
c === 12 /* FormFeed */ ||
c === 13 /* CarriageReturn */);
}
function isASCIIAlpha(c) {
return (c >= "a" && c <= "z") || (c >= "A" && c <= "Z");
return ((c >= 97 /* LowerA */ && c <= 122 /* LowerZ */) ||
(c >= 65 /* UpperA */ && c <= 90 /* UpperZ */));
}
function ifElseState(upper, SUCCESS, FAILURE) {
var lower = upper.toLowerCase();
if (upper === lower) {
return function (t, c) {
if (c === lower) {
t._state = SUCCESS;
}
else {
t._state = FAILURE;
t._index--;
}
};
}
var upperCode = upper.charCodeAt(0);
var lowerCode = upper.toLowerCase().charCodeAt(0);
return function (t, c) {
if (c === lower || c === upper) {
if (c === lowerCode || c === upperCode) {
t._state = SUCCESS;

@@ -40,14 +33,2 @@ }

}
function consumeSpecialNameChar(upper, NEXT_STATE) {
var lower = upper.toLowerCase();
return function (t, c) {
if (c === lower || c === upper) {
t._state = NEXT_STATE;
}
else {
t._state = 3 /* InTagName */;
t._index--; // Consume the token again
}
};
}
var stateBeforeCdata1 = ifElseState("C", 24 /* BeforeCdata2 */, 16 /* InDeclaration */);

@@ -58,6 +39,6 @@ var stateBeforeCdata2 = ifElseState("D", 25 /* BeforeCdata3 */, 16 /* InDeclaration */);

var stateBeforeCdata5 = ifElseState("A", 28 /* BeforeCdata6 */, 16 /* InDeclaration */);
var stateBeforeScript1 = consumeSpecialNameChar("R", 35 /* BeforeScript2 */);
var stateBeforeScript2 = consumeSpecialNameChar("I", 36 /* BeforeScript3 */);
var stateBeforeScript3 = consumeSpecialNameChar("P", 37 /* BeforeScript4 */);
var stateBeforeScript4 = consumeSpecialNameChar("T", 38 /* BeforeScript5 */);
var stateBeforeScript1 = ifElseState("R", 35 /* BeforeScript2 */, 3 /* InTagName */);
var stateBeforeScript2 = ifElseState("I", 36 /* BeforeScript3 */, 3 /* InTagName */);
var stateBeforeScript3 = ifElseState("P", 37 /* BeforeScript4 */, 3 /* InTagName */);
var stateBeforeScript4 = ifElseState("T", 38 /* BeforeScript5 */, 3 /* InTagName */);
var stateAfterScript1 = ifElseState("R", 40 /* AfterScript2 */, 1 /* Text */);

@@ -67,17 +48,16 @@ var stateAfterScript2 = ifElseState("I", 41 /* AfterScript3 */, 1 /* Text */);

var stateAfterScript4 = ifElseState("T", 43 /* AfterScript5 */, 1 /* Text */);
var stateBeforeStyle1 = consumeSpecialNameChar("Y", 45 /* BeforeStyle2 */);
var stateBeforeStyle2 = consumeSpecialNameChar("L", 46 /* BeforeStyle3 */);
var stateBeforeStyle3 = consumeSpecialNameChar("E", 47 /* BeforeStyle4 */);
var stateBeforeStyle1 = ifElseState("Y", 45 /* BeforeStyle2 */, 3 /* InTagName */);
var stateBeforeStyle2 = ifElseState("L", 46 /* BeforeStyle3 */, 3 /* InTagName */);
var stateBeforeStyle3 = ifElseState("E", 47 /* BeforeStyle4 */, 3 /* InTagName */);
var stateAfterStyle1 = ifElseState("Y", 49 /* AfterStyle2 */, 1 /* Text */);
var stateAfterStyle2 = ifElseState("L", 50 /* AfterStyle3 */, 1 /* Text */);
var stateAfterStyle3 = ifElseState("E", 51 /* AfterStyle4 */, 1 /* Text */);
var stateBeforeSpecialT = consumeSpecialNameChar("I", 54 /* BeforeTitle1 */);
var stateBeforeTitle1 = consumeSpecialNameChar("T", 55 /* BeforeTitle2 */);
var stateBeforeTitle2 = consumeSpecialNameChar("L", 56 /* BeforeTitle3 */);
var stateBeforeTitle3 = consumeSpecialNameChar("E", 57 /* BeforeTitle4 */);
var stateAfterSpecialTEnd = ifElseState("I", 58 /* AfterTitle1 */, 1 /* Text */);
var stateBeforeSpecialT = ifElseState("I", 54 /* BeforeTitle1 */, 3 /* InTagName */);
var stateBeforeTitle1 = ifElseState("T", 55 /* BeforeTitle2 */, 3 /* InTagName */);
var stateBeforeTitle2 = ifElseState("L", 56 /* BeforeTitle3 */, 3 /* InTagName */);
var stateBeforeTitle3 = ifElseState("E", 57 /* BeforeTitle4 */, 3 /* InTagName */);
var stateBeforeSpecialTEnd = ifElseState("I", 58 /* AfterTitle1 */, 1 /* Text */);
var stateAfterTitle1 = ifElseState("T", 59 /* AfterTitle2 */, 1 /* Text */);
var stateAfterTitle2 = ifElseState("L", 60 /* AfterTitle3 */, 1 /* Text */);
var stateAfterTitle3 = ifElseState("E", 61 /* AfterTitle4 */, 1 /* Text */);
var stateBeforeEntity = ifElseState("#", 63 /* BeforeNumericEntity */, 64 /* InNamedEntity */);
var stateBeforeNumericEntity = ifElseState("X", 66 /* InHexEntity */, 65 /* InNumericEntity */);

@@ -87,2 +67,3 @@ var Tokenizer = /** @class */ (function () {

var _a;
this.cbs = cbs;
/** The current state the tokenizer is in. */

@@ -109,5 +90,10 @@ this._state = 1 /* Text */;

this.ended = false;
this.trieIndex = 0;
this.trieCurrent = 0;
this.trieResult = null;
this.trieExcess = 0;
this.cbs = cbs;
this.xmlMode = !!(options === null || options === void 0 ? void 0 : options.xmlMode);
this.decodeEntities = (_a = options === null || options === void 0 ? void 0 : options.decodeEntities) !== null && _a !== void 0 ? _a : true;
this.entityTrie = this.xmlMode ? decode_1.xmlDecodeTree : decode_1.htmlDecodeTree;
}

@@ -153,2 +139,8 @@ Tokenizer.prototype.reset = function () {

/**
* The start of the current section.
*/
Tokenizer.prototype.getAbsoluteSectionStart = function () {
return this.sectionStart + this.bufferOffset;
};
/**
* The current index within all of the written data.

@@ -160,3 +152,3 @@ */

Tokenizer.prototype.stateText = function (c) {
if (c === "<") {
if (c === 60 /* Lt */) {
if (this._index > this.sectionStart) {

@@ -169,3 +161,3 @@ this.cbs.ontext(this.getSection());

else if (this.decodeEntities &&
c === "&" &&
c === 38 /* Amp */ &&
(this.special === 1 /* None */ || this.special === 4 /* Title */)) {

@@ -188,13 +180,16 @@ if (this._index > this.sectionStart) {

return (isASCIIAlpha(c) ||
(this.xmlMode && !whitespace(c) && c !== "/" && c !== ">"));
(this.xmlMode &&
!whitespace(c) &&
c !== 47 /* Slash */ &&
c !== 62 /* Gt */));
};
Tokenizer.prototype.stateBeforeTagName = function (c) {
if (c === "/") {
if (c === 47 /* Slash */) {
this._state = 5 /* BeforeClosingTagName */;
}
else if (c === "<") {
else if (c === 60 /* Lt */) {
this.cbs.ontext(this.getSection());
this.sectionStart = this._index;
}
else if (c === ">" ||
else if (c === 62 /* Gt */ ||
this.special !== 1 /* None */ ||

@@ -204,7 +199,7 @@ whitespace(c)) {

}
else if (c === "!") {
else if (c === 33 /* ExclamationMark */) {
this._state = 15 /* BeforeDeclaration */;
this.sectionStart = this._index + 1;
}
else if (c === "?") {
else if (c === 63 /* Questionmark */) {
this._state = 17 /* InProcessingInstruction */;

@@ -218,5 +213,7 @@ this.sectionStart = this._index + 1;

this._state =
!this.xmlMode && (c === "s" || c === "S")
!this.xmlMode &&
(c === 115 /* LowerS */ || c === 83 /* UpperS */)
? 32 /* BeforeSpecialS */
: !this.xmlMode && (c === "t" || c === "T")
: !this.xmlMode &&
(c === 116 /* LowerT */ || c === 84 /* UpperT */)
? 52 /* BeforeSpecialT */

@@ -228,6 +225,7 @@ : 3 /* InTagName */;

Tokenizer.prototype.stateInTagName = function (c) {
if (c === "/" || c === ">" || whitespace(c)) {
this.emitToken("onopentagname");
if (c === 47 /* Slash */ || c === 62 /* Gt */ || whitespace(c)) {
this.cbs.onopentagname(this.getSection());
this.sectionStart = -1;
this._state = 8 /* BeforeAttributeName */;
this._index--;
this.stateBeforeAttributeName(c);
}

@@ -239,11 +237,12 @@ };

}
else if (c === ">") {
else if (c === 62 /* Gt */) {
this._state = 1 /* Text */;
}
else if (this.special !== 1 /* None */) {
if (this.special !== 4 /* Title */ && (c === "s" || c === "S")) {
if (this.special !== 4 /* Title */ &&
(c === 115 /* LowerS */ || c === 83 /* UpperS */)) {
this._state = 33 /* BeforeSpecialSEnd */;
}
else if (this.special === 4 /* Title */ &&
(c === "t" || c === "T")) {
(c === 116 /* LowerT */ || c === 84 /* UpperT */)) {
this._state = 53 /* BeforeSpecialTEnd */;

@@ -253,3 +252,3 @@ }

this._state = 1 /* Text */;
this._index--;
this.stateText(c);
}

@@ -267,6 +266,7 @@ }

Tokenizer.prototype.stateInClosingTagName = function (c) {
if (c === ">" || whitespace(c)) {
this.emitToken("onclosetag");
if (c === 62 /* Gt */ || whitespace(c)) {
this.cbs.onclosetag(this.getSection());
this.sectionStart = -1;
this._state = 7 /* AfterClosingTagName */;
this._index--;
this.stateAfterClosingTagName(c);
}

@@ -276,3 +276,3 @@ };

// Skip everything until ">"
if (c === ">") {
if (c === 62 /* Gt */) {
this._state = 1 /* Text */;

@@ -283,3 +283,3 @@ this.sectionStart = this._index + 1;

Tokenizer.prototype.stateBeforeAttributeName = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
this.cbs.onopentagend();

@@ -289,3 +289,3 @@ this._state = 1 /* Text */;

}
else if (c === "/") {
else if (c === 47 /* Slash */) {
this._state = 4 /* InSelfClosingTag */;

@@ -299,3 +299,3 @@ }

Tokenizer.prototype.stateInSelfClosingTag = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
this.cbs.onselfclosingtag();

@@ -308,21 +308,24 @@ this._state = 1 /* Text */;

this._state = 8 /* BeforeAttributeName */;
this._index--;
this.stateBeforeAttributeName(c);
}
};
Tokenizer.prototype.stateInAttributeName = function (c) {
if (c === "=" || c === "/" || c === ">" || whitespace(c)) {
if (c === 61 /* Eq */ ||
c === 47 /* Slash */ ||
c === 62 /* Gt */ ||
whitespace(c)) {
this.cbs.onattribname(this.getSection());
this.sectionStart = -1;
this._state = 10 /* AfterAttributeName */;
this._index--;
this.stateAfterAttributeName(c);
}
};
Tokenizer.prototype.stateAfterAttributeName = function (c) {
if (c === "=") {
if (c === 61 /* Eq */) {
this._state = 11 /* BeforeAttributeValue */;
}
else if (c === "/" || c === ">") {
else if (c === 47 /* Slash */ || c === 62 /* Gt */) {
this.cbs.onattribend(undefined);
this._state = 8 /* BeforeAttributeName */;
this._index--;
this.stateBeforeAttributeName(c);
}

@@ -336,7 +339,7 @@ else if (!whitespace(c)) {

Tokenizer.prototype.stateBeforeAttributeValue = function (c) {
if (c === '"') {
if (c === 34 /* DoubleQuote */) {
this._state = 12 /* InAttributeValueDq */;
this.sectionStart = this._index + 1;
}
else if (c === "'") {
else if (c === 39 /* SingleQuote */) {
this._state = 13 /* InAttributeValueSq */;

@@ -346,5 +349,5 @@ this.sectionStart = this._index + 1;

else if (!whitespace(c)) {
this.sectionStart = this._index;
this._state = 14 /* InAttributeValueNq */;
this.sectionStart = this._index;
this._index--; // Reconsume token
this.stateInAttributeValueNoQuotes(c); // Reconsume token
}

@@ -354,8 +357,9 @@ };

if (c === quote) {
this.emitToken("onattribdata");
this.cbs.onattribend(quote);
this.cbs.onattribdata(this.getSection());
this.sectionStart = -1;
this.cbs.onattribend(String.fromCharCode(quote));
this._state = 8 /* BeforeAttributeName */;
}
else if (this.decodeEntities && c === "&") {
this.emitToken("onattribdata");
else if (this.decodeEntities && c === 38 /* Amp */) {
this.cbs.onattribdata(this.getSection());
this.baseState = this._state;

@@ -367,16 +371,17 @@ this._state = 62 /* BeforeEntity */;

Tokenizer.prototype.stateInAttributeValueDoubleQuotes = function (c) {
this.handleInAttributeValue(c, '"');
this.handleInAttributeValue(c, 34 /* DoubleQuote */);
};
Tokenizer.prototype.stateInAttributeValueSingleQuotes = function (c) {
this.handleInAttributeValue(c, "'");
this.handleInAttributeValue(c, 39 /* SingleQuote */);
};
Tokenizer.prototype.stateInAttributeValueNoQuotes = function (c) {
if (whitespace(c) || c === ">") {
this.emitToken("onattribdata");
if (whitespace(c) || c === 62 /* Gt */) {
this.cbs.onattribdata(this.getSection());
this.sectionStart = -1;
this.cbs.onattribend(null);
this._state = 8 /* BeforeAttributeName */;
this._index--;
this.stateBeforeAttributeName(c);
}
else if (this.decodeEntities && c === "&") {
this.emitToken("onattribdata");
else if (this.decodeEntities && c === 38 /* Amp */) {
this.cbs.onattribdata(this.getSection());
this.baseState = this._state;

@@ -389,5 +394,5 @@ this._state = 62 /* BeforeEntity */;

this._state =
c === "["
c === 91 /* OpeningSquareBracket */
? 23 /* BeforeCdata1 */
: c === "-"
: c === 45 /* Dash */
? 18 /* BeforeComment */

@@ -397,3 +402,3 @@ : 16 /* InDeclaration */;

Tokenizer.prototype.stateInDeclaration = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
this.cbs.ondeclaration(this.getSection());

@@ -405,3 +410,3 @@ this._state = 1 /* Text */;

Tokenizer.prototype.stateInProcessingInstruction = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
this.cbs.onprocessinginstruction(this.getSection());

@@ -413,3 +418,3 @@ this._state = 1 /* Text */;

Tokenizer.prototype.stateBeforeComment = function (c) {
if (c === "-") {
if (c === 45 /* Dash */) {
this._state = 19 /* InComment */;

@@ -423,7 +428,7 @@ this.sectionStart = this._index + 1;

Tokenizer.prototype.stateInComment = function (c) {
if (c === "-")
if (c === 45 /* Dash */)
this._state = 21 /* AfterComment1 */;
};
Tokenizer.prototype.stateInSpecialComment = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
this.cbs.oncomment(this.buffer.substring(this.sectionStart, this._index));

@@ -435,3 +440,3 @@ this._state = 1 /* Text */;

Tokenizer.prototype.stateAfterComment1 = function (c) {
if (c === "-") {
if (c === 45 /* Dash */) {
this._state = 22 /* AfterComment2 */;

@@ -444,3 +449,3 @@ }

Tokenizer.prototype.stateAfterComment2 = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
// Remove 2 trailing chars

@@ -451,3 +456,3 @@ this.cbs.oncomment(this.buffer.substring(this.sectionStart, this._index - 2));

}
else if (c !== "-") {
else if (c !== 45 /* Dash */) {
this._state = 19 /* InComment */;

@@ -458,3 +463,3 @@ }

Tokenizer.prototype.stateBeforeCdata6 = function (c) {
if (c === "[") {
if (c === 91 /* OpeningSquareBracket */) {
this._state = 29 /* InCdata */;

@@ -465,11 +470,11 @@ this.sectionStart = this._index + 1;

this._state = 16 /* InDeclaration */;
this._index--;
this.stateInDeclaration(c);
}
};
Tokenizer.prototype.stateInCdata = function (c) {
if (c === "]")
if (c === 93 /* ClosingSquareBracket */)
this._state = 30 /* AfterCdata1 */;
};
Tokenizer.prototype.stateAfterCdata1 = function (c) {
if (c === "]")
if (c === 93 /* ClosingSquareBracket */)
this._state = 31 /* AfterCdata2 */;

@@ -480,3 +485,3 @@ else

Tokenizer.prototype.stateAfterCdata2 = function (c) {
if (c === ">") {
if (c === 62 /* Gt */) {
// Remove 2 trailing chars

@@ -487,3 +492,3 @@ this.cbs.oncdata(this.buffer.substring(this.sectionStart, this._index - 2));

}
else if (c !== "]") {
else if (c !== 93 /* ClosingSquareBracket */) {
this._state = 29 /* InCdata */;

@@ -494,6 +499,6 @@ }

Tokenizer.prototype.stateBeforeSpecialS = function (c) {
if (c === "c" || c === "C") {
if (c === 99 /* LowerC */ || c === 67 /* UpperC */) {
this._state = 34 /* BeforeScript1 */;
}
else if (c === "t" || c === "T") {
else if (c === 116 /* LowerT */ || c === 84 /* UpperT */) {
this._state = 44 /* BeforeStyle1 */;

@@ -503,10 +508,12 @@ }

this._state = 3 /* InTagName */;
this._index--; // Consume the token again
this.stateInTagName(c); // Consume the token again
}
};
Tokenizer.prototype.stateBeforeSpecialSEnd = function (c) {
if (this.special === 2 /* Script */ && (c === "c" || c === "C")) {
if (this.special === 2 /* Script */ &&
(c === 99 /* LowerC */ || c === 67 /* UpperC */)) {
this._state = 39 /* AfterScript1 */;
}
else if (this.special === 3 /* Style */ && (c === "t" || c === "T")) {
else if (this.special === 3 /* Style */ &&
(c === 116 /* LowerT */ || c === 84 /* UpperT */)) {
this._state = 48 /* AfterStyle1 */;

@@ -518,14 +525,14 @@ }

Tokenizer.prototype.stateBeforeSpecialLast = function (c, special) {
if (c === "/" || c === ">" || whitespace(c)) {
if (c === 47 /* Slash */ || c === 62 /* Gt */ || whitespace(c)) {
this.special = special;
}
this._state = 3 /* InTagName */;
this._index--; // Consume the token again
this.stateInTagName(c); // Consume the token again
};
Tokenizer.prototype.stateAfterSpecialLast = function (c, sectionStartOffset) {
if (c === ">" || whitespace(c)) {
if (c === 62 /* Gt */ || whitespace(c)) {
this.sectionStart = this._index - sectionStartOffset;
this.special = 1 /* None */;
this._state = 6 /* InClosingTagName */;
this.sectionStart = this._index - sectionStartOffset;
this._index--; // Reconsume the token
this.stateInClosingTagName(c); // Reconsume the token
}

@@ -535,60 +542,56 @@ else

};
// For entities terminated with a semicolon
Tokenizer.prototype.parseFixedEntity = function (map) {
if (map === void 0) { map = this.xmlMode ? xml_json_1.default : entities_json_1.default; }
// Offset = 1
if (this.sectionStart + 1 < this._index) {
var entity = this.buffer.substring(this.sectionStart + 1, this._index);
if (Object.prototype.hasOwnProperty.call(map, entity)) {
this.emitPartial(map[entity]);
this.sectionStart = this._index + 1;
}
Tokenizer.prototype.stateBeforeEntity = function (c) {
if (c === 35 /* Num */) {
this._state = 63 /* BeforeNumericEntity */;
}
};
// Parses legacy entities (without trailing semicolon)
Tokenizer.prototype.parseLegacyEntity = function () {
var start = this.sectionStart + 1;
// The max length of legacy entities is 6
var limit = Math.min(this._index - start, 6);
while (limit >= 2) {
// The min length of legacy entities is 2
var entity = this.buffer.substr(start, limit);
if (Object.prototype.hasOwnProperty.call(legacy_json_1.default, entity)) {
this.emitPartial(legacy_json_1.default[entity]);
this.sectionStart += limit + 1;
return;
}
limit--;
else if (c === 38 /* Amp */) {
// We have two `&` characters in a row. Emit the first one.
this.emitPartial(this.getSection());
this.sectionStart = this._index;
}
else {
this._state = 64 /* InNamedEntity */;
this.trieIndex = 0;
this.trieCurrent = this.entityTrie[0];
this.trieResult = null;
// Start excess with 1 to include the '&'
this.trieExcess = 1;
this._index--;
}
};
Tokenizer.prototype.stateInNamedEntity = function (c) {
if (c === ";") {
this.parseFixedEntity();
// Retry as legacy entity if entity wasn't parsed
if (this.baseState === 1 /* Text */ &&
this.sectionStart + 1 < this._index &&
!this.xmlMode) {
this.parseLegacyEntity();
}
this._state = this.baseState;
this.trieExcess += 1;
this.trieIndex = decode_1.determineBranch(this.entityTrie, this.trieCurrent, this.trieIndex + 1, c);
if (this.trieIndex < 0) {
this.emitNamedEntity();
return;
}
else if ((c < "0" || c > "9") && !isASCIIAlpha(c)) {
if (this.xmlMode || this.sectionStart + 1 === this._index) {
// Ignore
this.trieCurrent = this.entityTrie[this.trieIndex];
// If the branch is a value, store it and continue
if (this.trieCurrent & decode_1.BinTrieFlags.HAS_VALUE) {
// If we have a legacy entity while parsing strictly, just skip the number of bytes
if (!this.allowLegacyEntity() && c !== 59 /* Semi */) {
// No need to consider multi-byte values, as the legacy entity is always a single byte
this.trieIndex += 1;
}
else if (this.baseState !== 1 /* Text */) {
if (c !== "=") {
// Parse as legacy entity, without allowing additional characters.
this.parseFixedEntity(legacy_json_1.default);
}
}
else {
this.parseLegacyEntity();
// If this is a surrogate pair, combine the higher bits from the node with the next byte
this.trieResult =
this.trieCurrent & decode_1.BinTrieFlags.MULTI_BYTE
? String.fromCharCode(this.entityTrie[++this.trieIndex], this.entityTrie[++this.trieIndex])
: String.fromCharCode(this.entityTrie[++this.trieIndex]);
this.trieExcess = 0;
}
this._state = this.baseState;
this._index--;
}
};
Tokenizer.prototype.decodeNumericEntity = function (offset, base, strict) {
var sectionStart = this.sectionStart + offset;
Tokenizer.prototype.emitNamedEntity = function () {
if (this.trieResult) {
this.emitPartial(this.trieResult);
}
this.sectionStart = this._index - this.trieExcess + 1;
this._state = this.baseState;
this._index--;
};
Tokenizer.prototype.decodeNumericEntity = function (base, strict) {
var sectionStart = this.sectionStart + 2 + (base >> 4);
if (sectionStart !== this._index) {

@@ -604,8 +607,8 @@ // Parse entity

Tokenizer.prototype.stateInNumericEntity = function (c) {
if (c === ";") {
this.decodeNumericEntity(2, 10, true);
if (c === 59 /* Semi */) {
this.decodeNumericEntity(10, true);
}
else if (c < "0" || c > "9") {
if (!this.xmlMode) {
this.decodeNumericEntity(2, 10, false);
else if (c < 48 /* Zero */ || c > 57 /* Nine */) {
if (this.allowLegacyEntity()) {
this.decodeNumericEntity(10, false);
}

@@ -619,10 +622,10 @@ else {

Tokenizer.prototype.stateInHexEntity = function (c) {
if (c === ";") {
this.decodeNumericEntity(3, 16, true);
if (c === 59 /* Semi */) {
this.decodeNumericEntity(16, true);
}
else if ((c < "a" || c > "f") &&
(c < "A" || c > "F") &&
(c < "0" || c > "9")) {
if (!this.xmlMode) {
this.decodeNumericEntity(3, 16, false);
else if ((c < 97 /* LowerA */ || c > 102 /* LowerF */) &&
(c < 65 /* UpperA */ || c > 70 /* UpperF */) &&
(c < 48 /* Zero */ || c > 57 /* Nine */)) {
if (this.allowLegacyEntity()) {
this.decodeNumericEntity(16, false);
}

@@ -635,2 +638,5 @@ else {

};
Tokenizer.prototype.allowLegacyEntity = function () {
return !this.xmlMode && this.baseState === 1 /* Text */;
};
Tokenizer.prototype.cleanup = function () {

@@ -673,3 +679,3 @@ if (this.sectionStart < 0) {

while (this._index < this.buffer.length && this.running) {
var c = this.buffer.charAt(this._index);
var c = this.buffer.charCodeAt(this._index);
if (this._state === 1 /* Text */) {

@@ -745,3 +751,3 @@ this.stateText(c);

else if (this._state === 53 /* BeforeSpecialTEnd */) {
stateAfterSpecialTEnd(this, c);
stateBeforeSpecialTEnd(this, c);
}

@@ -842,3 +848,3 @@ else if (this._state === 39 /* AfterScript1 */) {

else if (this._state === 62 /* BeforeEntity */) {
stateBeforeEntity(this, c);
this.stateBeforeEntity(c);
}

@@ -903,3 +909,5 @@ else if (this._state === 24 /* BeforeCdata2 */) {

else if (this._state === 64 /* InNamedEntity */ && !this.xmlMode) {
this.parseLegacyEntity();
// Increase excess for EOF
this.trieExcess++;
this.emitNamedEntity();
if (this.sectionStart < this._index) {

@@ -911,3 +919,3 @@ this._state = this.baseState;

else if (this._state === 65 /* InNumericEntity */ && !this.xmlMode) {
this.decodeNumericEntity(2, 10, false);
this.decodeNumericEntity(10, false);
if (this.sectionStart < this._index) {

@@ -919,3 +927,3 @@ this._state = this.baseState;

else if (this._state === 66 /* InHexEntity */ && !this.xmlMode) {
this.decodeNumericEntity(3, 16, false);
this.decodeNumericEntity(16, false);
if (this.sectionStart < this._index) {

@@ -945,6 +953,2 @@ this._state = this.baseState;

};
Tokenizer.prototype.emitToken = function (name) {
this.cbs[name](this.getSection());
this.sectionStart = -1;
};
Tokenizer.prototype.emitPartial = function (value) {

@@ -951,0 +955,0 @@ if (this.baseState !== 1 /* Text */) {

23

package.json
{
"name": "htmlparser2",
"description": "Fast & forgiving HTML/XML parser",
"version": "6.1.0",
"version": "7.0.0",
"author": "Felix Boehm <me@feedic.com>",

@@ -38,3 +38,4 @@ "funding": [

"scripts": {
"test": "jest --coverage",
"test": "npm run test:jest && npm run lint",
"test:jest": "jest",
"lint": "npm run lint:es && npm run lint:prettier",

@@ -54,15 +55,15 @@ "lint:es": "eslint src",

"domutils": "^2.5.2",
"entities": "^2.0.0"
"entities": "^3.0.1"
},
"devDependencies": {
"@types/jest": "^26.0.0",
"@types/node": "^14.0.5",
"@typescript-eslint/eslint-plugin": "^4.9.1",
"@typescript-eslint/parser": "^4.9.1",
"eslint": "^7.15.0",
"@types/jest": "^27.0.1",
"@types/node": "^16.6.2",
"@typescript-eslint/eslint-plugin": "^4.29.2",
"@typescript-eslint/parser": "^4.29.2",
"eslint": "^7.32.0",
"eslint-config-prettier": "^8.1.0",
"jest": "^26.0.1",
"jest": "^27.0.1",
"prettier": "^2.1.1",
"ts-jest": "^26.0.0",
"typescript": "^4.0.2"
"ts-jest": "^27.0.5",
"typescript": "^4.3.5"
},

@@ -69,0 +70,0 @@ "jest": {

lib/FeedHandler.d.ts.map

Sorry, the diff of this file is not supported yet

lib/Parser.d.ts.map

Sorry, the diff of this file is not supported yet

lib/Tokenizer.d.ts.map

Sorry, the diff of this file is not supported yet

SocketSocket SOC 2 Logo

Product

  • Package Alerts
  • Integrations
  • Docs
  • Pricing
  • FAQ
  • Roadmap

About

Packages

Stay in touch

Get open source security insights delivered straight into your inbox.

  • Terms
  • Privacy
  • Security

Made with ⚡️ by Socket Inc