Server environments

Reading a DOCX needs an XML parser. Browsers have one; anywhere else, provide one.

Passing a parser

import { JSDOM } from "jsdom";
import { importDocx } from "@portone/docx-editor/core";

const { DOMParser } = new JSDOM().window;

const { doc, session } = importDocx(source, { xmlParser: new DOMParser() });

importDocx, exportDocx, exportProblems, onlyCommentsChangedBy, and compareDocx take xmlParser in their options; documentNumbering and parseNumbering take it as their second argument. The parser must implement parseFromString(source, "application/xml") and return a real DOM document, since the readers use the standard DOM API: documentElement, children, childNodes, attributes, getAttribute and getElementsByTagName with their namespace-aware forms, localName, nodeName, nodeType, nodeValue, namespaceURI, textContent, parentElement, ownerDocument, and cloneNode. Unreadable markup must throw or return a document holding a parsererror element; either is reported as malformed-xml.

The package bundles no parser, and jsdom is the only implementation it tests. Serverless functions and edge runtimes are unsupported for now. Configure the parser to disable external entities and external fetching.

A DOMParser global

Installing a DOMParser global works too, and every call then leaves the option out:

globalThis.DOMParser = new JSDOM().window.DOMParser;

A call with neither is refused with DocxImportError and the code no-xml-parser. Nothing else from the browser is needed: no Node, document, or window global.