@manuscripts/transform 3.0.21-LEAN-4076.1 → 3.0.21

package/dist/es/jats/importer/jats-dom-parser.js

@@ -17,971 +17,994 @@ import { buildBibliographicDate, buildBibliographicName, buildContribution, Obje
 import mime from 'mime';
 import { DOMParser, Fragment } from 'prosemirror-model';
 import { dateToTimestamp, getTrimmedTextContent } from '../../lib/utils';
-import { schema, } from '../../schema';
-import { chooseSectionCategory } from '../../transformer';
 import { DEFAULT_PROFILE_ID } from './jats-comments';
 import { htmlFromJatsNode } from './jats-parser-utils';
-const XLINK_NAMESPACE = 'http://www.w3.org/1999/xlink';
-const chooseContentType = (graphicNode) => {
-    if (graphicNode) {
-        const mimetype = graphicNode.getAttribute('mimetype');
-        const subtype = graphicNode.getAttribute('mime-subtype');
-        if (mimetype && subtype) {
-            return [mimetype, subtype].join('/');
-        }
-        const href = graphicNode.getAttributeNS(XLINK_NAMESPACE, 'href');
-        if (href) {
-            return mime.getType(href) || undefined;
-        }
-    }
-};
-const parsePriority = (priority) => {
-    if (!priority) {
-        return undefined;
-    }
-    return parseInt(priority);
-};
-const getEquationContent = (p) => {
-    var _a;
-    const element = p;
-    const id = element.getAttribute('id');
-    const container = (_a = element.querySelector('alternatives')) !== null && _a !== void 0 ? _a : element;
-    let contents = '';
-    let format = '';
-    for (const child of container.childNodes) {
-        const nodeName = child.nodeName.replace(/^[a-z]:/, '');
-        switch (nodeName) {
-            case 'tex-math':
-                contents = child.textContent;
-                format = 'tex';
-                break;
-            case 'mml:math':
-                contents = child.outerHTML;
-                format = 'mathml';
-                break;
-        }
-    }
-    return { id, format, contents };
-};
-const parseDates = (historyNode) => {
-    if (!historyNode) {
-        return undefined;
-    }
-    const history = {};
-    for (const date of historyNode.children) {
-        const dateType = date.getAttribute('date-type');
-        switch (dateType) {
-            case 'received': {
-                history.receiveDate = dateToTimestamp(date);
-                break;
-            }
-            case 'rev-recd': {
-                history.revisionReceiveDate = dateToTimestamp(date);
-                break;
+export class JATSDOMParser {
+    constructor(sectionCategories, schema) {
+        this.sectionCategories = sectionCategories;
+        this.schema = schema;
+        this.XLINK_NAMESPACE = 'http://www.w3.org/1999/xlink';
+        this.chooseContentType = (graphicNode) => {
+            if (graphicNode) {
+                const mimetype = graphicNode.getAttribute('mimetype');
+                const subtype = graphicNode.getAttribute('mime-subtype');
+                if (mimetype && subtype) {
+                    return [mimetype, subtype].join('/');
+                }
+                const href = graphicNode.getAttributeNS(this.XLINK_NAMESPACE, 'href');
+                if (href) {
+                    return mime.getType(href) || undefined;
+                }
             }
-            case 'accepted': {
-                history.acceptanceDate = dateToTimestamp(date);
-                break;
+        };
+        this.parsePriority = (priority) => {
+            if (!priority) {
+                return undefined;
             }
-            case 'rev-request': {
-                history.revisionRequestDate = dateToTimestamp(date);
-                break;
+            return parseInt(priority);
+        };
+        this.getEquationContent = (p) => {
+            var _a;
+            const element = p;
+            const id = element.getAttribute('id');
+            const container = (_a = element.querySelector('alternatives')) !== null && _a !== void 0 ? _a : element;
+            let contents = '';
+            let format = '';
+            for (const child of container.childNodes) {
+                const nodeName = child.nodeName.replace(/^[a-z]:/, '');
+                switch (nodeName) {
+                    case 'tex-math':
+                        contents = child.textContent;
+                        format = 'tex';
+                        break;
+                    case 'mml:math':
+                        contents = child.outerHTML;
+                        format = 'mathml';
+                        break;
+                }
             }
-            case 'retracted': {
-                history.retractionDate = dateToTimestamp(date);
-                break;
+            return { id, format, contents };
+        };
+        this.parseDates = (historyNode) => {
+            if (!historyNode) {
+                return undefined;
             }
-            case 'corrected': {
-                history.correctionDate = dateToTimestamp(date);
-                break;
+            const history = {};
+            for (const date of historyNode.children) {
+                const dateType = date.getAttribute('date-type');
+                switch (dateType) {
+                    case 'received': {
+                        history.receiveDate = dateToTimestamp(date);
+                        break;
+                    }
+                    case 'rev-recd': {
+                        history.revisionReceiveDate = dateToTimestamp(date);
+                        break;
+                    }
+                    case 'accepted': {
+                        history.acceptanceDate = dateToTimestamp(date);
+                        break;
+                    }
+                    case 'rev-request': {
+                        history.revisionRequestDate = dateToTimestamp(date);
+                        break;
+                    }
+                    case 'retracted': {
+                        history.retractionDate = dateToTimestamp(date);
+                        break;
+                    }
+                    case 'corrected': {
+                        history.correctionDate = dateToTimestamp(date);
+                        break;
+                    }
+                }
             }
-        }
-    }
-    return history;
-};
-const getEmail = (element) => {
-    var _a, _b;
-    const email = element.querySelector('email');
-    if (email) {
-        return {
-            href: (_a = email.getAttributeNS(XLINK_NAMESPACE, 'href')) !== null && _a !== void 0 ? _a : '',
-            text: (_b = getTrimmedTextContent(email)) !== null && _b !== void 0 ? _b : '',
+            return history;
         };
-    }
-};
-const getInstitutionDetails = (element) => {
-    let department = '';
-    let institution = '';
-    for (const node of element.querySelectorAll('institution')) {
-        const content = getTrimmedTextContent(node);
-        if (!content) {
-            continue;
-        }
-        const type = node.getAttribute('content-type');
-        if (type === 'dept') {
-            department = content;
-        }
-        else {
-            institution = content;
-        }
-    }
-    return { department, institution };
-};
-const getAddressLine = (element, index) => {
-    return getTrimmedTextContent(element, `addr-line:nth-of-type(${index})`) || '';
-};
-const getHTMLContent = (node, querySelector) => {
-    return htmlFromJatsNode(node.querySelector(querySelector));
-};
-const chooseBibliographyItemType = (publicationType) => {
-    switch (publicationType) {
-        case 'book':
-        case 'thesis':
-            return publicationType;
-        case 'journal':
-        default:
-            return 'article-journal';
-    }
-};
-const parseRef = (element) => {
-    const publicationType = element.getAttribute('publication-type');
-    const authorNodes = [
-        ...element.querySelectorAll('person-group[person-group-type="author"] > *'),
-    ];
-    const id = element.id;
-    const attrs = {
-        id,
-        type: chooseBibliographyItemType(publicationType),
-    };
-    const title = getHTMLContent(element, 'article-title');
-    if (title) {
-        attrs.title = title;
-    }
-    const mixedCitation = element.querySelector('mixed-citation');
-    if (authorNodes.length <= 0) {
-        mixedCitation === null || mixedCitation === void 0 ? void 0 : mixedCitation.childNodes.forEach((item) => {
+        this.getEmail = (element) => {
             var _a, _b;
-            if (item.nodeType === Node.TEXT_NODE &&
-                ((_a = item.textContent) === null || _a === void 0 ? void 0 : _a.match(/[A-Za-z]+/g))) {
-                attrs.literal = (_b = getTrimmedTextContent(mixedCitation)) !== null && _b !== void 0 ? _b : '';
-                return attrs;
+            const email = element.querySelector('email');
+            if (email) {
+                return {
+                    href: (_a = email.getAttributeNS(this.XLINK_NAMESPACE, 'href')) !== null && _a !== void 0 ? _a : '',
+                    text: (_b = getTrimmedTextContent(email)) !== null && _b !== void 0 ? _b : '',
+                };
             }
-        });
-    }
-    const source = getHTMLContent(element, 'source');
-    if (source) {
-        attrs.containerTitle = source;
-    }
-    const volume = getTrimmedTextContent(element, 'volume');
-    if (volume) {
-        attrs.volume = volume;
-    }
-    const issue = getTrimmedTextContent(element, 'issue');
-    if (issue) {
-        attrs.issue = issue;
-    }
-    const supplement = getTrimmedTextContent(element, 'supplement');
-    if (supplement) {
-        attrs.supplement = supplement;
-    }
-    const fpage = getTrimmedTextContent(element, 'fpage');
-    const lpage = getTrimmedTextContent(element, 'lpage');
-    if (fpage) {
-        attrs.page = lpage ? `${fpage}-${lpage}` : fpage;
-    }
-    const year = getTrimmedTextContent(element, 'year');
-    if (year) {
-        attrs.issued = buildBibliographicDate({
-            'date-parts': [[year]],
-        });
-    }
-    const doi = getTrimmedTextContent(element, 'pub-id[pub-id-type="doi"]');
-    if (doi) {
-        attrs.doi = doi;
-    }
-    const authors = [];
-    authorNodes.forEach((authorNode) => {
-        const name = buildBibliographicName({});
-        const given = getTrimmedTextContent(authorNode, 'given-names');
-        if (given) {
-            name.given = given;
-        }
-        const family = getTrimmedTextContent(authorNode, 'surname');
-        if (family) {
-            name.family = family;
-        }
-        if (authorNode.nodeName === 'collab') {
-            name.literal = getTrimmedTextContent(authorNode);
-        }
-        authors.push(name);
-    });
-    if (authors.length) {
-        attrs.author = authors;
-    }
-    return attrs;
-};
-const marks = [
-    {
-        tag: 'bold',
-        mark: 'bold',
-    },
-    {
-        tag: 'code',
-        mark: 'code',
-    },
-    {
-        tag: 'italic',
-        mark: 'italic',
-    },
-    {
-        tag: 'sc',
-        mark: 'smallcaps',
-    },
-    {
-        tag: 'strike',
-        mark: 'strikethrough',
-    },
-    {
-        tag: 'styled-content',
-        mark: 'styled',
-        getAttrs: (node) => ({
-            style: node.getAttribute('style'),
-        }),
-    },
-    {
-        tag: 'sub',
-        mark: 'subscript',
-    },
-    {
-        tag: 'sup',
-        mark: 'superscript',
-    },
-    {
-        tag: 'underline',
-        mark: 'underline',
-    },
-];
-const nodes = [
-    {
-        tag: 'article',
-        node: 'manuscript',
-        getAttrs: (node) => {
-            var _a, _b;
-            const element = node;
-            const doi = element.querySelector('front > article-meta > article-id[pub-id-type="doi"]');
-            const history = element.querySelector('history');
-            const dates = parseDates(history);
-            return Object.assign({ doi: getTrimmedTextContent(doi), articleType: (_a = element.getAttribute('article-type')) !== null && _a !== void 0 ? _a : '', primaryLanguageCode: (_b = element.getAttribute('lang')) !== null && _b !== void 0 ? _b : '' }, dates);
-        },
-    },
-    {
-        tag: 'article-title',
-        node: 'title',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'highlight-marker',
-        node: 'highlight_marker',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.id,
-                position: element.getAttribute('position'),
-            };
-        },
-    },
-    {
-        tag: 'comment',
-        node: 'comment',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                target: element.getAttribute('target-id'),
-                contents: getTrimmedTextContent(element),
-                contributions: [buildContribution(DEFAULT_PROFILE_ID)],
-            };
-        },
-    },
-    {
-        tag: 'author-notes',
-        node: 'author_notes',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'funding-group',
-        node: 'awards',
-    },
-    {
-        tag: 'award-group',
-        node: 'award',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                recipient: getTrimmedTextContent(element, 'principal-award-recipient'),
-                code: Array.from(element.querySelectorAll('award-id'))
-                    .map((awardID) => getTrimmedTextContent(awardID))
-                    .reduce((acc, text) => (acc ? `${acc};${text}` : text), ''),
-                source: getTrimmedTextContent(element, 'funding-source'),
-            };
-        },
-    },
-    {
-        tag: 'fn:not([fn-type])',
-        node: 'footnote',
-        context: 'author_notes/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                kind: 'footnote',
-            };
-        },
-    },
-    {
-        tag: 'corresp',
-        node: 'corresp',
-        getAttrs: (node) => {
-            const element = node;
-            const label = element.querySelector('label');
-            if (label) {
-                label.remove();
-            }
-            return {
-                id: element.getAttribute('id'),
-                label: getTrimmedTextContent(label),
-            };
-        },
-        getContent: (node) => {
-            const element = node;
-            return Fragment.from(schema.text(getTrimmedTextContent(element) || ''));
-        },
-    },
-    {
-        tag: 'contrib[contrib-type="author"]',
-        node: 'contributor',
-        getAttrs: (node) => {
-            const element = node;
-            const footnote = [];
-            const affiliations = [];
-            const corresp = [];
-            const xrefs = element.querySelectorAll('xref');
-            for (const xref of xrefs) {
-                const rid = xref.getAttribute('rid');
-                const type = xref.getAttribute('ref-type');
-                if (!rid) {
+        };
+        this.getInstitutionDetails = (element) => {
+            let department = '';
+            let institution = '';
+            for (const node of element.querySelectorAll('institution')) {
+                const content = getTrimmedTextContent(node);
+                if (!content) {
                     continue;
                 }
-                switch (type) {
-                    case 'fn':
-                        footnote.push({
-                            noteID: rid,
-                            noteLabel: getTrimmedTextContent(xref) || '',
-                        });
-                        break;
-                    case 'corresp':
-                        corresp.push({
-                            correspID: rid,
-                            correspLabel: getTrimmedTextContent(xref) || '',
-                        });
-                        break;
-                    case 'aff':
-                        affiliations.push(rid);
-                        break;
+                const type = node.getAttribute('content-type');
+                if (type === 'dept') {
+                    department = content;
+                }
+                else {
+                    institution = content;
                 }
             }
-            return {
-                id: element.getAttribute('id'),
-                role: 'author',
-                affiliations,
-                corresp,
-                footnote,
-                isCorresponding: element.getAttribute('corresp')
-                    ? element.getAttribute('corresp') === 'yes'
-                    : undefined,
-                bibliographicName: {
-                    given: getTrimmedTextContent(element, 'name > given-names'),
-                    family: getTrimmedTextContent(element, 'name > surname'),
-                    ObjectType: ObjectTypes.BibliographicName,
-                },
-                ORCIDIdentifier: getTrimmedTextContent(element, 'contrib-id[contrib-id-type="orcid"]'),
-                priority: parsePriority(element.getAttribute('priority')),
-            };
-        },
-        getContent: () => {
-            return Fragment.from(schema.text('_'));
-        },
-    },
-    {
-        tag: 'affiliations',
-        node: 'affiliations',
-    },
-    {
-        tag: 'aff',
-        node: 'affiliation',
-        context: 'affiliations/',
-        getAttrs: (node) => {
-            var _a, _b;
-            const element = node;
-            const { department, institution } = getInstitutionDetails(element);
-            return {
-                id: element.getAttribute('id'),
-                institution: institution !== null && institution !== void 0 ? institution : '',
-                department: department !== null && department !== void 0 ? department : '',
-                addressLine1: getAddressLine(element, 1),
-                addressLine2: getAddressLine(element, 2),
-                addressLine3: getAddressLine(element, 3),
-                postCode: (_a = getTrimmedTextContent(element, 'postal-code')) !== null && _a !== void 0 ? _a : '',
-                country: (_b = getTrimmedTextContent(element, 'country')) !== null && _b !== void 0 ? _b : '',
-                email: getEmail(element),
-                priority: parsePriority(element.getAttribute('priority')),
-            };
-        },
-        getContent: () => {
-            return Fragment.from(schema.text('_'));
-        },
-    },
-    {
-        tag: 'attrib',
-        node: 'attribution',
-    },
-    {
-        tag: 'back',
-        ignore: true,
-    },
-    {
-        tag: 'history',
-        ignore: true,
-    },
-    {
-        tag: 'break',
-        node: 'hard_break',
-    },
-    {
-        tag: 'caption',
-        node: 'figcaption',
-        context: 'figure/',
-    },
-    {
-        tag: 'caption',
-        node: 'figcaption',
-        context: 'figure_element/',
-        getContent: (node, schema) => {
-            const element = node;
-            const content = [];
-            const title = element.querySelector('title');
+            return { department, institution };
+        };
+        this.getAddressLine = (element, index) => {
+            return (getTrimmedTextContent(element, `addr-line:nth-of-type(${index})`) || '');
+        };
+        this.getHTMLContent = (node, querySelector) => {
+            return htmlFromJatsNode(node.querySelector(querySelector));
+        };
+        this.chooseBibliographyItemType = (publicationType) => {
+            switch (publicationType) {
+                case 'book':
+                case 'thesis':
+                    return publicationType;
+                case 'journal':
+                default:
+                    return 'article-journal';
+            }
+        };
+        this.parseRef = (element) => {
+            const publicationType = element.getAttribute('publication-type');
+            const authorNodes = [
+                ...element.querySelectorAll('person-group[person-group-type="author"] > *'),
+            ];
+            const id = element.id;
+            const attrs = {
+                id,
+                type: this.chooseBibliographyItemType(publicationType),
+            };
+            const title = this.getHTMLContent(element, 'article-title');
             if (title) {
-                const captionTitle = schema.nodes.caption_title.create();
-                content.push(jatsDOMParser.parse(title, { topNode: captionTitle }));
+                attrs.title = title;
             }
-            const paragraphs = element.querySelectorAll('p');
-            if (paragraphs.length) {
-                const figcaption = schema.nodes.caption.create();
-                for (const paragraph of paragraphs) {
-                    content.push(jatsDOMParser.parse(paragraph, { topNode: figcaption }));
-                }
+            const mixedCitation = element.querySelector('mixed-citation');
+            if (authorNodes.length <= 0) {
+                mixedCitation === null || mixedCitation === void 0 ? void 0 : mixedCitation.childNodes.forEach((item) => {
+                    var _a, _b;
+                    if (item.nodeType === Node.TEXT_NODE &&
+                        ((_a = item.textContent) === null || _a === void 0 ? void 0 : _a.match(/[A-Za-z]+/g))) {
+                        attrs.literal = (_b = getTrimmedTextContent(mixedCitation)) !== null && _b !== void 0 ? _b : '';
+                        return attrs;
+                    }
+                });
             }
-            return Fragment.from(content);
-        },
-    },
-    {
-        tag: 'caption',
-        node: 'figcaption',
-        context: 'table_element/',
-    },
-    {
-        tag: 'caption',
-        node: 'figcaption',
-        context: 'box_element/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'code',
-        node: 'listing',
-        context: 'listing_element/',
-        getAttrs: (node) => {
-            var _a;
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                language: (_a = element.getAttribute('language')) !== null && _a !== void 0 ? _a : '',
-                contents: getTrimmedTextContent(element),
-            };
-        },
-    },
-    {
-        tag: 'inline-formula',
-        node: 'inline_equation',
-        getAttrs: (node) => {
-            const element = node;
-            return getEquationContent(element);
-        },
-    },
-    {
-        tag: 'disp-formula',
-        node: 'equation_element',
-        getAttrs: (node) => {
-            var _a;
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                label: (_a = getTrimmedTextContent(element, 'label')) !== null && _a !== void 0 ? _a : '',
-            };
-        },
-        getContent: (node, schema) => {
-            const element = node;
-            const attrs = getEquationContent(element);
-            return Fragment.from([
-                schema.nodes.equation.createChecked(Object.assign({}, attrs)),
-            ]);
-        },
-    },
-    {
-        tag: 'disp-quote[content-type=quote]',
-        node: 'blockquote_element',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'disp-quote[content-type=pullquote]',
-        node: 'pullquote_element',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'ext-link',
-        node: 'link',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                href: element.getAttributeNS(XLINK_NAMESPACE, 'href') || '',
-                title: element.getAttributeNS(XLINK_NAMESPACE, 'title') || '',
-            };
-        },
-    },
-    {
-        tag: 'fig[fig-type=equation]',
-        node: 'equation_element',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'fig[fig-type=listing]',
-        node: 'listing_element',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'graphic[specific-use=MISSING]',
-        node: 'missing_figure',
-        context: 'figure_element/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'graphic',
-        node: 'figure',
-        context: 'figure_element/',
-        getAttrs: (node) => {
-            const element = node;
-            const position = element.getAttribute('position');
-            const src = element.getAttributeNS(XLINK_NAMESPACE, 'href');
-            return {
-                id: element.getAttribute('id'),
-                contentType: chooseContentType(element || undefined) || '',
-                src,
-                position,
-            };
-        },
-    },
-    {
-        tag: 'fig',
-        node: 'figure_element',
-        getAttrs: (node) => {
-            var _a, _b;
-            const element = node;
-            const labelNode = element.querySelector('label');
-            if (labelNode) {
-                element.removeChild(labelNode);
+            const source = this.getHTMLContent(element, 'source');
+            if (source) {
+                attrs.containerTitle = source;
             }
-            const attrib = element.querySelector('attrib');
-            const attribution = attrib
-                ? {
-                    literal: (_a = getTrimmedTextContent(attrib)) !== null && _a !== void 0 ? _a : '',
-                }
-                : undefined;
-            return {
-                id: element.getAttribute('id'),
-                label: (_b = getTrimmedTextContent(labelNode)) !== null && _b !== void 0 ? _b : '',
-                attribution,
-                type: element.getAttribute('fig-type'),
-            };
-        },
-    },
-    {
-        tag: 'fn-group',
-        node: 'footnotes_element',
-        context: 'footnotes_section/|table_element_footer/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                kind: 'footnote',
-            };
-        },
-    },
-    {
-        tag: 'table-wrap-foot',
-        node: 'table_element_footer',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'general-table-footnote',
-        node: 'general_table_footnote',
-        context: 'table_element_footer/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-        getContent: (node) => {
-            const paragraphs = [];
-            node.childNodes.forEach((p) => {
-                const paragraph = schema.nodes.paragraph.create();
-                const content = jatsDOMParser.parse(p, {
-                    topNode: paragraph,
+            const volume = getTrimmedTextContent(element, 'volume');
+            if (volume) {
+                attrs.volume = volume;
+            }
+            const issue = getTrimmedTextContent(element, 'issue');
+            if (issue) {
+                attrs.issue = issue;
+            }
+            const supplement = getTrimmedTextContent(element, 'supplement');
+            if (supplement) {
+                attrs.supplement = supplement;
+            }
+            const fpage = getTrimmedTextContent(element, 'fpage');
+            const lpage = getTrimmedTextContent(element, 'lpage');
+            if (fpage) {
+                attrs.page = lpage ? `${fpage}-${lpage}` : fpage;
+            }
+            const year = getTrimmedTextContent(element, 'year');
+            if (year) {
+                attrs.issued = buildBibliographicDate({
+                    'date-parts': [[year]],
                 });
-                paragraphs.push(content);
+            }
+            const doi = getTrimmedTextContent(element, 'pub-id[pub-id-type="doi"]');
+            if (doi) {
+                attrs.doi = doi;
+            }
+            const authors = [];
+            authorNodes.forEach((authorNode) => {
+                const name = buildBibliographicName({});
+                const given = getTrimmedTextContent(authorNode, 'given-names');
+                if (given) {
+                    name.given = given;
+                }
+                const family = getTrimmedTextContent(authorNode, 'surname');
+                if (family) {
+                    name.family = family;
+                }
+                if (authorNode.nodeName === 'collab') {
+                    name.literal = getTrimmedTextContent(authorNode);
+                }
+                authors.push(name);
             });
-            return Fragment.from([...paragraphs]);
-        },
-    },
-    {
-        tag: 'fn',
-        node: 'footnote',
-        context: 'footnotes_element/|table_element_footer/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                kind: 'footnote',
-            };
-        },
-    },
-    {
-        tag: 'front',
-        ignore: true,
-    },
-    {
-        tag: 'list',
-        node: 'list',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                listStyleType: element.getAttribute('list-type'),
-            };
-        },
-    },
-    {
-        tag: 'list-item',
-        node: 'list_item',
-    },
-    {
-        tag: 'p',
-        node: 'paragraph',
-        context: 'section/',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'p',
-        node: 'paragraph',
-    },
-    {
-        tag: 'sec[sec-type="endnotes"]',
-        node: 'footnotes_section',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'sec[sec-type="keywords"]',
-        node: 'keywords',
-    },
-    {
-        tag: 'sec[sec-type="supplementary-material"]',
-        node: 'supplements',
-    },
-    {
-        tag: 'supplementary-material',
-        node: 'supplement',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                href: element.getAttributeNS(XLINK_NAMESPACE, 'href'),
-                mimeType: element.getAttribute('mimetype'),
-                mimeSubType: element.getAttribute('mime-subtype'),
-                title: getTrimmedTextContent(element, 'title'),
-            };
-        },
-    },
-    {
-        tag: 'sec[sec-type="abstracts"]',
-        node: 'abstracts',
-    },
-    {
-        tag: 'sec[sec-type="body"]',
-        node: 'body',
-    },
-    {
-        tag: 'sec[sec-type="backmatter"]',
-        node: 'backmatter',
-    },
-    {
-        tag: 'sec[sec-type="box-element"]',
-        node: 'box_element',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-                label: getTrimmedTextContent(element, 'label'),
-            };
-        },
-    },
-    {
-        tag: 'sec[sec-type="bibliography"]',
-        node: 'bibliography_section',
-    },
-    {
-        tag: 'ref-list',
-        context: 'bibliography_section/',
-        node: 'bibliography_element',
-    },
-    {
-        tag: 'ref',
-        context: 'bibliography_element/',
-        node: 'bibliography_item',
-        getAttrs: (node) => parseRef(node),
-    },
-    {
-        tag: 'sec[sec-type="abstract-graphical"]',
-        node: 'graphical_abstract_section',
-    },
-    {
-        tag: 'sec',
-        node: 'section',
-        getAttrs: (node) => {
-            var _a, _b;
-            const element = node;
-            const grandParentNodeName = (_b = (_a = element.parentNode) === null || _a === void 0 ? void 0 : _a.parentNode) === null || _b === void 0 ? void 0 : _b.nodeName.toLowerCase();
-            if (grandParentNodeName && grandParentNodeName !== 'body') {
-                element.setAttribute('sec-type', 'subsection');
+            if (authors.length) {
+                attrs.author = authors;
             }
-            return {
-                id: element.getAttribute('id'),
-                category: chooseSectionCategory(element),
-            };
-        },
-    },
-    {
-        tag: 'kwd-group-list',
-        context: 'keywords/',
-        node: 'keywords_element',
-    },
-    {
-        tag: 'kwd-group',
-        context: 'keywords_element/',
-        node: 'keyword_group',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.id,
-                type: element.getAttribute('kwd-group-type'),
-            };
-        },
-    },
-    {
-        tag: 'kwd',
-        context: 'keyword_group//',
-        node: 'keyword',
-    },
-    {
-        tag: 'label',
-        context: 'box_element/',
-        ignore: true,
-    },
-    {
-        tag: 'boxed-text',
-        ignore: true,
-    },
-    {
-        tag: 'label',
-        context: 'section/',
-        node: 'section_label',
-    },
-    {
-        tag: 'label',
-        context: 'table_element/',
-        ignore: true,
-    },
-    {
-        tag: 'label',
-        context: 'figure/',
-        ignore: true,
-    },
-    {
-        tag: 'table',
-        node: 'table',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'table-wrap',
-        node: 'table_element',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                id: element.getAttribute('id'),
-            };
-        },
-    },
-    {
-        tag: 'title',
-        node: 'section_title',
-        context: 'section/|footnotes_section/|bibliography_section/|keywords/|supplements/|author_notes/|graphical_abstract_section/',
-    },
-    {
-        tag: 'title',
-        node: 'caption_title',
-        context: 'figcaption/',
-    },
-    {
-        tag: 'tr',
-        node: 'table_row',
-    },
-    {
-        tag: 'td',
-        node: 'table_cell',
-        getAttrs: (node) => {
-            const element = node;
-            const colspan = parseInt(element.getAttribute('colspan') || '1');
-            const rowspan = parseInt(element.getAttribute('rowspan') || '1');
-            return Object.assign(Object.assign(Object.assign({}, (colspan && { colspan })), (rowspan && { rowspan })), { valign: element.getAttribute('valign'), align: element.getAttribute('align'), scope: element.getAttribute('scope'), style: element.getAttribute('style') });
-        },
-    },
-    {
-        tag: 'th',
-        node: 'table_header',
-        getAttrs: (node) => {
-            const element = node;
-            const colspan = parseInt(element.getAttribute('colspan') || '1');
-            const rowspan = parseInt(element.getAttribute('rowspan') || '1');
-            return Object.assign(Object.assign(Object.assign({}, (colspan && { colspan })), (rowspan && { rowspan })), { valign: element.getAttribute('valign'), align: element.getAttribute('align'), scope: element.getAttribute('scope'), style: element.getAttribute('style') });
-        },
-    },
-    {
-        tag: 'col',
-        node: 'table_col',
-        getAttrs: (node) => {
-            const element = node;
-            return {
-                width: element.getAttribute('width'),
-            };
-        },
-    },
-    {
-        tag: 'colgroup',
-        node: 'table_colgroup',
-    },
-    {
-        tag: 'xref[ref-type="bibr"]',
-        node: 'citation',
-        getAttrs: (node) => {
-            var _a;
-            const element = node;
-            return {
-                rids: ((_a = element.getAttribute('rid')) === null || _a === void 0 ? void 0 : _a.split(/\s+/)) || [],
-                contents: getTrimmedTextContent(element),
-            };
-        },
-    },
-    {
-        tag: 'xref[ref-type="fn"]',
-        node: 'inline_footnote',
-        getAttrs: (node) => {
-            var _a;
-            const element = node;
-            return {
-                rids: ((_a = element.getAttribute('rid')) === null || _a === void 0 ? void 0 : _a.split(/\s+/)) || [],
-            };
-        },
-    },
-    {
-        tag: 'xref',
-        node: 'cross_reference',
-        getAttrs: (node) => {
-            var _a;
-            const element = node;
-            return {
-                rids: ((_a = element.getAttribute('rid')) === null || _a === void 0 ? void 0 : _a.split(/\s+/)) || [],
-                label: getTrimmedTextContent(element),
-            };
-        },
-    },
-];
-export const jatsDOMParser = new DOMParser(schema, [...marks, ...nodes]);
+            return attrs;
+        };
+        this.nodes = [
+            {
+                tag: 'article',
+                node: 'manuscript',
+                getAttrs: (node) => {
+                    var _a, _b;
+                    const element = node;
+                    const doi = element.querySelector('front > article-meta > article-id[pub-id-type="doi"]');
+                    const history = element.querySelector('history');
+                    const dates = this.parseDates(history);
+                    return Object.assign({ doi: getTrimmedTextContent(doi), articleType: (_a = element.getAttribute('article-type')) !== null && _a !== void 0 ? _a : '', primaryLanguageCode: (_b = element.getAttribute('lang')) !== null && _b !== void 0 ? _b : '' }, dates);
+                },
+            },
+            {
+                tag: 'article-title',
+                node: 'title',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'highlight-marker',
+                node: 'highlight_marker',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.id,
+                        position: element.getAttribute('position'),
+                    };
+                },
+            },
+            {
+                tag: 'comment',
+                node: 'comment',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        target: element.getAttribute('target-id'),
+                        contents: getTrimmedTextContent(element),
+                        contributions: [buildContribution(DEFAULT_PROFILE_ID)],
+                    };
+                },
+            },
+            {
+                tag: 'author-notes',
+                node: 'author_notes',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'funding-group',
+                node: 'awards',
+            },
+            {
+                tag: 'award-group',
+                node: 'award',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        recipient: getTrimmedTextContent(element, 'principal-award-recipient'),
+                        code: Array.from(element.querySelectorAll('award-id'))
+                            .map((awardID) => getTrimmedTextContent(awardID))
+                            .reduce((acc, text) => (acc ? `${acc};${text}` : text), ''),
+                        source: getTrimmedTextContent(element, 'funding-source'),
+                    };
+                },
+            },
+            {
+                tag: 'fn:not([fn-type])',
+                node: 'footnote',
+                context: 'author_notes/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        kind: 'footnote',
+                    };
+                },
+            },
+            {
+                tag: 'corresp',
+                node: 'corresp',
+                getAttrs: (node) => {
+                    const element = node;
+                    const label = element.querySelector('label');
+                    if (label) {
+                        label.remove();
+                    }
+                    return {
+                        id: element.getAttribute('id'),
+                        label: getTrimmedTextContent(label),
+                    };
+                },
+                getContent: (node) => {
+                    const element = node;
+                    return Fragment.from(this.schema.text(getTrimmedTextContent(element) || ''));
+                },
+            },
+            {
+                tag: 'contrib[contrib-type="author"]',
+                node: 'contributor',
+                getAttrs: (node) => {
+                    const element = node;
+                    const footnote = [];
+                    const affiliations = [];
+                    const corresp = [];
+                    const xrefs = element.querySelectorAll('xref');
+                    for (const xref of xrefs) {
+                        const rid = xref.getAttribute('rid');
+                        const type = xref.getAttribute('ref-type');
+                        if (!rid) {
+                            continue;
+                        }
+                        switch (type) {
+                            case 'fn':
+                                footnote.push({
+                                    noteID: rid,
+                                    noteLabel: getTrimmedTextContent(xref) || '',
+                                });
+                                break;
+                            case 'corresp':
+                                corresp.push({
+                                    correspID: rid,
+                                    correspLabel: getTrimmedTextContent(xref) || '',
+                                });
+                                break;
+                            case 'aff':
+                                affiliations.push(rid);
+                                break;
+                        }
+                    }
+                    return {
+                        id: element.getAttribute('id'),
+                        role: 'author',
+                        affiliations,
+                        corresp,
+                        footnote,
+                        isCorresponding: element.getAttribute('corresp')
+                            ? element.getAttribute('corresp') === 'yes'
+                            : undefined,
+                        bibliographicName: {
+                            given: getTrimmedTextContent(element, 'name > given-names'),
+                            family: getTrimmedTextContent(element, 'name > surname'),
+                            ObjectType: ObjectTypes.BibliographicName,
+                        },
+                        ORCIDIdentifier: getTrimmedTextContent(element, 'contrib-id[contrib-id-type="orcid"]'),
+                        priority: this.parsePriority(element.getAttribute('priority')),
+                    };
+                },
+                getContent: () => {
+                    return Fragment.from(this.schema.text('_'));
+                },
+            },
+            {
+                tag: 'affiliations',
+                node: 'affiliations',
+            },
+            {
+                tag: 'aff',
+                node: 'affiliation',
+                context: 'affiliations/',
+                getAttrs: (node) => {
+                    var _a, _b;
+                    const element = node;
+                    const { department, institution } = this.getInstitutionDetails(element);
+                    return {
+                        id: element.getAttribute('id'),
+                        institution: institution !== null && institution !== void 0 ? institution : '',
+                        department: department !== null && department !== void 0 ? department : '',
+                        addressLine1: this.getAddressLine(element, 1),
+                        addressLine2: this.getAddressLine(element, 2),
+                        addressLine3: this.getAddressLine(element, 3),
+                        postCode: (_a = getTrimmedTextContent(element, 'postal-code')) !== null && _a !== void 0 ? _a : '',
+                        country: (_b = getTrimmedTextContent(element, 'country')) !== null && _b !== void 0 ? _b : '',
+                        email: this.getEmail(element),
+                        priority: this.parsePriority(element.getAttribute('priority')),
+                    };
+                },
+                getContent: () => {
+                    return Fragment.from(this.schema.text('_'));
+                },
+            },
+            {
+                tag: 'attrib',
+                node: 'attribution',
+            },
+            {
+                tag: 'back',
+                ignore: true,
+            },
+            {
+                tag: 'history',
+                ignore: true,
+            },
+            {
+                tag: 'break',
+                node: 'hard_break',
+            },
+            {
+                tag: 'caption',
+                node: 'figcaption',
+                context: 'figure/',
+            },
+            {
+                tag: 'caption',
+                node: 'figcaption',
+                context: 'figure_element/',
+                getContent: (node, schema) => {
+                    const element = node;
+                    const content = [];
+                    const title = element.querySelector('title');
+                    if (title) {
+                        const captionTitle = schema.nodes.caption_title.create();
+                        content.push(this.parse(title, { topNode: captionTitle }));
+                    }
+                    const paragraphs = element.querySelectorAll('p');
+                    if (paragraphs.length) {
+                        const figcaption = schema.nodes.caption.create();
+                        for (const paragraph of paragraphs) {
+                            content.push(this.parse(paragraph, { topNode: figcaption }));
+                        }
+                    }
+                    return Fragment.from(content);
+                },
+            },
+            {
+                tag: 'caption',
+                node: 'figcaption',
+                context: 'table_element/',
+            },
+            {
+                tag: 'caption',
+                node: 'figcaption',
+                context: 'box_element/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'code',
+                node: 'listing',
+                context: 'listing_element/',
+                getAttrs: (node) => {
+                    var _a;
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        language: (_a = element.getAttribute('language')) !== null && _a !== void 0 ? _a : '',
+                        contents: getTrimmedTextContent(element),
+                    };
+                },
+            },
+            {
+                tag: 'inline-formula',
+                node: 'inline_equation',
+                getAttrs: (node) => {
+                    const element = node;
+                    return this.getEquationContent(element);
+                },
+            },
+            {
+                tag: 'disp-formula',
+                node: 'equation_element',
+                getAttrs: (node) => {
+                    var _a;
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        label: (_a = getTrimmedTextContent(element, 'label')) !== null && _a !== void 0 ? _a : '',
+                    };
+                },
+                getContent: (node, schema) => {
+                    const element = node;
+                    const attrs = this.getEquationContent(element);
+                    return Fragment.from([
+                        schema.nodes.equation.createChecked(Object.assign({}, attrs)),
+                    ]);
+                },
+            },
+            {
+                tag: 'disp-quote[content-type=quote]',
+                node: 'blockquote_element',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'disp-quote[content-type=pullquote]',
+                node: 'pullquote_element',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'ext-link',
+                node: 'link',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        href: element.getAttributeNS(this.XLINK_NAMESPACE, 'href') || '',
+                        title: element.getAttributeNS(this.XLINK_NAMESPACE, 'title') || '',
+                    };
+                },
+            },
+            {
+                tag: 'fig[fig-type=equation]',
+                node: 'equation_element',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'fig[fig-type=listing]',
+                node: 'listing_element',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'graphic[specific-use=MISSING]',
+                node: 'missing_figure',
+                context: 'figure_element/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'graphic',
+                node: 'figure',
+                context: 'figure_element/',
+                getAttrs: (node) => {
+                    const element = node;
+                    const position = element.getAttribute('position');
+                    const src = element.getAttributeNS(this.XLINK_NAMESPACE, 'href');
+                    return {
+                        id: element.getAttribute('id'),
+                        contentType: this.chooseContentType(element || undefined) || '',
+                        src,
+                        position,
+                    };
+                },
+            },
+            {
+                tag: 'fig',
+                node: 'figure_element',
+                getAttrs: (node) => {
+                    var _a, _b;
+                    const element = node;
+                    const labelNode = element.querySelector('label');
+                    if (labelNode) {
+                        element.removeChild(labelNode);
+                    }
+                    const attrib = element.querySelector('attrib');
+                    const attribution = attrib
+                        ? {
+                            literal: (_a = getTrimmedTextContent(attrib)) !== null && _a !== void 0 ? _a : '',
+                        }
+                        : undefined;
+                    return {
+                        id: element.getAttribute('id'),
+                        label: (_b = getTrimmedTextContent(labelNode)) !== null && _b !== void 0 ? _b : '',
+                        attribution,
+                        type: element.getAttribute('fig-type'),
+                    };
+                },
+            },
+            {
+                tag: 'fn-group',
+                node: 'footnotes_element',
+                context: 'footnotes_section/|table_element_footer/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        kind: 'footnote',
+                    };
+                },
+            },
+            {
+                tag: 'table-wrap-foot',
+                node: 'table_element_footer',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'general-table-footnote',
+                node: 'general_table_footnote',
+                context: 'table_element_footer/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+                getContent: (node) => {
+                    const paragraphs = [];
+                    node.childNodes.forEach((p) => {
+                        const paragraph = this.schema.nodes.paragraph.create();
+                        const content = this.parse(p, {
+                            topNode: paragraph,
+                        });
+                        paragraphs.push(content);
+                    });
+                    return Fragment.from([...paragraphs]);
+                },
+            },
+            {
+                tag: 'fn',
+                node: 'footnote',
+                context: 'footnotes_element/|table_element_footer/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        kind: 'footnote',
+                    };
+                },
+            },
+            {
+                tag: 'front',
+                ignore: true,
+            },
+            {
+                tag: 'list',
+                node: 'list',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        listStyleType: element.getAttribute('list-type'),
+                    };
+                },
+            },
+            {
+                tag: 'list-item',
+                node: 'list_item',
+            },
+            {
+                tag: 'p',
+                node: 'paragraph',
+                context: 'section/',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'p',
+                node: 'paragraph',
+            },
+            {
+                tag: 'sec[sec-type="endnotes"]',
+                node: 'footnotes_section',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'sec[sec-type="keywords"]',
+                node: 'keywords',
+            },
+            {
+                tag: 'sec[sec-type="supplementary-material"]',
+                node: 'supplements',
+            },
+            {
+                tag: 'supplementary-material',
+                node: 'supplement',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        href: element.getAttributeNS(this.XLINK_NAMESPACE, 'href'),
+                        mimeType: element.getAttribute('mimetype'),
+                        mimeSubType: element.getAttribute('mime-subtype'),
+                        title: getTrimmedTextContent(element, 'title'),
+                    };
+                },
+            },
+            {
+                tag: 'sec[sec-type="abstracts"]',
+                node: 'abstracts',
+            },
+            {
+                tag: 'sec[sec-type="body"]',
+                node: 'body',
+            },
+            {
+                tag: 'sec[sec-type="backmatter"]',
+                node: 'backmatter',
+            },
+            {
+                tag: 'sec[sec-type="box-element"]',
+                node: 'box_element',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        label: getTrimmedTextContent(element, 'label'),
+                    };
+                },
+            },
+            {
+                tag: 'sec[sec-type="bibliography"]',
+                node: 'bibliography_section',
+            },
+            {
+                tag: 'ref-list',
+                context: 'bibliography_section/',
+                node: 'bibliography_element',
+            },
+            {
+                tag: 'ref',
+                context: 'bibliography_element/',
+                node: 'bibliography_item',
+                getAttrs: (node) => this.parseRef(node),
+            },
+            {
+                tag: 'sec[sec-type="abstract-graphical"]',
+                node: 'graphical_abstract_section',
+            },
+            {
+                tag: 'sec',
+                node: 'section',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                        category: this.chooseSectionCategory(element),
+                    };
+                },
+            },
+            {
+                tag: 'kwd-group-list',
+                context: 'keywords/',
+                node: 'keywords_element',
+            },
+            {
+                tag: 'kwd-group',
+                context: 'keywords_element/',
+                node: 'keyword_group',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.id,
+                        type: element.getAttribute('kwd-group-type'),
+                    };
+                },
+            },
+            {
+                tag: 'kwd',
+                context: 'keyword_group//',
+                node: 'keyword',
+            },
+            {
+                tag: 'label',
+                context: 'box_element/',
+                ignore: true,
+            },
+            {
+                tag: 'boxed-text',
+                ignore: true,
+            },
+            {
+                tag: 'label',
+                context: 'section/',
+                node: 'section_label',
+            },
+            {
+                tag: 'label',
+                context: 'table_element/',
+                ignore: true,
+            },
+            {
+                tag: 'label',
+                context: 'figure/',
+                ignore: true,
+            },
+            {
+                tag: 'table',
+                node: 'table',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'table-wrap',
+                node: 'table_element',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        id: element.getAttribute('id'),
+                    };
+                },
+            },
+            {
+                tag: 'title',
+                node: 'section_title',
+                context: 'section/|footnotes_section/|bibliography_section/|keywords/|supplements/|author_notes/|graphical_abstract_section/',
+            },
+            {
+                tag: 'title',
+                node: 'caption_title',
+                context: 'figcaption/',
+            },
+            {
+                tag: 'tr',
+                node: 'table_row',
+            },
+            {
+                tag: 'td',
+                node: 'table_cell',
+                getAttrs: (node) => {
+                    const element = node;
+                    const colspan = parseInt(element.getAttribute('colspan') || '1');
+                    const rowspan = parseInt(element.getAttribute('rowspan') || '1');
+                    return Object.assign(Object.assign(Object.assign({}, (colspan && { colspan })), (rowspan && { rowspan })), { valign: element.getAttribute('valign'), align: element.getAttribute('align'), scope: element.getAttribute('scope'), style: element.getAttribute('style') });
+                },
+            },
+            {
+                tag: 'th',
+                node: 'table_header',
+                getAttrs: (node) => {
+                    const element = node;
+                    const colspan = parseInt(element.getAttribute('colspan') || '1');
+                    const rowspan = parseInt(element.getAttribute('rowspan') || '1');
+                    return Object.assign(Object.assign(Object.assign({}, (colspan && { colspan })), (rowspan && { rowspan })), { valign: element.getAttribute('valign'), align: element.getAttribute('align'), scope: element.getAttribute('scope'), style: element.getAttribute('style') });
+                },
+            },
+            {
+                tag: 'col',
+                node: 'table_col',
+                getAttrs: (node) => {
+                    const element = node;
+                    return {
+                        width: element.getAttribute('width'),
+                    };
+                },
+            },
+            {
+                tag: 'colgroup',
+                node: 'table_colgroup',
+            },
+            {
+                tag: 'xref[ref-type="bibr"]',
+                node: 'citation',
+                getAttrs: (node) => {
+                    var _a;
+                    const element = node;
+                    return {
+                        rids: ((_a = element.getAttribute('rid')) === null || _a === void 0 ? void 0 : _a.split(/\s+/)) || [],
+                        contents: getTrimmedTextContent(element),
+                    };
+                },
+            },
+            {
+                tag: 'xref[ref-type="fn"]',
+                node: 'inline_footnote',
+                getAttrs: (node) => {
+                    var _a;
+                    const element = node;
+                    return {
+                        rids: ((_a = element.getAttribute('rid')) === null || _a === void 0 ? void 0 : _a.split(/\s+/)) || [],
+                    };
+                },
+            },
+            {
+                tag: 'xref',
+                node: 'cross_reference',
+                getAttrs: (node) => {
+                    var _a;
+                    const element = node;
+                    return {
+                        rids: ((_a = element.getAttribute('rid')) === null || _a === void 0 ? void 0 : _a.split(/\s+/)) || [],
+                        label: getTrimmedTextContent(element),
+                    };
+                },
+            },
+        ];
+        this.marks = [
+            {
+                tag: 'bold',
+                mark: 'bold',
+            },
+            {
+                tag: 'code',
+                mark: 'code',
+            },
+            {
+                tag: 'italic',
+                mark: 'italic',
+            },
+            {
+                tag: 'sc',
+                mark: 'smallcaps',
+            },
+            {
+                tag: 'strike',
+                mark: 'strikethrough',
+            },
+            {
+                tag: 'styled-content',
+                mark: 'styled',
+                getAttrs: (node) => ({
+                    style: node.getAttribute('style'),
+                }),
+            },
+            {
+                tag: 'sub',
+                mark: 'subscript',
+            },
+            {
+                tag: 'sup',
+                mark: 'superscript',
+            },
+            {
+                tag: 'underline',
+                mark: 'underline',
+            },
+        ];
+        this.parser = new DOMParser(this.schema, [...this.marks, ...this.nodes]);
+    }
+    parse(doc, options) {
+        return this.parser.parse(doc, options);
+    }
+    isMatchingCategory(secType, titleNode, category) {
+        if (secType && category.synonyms.includes(secType)) {
+            return true;
+        }
+        if (titleNode && titleNode.nodeName === 'title' && titleNode.textContent) {
+            const textContent = titleNode.textContent.trim().toLowerCase();
+            if (category.synonyms.includes(textContent)) {
+                return true;
+            }
+        }
+        return false;
+    }
+    chooseSectionCategory(section) {
+        const secType = section.getAttribute('sec-type');
+        const titleNode = section.firstElementChild;
+        for (const category of this.sectionCategories) {
+            if (this.isMatchingCategory(secType, titleNode, category)) {
+                return category.id;
+            }
+        }
+    }
+}