@manuscripts/transform 2.10.0-LEAN-3577 → 3.0.0

package/dist/cjs/jats/importer/{jats-body-dom-parser.js → jats-dom-parser.js}

@@ -18,11 +18,15 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.jatsBodyDOMParser = void 0;
+exports.jatsDOMParser = void 0;
+const json_schema_1 = require("@manuscripts/json-schema");
 const mime_1 = __importDefault(require("mime"));
 const prosemirror_model_1 = require("prosemirror-model");
+const utils_1 = require("../../lib/utils");
 const schema_1 = require("../../schema");
 const transformer_1 = require("../../transformer");
+const jats_comments_1 = require("./jats-comments");
+const jats_parser_utils_1 = require("./jats-parser-utils");
 const XLINK_NAMESPACE = 'http://www.w3.org/1999/xlink';
 const chooseContentType = (graphicNode) => {
     if (graphicNode) {
@@ -37,6 +41,12 @@ const chooseContentType = (graphicNode) => {
         }
     }
 };
+const parsePriority = (priority) => {
+    if (!priority) {
+        return undefined;
+    }
+    return parseInt(priority);
+};
 const getEquationContent = (p) => {
     var _a;
     const element = p;
@@ -48,7 +58,7 @@ const getEquationContent = (p) => {
         const nodeName = child.nodeName.replace(/^[a-z]:/, '');
         switch (nodeName) {
             case 'tex-math':
-                contents = child.innerHTML;
+                contents = child.textContent;
                 format = 'tex';
                 break;
             case 'mml:math':
@@ -59,6 +69,129 @@ const getEquationContent = (p) => {
     }
     return { id, format, contents };
 };
+const getEmail = (element) => {
+    var _a, _b, _c;
+    const email = element.querySelector('email');
+    if (email) {
+        return {
+            href: (_a = email.getAttributeNS(XLINK_NAMESPACE, 'href')) !== null && _a !== void 0 ? _a : '',
+            text: (_c = (_b = email.textContent) === null || _b === void 0 ? void 0 : _b.trim()) !== null && _c !== void 0 ? _c : '',
+        };
+    }
+};
+const getInstitutionDetails = (element) => {
+    var _a;
+    let department = '';
+    let institution = '';
+    for (const node of element.querySelectorAll('institution')) {
+        const content = (_a = node.textContent) === null || _a === void 0 ? void 0 : _a.trim();
+        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 (0, utils_1.getTrimmedTextContent)(element, `addr-line:nth-of-type(${index})`) || '';
+};
+const getHTMLContent = (node, querySelector) => {
+    return (0, jats_parser_utils_1.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) => {
+            var _a, _b, _c;
+            if (item.nodeType === Node.TEXT_NODE &&
+                ((_a = item.textContent) === null || _a === void 0 ? void 0 : _a.match(/[A-Za-z]+/g))) {
+                attrs.literal = (_c = (_b = mixedCitation.textContent) === null || _b === void 0 ? void 0 : _b.trim()) !== null && _c !== void 0 ? _c : '';
+                return attrs;
+            }
+        });
+    }
+    const source = getHTMLContent(element, 'source');
+    if (source) {
+        attrs.containerTitle = source;
+    }
+    const volume = (0, utils_1.getTrimmedTextContent)(element, 'volume');
+    if (volume) {
+        attrs.volume = volume;
+    }
+    const issue = (0, utils_1.getTrimmedTextContent)(element, 'issue');
+    if (issue) {
+        attrs.issue = issue;
+    }
+    const supplement = (0, utils_1.getTrimmedTextContent)(element, 'supplement');
+    if (supplement) {
+        attrs.supplement = supplement;
+    }
+    const fpage = (0, utils_1.getTrimmedTextContent)(element, 'fpage');
+    const lpage = (0, utils_1.getTrimmedTextContent)(element, 'lpage');
+    if (fpage) {
+        attrs.page = lpage ? `${fpage}-${lpage}` : fpage;
+    }
+    const year = (0, utils_1.getTrimmedTextContent)(element, 'year');
+    if (year) {
+        attrs.issued = (0, json_schema_1.buildBibliographicDate)({
+            'date-parts': [[year]],
+        });
+    }
+    const doi = (0, utils_1.getTrimmedTextContent)(element, 'pub-id[pub-id-type="doi"]');
+    if (doi) {
+        attrs.doi = doi;
+    }
+    const authors = [];
+    authorNodes.forEach((authorNode) => {
+        var _a;
+        const name = (0, json_schema_1.buildBibliographicName)({});
+        const given = (0, utils_1.getTrimmedTextContent)(authorNode, 'given-names');
+        if (given) {
+            name.given = given;
+        }
+        const family = (0, utils_1.getTrimmedTextContent)(authorNode, 'surname');
+        if (family) {
+            name.family = family;
+        }
+        if (authorNode.nodeName === 'collab') {
+            name.literal = (_a = authorNode.textContent) === null || _a === void 0 ? void 0 : _a.trim();
+        }
+        authors.push(name);
+    });
+    if (authors.length) {
+        attrs.author = authors;
+    }
+    return attrs;
+};
 const marks = [
     {
         tag: 'bold',
@@ -101,6 +234,182 @@ const marks = [
     },
 ];
 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"]');
+            return {
+                doi: doi === null || doi === void 0 ? void 0 : doi.textContent,
+                articleType: (_a = element.getAttribute('article-type')) !== null && _a !== void 0 ? _a : '',
+                primaryLanguageCode: (_b = element.getAttribute('lang')) !== null && _b !== void 0 ? _b : '',
+            };
+        },
+    },
+    {
+        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: element.textContent,
+                contributions: [(0, json_schema_1.buildContribution)(jats_comments_1.DEFAULT_PROFILE_ID)],
+            };
+        },
+    },
+    {
+        tag: 'author-notes',
+        node: 'author_notes',
+        getAttrs: (node) => {
+            const element = node;
+            return {
+                id: element.getAttribute('id'),
+            };
+        },
+    },
+    {
+        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) => {
+            var _a;
+            const element = node;
+            const label = element.querySelector('label');
+            if (label) {
+                label.remove();
+            }
+            return {
+                id: element.getAttribute('id'),
+                label: (_a = label === null || label === void 0 ? void 0 : label.textContent) === null || _a === void 0 ? void 0 : _a.trim(),
+            };
+        },
+        getContent: (node) => {
+            var _a;
+            const element = node;
+            return prosemirror_model_1.Fragment.from(schema_1.schema.text(((_a = element.textContent) === null || _a === void 0 ? void 0 : _a.trim()) || ''));
+        },
+    },
+    {
+        tag: 'contrib[contrib-type="author"]',
+        node: 'contributor',
+        getAttrs: (node) => {
+            var _a, _b;
+            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: ((_a = xref.textContent) === null || _a === void 0 ? void 0 : _a.trim()) || '',
+                        });
+                        break;
+                    case 'corresp':
+                        corresp.push({
+                            correspID: rid,
+                            correspLabel: ((_b = xref.textContent) === null || _b === void 0 ? void 0 : _b.trim()) || '',
+                        });
+                        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: (0, utils_1.getTrimmedTextContent)(element, 'name > given-names'),
+                    family: (0, utils_1.getTrimmedTextContent)(element, 'name > surname'),
+                    ObjectType: json_schema_1.ObjectTypes.BibliographicName,
+                },
+                ORCIDIdentifier: (0, utils_1.getTrimmedTextContent)(element, 'contrib-id[contrib-id-type="orcid"]'),
+                priority: parsePriority(element.getAttribute('priority')),
+            };
+        },
+        getContent: () => {
+            return prosemirror_model_1.Fragment.from(schema_1.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 = (0, utils_1.getTrimmedTextContent)(element, 'postal-code')) !== null && _a !== void 0 ? _a : '',
+                country: (_b = (0, utils_1.getTrimmedTextContent)(element, 'country')) !== null && _b !== void 0 ? _b : '',
+                email: getEmail(element),
+                priority: parsePriority(element.getAttribute('priority')),
+            };
+        },
+        getContent: () => {
+            return prosemirror_model_1.Fragment.from(schema_1.schema.text('_'));
+        },
+    },
     {
         tag: 'attrib',
         node: 'attribution',
@@ -109,10 +418,6 @@ const nodes = [
         tag: 'back',
         ignore: true,
     },
-    {
-        tag: 'body',
-        node: 'manuscript',
-    },
     {
         tag: 'break',
         node: 'hard_break',
@@ -132,13 +437,13 @@ const nodes = [
             const title = element.querySelector('title');
             if (title) {
                 const captionTitle = schema.nodes.caption_title.create();
-                content.push(exports.jatsBodyDOMParser.parse(title, { topNode: captionTitle }));
+                content.push(exports.jatsDOMParser.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(exports.jatsBodyDOMParser.parse(paragraph, { topNode: figcaption }));
+                    content.push(exports.jatsDOMParser.parse(paragraph, { topNode: figcaption }));
                 }
             }
             return prosemirror_model_1.Fragment.from(content);
@@ -149,6 +454,17 @@ const nodes = [
         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',
@@ -329,7 +645,7 @@ const nodes = [
             const paragraphs = [];
             node.childNodes.forEach((p) => {
                 const paragraph = schema_1.schema.nodes.paragraph.create();
-                const content = exports.jatsBodyDOMParser.parse(p, {
+                const content = exports.jatsDOMParser.parse(p, {
                     topNode: paragraph,
                 });
                 paragraphs.push(content);
@@ -428,6 +744,33 @@ const nodes = [
         tag: 'sec[sec-type="backmatter"]',
         node: 'backmatter',
     },
+    {
+        tag: 'sec[sec-type="box-element"]',
+        node: 'box_element',
+        getAttrs: (node) => {
+            var _a;
+            const element = node;
+            return {
+                id: element.getAttribute('id'),
+                label: (_a = element.querySelector('label')) === null || _a === void 0 ? void 0 : _a.textContent,
+            };
+        },
+    },
+    {
+        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',
         node: 'section',
@@ -451,6 +794,7 @@ const nodes = [
         getAttrs: (node) => {
             const element = node;
             return {
+                id: element.id,
                 type: element.getAttribute('kwd-group-type'),
             };
         },
@@ -460,6 +804,15 @@ const nodes = [
         context: 'keyword_group//',
         node: 'keyword',
     },
+    {
+        tag: 'label',
+        context: 'box_element/',
+        ignore: true,
+    },
+    {
+        tag: 'boxed-text',
+        ignore: true,
+    },
     {
         tag: 'label',
         context: 'section/',
@@ -498,7 +851,7 @@ const nodes = [
     {
         tag: 'title',
         node: 'section_title',
-        context: 'section/|footnotes_section/|bibliography_section/|keywords/|supplements/',
+        context: 'section/|footnotes_section/|bibliography_section/|keywords/|supplements/|author_notes/',
     },
     {
         tag: 'title',
@@ -580,4 +933,4 @@ const nodes = [
         },
     },
 ];
-exports.jatsBodyDOMParser = new prosemirror_model_1.DOMParser(schema_1.schema, [...marks, ...nodes]);
+exports.jatsDOMParser = new prosemirror_model_1.DOMParser(schema_1.schema, [...marks, ...nodes]);

package/dist/cjs/jats/importer/jats-journal-meta-parser.js

@@ -15,7 +15,8 @@
  * limitations under the License.
  */
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.parseJournalMeta = exports.parseJournalISSNs = exports.parseJournalAbbreviatedTitles = exports.parseJournalIdentifiers = void 0;
+exports.parseJournal = exports.parseJournalMeta = exports.parseJournalISSNs = exports.parseJournalAbbreviatedTitles = exports.parseJournalIdentifiers = void 0;
+const json_schema_1 = require("@manuscripts/json-schema");
 const utils_1 = require("../../lib/utils");
 const parseJournalIdentifiers = (element) => {
     var _a;
@@ -91,3 +92,12 @@ const parseJournalMeta = (element) => {
     };
 };
 exports.parseJournalMeta = parseJournalMeta;
+const parseJournal = (doc) => {
+    const journalMeta = doc.querySelector('journal-meta');
+    if (!journalMeta) {
+        return;
+    }
+    const meta = (0, exports.parseJournalMeta)(journalMeta);
+    return Object.assign(Object.assign({}, meta), { objectType: json_schema_1.ObjectTypes.Journal });
+};
+exports.parseJournal = parseJournal;

package/dist/cjs/jats/importer/jats-parser-utils.js

@@ -19,18 +19,21 @@ exports.htmlFromJatsNode = exports.updateDocumentIDs = void 0;
 const uuid_1 = require("uuid");
 const schema_1 = require("../../schema");
 const transformer_1 = require("../../transformer");
-const updateDocumentIDs = (node, replacements) => {
+const updateDocumentIDs = (node) => {
+    const replacements = new Map();
     const warnings = [];
     recurseDoc(node, (n) => updateNodeID(n, replacements, warnings));
     recurseDoc(node, (n) => updateNodeRID(n, replacements, warnings));
     recurseDoc(node, (n) => updateNodeRIDS(n, replacements, warnings));
+    recurseDoc(node, (n) => updateContributorNodesIDS(n, replacements, warnings));
+    recurseDoc(node, (n) => updateCommentTarget(n, replacements, warnings));
     return warnings;
 };
 exports.updateDocumentIDs = updateDocumentIDs;
-function recurseDoc(node, fn) {
+const recurseDoc = (node, fn) => {
     fn(node);
     node.descendants((n) => fn(n));
-}
+};
 const updateNodeID = (node, replacements, warnings) => {
     if (node.type === schema_1.schema.nodes.inline_equation) {
         node.attrs = Object.assign(Object.assign({}, node.attrs), { id: `InlineMathFragment:${(0, uuid_1.v4)()}` });
@@ -40,6 +43,10 @@ const updateNodeID = (node, replacements, warnings) => {
         node.attrs = Object.assign(Object.assign({}, node.attrs), { id: `GeneralTableFootnote:${(0, uuid_1.v4)()}` });
         return;
     }
+    if (node.type === schema_1.schema.nodes.comment ||
+        node.type === schema_1.schema.nodes.highlight_marker) {
+        return;
+    }
     if (!('id' in node.attrs)) {
         return;
     }
@@ -49,7 +56,7 @@ const updateNodeID = (node, replacements, warnings) => {
         return;
     }
     const previousID = node.attrs.id;
-    const nextID = (0, transformer_1.generateID)(objectType);
+    const nextID = (0, transformer_1.generateNodeID)(node.type);
     if (previousID) {
         if (replacements.has(previousID) ||
             Array.from(replacements.values()).includes(previousID)) {
@@ -78,6 +85,44 @@ const updateNodeRIDS = (node, replacements, warnings) => {
     }
     node.attrs = Object.assign(Object.assign({}, node.attrs), { rids: previousRIDs.map((r) => replacements.get(r) || r) });
 };
+const updateContributorNodesIDS = (node, replacements, warnings) => {
+    var _a, _b;
+    if (node.type === schema_1.schema.nodes.contributor) {
+        const footnote = (_a = node.attrs.footnote) === null || _a === void 0 ? void 0 : _a.map((fn) => {
+            return replacements.get(fn.noteID)
+                ? Object.assign(Object.assign({}, fn), { noteID: replacements.get(fn.noteID) }) : undefined;
+        }).filter(Boolean);
+        const corresp = (_b = node.attrs.corresp) === null || _b === void 0 ? void 0 : _b.map((corresp) => {
+            return replacements.get(corresp.correspID)
+                ? Object.assign(Object.assign({}, corresp), { correspID: replacements.get(corresp.correspID) }) : undefined;
+        }).filter(Boolean);
+        const affiliations = node.attrs.affiliations
+            .map((affiliation) => {
+            return replacements.get(affiliation);
+        })
+            .filter(Boolean);
+        node.attrs = Object.assign(Object.assign({}, node.attrs), { footnote,
+            corresp,
+            affiliations });
+    }
+    if (node.type !== schema_1.schema.nodes.contributors) {
+        return false;
+    }
+};
+const updateCommentTarget = (node, replacements, warnings) => {
+    if (node.type !== schema_1.schema.nodes.comment) {
+        return;
+    }
+    const target = node.attrs.target;
+    if (!target) {
+        return;
+    }
+    if (!replacements.has(target)) {
+    }
+    else {
+        node.attrs = Object.assign(Object.assign({}, node.attrs), { target: replacements.get(target) });
+    }
+};
 const JATS_TO_HTML_MAPPING = new Map([
     ['bold', 'b'],
     ['italic', 'i'],
@@ -113,6 +158,9 @@ const htmlFromJatsNode = (element, createElement) => {
     if (!element) {
         return undefined;
     }
+    if (!createElement) {
+        createElement = (tagName) => element.ownerDocument.createElement(tagName);
+    }
     const temp = createElement('template');
     renameJatsNodesToHTML(element, temp, createElement);
     return temp.innerHTML;