jats-xml 0.0.3 → 0.0.5

package/README.md

@@ -41,7 +41,7 @@ jats summary /local/article.jats
 
 This will provide a summary, including a list of what the JATS file contains.
 
-![Output of `jats summary`](images/jats-output.png)
+![Output of `jats summary`](/packages/jats-xml/images/jats-output.png)
 
 ## Working in Typescript
 

package/dist/cjs/cli/parse.js

@@ -30,7 +30,7 @@ const utils_1 = require("../utils");
 function hasValidExtension(output) {
     return ['.xml', '.jats'].includes((0, path_1.extname)(output).toLowerCase());
 }
-function downloadJats(session, urlOrDoi, output) {
+function downloadAndSaveJats(session, urlOrDoi, output) {
     return __awaiter(this, void 0, void 0, function* () {
         if (fs_1.default.existsSync(urlOrDoi)) {
             throw new Error(`File "${urlOrDoi}" is local and cannot be downloaded!`);
@@ -41,7 +41,7 @@ function downloadJats(session, urlOrDoi, output) {
         if (!hasValidExtension(output)) {
             session.log.warn(`The extension ${(0, path_1.extname)(output)} is not a valid extension for JATS, try using ".xml" or ".jats"`);
         }
-        const data = yield (0, download_1.downloadJatsFromUrl)(session, urlOrDoi, resolvers_1.DEFAULT_RESOLVERS);
+        const { data } = yield (0, download_1.downloadJatsFromUrl)(session, urlOrDoi, resolvers_1.DEFAULT_RESOLVERS);
         (0, myst_cli_utils_1.writeFileToFolder)(output, data);
         return data;
     });
@@ -52,15 +52,12 @@ function parseJats(session, file) {
         if (fs_1.default.existsSync(file)) {
             session.log.debug(`Found ${file} locally, parsing`);
             const data = fs_1.default.readFileSync(file).toString();
-            session.log.debug(toc(`Parsed JATS file from disk in %s`));
-            return new jats_1.Jats(data);
+            return new jats_1.Jats(data, { log: session.log });
         }
-        if (doi_utils_1.default.validate(file) || (0, myst_cli_utils_1.isUrl)(file)) {
-            const data = yield (0, download_1.downloadJatsFromUrl)(session, file, resolvers_1.DEFAULT_RESOLVERS);
-            session.log.debug(toc(`Downloaded and parsed JATS file in %s`));
-            return new jats_1.Jats(data);
-        }
-        throw new Error(`Could not find ${file} locally, and it doesn't look like a URL or DOI`);
+        const { source, data } = yield (0, download_1.downloadJatsFromUrl)(session, file, resolvers_1.DEFAULT_RESOLVERS);
+        const jats = new jats_1.Jats(data, { source, log: session.log });
+        session.log.debug(toc(`Downloaded and parsed JATS file in %s`));
+        return jats;
     });
 }
 function formatLongString(data, offset = 0, length = 88 - offset) {
@@ -79,6 +76,9 @@ function formatDictionary(dict, opts) {
     return Object.entries(dict)
         .map(([k, t]) => {
         var _a;
+        if (!t)
+            return null;
+        let wrap = typeof (opts === null || opts === void 0 ? void 0 : opts.wrap) === 'boolean' ? opts.wrap : true;
         let value = t;
         let color = chalk_1.default.yellow.bold;
         if (t && typeof t === 'object') {
@@ -86,21 +86,21 @@ function formatDictionary(dict, opts) {
                 return null;
             color = (_a = t.label) !== null && _a !== void 0 ? _a : color;
             value = t.value;
+            wrap = typeof t.wrap === 'boolean' ? t.wrap : wrap;
         }
-        const wrapped = (opts === null || opts === void 0 ? void 0 : opts.wrap) === false
-            ? String(value)
-            : formatLongString(String(value), maxLabel + 2, opts === null || opts === void 0 ? void 0 : opts.wrap);
+        const wrapped = wrap ? formatLongString(String(value), maxLabel + 2) : String(value);
         return `${color(k)}:${' '.repeat(maxLabel - k.length + 1)}${wrapped}`;
     })
         .filter((o) => !!o)
         .join('\n');
 }
 function jatsSummaryCLI(session, file) {
-    var _a, _b;
+    var _a, _b, _c;
     return __awaiter(this, void 0, void 0, function* () {
         const jats = yield parseJats(session, file);
         const summary = {
-            DOI: jats.doi ? doi_utils_1.default.buildUrl(jats.doi) : null,
+            Source: { value: jats.source, wrap: false },
+            DOI: jats.doi ? { value: doi_utils_1.default.buildUrl(jats.doi), wrap: false } : null,
             Title: (_a = (0, myst_common_1.toText)(jats.articleTitle)) === null || _a === void 0 ? void 0 : _a.replace(/\n/g, ' '),
             Date: (0, utils_1.formatDate)((0, utils_1.toDate)(jats.publicationDate)),
             Authors: jats.articleAuthors
@@ -108,36 +108,44 @@ function jatsSummaryCLI(session, file) {
                 .join(', '),
             Abstract: (_b = (0, myst_common_1.toText)(jats.abstract)) === null || _b === void 0 ? void 0 : _b.replace(/\n/g, ' '),
             Keywords: jats.keywords.map((k) => (0, myst_common_1.toText)(k)).join(', '),
-            License: jats.license['xlink:href'],
-            Figures: { label: chalk_1.default.blue.bold, value: String((0, unist_util_select_1.selectAll)(types_1.Tags.fig, jats.body).length) },
-            Equations: {
+            License: (_c = jats.license) === null || _c === void 0 ? void 0 : _c['xlink:href'],
+        };
+        if (jats.body) {
+            summary.Figures = {
+                label: chalk_1.default.blue.bold,
+                value: String((0, unist_util_select_1.selectAll)(types_1.Tags.fig, jats.body).length),
+            };
+            summary.Equations = {
                 label: chalk_1.default.blue.bold,
                 value: String((0, unist_util_select_1.selectAll)(types_1.Tags.dispFormula, jats.body).length),
-            },
-            Tables: {
+            };
+            summary.Tables = {
                 label: chalk_1.default.blue.bold,
                 value: String((0, unist_util_select_1.selectAll)(types_1.Tags.table, jats.body).length),
-            },
-            Code: {
+            };
+            summary.Code = {
                 label: chalk_1.default.blue.bold,
                 value: String((0, unist_util_select_1.selectAll)(types_1.Tags.code, jats.body).length),
-            },
-            Sections: {
+            };
+            summary.Sections = {
                 label: chalk_1.default.blue.bold,
                 value: String((0, unist_util_select_1.selectAll)(types_1.Tags.sec, jats.body).length),
-            },
-            Paragraphs: {
+            };
+            summary.Paragraphs = {
                 label: chalk_1.default.blue.bold,
                 value: String((0, unist_util_select_1.selectAll)(types_1.Tags.p, jats.body).length),
-            },
-            Citations: { label: chalk_1.default.blue.bold, value: String(jats.references.length) },
-            'Cross-References': {
+            };
+            summary.Citations = { label: chalk_1.default.blue.bold, value: String(jats.references.length) };
+            summary['Cross-References'] = {
                 label: chalk_1.default.blue.bold,
                 value: String((0, unist_util_select_1.selectAll)(types_1.Tags.xref, jats.body).length),
-            },
-            'Sub Articles': { label: chalk_1.default.blue.bold, value: String(jats.subArticles.length) },
-        };
+            };
+            summary['Sub Articles'] = { label: chalk_1.default.blue.bold, value: String(jats.subArticles.length) };
+        }
         session.log.info(formatDictionary(summary));
+        if (!jats.body) {
+            session.log.warn('\nThis is a partial JATS record that does not have <body>.');
+        }
     });
 }
 function jatsReferencesCLI(session, file) {
@@ -145,7 +153,7 @@ function jatsReferencesCLI(session, file) {
         const jats = yield parseJats(session, file);
         const sorted = jats.references
             .map((ref) => {
-            const doi = (0, utils_1.findDoi)(ref);
+            const doiString = (0, utils_1.findArticleId)(ref, 'doi');
             const title = (0, myst_common_1.toText)((0, unist_util_select_1.select)(types_1.Tags.articleTitle, ref));
             const year = (0, myst_common_1.toText)((0, unist_util_select_1.select)(types_1.Tags.year, ref));
             const surnames = (0, unist_util_select_1.selectAll)(types_1.Tags.surname, ref);
@@ -158,7 +166,7 @@ function jatsReferencesCLI(session, file) {
             return {
                 Citation: `${short} (${year})`,
                 Title: title,
-                DOI: doi ? doi_utils_1.default.buildUrl(doi) : null,
+                DOI: doiString ? doi_utils_1.default.buildUrl(doiString) : null,
                 Count: s.length,
             };
         })
@@ -188,7 +196,7 @@ function makeDownloadCLI(program) {
         .description('Parse a JATS file and provide a summary')
         .argument('<url>', 'The JATS url or a DOI')
         .argument('<output>', 'The JATS url or a DOI')
-        .action((0, myst_cli_utils_1.clirun)(downloadJats, { program, getSession: session_1.getSession }));
+        .action((0, myst_cli_utils_1.clirun)(downloadAndSaveJats, { program, getSession: session_1.getSession }));
     return command;
 }
 function addDownloadCLI(program) {

package/dist/cjs/download.js

@@ -12,35 +12,186 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.downloadJatsFromUrl = void 0;
+exports.downloadJatsFromUrl = exports.checkIfPubMedCentralHasJats = exports.convertPMID2PMCID = void 0;
 const doi_utils_1 = __importDefault(require("doi-utils"));
+const chalk_1 = __importDefault(require("chalk"));
 const node_fetch_1 = __importDefault(require("node-fetch"));
 const myst_cli_utils_1 = require("myst-cli-utils");
+const fair_principles_1 = require("fair-principles");
 const resolvers_1 = require("./resolvers");
+function logAboutJatsFailing(session, jatsUrls) {
+    session.log.warn('⛔️ JATS may not be Open Access 😭, you should petition your local representative 🪧');
+    session.log.info(`${chalk_1.default.green(`\nThe XML ${chalk_1.default.bold('should')} be here:\n\n${jatsUrls.join('\n')}`)}\n`);
+    const FAIR = (0, fair_principles_1.highlightFAIR)('A', { chalk: chalk_1.default });
+    session.log.info(`Some publishers aggressively block programmatic access, which isn't ${FAIR}.`);
+    session.log.debug((0, fair_principles_1.formatPrinciples)('A*', { chalk: chalk_1.default }));
+    session.log.info(`${chalk_1.default.blue('The link may work in a browser.')}\n`);
+}
 function dowloadFromUrl(session, jatsUrl) {
     return __awaiter(this, void 0, void 0, function* () {
+        const toc = (0, myst_cli_utils_1.tic)();
         session.log.debug(`Fetching JATS from ${jatsUrl}`);
-        const jatsResp = yield (0, node_fetch_1.default)(jatsUrl);
-        if (!jatsResp.ok)
-            throw new Error(`Problem fetching JATS from ${jatsUrl}`);
-        const contentType = jatsResp.headers.get('content-type');
-        if (!(contentType === 'application/xml' || (contentType === null || contentType === void 0 ? void 0 : contentType.includes('text/plain')))) {
-            throw new Error(`Expected content-type "application/xml" instead we got ${contentType} for ${jatsUrl}`);
-        }
-        const data = yield jatsResp.text();
+        const resp = yield (0, node_fetch_1.default)(jatsUrl, {
+            headers: [
+                ['accept', 'application/xml'],
+                [
+                    'user-agent',
+                    // A bunch of publishers just show the login screen or quickly block you.
+                    // We don't want to DDOS these publishers, they are the _good ones_ for sharing the XML!!
+                    // But some block on the second request?!
+                    // So we can pretend to be a random browser, I guess. How silly. 🤷‍♂️
+                    `Mozilla/5.0 (Macintosh; Intel Mac OS X ${Math.floor(Math.random() * 100)})`,
+                ],
+            ],
+        });
+        if (!resp.ok) {
+            session.log.debug(`JATS failed to download from "${jatsUrl}"`);
+            throw new Error(`STATUS ${resp.status}: ${resp.statusText}`);
+        }
+        const contentType = resp.headers.get('content-type');
+        if (!((contentType === null || contentType === void 0 ? void 0 : contentType.includes('application/xml')) ||
+            (contentType === null || contentType === void 0 ? void 0 : contentType.includes('text/xml')) ||
+            (contentType === null || contentType === void 0 ? void 0 : contentType.includes('text/plain')))) {
+            session.log.warn(`Expected content-type "application/xml" instead we got "${contentType}" for ${jatsUrl}\n${chalk_1.default.dim('Things may not work, but we are going to try our best...')}`);
+        }
+        const data = yield resp.text();
+        session.log.debug(toc(`Fetched document with content-type "${contentType}" in %s`));
         return data;
     });
 }
+/**
+ * There are 5.8M or so DOIs that have a full XML record:
+ *
+ * https://api.crossref.org/works?filter=full-text.type:application/xml,full-text.application:text-mining&facet=publisher-name:*&rows=0
+ *
+ * This function tries to find the correct URL for the record.
+ */
+function checkIfDoiHasJats(session, urlOrDoi) {
+    var _a, _b, _c, _d;
+    return __awaiter(this, void 0, void 0, function* () {
+        if (!doi_utils_1.default.validate(urlOrDoi))
+            return;
+        const toc = (0, myst_cli_utils_1.tic)();
+        const doiUrl = doi_utils_1.default.buildUrl(urlOrDoi);
+        session.log.debug(`Attempting to resolving full XML from DOI ${doiUrl}`);
+        const resp = yield (0, node_fetch_1.default)(doiUrl, { headers: [['Accept', 'application/json']] });
+        if (!resp.ok) {
+            // Silently return -- other functions can try!
+            session.log.debug(`DOI failed to resolve: ${doiUrl}`);
+            return;
+        }
+        const data = (yield resp.json());
+        session.log.debug(toc(`DOI resolved in %s with ${(_b = (_a = data.link) === null || _a === void 0 ? void 0 : _a.length) !== null && _b !== void 0 ? _b : 0} links to content`));
+        if (data.link) {
+            session.log.debug(['', ...data.link.map((link) => `content-type: ${link['content-type']}, ${link.URL}\n`)].join('  - '));
+        }
+        const fullXml = (_d = (_c = data.link) === null || _c === void 0 ? void 0 : _c.find((link) => { var _a; return ['text/xml', 'application/xml'].includes((_a = link['content-type']) !== null && _a !== void 0 ? _a : ''); })) === null || _d === void 0 ? void 0 : _d.URL;
+        if (fullXml)
+            return fullXml;
+        session.log.debug(`Could not find XML in DOI record ${doiUrl}`);
+        return undefined;
+    });
+}
+/**
+ * https://www.ncbi.nlm.nih.gov/pmc/tools/id-converter-api/
+ */
+function convertPMID2PMCID(session, PMID) {
+    var _a, _b;
+    return __awaiter(this, void 0, void 0, function* () {
+        if (PMID.startsWith('https://')) {
+            const idPart = new URL(PMID).pathname.slice(1);
+            session.log.debug(`Extract ${PMID} to ${idPart}`);
+            return convertPMID2PMCID(session, idPart);
+        }
+        const toc = (0, myst_cli_utils_1.tic)();
+        const converter = 'https://www.ncbi.nlm.nih.gov/pmc/utils/idconv/v1.0/';
+        const resp = yield (0, node_fetch_1.default)(`${converter}?tool=jats-xml&format=json&ids=${PMID}`, {
+            headers: [['Accept', 'application/json']],
+        });
+        if (!resp.ok) {
+            // Silently return -- other functions can try!
+            session.log.debug(`Failed to convert PubMedID: ${PMID}`);
+            return;
+        }
+        const data = yield resp.json();
+        const PMCID = (_b = (_a = data === null || data === void 0 ? void 0 : data.records) === null || _a === void 0 ? void 0 : _a[0]) === null || _b === void 0 ? void 0 : _b.pmcid;
+        session.log.debug(toc(`Used nih.gov to transform ${PMID} to ${PMCID} in %s.`));
+        return PMCID;
+    });
+}
+exports.convertPMID2PMCID = convertPMID2PMCID;
+function pubMedCentralJats(PMCID) {
+    const normalized = PMCID.replace(/^PMC:?/, '');
+    return `https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pmc&id=${normalized}`;
+}
+function checkIfPubMedCentralHasJats(session, urlOrDoi) {
+    var _a, _b;
+    return __awaiter(this, void 0, void 0, function* () {
+        if (urlOrDoi.match(/^PMC:?([0-9]+)$/))
+            return pubMedCentralJats(urlOrDoi);
+        if (!doi_utils_1.default.validate(urlOrDoi))
+            return;
+        const toc = (0, myst_cli_utils_1.tic)();
+        const doiUrl = doi_utils_1.default.buildUrl(urlOrDoi);
+        session.log.debug(`Attempting to resolve PMCID using OpenAlex from ${doiUrl}`);
+        const openAlexUrl = `https://api.openalex.org/works/${doiUrl}`;
+        const resp = yield (0, node_fetch_1.default)(openAlexUrl, { headers: [['Accept', 'application/json']] });
+        if (!resp.ok) {
+            // Silently return -- other functions can try!
+            session.log.debug(`Failed to lookup on OpenAlex: ${openAlexUrl}`);
+            return;
+        }
+        const data = (yield resp.json());
+        const PMID = (_a = data === null || data === void 0 ? void 0 : data.ids) === null || _a === void 0 ? void 0 : _a.pmid;
+        let PMCID = (_b = data === null || data === void 0 ? void 0 : data.ids) === null || _b === void 0 ? void 0 : _b.pmcid;
+        if (!PMCID && !!PMID) {
+            session.log.debug(toc(`OpenAlex resolved ${data === null || data === void 0 ? void 0 : data.ids.openalex} in %s. There is no PMCID, but there is a PMID`));
+            PMCID = yield convertPMID2PMCID(session, PMID);
+            if (!PMCID) {
+                session.log.debug(toc(`PubMed does not have a record of ${PMID}`));
+                return;
+            }
+        }
+        if (!PMCID) {
+            session.log.debug(toc(`OpenAlex resolved ${data === null || data === void 0 ? void 0 : data.ids.openalex} in %s, but there is no PMCID`));
+            return;
+        }
+        session.log.debug(toc(`OpenAlex resolved in %s, with a PMCID of ${PMCID}`));
+        return pubMedCentralJats(PMCID);
+    });
+}
+exports.checkIfPubMedCentralHasJats = checkIfPubMedCentralHasJats;
 function downloadJatsFromUrl(session, urlOrDoi, resolvers) {
     return __awaiter(this, void 0, void 0, function* () {
+        const expectedUrls = (yield Promise.all([
+            checkIfPubMedCentralHasJats(session, urlOrDoi),
+            checkIfDoiHasJats(session, urlOrDoi),
+        ])).filter((u) => !!u);
+        if (expectedUrls.length > 0) {
+            session.log.debug(['Trying URLs:\n', ...expectedUrls.map((url) => ` ${url}\n`)].join('  - '));
+            for (let index = 0; index < expectedUrls.length; index++) {
+                const url = expectedUrls[index];
+                try {
+                    const data = yield dowloadFromUrl(session, url);
+                    if (data)
+                        return { source: url, data };
+                }
+                catch (error) {
+                    session.log.debug(error.message);
+                }
+            }
+            // If there are expected URLs that don't work: see something, say something, etc.
+            logAboutJatsFailing(session, expectedUrls);
+        }
         if (doi_utils_1.default.validate(urlOrDoi)) {
-            const jatsUrl = yield (0, resolvers_1.resolveJatsUrlFromDoi)(session, urlOrDoi, resolvers);
+            const jatsUrl = yield (0, resolvers_1.customResolveJatsUrlFromDoi)(session, urlOrDoi, resolvers);
             const data = yield dowloadFromUrl(session, jatsUrl);
-            return data;
+            return { source: jatsUrl, data };
         }
         if ((0, myst_cli_utils_1.isUrl)(urlOrDoi)) {
+            session.log.debug("No resolver matched, and the URL doesn't look like a DOI. We will attempt to download it directly.");
             const data = yield dowloadFromUrl(session, urlOrDoi);
-            return data;
+            return { source: urlOrDoi, data };
         }
         throw new Error(`Could not find ${urlOrDoi} locally, and it doesn't look like a URL or DOI`);
     });

package/dist/cjs/index.js

@@ -17,9 +17,11 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.Jats = exports.version = void 0;
+exports.downloadJatsFromUrl = exports.Jats = exports.version = void 0;
 var version_1 = require("./version");
 Object.defineProperty(exports, "version", { enumerable: true, get: function () { return __importDefault(version_1).default; } });
 var jats_1 = require("./jats");
 Object.defineProperty(exports, "Jats", { enumerable: true, get: function () { return jats_1.Jats; } });
+var download_1 = require("./download");
+Object.defineProperty(exports, "downloadJatsFromUrl", { enumerable: true, get: function () { return download_1.downloadJatsFromUrl; } });
 __exportStar(require("./types"), exports);

package/dist/cjs/jats.js

@@ -1,41 +1,90 @@
 "use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.Jats = void 0;
+const myst_common_1 = require("myst-common");
 const xml_js_1 = require("xml-js");
+const doi_utils_1 = __importDefault(require("doi-utils"));
 const utils_1 = require("./utils");
 const unist_util_select_1 = require("unist-util-select");
 const types_1 = require("./types");
+const myst_cli_utils_1 = require("myst-cli-utils");
+function select(selector, node) {
+    var _a;
+    return ((_a = (0, unist_util_select_1.select)(selector, node)) !== null && _a !== void 0 ? _a : undefined);
+}
 class Jats {
-    constructor(data) {
-        this.raw = (0, xml_js_1.xml2js)(data, { compact: false });
+    constructor(data, opts) {
+        var _a;
+        const toc = (0, myst_cli_utils_1.tic)();
+        this.log = opts === null || opts === void 0 ? void 0 : opts.log;
+        if (opts === null || opts === void 0 ? void 0 : opts.source)
+            this.source = opts.source;
+        try {
+            this.raw = (0, xml_js_1.xml2js)(data, { compact: false });
+        }
+        catch (error) {
+            throw new Error('Problem parsing the JATS document, please ensure it is XML');
+        }
         const { declaration, elements } = this.raw;
         this.declaration = declaration === null || declaration === void 0 ? void 0 : declaration.attributes;
-        if (!((elements === null || elements === void 0 ? void 0 : elements.length) === 2 && elements[0].type === 'doctype' && elements[1].name === 'article')) {
-            throw new Error('article is not the only element of the JATS');
+        if (!((elements === null || elements === void 0 ? void 0 : elements.length) === 2 && elements[0].type === 'doctype' && hasSingleArticle(elements[1]))) {
+            throw new Error('Element <article> is not the only element of the JATS');
         }
         this.doctype = elements[0].doctype;
-        this.tree = (0, utils_1.convertToUnist)(elements[1]);
+        const converted = (0, utils_1.convertToUnist)(elements[1]);
+        this.tree = select('article', converted);
+        (_a = this.log) === null || _a === void 0 ? void 0 : _a.debug(toc('Parsed and converted JATS to unist tree in %s'));
+    }
+    get frontmatter() {
+        var _a, _b, _c, _d;
+        const title = this.articleTitle;
+        const subtitle = this.articleSubtitle;
+        const date = this.publicationDate;
+        const authors = this.articleAuthors;
+        const firstSubject = select(types_1.Tags.subject, (_a = this.articleCategories) !== null && _a !== void 0 ? _a : this.front);
+        const journalTitle = select(types_1.Tags.journalTitle, this.front);
+        return {
+            title: title ? (0, myst_common_1.toText)(title) : undefined,
+            subtitle: subtitle ? (0, myst_common_1.toText)(subtitle) : undefined,
+            doi: (_b = this.doi) !== null && _b !== void 0 ? _b : undefined,
+            date: date ? (_c = (0, utils_1.toDate)(date)) === null || _c === void 0 ? void 0 : _c.toISOString() : undefined,
+            authors: authors === null || authors === void 0 ? void 0 : authors.map((a) => (0, utils_1.authorAndAffiliation)(a, this.tree)),
+            keywords: (_d = this.keywords) === null || _d === void 0 ? void 0 : _d.map((k) => (0, myst_common_1.toText)(k)),
+            venue: journalTitle ? { title: (0, myst_common_1.toText)(journalTitle) } : undefined,
+            subject: firstSubject ? (0, myst_common_1.toText)(firstSubject) : undefined,
+        };
     }
     get front() {
-        return (0, unist_util_select_1.select)(types_1.Tags.front, this.tree);
+        return select(types_1.Tags.front, this.tree);
     }
     get premissions() {
-        return (0, unist_util_select_1.select)(types_1.Tags.permissions, this.front);
+        return select(types_1.Tags.permissions, this.front);
     }
     get doi() {
-        return (0, utils_1.findDoi)(this.front);
+        var _a;
+        return doi_utils_1.default.normalize((_a = (0, utils_1.findArticleId)(this.front, 'doi')) !== null && _a !== void 0 ? _a : '');
+    }
+    get pmc() {
+        var _a;
+        return (_a = (0, utils_1.findArticleId)(this.front, 'pmc')) === null || _a === void 0 ? void 0 : _a.replace(/^PMC:?/, '');
+    }
+    get pmid() {
+        return (0, utils_1.findArticleId)(this.front, 'pmid');
     }
     get publicationDates() {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.pubDate, this.front);
     }
     get publicationDate() {
-        return this.publicationDates.find((d) => !!(0, unist_util_select_1.select)(types_1.Tags.day, d));
+        return this.publicationDates.find((d) => !!select(types_1.Tags.day, d));
     }
     get license() {
-        return (0, unist_util_select_1.select)(types_1.Tags.license, this.premissions);
+        return select(types_1.Tags.license, this.premissions);
     }
     get keywordGroup() {
-        return (0, unist_util_select_1.select)(types_1.Tags.kwdGroup, this.front);
+        return select(types_1.Tags.kwdGroup, this.front);
     }
     /** The first keywords */
     get keywords() {
@@ -44,23 +93,26 @@ class Jats {
     get keywordGroups() {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.kwdGroup, this.front);
     }
+    get articleCategories() {
+        return select(types_1.Tags.articleCategories, this.front);
+    }
     get titleGroup() {
-        return (0, unist_util_select_1.select)(types_1.Tags.titleGroup, this.front);
+        return select(types_1.Tags.titleGroup, this.front);
     }
     get articleTitle() {
-        return (0, unist_util_select_1.select)(types_1.Tags.articleTitle, this.titleGroup);
+        return select(types_1.Tags.articleTitle, this.titleGroup);
     }
     get articleSubtitle() {
-        return (0, unist_util_select_1.select)(types_1.Tags.subtitle, this.titleGroup);
+        return select(types_1.Tags.subtitle, this.titleGroup);
     }
     get abstract() {
-        return (0, unist_util_select_1.select)(types_1.Tags.abstract, this.front);
+        return select(types_1.Tags.abstract, this.front);
     }
     get abstracts() {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.abstract, this.front);
     }
     get contribGroup() {
-        return (0, unist_util_select_1.select)(types_1.Tags.contribGroup, this.front);
+        return select(types_1.Tags.contribGroup, this.front);
     }
     get contribGroups() {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.contribGroup, this.front);
@@ -69,19 +121,31 @@ class Jats {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.contrib, this.contribGroup);
     }
     get body() {
-        return (0, unist_util_select_1.select)(types_1.Tags.body, this.tree);
+        return select(types_1.Tags.body, this.tree);
     }
     get back() {
-        return (0, unist_util_select_1.select)(types_1.Tags.back, this.tree);
+        return select(types_1.Tags.back, this.tree);
     }
     get subArticles() {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.subArticle, this.tree);
     }
     get refList() {
-        return (0, unist_util_select_1.select)(types_1.Tags.refList, this.back);
+        return select(types_1.Tags.refList, this.back);
     }
     get references() {
         return (0, unist_util_select_1.selectAll)(types_1.Tags.ref, this.refList);
     }
 }
 exports.Jats = Jats;
+function hasSingleArticle(element) {
+    var _a;
+    if (element.name === 'article') {
+        return true;
+    }
+    if (element.name === 'pmc-articleset' &&
+        ((_a = element.elements) === null || _a === void 0 ? void 0 : _a.length) === 1 &&
+        element.elements[0].name === 'article') {
+        return true;
+    }
+    return false;
+}

package/dist/cjs/resolvers.js

@@ -12,7 +12,7 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.resolveJatsUrlFromDoi = exports.DEFAULT_RESOLVERS = exports.joss = exports.plos = exports.elife = void 0;
+exports.customResolveJatsUrlFromDoi = exports.DEFAULT_RESOLVERS = exports.joss = exports.plos = exports.elife = void 0;
 const doi_utils_1 = __importDefault(require("doi-utils"));
 const node_fetch_1 = __importDefault(require("node-fetch"));
 exports.elife = {
@@ -33,22 +33,25 @@ exports.plos = {
 };
 exports.joss = {
     test(url) {
-        return new URL(url).hostname === 'joss.theoj.org';
+        return new URL(url).hostname === 'joss.theoj.org' && doi_utils_1.default.validate(url);
     },
     jatsUrl(url) {
         // Probably a better way to do this, the joss papers on on github!
-        const doi = new URL(url).pathname.replace('/papers/', '');
-        const [org, jossId] = doi.split('/');
+        const doiString = doi_utils_1.default.normalize(url);
+        const [org, jossId] = doiString.split('/');
         const id = jossId.split('.')[1];
         return `https://raw.githubusercontent.com/openjournals/joss-papers/master/joss.${id}/${org}.${jossId}.jats`;
     },
 };
 exports.DEFAULT_RESOLVERS = [exports.elife, exports.plos, exports.joss];
-function resolveJatsUrlFromDoi(session, doi, resolvers = exports.DEFAULT_RESOLVERS) {
+/**
+ * Use the known custom resolvers to pick where the JATS should be downloaded from.
+ */
+function customResolveJatsUrlFromDoi(session, doiString, resolvers = exports.DEFAULT_RESOLVERS) {
     return __awaiter(this, void 0, void 0, function* () {
-        if (!doi_utils_1.default.validate(doi))
-            throw new Error(`The doi ${doi} is not valid`);
-        const doiUrl = doi_utils_1.default.buildUrl(doi);
+        if (!doi_utils_1.default.validate(doiString))
+            throw new Error(`The doi ${doiString} is not valid`);
+        const doiUrl = doi_utils_1.default.buildUrl(doiString);
         session.log.debug(`Resolving DOI ${doiUrl}`);
         const resp = yield (0, node_fetch_1.default)(doiUrl);
         const articleUrl = resp.url;
@@ -60,4 +63,4 @@ function resolveJatsUrlFromDoi(session, doi, resolvers = exports.DEFAULT_RESOLVE
         return jatsUrl;
     });
 }
-exports.resolveJatsUrlFromDoi = resolveJatsUrlFromDoi;
+exports.customResolveJatsUrlFromDoi = customResolveJatsUrlFromDoi;

package/dist/cjs/session.js

@@ -5,7 +5,6 @@ const myst_cli_utils_1 = require("myst-cli-utils");
 class Session {
     constructor(opts) {
         var _a;
-        this.API_URL = 'https://api.myst.tools';
         this.log = (_a = opts === null || opts === void 0 ? void 0 : opts.logger) !== null && _a !== void 0 ? _a : (0, myst_cli_utils_1.chalkLogger)(myst_cli_utils_1.LogLevel.debug);
     }
 }

package/dist/cjs/types/elementTags.js

@@ -1282,4 +1282,20 @@ var Tags;
      * See: https://jats.nlm.nih.gov/publishing/tag-library/1.3/element/pub-date.html
      */
     Tags["pubDate"] = "pub-date";
+    /**
+     * Article Grouping Data
+     *
+     * Not available in articleauthoring!
+     *
+     * See: https://jats.nlm.nih.gov/archiving/tag-library/1.3/element/article-categories.html
+     */
+    Tags["articleCategories"] = "article-categories";
+    /**
+     * Journal Title
+     *
+     * Not available in articleauthoring!
+     *
+     * See: https://jats.nlm.nih.gov/archiving/tag-library/1.3/element/journal-title.html
+     */
+    Tags["journalTitle"] = "journal-title";
 })(Tags = exports.Tags || (exports.Tags = {}));