jats-cli 1.0.12 → 1.0.14

package/README.md

@@ -1,8 +1,8 @@
 # jats-cli
 
 [![jats-cli on npm](https://img.shields.io/npm/v/jats-cli.svg)](https://www.npmjs.com/package/jats-cli)
-[![MIT License](https://img.shields.io/badge/license-MIT-blue.svg)](https://github.com/curvenote/jats/blob/main/LICENSE)
-[![CI](https://github.com/curvenote/jats/workflows/CI/badge.svg)](https://github.com/curvenote/jats-cli/actions)
+[![MIT License](https://img.shields.io/badge/license-MIT-blue.svg)](https://github.com/continuous-foundation/jats/blob/main/LICENSE)
+[![CI](https://github.com/continuous-foundation/jats/workflows/CI/badge.svg)](https://github.com/continuous-foundation/jats-cli/actions)
 
 Node CLI for working with JATS XML documents.
 
@@ -93,7 +93,7 @@ As of v1.0.0 this package is [ESM only](https://gist.github.com/sindresorhus/a39
 
 <p style="text-align: center; color: #aaa; padding-top: 50px">
   Made with love by
-  <a href="https://curvenote.com" target="_blank" style="color: #aaa">
-    <img src="https://cdn.curvenote.com/brand/logo-blue-icon.png" style="height: 1em" /> Curvenote
+  <a href="https://continuous.foundation" target="_blank" style="color: #aaa">
+    Continuous Science Foundation <img src="https://continuous.foundation/images/logo-small.svg" style="height: 1em" />
   </a>
 </p>

package/dist/convert.js

@@ -4,7 +4,7 @@ import { clirun, getSession } from 'myst-cli-utils';
 function makeConvertCLI(program) {
     const command = new Command('convert')
         .description('Convert JATS file to MyST mdast json')
-        .argument('<input>', 'The JATS file')
+        .argument('[input]', 'The JATS file')
         .addOption(new Option('--frontmatter <frontmatter>', 'Treat JATS frontmatter fields as page or project, or ignore if not specified').choices(['page', 'project']))
         .addOption(new Option('--no-doi, --no-dois', 'By default, DOIs are used for references when available, to be later resolved against doi.org. This option disables that behavior and creates bibtex entries for citations with DOIs.'))
         .addOption(new Option('--no-bib, --no-bibtex', 'By default, a bibtex file will be written with referenced citations. This option prevents writing that file'))

package/dist/jats.cjs

@@ -6561,7 +6561,7 @@ var require_xml2js = __commonJS({
 var require_xml2json = __commonJS({
   "../../node_modules/xml-js/lib/xml2json.js"(exports, module2) {
     var helper = require_options_helper();
-    var xml2js2 = require_xml2js();
+    var xml2js3 = require_xml2js();
     function validateOptions2(userOptions) {
       var options = helper.copyOptions(userOptions);
       helper.ensureSpacesExists(options);
@@ -6570,7 +6570,7 @@ var require_xml2json = __commonJS({
     module2.exports = function(xml, userOptions) {
       var options, js, json2, parentKey;
       options = validateOptions2(userOptions);
-      js = xml2js2(xml, options);
+      js = xml2js3(xml, options);
       parentKey = "compact" in options && options.compact ? "_parent" : "parent";
       if ("addParent" in options && options.addParent) {
         json2 = JSON.stringify(js, function(k, v) {
@@ -6942,12 +6942,12 @@ var require_json2xml = __commonJS({
 // ../../node_modules/xml-js/lib/index.js
 var require_lib2 = __commonJS({
   "../../node_modules/xml-js/lib/index.js"(exports, module2) {
-    var xml2js2 = require_xml2js();
+    var xml2js3 = require_xml2js();
     var xml2json = require_xml2json();
     var js2xml3 = require_js2xml();
     var json2xml = require_json2xml();
     module2.exports = {
-      xml2js: xml2js2,
+      xml2js: xml2js3,
       xml2json,
       js2xml: js2xml3,
       json2xml
@@ -48999,7 +48999,7 @@ var {
 } = import_index.default;
 
 // src/version.ts
-var version = "1.0.12";
+var version = "1.0.14";
 var version_default = version;
 
 // src/parse.ts
@@ -56425,6 +56425,7 @@ var RefType;
   RefType2["table"] = "table";
   RefType2["tableFn"] = "table-fn";
   RefType2["ref"] = "ref";
+  RefType2["media"] = "media";
 })(RefType || (RefType = {}));
 
 // ../jats-fetch/dist/download.js
@@ -56436,6 +56437,7 @@ var import_node_fs4 = __toESM(require("node:fs"), 1);
 var import_node_path3 = __toESM(require("node:path"), 1);
 var import_node_readline = __toESM(require("node:readline"), 1);
 var import_client_s32 = __toESM(require_dist_cjs71(), 1);
+var import_xml_js2 = __toESM(require_lib2(), 1);
 
 // ../jats-fetch/dist/utils.js
 var import_node_fs3 = __toESM(require("node:fs"), 1);
@@ -57789,6 +57791,7 @@ async function streamToFile(url, dest, fetcher) {
 // ../jats-fetch/dist/pubmed.js
 var EFETCH_URL = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi";
 var ESUMMARY_URL = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/esummary.fcgi";
+var OA_URL = "https://www.ncbi.nlm.nih.gov/pmc/utils/oa/oa.fcgi";
 var IDCONV_URL = "https://www.ncbi.nlm.nih.gov/pmc/utils/idconv/v1.0/";
 var LISTING_BASE_URL = "https://ftp.ncbi.nlm.nih.gov/pub/pmc/";
 var LISTING_URL = `${LISTING_BASE_URL}oa_file_list.csv`;
@@ -57811,42 +57814,52 @@ function normalizePMID(session, pmid) {
   }
   return pmid;
 }
-async function convertPMID2PMCID(session, pmid, opts) {
+async function convertId(session, id, from, to, opts) {
+  var _a, _b, _c;
+  const toc = tic();
+  const resp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : defaultFetcher)(`${IDCONV_URL}?tool=jats-xml&format=json&ids=${id}`, "json");
+  if (!resp.ok) {
+    session.log.debug(`Failed to convert ${from} ID: ${id}`);
+    return;
+  }
+  const data = await resp.json();
+  const newId = (_c = (_b = data === null || data === void 0 ? void 0 : data.records) === null || _b === void 0 ? void 0 : _b[0]) === null || _c === void 0 ? void 0 : _c[to];
+  if (newId) {
+    session.log.debug(toc(`Used nih.gov to transform ${id} to ${newId} in %s.`));
+  }
+  return newId;
+}
+async function convertIds(session, ids, from, to, opts) {
   var _a, _b, _c;
-  pmid = normalizePMID(session, pmid);
   const toc = tic();
-  const resp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : defaultFetcher)(`${IDCONV_URL}?tool=jats-xml&format=json&ids=${pmid}`, "json");
+  const resp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : defaultFetcher)(`${IDCONV_URL}?tool=jats-xml&format=json&ids=${ids.join(",")}`, "json");
   if (!resp.ok) {
-    session.log.debug(`Failed to convert PubMedID: ${pmid}`);
+    session.log.debug(`Failed to convert ${from} ${ids.length} IDs`);
     return;
   }
   const data = await resp.json();
-  const pmcid = (_c = (_b = data === null || data === void 0 ? void 0 : data.records) === null || _b === void 0 ? void 0 : _b[0]) === null || _c === void 0 ? void 0 : _c.pmcid;
-  session.log.debug(toc(`Used nih.gov to transform ${pmid} to ${pmcid} in %s.`));
+  const entries = (_b = data === null || data === void 0 ? void 0 : data.records) === null || _b === void 0 ? void 0 : _b.filter((record) => !!record[from] && !!record[to]).map((record) => [record[from], record[to]]);
+  const newIds = entries ? Object.fromEntries(entries) : {};
+  session.log.debug(toc(`Used nih.gov to transform ${(_c = entries === null || entries === void 0 ? void 0 : entries.length) !== null && _c !== void 0 ? _c : 0}/${ids.length} ${from} to ${to} in %s.`));
+  return newIds;
+}
+async function convertPMID2PMCID(session, pmid, opts) {
+  pmid = normalizePMID(session, pmid);
+  const pmcid = await convertId(session, pmid, "pmid", "pmcid", opts);
   return pmcid;
 }
 async function convertPMIDs2DOIs(session, pmids, opts) {
-  var _a, _b, _c, _d;
+  var _a, _b, _c;
   pmids = [...new Set(pmids.map((pmid) => normalizePMID(session, pmid)))];
-  const pmDois = {};
   const toc = tic();
-  const idconvResp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : defaultFetcher)(`${IDCONV_URL}?tool=jats-xml&format=json&ids=${pmids.join(",")}`, "json");
-  if (idconvResp.ok) {
-    const data = await idconvResp.json();
-    (_b = data === null || data === void 0 ? void 0 : data.records) === null || _b === void 0 ? void 0 : _b.forEach((record) => {
-      if (record.doi)
-        pmDois[record.pmid] = record.doi;
-    });
-    const pmDoiCount = Object.keys(pmDois).length;
-    if (pmDoiCount === pmids.length) {
-      session.log.debug(toc(`Used nih.gov to convert ${pmDoiCount} PMIDs to DOIs in %s.`));
-      return pmDois;
-    }
+  const pmDois = (_a = await convertIds(session, pmids, "pmid", "doi", opts)) !== null && _a !== void 0 ? _a : {};
+  if (Object.keys(pmDois).length === pmids.length) {
+    return pmDois;
   }
-  const esummaryResp = await ((_c = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _c !== void 0 ? _c : defaultFetcher)(`${ESUMMARY_URL}?db=pubmed&format=json&id=${pmids.filter((pmid) => !pmDois[pmid]).join(",")}`, "json");
+  const esummaryResp = await ((_b = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _b !== void 0 ? _b : defaultFetcher)(`${ESUMMARY_URL}?db=pubmed&format=json&id=${pmids.filter((pmid) => !pmDois[pmid]).join(",")}`, "json");
   if (esummaryResp.ok) {
     const data = await esummaryResp.json();
-    Object.entries((_d = data === null || data === void 0 ? void 0 : data.result) !== null && _d !== void 0 ? _d : {}).filter(([pmid]) => pmid !== "uids").forEach(([pmid, record]) => {
+    Object.entries((_c = data === null || data === void 0 ? void 0 : data.result) !== null && _c !== void 0 ? _c : {}).filter(([pmid]) => pmid !== "uids").forEach(([pmid, record]) => {
       var _a2, _b2;
       const pmDoi = (_b2 = (_a2 = record.articleids) === null || _a2 === void 0 ? void 0 : _a2.find((articleid) => {
         return articleid.idtype === "doi";
@@ -57917,8 +57930,6 @@ async function getPubMedJatsFromS3(session, pmcid) {
   }
 }
 async function getListingsFile(session, dest, fetcher) {
-  if (!dest)
-    dest = __dirname;
   if (!import_node_path3.default.extname(dest))
     dest = import_node_path3.default.join(dest, LISTING_FILENAME);
   if (import_node_path3.default.extname(dest) !== ".csv") {
@@ -57963,19 +57974,37 @@ async function searchListingForPMC(listingFile, pmcid) {
   }
   throw new Error(`Article ${pmcid} not found in ${listingFile}`);
 }
-async function downloadAndUnzipPMC(session, entry, outputDir, fetcher) {
-  const urlParts = entry.url.split("/");
+async function getDownloadMetadata(pmcid, fetcher) {
+  var _a, _b, _c, _d, _e, _f, _g, _h, _j, _k, _l, _m;
+  const resp = await (fetcher !== null && fetcher !== void 0 ? fetcher : defaultFetcher)(`${OA_URL}?format=tgz&id=${pmcid}`, "xml");
+  if (!resp.ok) {
+    throw new Error(`Bad response from ${OA_URL}`);
+  }
+  const oaMeta = (0, import_xml_js2.xml2js)(await resp.text(), { compact: true });
+  if (((_c = (_b = (_a = oaMeta.OA) === null || _a === void 0 ? void 0 : _a.records) === null || _b === void 0 ? void 0 : _b._attributes) === null || _c === void 0 ? void 0 : _c["returned-count"]) !== "1") {
+    throw new Error(`Bad response from ${OA_URL} - returned count is not 1`);
+  }
+  const url = (_h = (_g = (_f = (_e = (_d = oaMeta === null || oaMeta === void 0 ? void 0 : oaMeta.OA) === null || _d === void 0 ? void 0 : _d.records) === null || _e === void 0 ? void 0 : _e.record) === null || _f === void 0 ? void 0 : _f.link) === null || _g === void 0 ? void 0 : _g._attributes) === null || _h === void 0 ? void 0 : _h.href;
+  if (!url) {
+    throw new Error(`Bad response from ${OA_URL} - href is not available`);
+  }
+  const { citation, license } = (_m = (_l = (_k = (_j = oaMeta.OA) === null || _j === void 0 ? void 0 : _j.records) === null || _k === void 0 ? void 0 : _k.record) === null || _l === void 0 ? void 0 : _l._attributes) !== null && _m !== void 0 ? _m : {};
+  return { url, citation, license };
+}
+async function downloadAndUnzipPMC(session, url, outputDir, fetcher) {
+  url = url.replace(/^ftp:/, "https:");
+  const urlParts = url.split("/");
   const filename = urlParts[urlParts.length - 1];
   const dest = import_node_path3.default.join(outputDir, filename);
   if (!import_node_fs4.default.existsSync(outputDir)) {
     import_node_fs4.default.mkdirSync(outputDir, { recursive: true });
   }
   if (!import_node_fs4.default.existsSync(dest)) {
-    session.log.info(`Downloading PMC data from ${entry.url}`);
-    await streamToFile(`${LISTING_BASE_URL}${entry.url}`, dest, fetcher);
+    session.log.info(`Downloading PMC data from ${url}`);
+    await streamToFile(url, dest, fetcher);
   }
   if (!import_node_fs4.default.existsSync(dest)) {
-    throw new Error(`Unable to download ${entry.url}`);
+    throw new Error(`Unable to download ${url}`);
   }
   session.log.info(`Extracting PMC data from ${dest} to ${outputDir}`);
   const unzip = makeExecutable(`tar -xf ${dest} -C ${outputDir}`, session.log);
@@ -57995,9 +58024,28 @@ async function downloadAndUnzipPMC(session, entry, outputDir, fetcher) {
   import_node_fs4.default.rmdirSync(zipDir);
 }
 async function getDataFromPMC(session, pmcid, outputDir, listing, fetcher) {
-  const listingFile = await getListingsFile(session, listing, fetcher);
-  const entry = await searchListingForPMC(listingFile, pmcid);
-  await downloadAndUnzipPMC(session, entry, outputDir, fetcher);
+  let url;
+  if (pmcid.endsWith(".tar.gz")) {
+    url = pmcid;
+  } else {
+    if (!pmcid.startsWith("PMC")) {
+      throw new Error("Data may only be downloaded for PMC articles");
+    }
+    try {
+      const metadata = await getDownloadMetadata(pmcid);
+      url = metadata.url;
+    } catch {
+      if (listing) {
+        const listingFile = await getListingsFile(session, listing, fetcher);
+        const entry = await searchListingForPMC(listingFile, pmcid);
+        url = `${LISTING_BASE_URL}${entry.url}`;
+      }
+    }
+  }
+  if (!url) {
+    throw new Error(`Unable to find PMC data download url for: ${pmcid}`);
+  }
+  await downloadAndUnzipPMC(session, url, outputDir, fetcher);
 }
 async function getPubMedJatsFromData(session, pmcid, outputDir, listing, fetcher) {
   await getDataFromPMC(session, pmcid, outputDir, listing, fetcher);
@@ -58149,22 +58197,40 @@ async function downloadJatsFromUrl(session, urlOrDoi, opts = {}) {
   return { success: false, source: urlOrDoi };
 }
 async function jatsFetch(session, input, opts) {
-  var _a, _b;
+  var _a;
   if (input === "listing" && !opts.data && !(opts.output && opts.listing)) {
-    const dest = await getListingsFile(session, (_a = opts.output) !== null && _a !== void 0 ? _a : opts.listing);
+    const inputDest = (_a = opts.output) !== null && _a !== void 0 ? _a : opts.listing;
+    if (!inputDest) {
+      throw new Error("Destination for listing file must be specified");
+    }
+    const dest = await getListingsFile(session, inputDest);
     session.log.info(`PMC Open Access listing saved to ${dest}`);
     return;
   }
-  let output = (_b = opts.output) !== null && _b !== void 0 ? _b : opts.data ? `${input}` : ".";
+  let output = opts.output;
+  let filename;
+  if (input.endsWith(".tar.gz")) {
+    opts.data = true;
+    const foldername = input.split("/").slice(-1)[0].slice(0, -".tar.gz".length);
+    filename = `${foldername}.xml`;
+    if (!output) {
+      output = foldername;
+    }
+  }
+  if (!output)
+    output = opts.data ? `${input}` : ".";
   if (!import_node_path4.default.extname(output)) {
-    const filename = input.startsWith("PMC") ? `${input}.xml` : "jats.xml";
+    filename = filename !== null && filename !== void 0 ? filename : input.startsWith("PMC") ? `${input}.xml` : "jats.xml";
     output = import_node_path4.default.join(output, filename);
   }
   if (import_node_path4.default.extname(output) && ![".xml", ".jats"].includes(import_node_path4.default.extname(output).toLowerCase())) {
     throw new Error(`Output must be an XML file or a directory`);
   }
   let result;
-  if (input.startsWith("PMC")) {
+  if (opts.data) {
+    result = await getPubMedJatsFromData(session, input, import_node_path4.default.dirname(output), opts.listing);
+  }
+  if (!(result === null || result === void 0 ? void 0 : result.data) && input.startsWith("PMC")) {
     result = await getPubMedJatsFromS3(session, input);
   }
   if (!(result === null || result === void 0 ? void 0 : result.data)) {
@@ -58183,17 +58249,10 @@ async function jatsFetch(session, input, opts) {
   }
   import_node_fs5.default.writeFileSync(output, result.data);
   session.log.info(`JATS written to ${output}`);
-  if (!opts.data)
-    return;
-  if (input.startsWith("PMC")) {
-    await getDataFromPMC(session, input, import_node_path4.default.dirname(output), opts.listing);
-  } else {
-    session.log.error("Data may only be downloaded for PMC articles");
-  }
 }
 
 // ../jats-xml/dist/jats.js
-var import_xml_js2 = __toESM(require_lib2(), 1);
+var import_xml_js3 = __toESM(require_lib2(), 1);
 
 // ../jats-xml/node_modules/unist-util-is/lib/index.js
 var convert2 = (
@@ -58525,7 +58584,7 @@ var Jats = class {
     if (opts === null || opts === void 0 ? void 0 : opts.source)
       this.source = opts.source;
     try {
-      this.raw = (0, import_xml_js2.xml2js)(data, { compact: false });
+      this.raw = (0, import_xml_js3.xml2js)(data, { compact: false });
     } catch (error) {
       throw new Error("Problem parsing the JATS document, please ensure it is XML");
     }
@@ -64768,7 +64827,7 @@ var u = (
 
 // ../jats-convert/dist/index.js
 var import_mathml_to_latex = __toESM(require_bundle_min(), 1);
-var import_xml_js3 = __toESM(require_lib2(), 1);
+var import_xml_js4 = __toESM(require_lib2(), 1);
 
 // ../jats-convert/dist/transforms/admonitions.js
 function admonitionTransform(tree, file) {
@@ -64970,8 +65029,15 @@ function blockNestingTransform(tree) {
 }
 
 // ../jats-convert/dist/transforms/sections.js
+var SECTION_TYPES = ["sec", "ack", "app"];
 function isSection(node) {
-  return node.type === "sec" || node.type === "ack";
+  return SECTION_TYPES.includes(node.type);
+}
+function liftApps(tree) {
+  selectAll2("app-group", tree).forEach((group) => {
+    group.type = "__lift__";
+  });
+  liftChildren(tree, "__lift__");
 }
 function recurseSections(tree, depth = 1, titleType) {
   var _a;
@@ -64999,11 +65065,15 @@ function recurseSections(tree, depth = 1, titleType) {
     }
     if (sec.type === "ack")
       sec.part = "acknowledgments";
+    if (sec.type === "app")
+      sec.part = "appendix";
+    sec.type = "sec";
     recurseSections(sec, depth + 1, titleType);
   });
 }
 function sectionTransform(tree, titleType) {
   var _a;
+  liftApps(tree);
   recurseSections(tree, 1, titleType);
   remove(tree, "__delete__");
   const topSections = (_a = tree.children) === null || _a === void 0 ? void 0 : _a.filter((n) => isSection(n));
@@ -65051,6 +65121,38 @@ function abbreviationSectionTransform(tree, frontmatter) {
   });
   remove(tree, "__delete__");
 }
+function abbreviationFootnoteTransform(tree, frontmatter) {
+  const fnDefs = selectAll2("footnoteDefinition", tree);
+  const fnRefs = selectAll2("footnoteReference", tree).map(({ identifier }) => identifier);
+  fnDefs.forEach((fnDef) => {
+    var _a;
+    if (fnDef.identifier && fnRefs.includes(fnDef.identifier))
+      return;
+    if (((_a = fnDef.children) === null || _a === void 0 ? void 0 : _a.length) !== 1)
+      return;
+    if (fnDef.children[0].type !== "paragraph")
+      return;
+    const fnText = toText2(fnDef.children[0]);
+    const abbrPrefix = "abbreviations: ";
+    if (!fnText.toLowerCase().startsWith(abbrPrefix))
+      return;
+    const abbreviations = fnText.slice(abbrPrefix.length).replace(/\.$/, "").split(/;\s*/g);
+    const entries = abbreviations.map((abbr) => {
+      const parts = abbr.split(/[,:]\s*/g);
+      if (parts.length !== 2)
+        return void 0;
+      if (parts[0].match(/\s/))
+        return void 0;
+      return [parts[0], parts[1]];
+    });
+    if (entries.findIndex((entry) => !entry) !== -1)
+      return;
+    const newAbbreviations = Object.fromEntries(entries);
+    frontmatter.abbreviations = { ...frontmatter.abbreviations, ...newAbbreviations };
+    fnDef.type = "__delete__";
+  });
+  remove(tree, "__delete__");
+}
 
 // ../jats-convert/dist/transforms/abstract.js
 function abstractTransform(tree) {
@@ -65090,28 +65192,26 @@ function descriptionFromAbstract(abstract) {
   return `${firstSentence} ${secondSentence}`;
 }
 
-// ../jats-convert/dist/transforms/citations.js
-function citationToMixedCitation(tree) {
-  selectAll2("citation", tree).forEach((node) => {
-    node.type = "mixed-citation";
-  });
-}
-
 // ../jats-convert/dist/transforms/footnotes.js
-function backToBodyTransform(jats) {
-  var _a, _b;
-  if (!jats.back)
+function backToBodyTransform(body, back) {
+  var _a;
+  if (!back)
     return;
-  const backNodes = (_a = jats.back.children) === null || _a === void 0 ? void 0 : _a.filter((node) => {
-    return ["fn-group", "sec", "ack"].includes(node.type);
+  const backNodes = (_a = back.children) === null || _a === void 0 ? void 0 : _a.filter((node) => {
+    return ["fn-group", "sec", "ack", "app-group"].includes(node.type);
   });
-  if (!((_b = jats.body) === null || _b === void 0 ? void 0 : _b.children) || backNodes.length === 0)
+  if (!(body === null || body === void 0 ? void 0 : body.children) || backNodes.length === 0)
     return;
-  jats.body.children.push({ type: "hr" }, ...copyNode(backNodes));
-  backNodes.forEach((node) => {
-    node.type = "__delete__";
+  body.children.push({ type: "hr" }, ...copyNode(backNodes));
+}
+function tableFootnotesToLegend(tree) {
+  const tableFns = selectAll2("legend > footnoteDefinition", tree);
+  const fnRefs = selectAll2("footnoteReference", tree).map(({ identifier }) => identifier);
+  tableFns.forEach((tableFn) => {
+    if (tableFn.identifier && fnRefs.includes(tableFn.identifier))
+      return;
+    tableFn.type = "paragraph";
   });
-  remove(jats.back, "__delete__");
 }
 
 // ../../node_modules/dotenv/config.js
@@ -65270,6 +65370,7 @@ function bibtexFromCite(key, cite, counts, doi2) {
   }
   if (bibtexLines.length === 1) {
     counts.unprocessed += 1;
+    bibtexLines.push(`  note = {${toText2(cite)}}`);
   } else {
     counts.bibtex += 1;
     counts.lostRefItems.push(...skipped);
@@ -65322,7 +65423,7 @@ function processRef(ref, pmidCache, fnCount, counts, dois) {
   const footnotes = [];
   const bibtexEntries = [];
   (_b = ref.children) === null || _b === void 0 ? void 0 : _b.forEach((child) => {
-    if (["element-citation", "mixed-citation"].includes(child.type)) {
+    if (["element-citation", "mixed-citation", "citation"].includes(child.type)) {
       if (!toText2(child))
         return;
       const cite = processRefCite(child, identifier, pmidCache, counts, dois);
@@ -65343,15 +65444,14 @@ function processRef(ref, pmidCache, fnCount, counts, dois) {
   });
   return { refLookup, footnotes, bibtexEntries };
 }
-async function processJatsReferences(jats, opts) {
+function processJatsReferences(body, references, opts) {
   var _a, _b;
   const dir = (_a = opts === null || opts === void 0 ? void 0 : opts.dir) !== null && _a !== void 0 ? _a : ".";
   const bibfile = import_node_path5.default.join(dir, "main.bib");
-  const refs = jats.references;
+  const pmidCache = (_b = opts === null || opts === void 0 ? void 0 : opts.pmidCache) !== null && _b !== void 0 ? _b : {};
   let refLookup = {};
   const footnotes = [];
   const bibtexEntries = [];
-  const pmidCache = await getPMIDLookup(refs, dir);
   const counts = {
     dois: 0,
     bibtex: 0,
@@ -65359,7 +65459,7 @@ async function processJatsReferences(jats, opts) {
     lostRefs: [],
     lostRefItems: []
   };
-  refs.forEach((ref) => {
+  references.forEach((ref) => {
     const { refLookup: newRefLookup, footnotes: newFootnotes, bibtexEntries: newBibtexEntries } = processRef(ref, pmidCache, footnotes.length + 1, counts, opts === null || opts === void 0 ? void 0 : opts.dois);
     refLookup = { ...refLookup, ...newRefLookup };
     bibtexEntries.push(...newBibtexEntries);
@@ -65367,7 +65467,7 @@ async function processJatsReferences(jats, opts) {
   });
   if (opts === null || opts === void 0 ? void 0 : opts.logInfo) {
     opts.logInfo.references = {
-      total: refs.length,
+      total: references.length,
       dois: counts.dois,
       bibtex: counts.bibtex,
       footnotes: footnotes.length,
@@ -65396,7 +65496,7 @@ async function processJatsReferences(jats, opts) {
     import_node_fs6.default.writeFileSync(bibfile, bibtexEntries.join("\n\n"));
   }
   if (footnotes.length) {
-    (_b = jats.body) === null || _b === void 0 ? void 0 : _b.children.push({ type: "fn-group", children: footnotes });
+    body.children.push({ type: "fn-group", children: footnotes });
   }
   return refLookup;
 }
@@ -65462,18 +65562,18 @@ async function resolveJatsCitations(tree, refLookup) {
 }
 
 // ../jats-convert/dist/transforms/index.js
-function basicTransformations(tree, file) {
-  sectionTransform(tree);
-  typographyTransform(tree);
-  admonitionTransform(tree, file);
-  figCaptionTitleTransform(tree);
+function basicTransformations(body, file) {
+  sectionTransform(body);
+  typographyTransform(body);
+  admonitionTransform(body, file);
+  figCaptionTitleTransform(body);
 }
-function journalTransforms(tree) {
-  graphicToBioRxivUrl(tree);
+function journalTransforms(body) {
+  graphicToBioRxivUrl(body);
 }
 
 // ../jats-convert/dist/version.js
-var version3 = "1.0.12";
+var version3 = "1.0.14";
 var version_default3 = version3;
 
 // ../jats-convert/dist/myst/inlineCitations.js
@@ -65613,6 +65713,35 @@ function expandHyphenatedCites(tree, referenceList) {
   });
   remove(tree, "__delete__");
 }
+function removeCiteSuperscript(tree) {
+  const citeGroupParents = selectAll2(":has(> citeGroup)", tree);
+  citeGroupParents.forEach((parent3) => {
+    if (parent3.type !== "superscript")
+      return;
+    if (parent3.children.length !== 1)
+      return;
+    parent3.type = "__lift__";
+  });
+  liftChildren(tree, "__lift__");
+}
+function ensureSpaceBeforeCite(tree) {
+  const citeGroupParents = selectAll2(":has(> text + citeGroup)", tree);
+  citeGroupParents.forEach((parent3) => {
+    parent3.children.forEach((child, index2) => {
+      var _a, _b;
+      if (child.type !== "text")
+        return;
+      const citeChild = parent3.children[index2 + 1];
+      if ((citeChild === null || citeChild === void 0 ? void 0 : citeChild.type) !== "citeGroup")
+        return;
+      if ((_a = child.value) === null || _a === void 0 ? void 0 : _a.match(/\s+$/))
+        return;
+      if ((_b = child.value) === null || _b === void 0 ? void 0 : _b.match(/[[(]$/))
+        return;
+      child.value = `${child.value} `;
+    });
+  });
+}
 function inlineCitationsTransform(tree, referenceIds) {
   let before = "";
   let current = JSON.stringify(tree);
@@ -65625,27 +65754,112 @@ function inlineCitationsTransform(tree, referenceIds) {
     removeCiteSeparators(tree);
     expandHyphenatedCites(tree, referenceIds);
     removeCiteParentheses(tree);
+    removeCiteSuperscript(tree);
+    ensureSpaceBeforeCite(tree);
     current = JSON.stringify(tree);
   }
 }
 
 // ../jats-convert/dist/transforms/supplementary.js
-function floatToEndTransform(jats) {
+function floatToEndTransform(body) {
   var _a;
-  if (!jats.body)
-    return;
-  const floatSupplementaryMaterial = selectAll2("supplementary-material[position=float]", jats.body);
+  const floatSupplementaryMaterial = selectAll2("supplementary-material[position=float]", body);
   if (floatSupplementaryMaterial.length === 0)
     return;
-  (_a = jats.body.children) === null || _a === void 0 ? void 0 : _a.push({ type: "hr" });
+  (_a = body.children) === null || _a === void 0 ? void 0 : _a.push({ type: "hr" });
   floatSupplementaryMaterial.forEach((node) => {
-    var _a2, _b;
+    var _a2;
     const copy = copyNode2(node);
     delete copy.position;
-    (_b = (_a2 = jats.body) === null || _a2 === void 0 ? void 0 : _a2.children) === null || _b === void 0 ? void 0 : _b.push(copy);
+    (_a2 = body === null || body === void 0 ? void 0 : body.children) === null || _a2 === void 0 ? void 0 : _a2.push(copy);
     node.type = "__delete__";
   });
-  remove(jats.body, "__delete__");
+  remove(body, "__delete__");
+}
+
+// ../jats-convert/dist/transforms/parts.js
+function dataAvailabilityTransform(tree) {
+  const dataAvailability = selectAll2("sec[sec-type=data-availability]", tree);
+  dataAvailability.forEach((node) => {
+    const titles = selectAll2("title", node);
+    titles.forEach((title) => {
+      if (toText(title).toLowerCase() === "data availability") {
+        title.type = "__delete__";
+      }
+    });
+  });
+  remove(tree, "__delete__");
+}
+
+// ../jats-convert/dist/myst/abbreviations.js
+function abbreviationsFromTree(tree, frontmatter) {
+  let abbreviations = {};
+  const paragraphs = selectAll2("paragraph", tree);
+  paragraphs.forEach((paragraph) => {
+    const text = toText2(paragraph);
+    abbreviations = {
+      ...abbreviations,
+      ...abbreviationsFromText(text)
+    };
+  });
+  frontmatter.abbreviations = { ...frontmatter.abbreviations, ...abbreviations };
+}
+function maybeStopWord(word) {
+  return word.length < 5;
+}
+function exploreAbbrPossibilities(letter, possibilities) {
+  const newPossibilities = [];
+  possibilities.forEach(({ prev, next }) => {
+    if (prev === null || prev === void 0 ? void 0 : prev.includes(letter)) {
+      newPossibilities.push({
+        prev: prev.slice(prev.indexOf(letter) + 1),
+        next
+      });
+    }
+    for (const [i2, n] of next.entries()) {
+      if (n.startsWith(letter)) {
+        newPossibilities.push({
+          prev: next[i2].slice(1),
+          next: next.slice(i2 + 1)
+        });
+      }
+      if (!prev || !maybeStopWord(n)) {
+        break;
+      }
+    }
+  });
+  return newPossibilities;
+}
+function abbreviationsFromText(text) {
+  const abbreviations = {};
+  const textList = text.split(" ");
+  textList.forEach((word, index2) => {
+    var _a;
+    const abbr = (_a = word.match(/^\(([^\s]{2,})\).{0,1}/)) === null || _a === void 0 ? void 0 : _a[1];
+    if (!abbr)
+      return;
+    const possibleWords = [];
+    let wordIndex = index2 - 1;
+    while (textList[wordIndex] && possibleWords.filter((w) => w.length > 4).length < abbr.length) {
+      possibleWords.unshift(textList[wordIndex]);
+      wordIndex--;
+    }
+    for (const i2 of Array(possibleWords.length).keys()) {
+      let possibilities = [
+        {
+          next: possibleWords.slice(i2).map((w) => w.toLowerCase())
+        }
+      ];
+      abbr.split("").filter((letter) => letter.match(/^[a-zA-Z]$/)).forEach((letter) => {
+        possibilities = exploreAbbrPossibilities(letter.toLowerCase(), possibilities);
+      });
+      if (possibilities.filter(({ next }) => next.length === 0).length) {
+        abbreviations[abbr] = possibleWords.slice(i2).join(" ");
+        break;
+      }
+    }
+  });
+  return abbreviations;
 }
 
 // ../jats-convert/dist/index.js
@@ -65694,7 +65908,7 @@ function texMathFromNode(node) {
       });
     }
   });
-  return import_mathml_to_latex.MathMLToLaTeX.convert((0, import_xml_js3.js2xml)({ type: "element", name: "root", elements: [math] }));
+  return import_mathml_to_latex.MathMLToLaTeX.convert((0, import_xml_js4.js2xml)({ type: "element", name: "root", elements: [math] }));
 }
 var handlers = {
   body(node, state2) {
@@ -65707,10 +65921,12 @@ var handlers = {
     state2.renderInline(node, "paragraph");
   },
   heading(node, state2) {
+    var _a;
+    const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
     state2.renderInline(node, "heading", {
       enumerated: true,
-      label: node.id,
-      identifier: node.id,
+      label,
+      identifier,
       depth: node.depth
     });
   },
@@ -65746,27 +65962,39 @@ var handlers = {
     state2.renderInline(node, "listItem");
   },
   ["inline-formula"](node, state2) {
+    var _a;
     const texMath = texMathFromNode(node);
     if (texMath) {
+      const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
       state2.addLeaf("inlineMath", {
         value: texMath,
-        label: node.id,
-        identifier: node.id
+        label,
+        identifier
       });
     } else {
       state2.renderChildren(node);
     }
   },
   ["disp-formula"](node, state2) {
+    var _a;
     const texMath = texMathFromNode(node);
+    const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
     if (texMath) {
       state2.addLeaf("math", {
         value: texMath,
-        label: node.id,
-        identifier: node.id
+        label,
+        identifier
       });
     } else {
+      if (node.id) {
+        state2.openNode("div", { label, identifier });
+      }
+      state2.openNode("paragraph");
       state2.renderChildren(node);
+      state2.closeNode();
+      if (node.id) {
+        state2.closeNode();
+      }
     }
   },
   bold(node, state2) {
@@ -65809,7 +66037,15 @@ var handlers = {
     state2.renderInline(node, "link", { url: node["xlink:href"] });
   },
   ["boxed-text"](node, state2) {
+    var _a;
+    if (node.id) {
+      const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
+      state2.openNode("div", { label, identifier });
+    }
     state2.renderInline(node, "admonition", { kind: "info" });
+    if (node.id) {
+      state2.closeNode();
+    }
   },
   admonitionTitle(node, state2) {
     state2.renderInline(node, "admonitionTitle");
@@ -65980,6 +66216,9 @@ var handlers = {
           kind: "narrative"
         });
         return;
+      case RefType.app:
+      case RefType.boxedText:
+      case RefType.media:
       case RefType.supplementaryMaterial:
       case RefType.sec:
       case RefType.fig:
@@ -66012,6 +66251,20 @@ var handlers = {
       state2.closeNode();
     }
   },
+  ["app-group"](node, state2) {
+    state2.renderChildren(node);
+  },
+  app(node, state2) {
+    var _a;
+    if (node.id) {
+      const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
+      state2.openNode("div", { label, identifier });
+    }
+    state2.renderChildren(node);
+    if (node.id) {
+      state2.closeNode();
+    }
+  },
   media(node, state2) {
     state2.renderInline(node, "link", { url: node["xlink:href"] });
   },
@@ -66021,6 +66274,9 @@ var handlers = {
   caption(node, state2) {
     state2.renderChildren(node);
   },
+  email(node, state2) {
+    state2.renderInline(node, "link", { url: `mailto:${toText2(node)}` });
+  },
   // These nodes can be safely ignored
   label() {
   },
@@ -66123,63 +66379,54 @@ var JatsParser = class {
   }
 };
 var jatsConvertPlugin = function(jats, opts) {
-  this.Compiler = (node, file) => {
-    if (jats.abstract)
+  this.Compiler = (body, file) => {
+    if (jats.abstract) {
       abstractTransform(jats.abstract);
-    const tree = jats.abstract ? {
-      type: "root",
-      children: [
+      body.children = [
         u("block", { part: "abstract" }, copyNode(jats.abstract).children),
-        ...copyNode(node).children
-      ]
-    } : copyNode(node);
+        ...body.children
+      ];
+    }
     const floatsGroup = selectAll2("floats-group", jats.tree);
     if (floatsGroup.length > 0) {
       floatsGroup.forEach((g) => {
-        tree.children.push(...g.children);
+        body.children.push(...g.children);
       });
     }
-    basicTransformations(tree, file);
-    citationToMixedCitation(tree);
-    journalTransforms(tree);
-    const state2 = new JatsParser(file, jats, opts !== null && opts !== void 0 ? opts : { handlers });
-    state2.renderChildren(tree);
+    floatToEndTransform(body);
+    backToBodyTransform(body, jats.back);
+    dataAvailabilityTransform(body);
+    const refLookup = processJatsReferences(body, jats.references, opts);
+    basicTransformations(body, file);
+    journalTransforms(body);
+    const state2 = new JatsParser(file, jats, opts);
+    state2.renderChildren(body);
     while (state2.stack.length > 1)
       state2.closeNode();
+    const tree = state2.stack[0];
     if (state2.unhandled.length && (opts === null || opts === void 0 ? void 0 : opts.logInfo)) {
       opts.logInfo.unhandled = [...new Set(state2.unhandled)];
     }
-    const referenceData = Object.fromEntries(jats.references.map((bibr) => {
-      const id = bibr.id;
-      const names = selectAll2("name,string-name", bibr).map((n) => `${toText2(select3("surname", n))}, ${toText2(select3("given-names", n))}`).join(", ");
-      const year = toText2(select3("year", bibr));
-      const title = toText2(select3("article-title", bibr));
-      const source = toText2(select3("source", bibr));
-      const volume = toText2(select3("volume", bibr));
-      const fpage = toText2(select3("fpage", bibr));
-      const lpage = toText2(select3("lpage", bibr));
-      const doiElement = selectAll2("ext-link,[pub-id-type=doi]", bibr).find((e2) => doi.validate(toText2(e2)));
-      const doiString = doiElement ? toText2(doiElement) : void 0;
-      const doiLink = doiString ? ` <a href=${doi.buildUrl(doiString)}>${doiString}</a>` : "";
-      return [
-        id,
-        {
-          html: `${names}. (${year}). ${title}. <i>${source}</i>, <i>${volume}</i>, ${fpage}-${lpage}.${doiLink}`,
-          doi: doiString
-        }
-      ];
-    }));
-    const referenceOrder = [];
-    const xrefs = selectAll2("xref[ref-type=bibr]", jats.body);
-    xrefs.forEach((xref) => {
-      const rid = xref.rid;
-      if (!referenceOrder.includes(rid)) {
-        referenceOrder.push(rid);
-      }
+    resolveJatsCitations(tree, refLookup);
+    inlineCitationsTransform(tree, jats.references.map(({ id }) => {
+      var _a;
+      const { identifier } = (_a = normalizeLabel(id)) !== null && _a !== void 0 ? _a : {};
+      return identifier;
+    }).filter((id) => !!id));
+    const { frontmatter } = jats;
+    abbreviationSectionTransform(tree, frontmatter);
+    abbreviationFootnoteTransform(tree, frontmatter);
+    abbreviationsFromTree(tree, frontmatter);
+    tableFootnotesToLegend(tree);
+    const abstract = selectAll2("block", tree).find((block) => {
+      return block.data && block.data.part === "abstract";
     });
+    if (abstract) {
+      frontmatter.description = descriptionFromAbstract(toText2(abstract));
+    }
     const result = {
-      references: { order: referenceOrder, data: referenceData },
-      tree: state2.stack[0]
+      tree,
+      frontmatter
     };
     file.result = result;
     return file;
@@ -66188,9 +66435,8 @@ var jatsConvertPlugin = function(jats, opts) {
     return node;
   };
 };
-async function jatsConvertTransform(data, opts) {
+function jatsConvertTransform(jats, opts) {
   var _a, _b, _c;
-  const jats = typeof data === "string" ? new Jats(data) : data;
   if (opts === null || opts === void 0 ? void 0 : opts.logInfo) {
     opts.logInfo.publisher = toText2(select3("publisher-name", jats.tree)) || null;
     opts.logInfo.journal = toText2(select3("journal-title", jats.tree)) || null;
@@ -66211,24 +66457,10 @@ async function jatsConvertTransform(data, opts) {
     }
     opts.logInfo.license = licenseString;
   }
-  const { frontmatter } = jats;
   const file = (_b = opts === null || opts === void 0 ? void 0 : opts.vfile) !== null && _b !== void 0 ? _b : new VFile();
-  const refLookup = await processJatsReferences(jats, opts);
-  floatToEndTransform(jats);
-  backToBodyTransform(jats);
   const pipe = unified().use(jatsConvertPlugin, jats, opts);
-  pipe.stringify((_c = jats.body) !== null && _c !== void 0 ? _c : { type: "body", children: [] }, file);
-  const references = file.result.references;
-  const tree = file.result.tree;
-  resolveJatsCitations(tree, refLookup);
-  inlineCitationsTransform(tree, [...Object.keys(references.data)]);
-  abbreviationSectionTransform(tree, frontmatter);
-  const abstract = selectAll2("block", tree).find((node) => {
-    return node.data && node.data.part === "abstract";
-  });
-  if (abstract) {
-    frontmatter.description = descriptionFromAbstract(toText2(abstract));
-  }
+  pipe.stringify(copyNode((_c = jats.body) !== null && _c !== void 0 ? _c : { type: "body", children: [] }), file);
+  const { tree, frontmatter } = file.result;
   if (opts === null || opts === void 0 ? void 0 : opts.logInfo) {
     opts.logInfo.figures = {
       body: selectAll2("fig", jats.body).length,
@@ -66258,17 +66490,30 @@ async function jatsConvertTransform(data, opts) {
       myst: selectAll2("footnoteDefinition", tree).length
     };
   }
-  return { tree, jats, references, frontmatter };
+  return { tree, frontmatter };
 }
 async function jatsConvert(session, input, opts) {
   const logInfo = { jatsVersion: version_default3 };
+  if (!input) {
+    const xmls = import_node_fs7.default.readdirSync(".").filter((file) => import_node_path6.default.extname(file) === ".xml");
+    if (xmls.length === 0) {
+      throw new Error("No JATS xml file found in current directory; please specify input");
+    }
+    if (xmls.length > 1) {
+      throw new Error("Multiple XML files found in current directory; please specify input");
+    }
+    input = xmls[0];
+  }
   const dir = import_node_path6.default.dirname(input);
   const vfile2 = new VFile();
   vfile2.path = input;
-  const { tree, frontmatter } = await jatsConvertTransform(import_node_fs7.default.readFileSync(input).toString(), {
+  const jats = new Jats(import_node_fs7.default.readFileSync(input).toString());
+  const pmidCache = await getPMIDLookup(jats.references, dir);
+  const { tree, frontmatter } = jatsConvertTransform(jats, {
     vfile: vfile2,
     dir,
     logInfo,
+    pmidCache,
     dois: opts === null || opts === void 0 ? void 0 : opts.dois,
     bibtex: opts === null || opts === void 0 ? void 0 : opts.bibtex
   });
@@ -66297,7 +66542,7 @@ async function jatsConvert(session, input, opts) {
 
 // src/convert.ts
 function makeConvertCLI(program3) {
-  const command = new Command("convert").description("Convert JATS file to MyST mdast json").argument("<input>", "The JATS file").addOption(
+  const command = new Command("convert").description("Convert JATS file to MyST mdast json").argument("[input]", "The JATS file").addOption(
     new Option(
       "--frontmatter <frontmatter>",
       "Treat JATS frontmatter fields as page or project, or ignore if not specified"

package/dist/version.d.ts

@@ -1,3 +1,3 @@
-declare const version = "1.0.12";
+declare const version = "1.0.14";
 export default version;
 //# sourceMappingURL=version.d.ts.map

package/dist/version.js

@@ -1,2 +1,2 @@
-const version = '1.0.12';
+const version = '1.0.14';
 export default version;

package/package.json

@@ -1,9 +1,9 @@
 {
   "name": "jats-cli",
-  "version": "1.0.12",
+  "version": "1.0.14",
   "description": "Typescript CLI for with JATS",
-  "author": "Rowan Cockett <rowan@curvenote.com>",
-  "homepage": "https://github.com/curvenote/jats",
+  "author": "Rowan Cockett <rowan@continuous.foundation>",
+  "homepage": "https://github.com/continuous-foundation/jats",
   "license": "MIT",
   "sideEffects": false,
   "type": "module",
@@ -22,7 +22,7 @@
   },
   "repository": {
     "type": "git",
-    "url": "git+https://github.com/curvenote/jats.git"
+    "url": "git+https://github.com/continuous-foundation/jats.git"
   },
   "bin": {
     "jats": "./dist/jats.cjs"
@@ -40,16 +40,16 @@
     "build": "npm-run-all -l clean copy:version -p build:esm build:cli"
   },
   "bugs": {
-    "url": "https://github.com/curvenote/jats/issues"
+    "url": "https://github.com/continuous-foundation/jats/issues"
   },
   "dependencies": {
     "doi-utils": "^2.0.0",
     "fair-principles": "^2.0.0",
-    "jats-convert": "^1.0.12",
-    "jats-fetch": "^1.0.12",
-    "jats-tags": "^1.0.12",
-    "jats-utils": "^1.0.12",
-    "jats-xml": "^1.0.12",
+    "jats-convert": "^1.0.14",
+    "jats-fetch": "^1.0.14",
+    "jats-tags": "^1.0.14",
+    "jats-utils": "^1.0.14",
+    "jats-xml": "^1.0.14",
     "js-yaml": "^4.1.0",
     "unist-util-is": "^5.2.1",
     "unist-util-select": "^4.0.0"