jats-cli 1.0.11 → 1.0.13

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 = "1.0.11";
+var version = "1.0.13";
 var version_default = version;
 
 // src/parse.ts
@@ -56436,6 +56436,7 @@ 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 +57790,7 @@ 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 = `${LISTING_BASE_URL}oa_file_list.csv`;
@@ -57811,42 +57813,52 @@ function normalizePMID(session, pmid) {
   }
   return pmid;
 }
-async function convertPMID2PMCID(session, pmid, opts) {
+async function convertId(session, id, from, to, opts) {
   var _a, _b, _c;
-  pmid = normalizePMID(session, pmid);
   const toc = tic();
-  const resp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : defaultFetcher)(`${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=${id}`, "json");
   if (!resp.ok) {
-    session.log.debug(`Failed to convert PubMedID: ${pmid}`);
+    session.log.debug(`Failed to convert ${from} ID: ${id}`);
     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 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 ${newId} in %s.`));
+  }
+  return newId;
+}
+async function convertIds(session, ids, from, to, opts) {
+  var _a, _b, _c;
+  const toc = tic();
+  const resp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : defaultFetcher)(`${IDCONV_URL}?tool=jats-xml&format=json&ids=${ids.join(",")}`, "json");
+  if (!resp.ok) {
+    session.log.debug(`Failed to convert ${from} ${ids.length} IDs`);
+    return;
+  }
+  const data = await resp.json();
+  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";
@@ -57911,13 +57923,12 @@ async function getPubMedJatsFromS3(session, pmcid) {
     session.log.debug(`Not available from open-access S3 bucket: ${pmcid}`);
     return { success: false, source: pmcid };
   } else {
+    session.log.debug(`Fetching PMC JATS from S3 bucket: ${pmcid}`);
     const result = await downloadFileFromS3(client, found.path, OA_CONFIG);
     return result;
   }
 }
 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") {
@@ -57962,19 +57973,37 @@ 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 ${outputDir}`, session.log);
@@ -57994,9 +58023,28 @@ async function downloadAndUnzipPMC(session, entry, outputDir, fetcher) {
   import_node_fs4.default.rmdirSync(zipDir);
 }
 async function getDataFromPMC(session, pmcid, outputDir, listing, fetcher) {
-  const listingFile = await getListingsFile(session, listing, fetcher);
-  const entry = await searchListingForPMC(listingFile, pmcid);
-  await downloadAndUnzipPMC(session, entry, outputDir, fetcher);
+  let url;
+  if (pmcid.endsWith(".tar.gz")) {
+    url = pmcid;
+  } else {
+    if (!pmcid.startsWith("PMC")) {
+      throw new Error("Data may only be downloaded for PMC articles");
+    }
+    try {
+      const metadata = await getDownloadMetadata(pmcid);
+      url = metadata.url;
+    } catch {
+      if (listing) {
+        const listingFile = await getListingsFile(session, listing, fetcher);
+        const entry = await searchListingForPMC(listingFile, pmcid);
+        url = `${LISTING_BASE_URL}${entry.url}`;
+      }
+    }
+  }
+  if (!url) {
+    throw new Error(`Unable to find PMC data download url for: ${pmcid}`);
+  }
+  await downloadAndUnzipPMC(session, url, outputDir, fetcher);
 }
 async function getPubMedJatsFromData(session, pmcid, outputDir, listing, fetcher) {
   await getDataFromPMC(session, pmcid, outputDir, listing, fetcher);
@@ -58071,6 +58119,9 @@ async function downloadFromUrl(session, jatsUrl, opts) {
     session.log.warn(`Expected content-type "application/xml" instead we got "${contentType}" for ${jatsUrl}`);
   }
   const data = await resp.text();
+  if (!data.match(/<article/)) {
+    throw new Error(`XML downloaded from ${jatsUrl} does not look like a JATS article`);
+  }
   session.log.debug(toc(`Fetched document with content-type "${contentType}" in %s`));
   return data;
 }
@@ -58126,43 +58177,66 @@ async function downloadJatsFromUrl(session, urlOrDoi, opts = {}) {
     }
     return { success: false, source: expectedUrls[0] };
   }
-  if (doi.validate(urlOrDoi)) {
-    const jatsUrl = await customResolveJatsUrlFromDoi(session, urlOrDoi, opts);
-    const data = await downloadFromUrl(session, jatsUrl, opts);
-    return { success: true, source: jatsUrl, data };
-  }
-  if (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 = await downloadFromUrl(session, urlOrDoi, opts);
-    return { success: true, source: urlOrDoi, data };
+  try {
+    if (doi.validate(urlOrDoi)) {
+      const jatsUrl = await customResolveJatsUrlFromDoi(session, urlOrDoi, opts);
+      const data = await downloadFromUrl(session, jatsUrl, opts);
+      return { success: true, source: jatsUrl, data };
+    }
+    if (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 = await downloadFromUrl(session, urlOrDoi, opts);
+      return { success: true, source: urlOrDoi, data };
+    }
+  } catch (error) {
+    session.log.debug(error.message);
+    return { success: false, source: urlOrDoi };
   }
-  throw new Error(`Could not find ${urlOrDoi} locally, and it doesn't look like a URL or DOI`);
+  session.log.debug(`Could not find ${urlOrDoi} locally, and it doesn't look like a URL or DOI`);
+  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;
-  try {
-    result = await downloadJatsFromUrl(session, input);
-  } catch {
+  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)) {
-      result = await getPubMedJatsFromData(session, input, import_node_path4.default.dirname(output), opts.listing);
-    }
+  }
+  if (!(result === null || result === void 0 ? void 0 : result.data)) {
+    result = await downloadJatsFromUrl(session, input);
+  }
+  if (!(result === null || result === void 0 ? void 0 : result.data) && input.startsWith("PMC")) {
+    result = await getPubMedJatsFromData(session, input, import_node_path4.default.dirname(output), opts.listing);
   }
   if (!(result === null || result === void 0 ? void 0 : result.data)) {
     throw new Error(`Unable to resolve JATS XML content from ${input}`);
@@ -58174,17 +58248,10 @@ 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 = (
@@ -58516,7 +58583,7 @@ 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");
     }
@@ -64759,7 +64826,7 @@ 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) {
@@ -64977,7 +65044,9 @@ function recurseSections(tree, depth = 1, titleType) {
       firstChild3 = (_b = sec.children) === null || _b === void 0 ? void 0 : _b[1];
     }
     if ((firstChild3 === null || firstChild3 === void 0 ? void 0 : firstChild3.type) === "title") {
-      if (titleType === "strong") {
+      if (sec.type === "ack" && toText(firstChild3).toLowerCase().startsWith("ack")) {
+        firstChild3.type = "__delete__";
+      } else if (titleType === "strong") {
         firstChild3.type = "p";
         firstChild3.children = [{ type: "bold", children: firstChild3.children }];
       } else {
@@ -64986,6 +65055,8 @@ function recurseSections(tree, depth = 1, titleType) {
         firstChild3.depth = depth;
       }
     }
+    if (sec.type === "ack")
+      sec.part = "acknowledgments";
     recurseSections(sec, depth + 1, titleType);
   });
 }
@@ -64995,8 +65066,6 @@ function sectionTransform(tree, titleType) {
   remove(tree, "__delete__");
   const topSections = (_a = tree.children) === null || _a === void 0 ? void 0 : _a.filter((n) => isSection(n));
   topSections.forEach((sec) => {
-    if (sec.type === "ack")
-      sec.part = "acknowledgments";
     sec.type = "block";
   });
   while (select3("sec", tree))
@@ -65462,7 +65531,7 @@ function journalTransforms(tree) {
 }
 
 // ../jats-convert/dist/version.js
-var version3 = "1.0.11";
+var version3 = "1.0.13";
 var version_default3 = version3;
 
 // ../jats-convert/dist/myst/inlineCitations.js
@@ -65683,7 +65752,7 @@ 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) {
@@ -65980,7 +66049,7 @@ var handlers = {
       }
       case RefType.fn:
       case RefType.tableFn: {
-        state2.renderInline(node, "footnoteReference", { label, identifier });
+        state2.addLeaf("footnoteReference", { label, identifier });
         return;
       }
       default: {

package/dist/version.d.ts

@@ -1,3 +1,3 @@
-declare const version = "1.0.11";
+declare const version = "1.0.13";
 export default version;
 //# sourceMappingURL=version.d.ts.map

package/dist/version.js

@@ -1,2 +1,2 @@
-const version = '1.0.11';
+const version = '1.0.13';
 export default version;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jats-cli",
-  "version": "1.0.11",
+  "version": "1.0.13",
   "description": "Typescript CLI for with JATS",
   "author": "Rowan Cockett <rowan@curvenote.com>",
   "homepage": "https://github.com/curvenote/jats",
@@ -45,11 +45,11 @@
   "dependencies": {
     "doi-utils": "^2.0.0",
     "fair-principles": "^2.0.0",
-    "jats-convert": "^1.0.11",
-    "jats-fetch": "^1.0.11",
-    "jats-tags": "^1.0.11",
-    "jats-utils": "^1.0.11",
-    "jats-xml": "^1.0.11",
+    "jats-convert": "^1.0.13",
+    "jats-fetch": "^1.0.13",
+    "jats-tags": "^1.0.13",
+    "jats-utils": "^1.0.13",
+    "jats-xml": "^1.0.13",
     "js-yaml": "^4.1.0",
     "unist-util-is": "^5.2.1",
     "unist-util-select": "^4.0.0"