jats-cli 1.0.14 → 1.0.16

package/dist/jats.cjs

@@ -48999,7 +48999,7 @@ var {
 } = import_index.default;
 
 // src/version.ts
-var version = "1.0.14";
+var version = "1.0.16";
 var version_default = version;
 
 // src/parse.ts
@@ -56426,6 +56426,7 @@ var RefType;
   RefType2["tableFn"] = "table-fn";
   RefType2["ref"] = "ref";
   RefType2["media"] = "media";
+  RefType2["video"] = "video";
 })(RefType || (RefType = {}));
 
 // ../jats-fetch/dist/download.js
@@ -57848,6 +57849,10 @@ async function convertPMID2PMCID(session, pmid, opts) {
   const pmcid = await convertId(session, pmid, "pmid", "pmcid", opts);
   return pmcid;
 }
+async function convertDOI2PMCID(session, input, opts) {
+  const pmDoi = await convertId(session, input, "doi", "pmcid", opts);
+  return pmDoi;
+}
 async function convertPMIDs2DOIs(session, pmids, opts) {
   var _a, _b, _c;
   pmids = [...new Set(pmids.map((pmid) => normalizePMID(session, pmid)))];
@@ -58088,8 +58093,21 @@ var joss = {
     return `https://raw.githubusercontent.com/openjournals/joss-papers/master/joss.${id}/${org}.${jossId}.jats`;
   }
 };
-var DEFAULT_RESOLVERS2 = [elife2, plos, joss];
-async function customResolveJatsUrlFromDoi(session, doiString, opts = { resolvers: DEFAULT_RESOLVERS2 }) {
+var biorxiv2 = {
+  test(url) {
+    return new URL(url).hostname === "www.biorxiv.org";
+  },
+  async jatsUrl(url, fetcher) {
+    const resp = await (fetcher !== null && fetcher !== void 0 ? fetcher : fetch2)(url);
+    const xmlUrl = (await resp.text()).match(/https:\/\/www\.biorxiv\.org\/content\/biorxiv\/[^\s]+\.full\.pdf/);
+    if (!xmlUrl) {
+      throw new Error(`Could not resolve JATS for biorxiv URL: ${url}`);
+    }
+    return xmlUrl[0].replace(".full.pdf", ".source.xml");
+  }
+};
+var DEFAULT_RESOLVERS2 = [biorxiv2, elife2, plos, joss];
+async function customResolveJatsUrlFromDoi(session, doiString, opts) {
   var _a, _b;
   if (!doi.validate(doiString))
     throw new Error(`The doi ${doiString} is not valid`);
@@ -58098,10 +58116,11 @@ async function customResolveJatsUrlFromDoi(session, doiString, opts = { resolver
   const resp = await ((_a = opts === null || opts === void 0 ? void 0 : opts.fetcher) !== null && _a !== void 0 ? _a : fetch2)(doiUrl);
   const articleUrl = resp.url;
   session.log.debug(`Found resolved URL for DOI at ${articleUrl}`);
-  const resolver = (_b = opts === null || opts === void 0 ? void 0 : opts.resolvers) === null || _b === void 0 ? void 0 : _b.find((r2) => r2.test(articleUrl));
+  const resolvers = (_b = opts === null || opts === void 0 ? void 0 : opts.resolvers) !== null && _b !== void 0 ? _b : DEFAULT_RESOLVERS2;
+  const resolver = resolvers === null || resolvers === void 0 ? void 0 : resolvers.find((r2) => r2.test(articleUrl));
   if (!resolver)
     throw new Error(`Could not resolve JATS for ${articleUrl}, no resolver matched`);
-  const jatsUrl = resolver.jatsUrl(articleUrl);
+  const jatsUrl = await resolver.jatsUrl(articleUrl, opts === null || opts === void 0 ? void 0 : opts.fetcher);
   return jatsUrl;
 }
 
@@ -58217,6 +58236,20 @@ async function jatsFetch(session, input, opts) {
       output = foldername;
     }
   }
+  if (input.match(/^[0-9]+$/)) {
+    const pmcid = await convertPMID2PMCID(session, input);
+    if (pmcid) {
+      session.log.debug(`Resolved input ${input} to PMC ID: ${pmcid}`);
+      input = pmcid;
+    }
+  }
+  if (doi.validate(input)) {
+    const pmcid = await convertDOI2PMCID(session, input);
+    if (pmcid) {
+      session.log.debug(`Resolved input ${input} to PMC ID: ${pmcid}`);
+      input = pmcid;
+    }
+  }
   if (!output)
     output = opts.data ? `${input}` : ".";
   if (!import_node_path4.default.extname(output)) {
@@ -58590,15 +58623,16 @@ var Jats = class {
     }
     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) && elements[0].type !== "doctype") {
+    const filteredElements = elements === null || elements === void 0 ? void 0 : elements.filter((elem) => elem.type !== "comment");
+    if ((filteredElements === null || filteredElements === void 0 ? void 0 : filteredElements.length) && filteredElements[0].type !== "doctype") {
       (_a = this.log) === null || _a === void 0 ? void 0 : _a.warn("JATS is missing DOCTYPE declaration");
-      elements.unshift({ type: "doctype" });
+      filteredElements.unshift({ type: "doctype" });
     }
-    if (!((elements === null || elements === void 0 ? void 0 : elements.length) === 2 && elements[0].type === "doctype" && hasSingleArticle(elements[1]))) {
+    if (!((filteredElements === null || filteredElements === void 0 ? void 0 : filteredElements.length) === 2 && filteredElements[0].type === "doctype" && hasSingleArticle(filteredElements[1]))) {
       throw new Error("JATS must be structured as <!DOCTYPE><article>...</article>");
     }
-    this.doctype = elements[0].doctype;
-    const converted = convertToUnist(elements[1]);
+    this.doctype = filteredElements[0].doctype;
+    const converted = convertToUnist(filteredElements[1]);
     this.tree = select2("article", converted);
     (_b = this.log) === null || _b === void 0 ? void 0 : _b.debug(toc("Parsed and converted JATS to unist tree in %s"));
   }
@@ -64835,7 +64869,10 @@ function admonitionTransform(tree, file) {
   captions.forEach((caption) => {
     const title = select3(Tags.title, caption);
     if (!title) {
-      fileWarn(file, "", { node: caption, ruleId: RuleId.jatsParses });
+      fileWarn(file, "Encountered boxed-text without title", {
+        node: caption,
+        ruleId: RuleId.jatsParses
+      });
       return;
     }
     caption.type = "admonitionTitle";
@@ -64890,19 +64927,19 @@ function isBioRxiv(tree) {
   const journalId = select3("journal-id[journal-id-type=hwp]", tree);
   return toText2(journalId) === "biorxiv";
 }
-function graphicToBioRxivUrl(tree) {
-  if (!isBioRxiv(tree))
+function graphicToBioRxivUrl(fullTree, body) {
+  if (!isBioRxiv(fullTree))
     return;
   console.log("inside graphics transform");
-  const accepted = select3("date[date-type=accepted]", tree);
+  const accepted = select3("date[date-type=accepted]", fullTree);
   if (!accepted)
     return;
   const year = toText2(select3("year", accepted));
   const month = toText2(select3("month", accepted)).padStart(2, "0");
   const day = toText2(select3("day", accepted)).padStart(2, "0");
-  const slug = toText2(select3("article-id[pub-id-type=doi]", tree)).split("/").slice(1).join("/");
+  const slug = toText2(select3("article-id[pub-id-type=doi]", fullTree)).split("/").slice(1).join("/");
   const urlBase = `https://www.biorxiv.org/content/biorxiv/early/${year}/${month}/${day}/${slug}`;
-  selectAll2("fig,table-wrap", tree).forEach((node) => {
+  selectAll2("fig,table-wrap", body).forEach((node) => {
     const figId = node["hwp:id"];
     if (!figId)
       return;
@@ -65568,12 +65605,12 @@ function basicTransformations(body, file) {
   admonitionTransform(body, file);
   figCaptionTitleTransform(body);
 }
-function journalTransforms(body) {
-  graphicToBioRxivUrl(body);
+function journalTransforms(fullTree, body) {
+  graphicToBioRxivUrl(fullTree, body);
 }
 
 // ../jats-convert/dist/version.js
-var version3 = "1.0.14";
+var version3 = "1.0.16";
 var version_default3 = version3;
 
 // ../jats-convert/dist/myst/inlineCitations.js
@@ -65863,6 +65900,20 @@ function abbreviationsFromText(text) {
 }
 
 // ../jats-convert/dist/index.js
+var MEDIA_FIGURE_EXTENSIONS = [
+  ".png",
+  ".jpg",
+  ".jpeg",
+  ".svg",
+  ".gif",
+  ".tiff",
+  ".tif",
+  ".eps",
+  ".webp",
+  ".mp4",
+  ".mov",
+  ".avi"
+];
 function refTypeToReferenceKind(kind) {
   switch (kind) {
     case RefType.sec:
@@ -66042,7 +66093,7 @@ var handlers = {
       const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
       state2.openNode("div", { label, identifier });
     }
-    state2.renderInline(node, "admonition", { kind: "info" });
+    state2.renderInline(node, "admonition", { kind: "note" });
     if (node.id) {
       state2.closeNode();
     }
@@ -66086,7 +66137,7 @@ var handlers = {
   fig(node, state2) {
     var _a;
     const caption = select3("caption", node);
-    const graphic = select3("graphic", node);
+    const graphic = select3("graphic,media", node);
     const title = select3("title", node);
     const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
     state2.openNode("container", { label, identifier, kind: "figure" });
@@ -66216,6 +66267,7 @@ var handlers = {
           kind: "narrative"
         });
         return;
+      case RefType.video:
       case RefType.app:
       case RefType.boxedText:
       case RefType.media:
@@ -66241,14 +66293,49 @@ var handlers = {
     }
   },
   ["supplementary-material"](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) {
+    var _a, _b, _c, _d, _e;
+    const { label, identifier } = (_a = normalizeLabel(node.id)) !== null && _a !== void 0 ? _a : {};
+    let maybeCaption;
+    let media;
+    if (((_b = node.children) === null || _b === void 0 ? void 0 : _b.length) === 1 && node.children[0].type === "media") {
+      media = node.children[0];
+    } else if (((_c = node.children) === null || _c === void 0 ? void 0 : _c.length) === 2 && node.children[0].type === "label" && node.children[1].type === "media") {
+      maybeCaption = node.children[0];
+      media = node.children[1];
+    }
+    const url = media === null || media === void 0 ? void 0 : media["xlink:href"];
+    let caption = (_d = select3("caption", media)) !== null && _d !== void 0 ? _d : maybeCaption;
+    if (((_e = caption === null || caption === void 0 ? void 0 : caption.children) === null || _e === void 0 ? void 0 : _e.length) === 1 && caption.children[0].type === "p") {
+      caption = caption.children[0];
+    }
+    if (url && MEDIA_FIGURE_EXTENSIONS.find((ext) => url.endsWith(ext))) {
+      const title = select3("title", media);
+      state2.openNode("container", { label, identifier, kind: "figure" });
+      const wasInContainer = state2.data.isInContainer;
+      state2.data.isInContainer = true;
+      state2.addLeaf("image", { url });
+      state2.openNode("caption");
+      state2.openNode("link", { url });
+      if (title) {
+        state2.openNode("strong");
+        state2.renderChildren(title);
+        state2.closeNode();
+      }
+      if (caption) {
+        state2.renderChildren(caption);
+      }
+      state2.closeNode();
+      state2.closeNode();
       state2.closeNode();
+      state2.data.isInContainer = wasInContainer;
+    } else {
+      if (node.id) {
+        state2.openNode("div", { label, identifier });
+      }
+      state2.renderChildren(node);
+      if (node.id) {
+        state2.closeNode();
+      }
     }
   },
   ["app-group"](node, state2) {
@@ -66398,7 +66485,7 @@ var jatsConvertPlugin = function(jats, opts) {
     dataAvailabilityTransform(body);
     const refLookup = processJatsReferences(body, jats.references, opts);
     basicTransformations(body, file);
-    journalTransforms(body);
+    journalTransforms(jats.tree, body);
     const state2 = new JatsParser(file, jats, opts);
     state2.renderChildren(body);
     while (state2.stack.length > 1)

package/dist/version.d.ts

@@ -1,3 +1,3 @@
-declare const version = "1.0.14";
+declare const version = "1.0.16";
 export default version;
 //# sourceMappingURL=version.d.ts.map

package/dist/version.js

@@ -1,2 +1,2 @@
-const version = '1.0.14';
+const version = '1.0.16';
 export default version;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jats-cli",
-  "version": "1.0.14",
+  "version": "1.0.16",
   "description": "Typescript CLI for with JATS",
   "author": "Rowan Cockett <rowan@continuous.foundation>",
   "homepage": "https://github.com/continuous-foundation/jats",
@@ -45,11 +45,11 @@
   "dependencies": {
     "doi-utils": "^2.0.0",
     "fair-principles": "^2.0.0",
-    "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",
+    "jats-convert": "^1.0.16",
+    "jats-fetch": "^1.0.16",
+    "jats-tags": "^1.0.16",
+    "jats-utils": "^1.0.16",
+    "jats-xml": "^1.0.16",
     "js-yaml": "^4.1.0",
     "unist-util-is": "^5.2.1",
     "unist-util-select": "^4.0.0"