Abstract from parts and funding for preprints (#17)

Co-authored-by: Rowan Cockett <rowanc1@gmail.com>

Author: fwkoch

.changeset/eighty-nails-press.md

@@ -0,0 +1,5 @@
+---
+"crossref-utils": patch
+---
+
+Support abstract part

src/cli/deposit.ts

@@ -13,6 +13,7 @@ import {
   loadProject,
   parseMyst,
   processProject,
+  resolveFrontmatterParts,
   selectors,
 } from 'myst-cli';
 import type { ISession } from 'myst-cli';
@@ -85,7 +86,9 @@ export async function depositArticleFromSource(session: ISession, depositSource:
     } else {
       fileContents.forEach(({ mdast }) => {
         if (abstractPart) return;
-        abstractPart = extractPart(mdast, 'abstract');
+        abstractPart = extractPart(mdast, 'abstract', {
+          frontmatterParts: resolveFrontmatterParts(session, projectFrontmatter),
+        });
       });
     }
     fileContents.forEach(({ references }) => {
@@ -108,7 +111,9 @@ export async function depositArticleFromSource(session: ISession, depositSource:
       ? projectFrontmatter?.subtitle ?? undefined
       : frontmatter?.subtitle;
     frontmatter = { ...fileContent.frontmatter, title, subtitle };
-    abstractPart = extractPart(fileContent.mdast, 'abstract');
+    abstractPart = extractPart(fileContent.mdast, 'abstract', {
+      frontmatterParts: resolveFrontmatterParts(session, frontmatter),
+    });
     fileContent.references.cite?.order.forEach((key) => {
       const value = fileContent.references.cite?.data[key].doi;
       if (value) dois[key] = value;
@@ -117,6 +122,7 @@ export async function depositArticleFromSource(session: ISession, depositSource:
   }
 
   let abstract: Element | undefined;
+  const description = (frontmatter?.description || projectFrontmatter?.description)?.trim();
   if (abstractPart) {
     transformXrefToLink(abstractPart);
     transformCiteToText(abstractPart);
@@ -128,6 +134,20 @@ export async function depositArticleFromSource(session: ISession, depositSource:
       { name: 'jats:abstract' },
       jats.map((e) => element2JatsUnist(e)),
     ) as Element;
+  } else if (description) {
+    // Use the project description as the fallback for the abstract
+    abstractPart = {
+      type: 'root',
+      children: [{ type: 'paragraph', children: [{ type: 'text', value: description }] }],
+    };
+    transformNewlineToSpace(abstractPart);
+    const serializer = new JatsSerializer(new VFile(), abstractPart as any);
+    const jats = serializer.render(true).elements();
+    abstract = u(
+      'element',
+      { name: 'jats:abstract' },
+      jats.map((e) => element2JatsUnist(e)),
+    ) as Element;
   }
   return { frontmatter: frontmatter ?? {}, dois, abstract, configFile };
 }
@@ -612,7 +632,7 @@ export async function deposit(session: ISession, opts: DepositOptions) {
       throw new Error('preprint deposit may only use a single article');
     }
     const { frontmatter, dois, abstract } = depositArticles[0];
-    body = preprintFromMyst(frontmatter, dois, abstract);
+    body = preprintFromMyst(session, frontmatter, dois, abstract);
   }
   const batch = new DoiBatch(
     { id: opts.id ?? uuid(), depositor: { name, email }, registrant },

src/funding.ts

@@ -1,6 +1,7 @@
 import type { Award, ProjectFrontmatter } from 'myst-frontmatter';
 import type { Fundref } from './types.js';
 import type { ISession } from 'myst-cli-utils';
+import { e, t } from './utils.js';
 
 export function fundrefFromMyst(
   session: ISession,
@@ -35,3 +36,29 @@ export function fundrefFromMyst(
     return { sources, awardNumbers };
   });
 }
+
+export function createFundingXml(funding?: Fundref[]) {
+  if (!funding || funding.length === 0) return null;
+  return e(
+    'fr:program',
+    { name: 'fundref' },
+    funding.map((fundingItem) => {
+      if (!fundingItem.sources.length) {
+        throw new Error('Fundref entry must have at least one source');
+      }
+      return e('fr:assertion', { name: 'fundgroup' }, [
+        ...fundingItem.sources.map((source) => {
+          return e('fr:assertion', { name: 'funder_name' }, [
+            t(source.name),
+            ...source.identifiers.map((id) => {
+              return e('fr:assertion', { name: 'funder_identifier' }, id);
+            }),
+          ]);
+        }),
+        ...fundingItem.awardNumbers.map((awardNumber) => {
+          return e('fr:assertion', { name: 'award_number' }, awardNumber);
+        }),
+      ]);
+    }),
+  );
+}

src/journal.ts

@@ -1,11 +1,11 @@
 import type { Element } from 'xast';
-import { e, t } from './utils.js';
+import { e } from './utils.js';
 import type { JournalArticle, JournalIssue, JournalMetadata } from './types.js';
 import { publicationDateXml } from './dates.js';
 import type { ProjectFrontmatter } from 'myst-frontmatter';
 import { contributorsXmlFromMystAuthors } from './contributors.js';
 import { normalize } from 'doi-utils';
-import { fundrefFromMyst } from './funding.js';
+import { createFundingXml, fundrefFromMyst } from './funding.js';
 import type { ISession } from 'myst-cli-utils';
 
 /**
@@ -172,32 +172,9 @@ export function journalArticleXml({
     children.push(e('pages', pageChildren));
   }
   // publisher_item
-  if (funding?.length) {
-    children.push(
-      e(
-        'fr:program',
-        { name: 'fundref' },
-        funding.map((fundingItem) => {
-          if (!fundingItem.sources.length) {
-            throw new Error('Fundref entry must have at least one source');
-          }
-          return e('fr:assertion', { name: 'fundgroup' }, [
-            ...fundingItem.sources.map((source) => {
-              return e('fr:assertion', { name: 'funder_name' }, [
-                t(source.name),
-                ...source.identifiers.map((id) => {
-                  return e('fr:assertion', { name: 'funder_identifier' }, id);
-                }),
-              ]);
-            }),
-            ...fundingItem.awardNumbers.map((awardNumber) => {
-              return e('fr:assertion', { name: 'award_number' }, awardNumber);
-            }),
-          ]);
-        }),
-      ),
-    );
-  }
+  const fundingXml = createFundingXml(funding);
+  if (fundingXml) children.push(fundingXml);
+
   if (license) {
     children.push(
       e('ai:program', { name: 'AccessIndicators' }, [

src/preprint.ts

@@ -5,6 +5,8 @@ import type { ProjectFrontmatter } from 'myst-frontmatter';
 import { normalize } from 'doi-utils';
 import { contributorsXmlFromMystAuthors } from './contributors.js';
 import { dateXml } from './dates.js';
+import { createFundingXml, fundrefFromMyst } from './funding.js';
+import type { ISession } from 'myst-cli';
 
 /**
  * Create posted content xml
@@ -19,6 +21,7 @@ export function preprintXml({
   title,
   subtitle,
   abstract,
+  funding,
   doi_data,
   citations,
   license,
@@ -35,6 +38,10 @@ export function preprintXml({
   children.push(e('titles', titles));
   children.push(posted_date);
   if (abstract) children.push(abstract);
+
+  const fundingXml = createFundingXml(funding);
+  if (fundingXml) children.push(fundingXml);
+
   if (license) {
     children.push(
       e('ai:program', { name: 'AccessIndicators' }, [
@@ -85,6 +92,7 @@ export function preprintXml({
 }
 
 export function preprintFromMyst(
+  session: ISession,
   myst: ProjectFrontmatter,
   citations?: Record<string, string>,
   abstract?: Element,
@@ -98,6 +106,7 @@ export function preprintFromMyst(
     date: typeof date === 'string' ? new Date(date) : undefined,
     license: license?.content?.url,
     abstract,
+    funding: fundrefFromMyst(session, myst),
   };
   if (license && license.content?.CC) {
     // Only put in CC licenses at this time