Merge pull request #26 from elifesciences/investigate-suspicious-vers…

…ion-parsing

Investigate suspicious version parsing

src/docmap-generator.ts

@@ -14,6 +14,7 @@ import {
   JsonLDFrameUrl,
   Manifestation,
   ManifestationType,
+  Manuscript,
   Output,
   Participant,
   PeerReview,
@@ -25,23 +26,22 @@ import {
   Url,
   VersionOfRecord,
   WebPage,
-  Work,
 } from './docmap';
 
 type Steps = {
   'first-step': string,
   steps: Map<string, Step>
 };
 
-export const generatePreprint = (doi: DOI, published?: Date, url?: Url, version?: string, content?: Manifestation[], license?: string, work?: Work): Preprint => ({
+export const generatePreprint = (doi: DOI, published?: Date, url?: Url, version?: string, content?: Manifestation[], license?: string, manuscript?: Manuscript): Preprint => ({
   type: ExpressionType.Preprint,
   doi,
   url,
   published,
   versionIdentifier: version,
   content,
   license,
-  embodimentOf: work,
+  partOf: manuscript,
 });
 
 export const generateRevisedPreprint = (doi: DOI, published?: Date, url?: Url, version?: string, content?: Manifestation[]): RevisedPreprint => ({
@@ -114,7 +114,7 @@ export const generateWebContent = (url: Url): WebPage => ({
   url,
 });
 
-export const generateWork = (doi?: DOI, identifier?: string, volumeIdentifier?: string, electronicArticleIdentifier?: string): Work => ({
+export const generateManuscript = (doi?: DOI, identifier?: string, volumeIdentifier?: string, electronicArticleIdentifier?: string): Manuscript => ({
   type: 'manuscript',
   doi,
   identifier,

src/docmap-parser.test.ts

@@ -386,8 +386,8 @@ describe('docmap-parser', () => {
     expect(parsedData.versions[0].license).toStrictEqual('http://creativecommons.org/licenses/by/4.0/');
   });
 
-  it('extracts embodimentOf, if present', () => {
-    const parsedData = parseDocMap(fixtures.preprintWithWorkAsOutput());
+  it('extracts partOf, if present', () => {
+    const parsedData = parseDocMap(fixtures.preprintWithManuscriptAsOutput());
 
     expect(parsedData.manuscript).toStrictEqual({
       doi: '10.1101/123456',
@@ -397,7 +397,7 @@ describe('docmap-parser', () => {
   });
 
   it('extracts partial embodimentOf, if present', () => {
-    const parsedData = parseDocMap(fixtures.preprintWithPartialWorkAsOutput());
+    const parsedData = parseDocMap(fixtures.preprintWithPartialManuscriptAsOutput());
 
     expect(parsedData.manuscript).toStrictEqual({
       doi: '10.1101/123456',
@@ -420,4 +420,17 @@ describe('docmap-parser', () => {
       versionIdentifier: '1',
     });
   });
+
+  it('parses umbrella expressions from multiple locations', () => {
+    const docmap = fixtures.preprintWithUmbrellaExpressionsFromMultipleLocations();
+    const parsedData = parseDocMap(docmap);
+
+    expect(parsedData.versions.length).toEqual(1);
+
+    expect(parsedData.manuscript).toStrictEqual({
+      doi: '10.1101/123456',
+      volume: '1',
+      eLocationId: 'RP123456',
+    });
+  });
 });

src/docmap-parser.ts

@@ -93,14 +93,14 @@ export type ManuscriptData = {
 };
 
 const getManuscriptFromExpression = (expression: Expression): Manuscript | false => {
-  if (!expression.embodimentOf) {
+  if (!expression.partOf) {
     return false;
   }
 
   return {
-    doi: expression.embodimentOf.doi,
-    volume: expression.embodimentOf.volumeIdentifier,
-    eLocationId: expression.embodimentOf.electronicArticleIdentifier,
+    doi: expression.partOf.doi,
+    volume: expression.partOf.volumeIdentifier,
+    eLocationId: expression.partOf.electronicArticleIdentifier,
   };
 };
 
@@ -214,11 +214,23 @@ const findAndUpdateOrAddPreprintDescribedBy = (expression: Expression, preprintC
   return foundPreprint;
 };
 
-const republishPreprintAs = (expression: Expression, preprint: ReviewedPreprint) => {
+const republishPreprintAs = (expression: Expression, preprint: ReviewedPreprint, manuscript: Manuscript) => {
   if (!expression.doi) {
     throw Error('Cannot identify Expression by DOI');
   }
-
+  const foundManuscriptData = getManuscriptFromExpression(expression);
+  const existingManuscript = manuscript;
+  if (foundManuscriptData) {
+    if (foundManuscriptData.doi) {
+      existingManuscript.doi = foundManuscriptData.doi;
+    }
+    if (foundManuscriptData.eLocationId) {
+      existingManuscript.eLocationId = foundManuscriptData.eLocationId;
+    }
+    if (foundManuscriptData.volume) {
+      existingManuscript.volume = foundManuscriptData.volume;
+    }
+  }
   const newPreprint = preprint;
 
   newPreprint.id = expression.identifier ?? expression.doi;
@@ -371,7 +383,7 @@ const parseStep = (step: Step, preprints: Array<ReviewedPreprint>, manuscript: M
   if (inferredRepublished) {
     // preprint input, preprint output, but no evaluations = superceed input preprint with output Reviewed Preprint
     const preprint = findAndUpdateOrAddPreprintDescribedBy(inferredRepublished.originalExpression, preprints, manuscript);
-    republishPreprintAs(inferredRepublished.republishedExpression, preprint);
+    republishPreprintAs(inferredRepublished.republishedExpression, preprint, manuscript);
   }
 
   const inferredPeerReviewed = getPeerReviewedPreprint(step);
@@ -382,7 +394,7 @@ const parseStep = (step: Step, preprints: Array<ReviewedPreprint>, manuscript: M
 
     // sometimes a new reviewed preprint is published as an output
     if (inferredPeerReviewed.republishedPreprint) {
-      republishPreprintAs(inferredPeerReviewed.republishedPreprint, preprint);
+      republishPreprintAs(inferredPeerReviewed.republishedPreprint, preprint, manuscript);
     }
   }
 

src/docmap.ts

@@ -38,7 +38,7 @@ export type Expression = {
   doi?: DOI,
   content?: Manifestation[],
   license?: string,
-  embodimentOf?: Work,
+  partOf?: Manuscript,
 };
 
 export type Manifestation = {
@@ -116,7 +116,7 @@ export type Assertion = {
   happened?: Date,
 };
 
-export type Work = {
+export type Manuscript = {
   type: 'manuscript',
   doi?: DOI,
   identifier?: string,

src/test-fixtures/docmapGenerators.ts

@@ -16,7 +16,7 @@ import {
   generateStep,
   generateUnderReviewAssertion,
   generateWebContent,
-  generateWork,
+  generateManuscript,
 } from '../docmap-generator';
 
 const publisher = {
@@ -48,20 +48,30 @@ export const fixtures = {
     return generateDocMap('test', publisher, firstStep);
   },
 
-  preprintWithWorkAsOutput: (): DocMap => {
-    const work = generateWork('10.1101/123456', '123456', '1', 'RP123456');
-    const preprint = generatePreprint('preprint/article1', new Date('2022-03-01'), undefined, undefined, undefined, undefined, work);
+  preprintWithManuscriptAsOutput: (): DocMap => {
+    const manuscript = generateManuscript('10.1101/123456', '123456', '1', 'RP123456');
+    const preprint = generatePreprint('preprint/article1', new Date('2022-03-01'), undefined, undefined, undefined, undefined, manuscript);
     const firstStep = generateStep([], [generateAction([], [preprint])], []);
     return generateDocMap('test', publisher, firstStep);
   },
 
-  preprintWithPartialWorkAsOutput: (): DocMap => {
-    const work = generateWork('10.1101/123456', '123456', undefined, 'RP123456');
-    const preprint = generatePreprint('preprint/article1', new Date('2022-03-01'), undefined, undefined, undefined, undefined, work);
+  preprintWithPartialManuscriptAsOutput: (): DocMap => {
+    const manuscript = generateManuscript('10.1101/123456', '123456', undefined, 'RP123456');
+    const preprint = generatePreprint('preprint/article1', new Date('2022-03-01'), undefined, undefined, undefined, undefined, manuscript);
     const firstStep = generateStep([], [generateAction([], [preprint])], []);
     return generateDocMap('test', publisher, firstStep);
   },
 
+  preprintWithUmbrellaExpressionsFromMultipleLocations: (): DocMap => {
+    const manuscript = generateManuscript('10.1101/123456', '123456', undefined, 'RP123456');
+    const manuscript2 = generateManuscript('10.1101/123456', '123456', "1");
+    const preprint = generatePreprint('preprint/article1', new Date('2022-03-01'), undefined, undefined, undefined, undefined, manuscript);
+    const preprint2 = generatePreprint('preprint/article1', new Date('2022-03-01'), undefined, undefined, undefined, undefined, manuscript2);
+    const assertion = generateDraftAssertion(preprint2);
+    const firstStep = generateStep([], [generateAction([], [preprint])], [assertion]);
+    return generateDocMap('test', publisher, firstStep);
+  },
+
   simplePreprintWithUrlAsOutput: (): DocMap => {
     const preprint = generatePreprint('preprint/article1', new Date('2022-03-01'), 'https://somewhere.org/preprint/article1');
     const firstStep = generateStep([], [generateAction([], [preprint])], []);