@financial-times/cp-content-pipeline-schema 2.15.0 → 3.0.0

package/src/helpers/decorateHeadshotUrl.ts

@@ -1,6 +1,6 @@
-import { CapiPerson } from '../types/internal-content'
+import { CapiPerson } from '../model/schemas/capi/internal-content'
 
-export const UUID_REGEX =
+const UUID_REGEX =
   /\b[0-9a-f]{8}-[0-9a-f]{4}-[1-9][0-9a-f]{3}-[89ab][0-9a-f]{3}-[0-9a-f]{12}\b/i
 
 const IMAGESET_REGEX = /fthead(?:-v\d)?\:[^?]+/

package/src/model/Byline.test.ts

@@ -1,107 +1,188 @@
 import { Byline } from './Byline'
 import context from '../fixtures/dummyContext'
+import { Concept } from './Concept'
+
 const vanity = true
 
 describe('byline transformation', () => {
   test('adds a link around a single known author', async () => {
     const bylineText = 'Chris Giles in London'
-    const authorUrlMapping = new Map([
-      [
-        'Chris Giles',
-        'https://www.ft.com/stream/uuid/1d556016-ad16-4fe7-8724-42b3fb15ad28',
-      ],
-    ])
+    const authors = [
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/1d556016-ad16-4fe7-8724-42b3fb15ad28',
+          apiUrl:
+            'https://api.ft.com/people/1d556016-ad16-4fe7-8724-42b3fb15ad28',
+          prefLabel: 'Chris Giles',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+    ]
 
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
+    const byline = new Byline(bylineText, vanity, authors)
     const result = await byline.buildBylineTree()
     expect(result).toMatchSnapshot()
   })
 
   test('adds links around multiple known authors', async () => {
     const bylineText = 'Chris Giles and Martin Wolf in London'
-    const authorUrlMapping = new Map([
-      [
-        'Chris Giles',
-        'https://www.ft.com/stream/uuid/1d556016-ad16-4fe7-8724-42b3fb15ad28',
-      ],
-      [
-        'Martin Wolf',
-        'https://www.ft.com/stream/uuid/7c1e1e72-57ae-4461-862a-f8d24dd42e22',
-      ],
-    ])
+    const authors = [
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/1d556016-ad16-4fe7-8724-42b3fb15ad28',
+          apiUrl:
+            'https://api.ft.com/people/1d556016-ad16-4fe7-8724-42b3fb15ad28',
+          prefLabel: 'Chris Giles',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/7c1e1e72-57ae-4461-862a-f8d24dd42e22',
+          apiUrl:
+            'https://api.ft.com/people/7c1e1e72-57ae-4461-862a-f8d24dd42e22',
+          prefLabel: 'Martin Wolf',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+    ]
 
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
+    const byline = new Byline(bylineText, vanity, authors)
     const result = await byline.buildBylineTree()
     expect(result).toMatchSnapshot()
   })
 
   test('ignores unknown authors in byline text', async () => {
     const bylineText = 'Chris Giles and Nick Ramsbottom in London'
-    const authorUrlMapping = new Map([
-      [
-        'Chris Giles',
-        'https://www.ft.com/stream/uuid/1d556016-ad16-4fe7-8724-42b3fb15ad28',
-      ],
-    ])
+    const authors = [
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/1d556016-ad16-4fe7-8724-42b3fb15ad28',
+          apiUrl:
+            'https://api.ft.com/people/1d556016-ad16-4fe7-8724-42b3fb15ad28',
+          prefLabel: 'Chris Giles',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+    ]
 
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
+    const byline = new Byline(bylineText, vanity, authors)
     const result = await byline.buildBylineTree()
     expect(result).toMatchSnapshot()
   })
 
   test('ignores extra authors in annotations array', async () => {
     const bylineText = 'Martin Wolf in London'
-    const authorUrlMapping = new Map([
-      [
-        'Chris Giles',
-        'https://www.ft.com/stream/uuid/1d556016-ad16-4fe7-8724-42b3fb15ad28',
-      ],
-      [
-        'Martin Wolf',
-        'https://www.ft.com/stream/uuid/7c1e1e72-57ae-4461-862a-f8d24dd42e22',
-      ],
-    ])
+    const authors = [
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/7c1e1e72-57ae-4461-862a-f8d24dd42e22',
+          apiUrl:
+            'https://api.ft.com/people/7c1e1e72-57ae-4461-862a-f8d24dd42e22',
+          prefLabel: 'Martin Wolf',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+    ]
 
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
+    const byline = new Byline(bylineText, vanity, authors)
     const result = await byline.buildBylineTree()
     expect(result).toMatchSnapshot()
   })
 
   test('handles bylines without any matching authors', async () => {
     const bylineText = 'Chris Giles and Nick Ramsbottom in London'
-    const authorUrlMapping: Map<string, string> = new Map([])
+    const authors: Concept[] = []
 
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
+    const byline = new Byline(bylineText, vanity, authors)
     const result = await byline.buildBylineTree()
-
     expect(result).toMatchSnapshot()
   })
 
   test('converts straight apostrophes in byline', async () => {
     const bylineText = "Sarah O'Connor in Bali"
-    const authorUrlMapping = new Map([
-      [
-        'Sarah O’Connor',
-        'https://www.ft.com/stream/uuid/1d556016-ad16-4fe7-8724-eeeeeeeeeeee',
-      ],
-    ])
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
-    const result = await byline.buildBylineTree()
+    const authors = [
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/1d556016-ad16-4fe7-8724-eeeeeeeeeeee',
+          apiUrl:
+            'https://api.ft.com/people/1d556016-ad16-4fe7-8724-eeeeeeeeeeee',
+          prefLabel: 'Sarah O’Connor',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+    ]
 
+    const byline = new Byline(bylineText, vanity, authors)
+    const result = await byline.buildBylineTree()
     expect(result).toMatchSnapshot()
   })
 
   test('leaves curly apostrophes in byline', async () => {
     const bylineText = 'Sarah O’Connor in Bali'
-    const authorUrlMapping = new Map([
-      [
-        'Sarah O’Connor',
-        'https://www.ft.com/stream/uuid/1d556016-ad16-4fe7-8724-eeeeeeeeeeee',
-      ],
-    ])
-    const byline = new Byline(bylineText, vanity, authorUrlMapping, context)
-    const result = await byline.buildBylineTree()
+    const authors = [
+      new Concept(
+        {
+          id: 'https://api.ft.com/things/1d556016-ad16-4fe7-8724-eeeeeeeeeeee',
+          apiUrl:
+            'https://api.ft.com/people/1d556016-ad16-4fe7-8724-eeeeeeeeeeee',
+          prefLabel: 'Sarah O’Connor',
+          directType: 'http://www.ft.com/ontology/person/Person',
+          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
+          types: [
+            'http://www.ft.com/ontology/core/Thing',
+            'http://www.ft.com/ontology/concept/Concept',
+            'http://www.ft.com/ontology/person/Person',
+          ],
+        },
+        context
+      ),
+    ]
 
+    const byline = new Byline(bylineText, vanity, authors)
+    const result = await byline.buildBylineTree()
     expect(result).toMatchSnapshot()
   })
 })

package/src/model/Byline.ts

@@ -1,58 +1,68 @@
 import { ContentTree } from '@financial-times/content-tree'
-import { AuthorLink } from '../resolvers/content-tree/Workarounds'
-import { QueryContext } from '..'
+import { Concept } from './Concept'
+import {
+  Byline as BylineNode,
+  Author as AuthorNode,
+} from '../resolvers/content-tree/Workarounds'
+import updateTreeWithReferenceIds from '../resolvers/content-tree/updateTreeWithReferenceIds'
+import { CapiResponse } from './CapiResponse'
 
 export class Byline {
+  static normaliseQuotes(str: string) {
+    return str.replaceAll("'", '’')
+  }
+
   constructor(
     private byline: string,
     private vanity: boolean,
-    private authorUrlMapping: Map<string, string>,
-    private context: QueryContext
+    private authors: Concept[],
+    private contentApiData?: CapiResponse
   ) {}
 
   async buildBylineTree() {
     // Normalise apostrophes in byline string
-    const bylineWithCorrectApostrophes = this.byline.replace("'", '’')
+    const normalisedByline = Byline.normaliseQuotes(this.byline)
+
+    const bylineParts = this.#splitBylineByName(
+      normalisedByline,
+      this.authors.map((author) => Byline.normaliseQuotes(author.prefLabel()))
+    )
+
+    const authorNameMapping = Object.fromEntries(
+      this.authors.map((author) => {
+        return [Byline.normaliseQuotes(author.prefLabel()), author] as const
+      })
+    )
+
+    const children = bylineParts.map((part): AuthorNode | ContentTree.Text => {
+      const author = authorNameMapping[part]
+      const text: ContentTree.Text = {
+        type: 'text',
+        value: part,
+      }
+
+      return author
+        ? {
+            type: 'author',
+            id: author.uuid(),
+            children: [text],
+          }
+        : text
+    })
 
-    const split = this.#splitBylineByName(bylineWithCorrectApostrophes, [
-      ...this.authorUrlMapping.keys(),
-    ])
-    return this.#buildUrlTree(this.authorUrlMapping, split)
+    return updateTreeWithReferenceIds(
+      {
+        type: 'byline',
+        children,
+      } satisfies BylineNode,
+      this.contentApiData
+    )
   }
 
   #splitBylineByName(byline: string, names: string[]) {
     if (!names.length) return [byline]
+
     const regex = new RegExp(`(${names.join('|')})`, 'ig')
     return byline.split(regex).filter((string) => string !== '')
   }
-
-  async #buildUrlTree(urlMapping: Map<string, string>, parts: string[]) {
-    const children: (AuthorLink | ContentTree.Text)[] = await Promise.all(
-      parts.map(async (part) => {
-        const fullUrl = urlMapping.get(part)
-        const vanityUrl =
-          this.vanity && fullUrl
-            ? await this.context.dataSources.vanityUrls.get(fullUrl)
-            : null
-
-        if (fullUrl) {
-          return {
-            type: 'author-link',
-            href: vanityUrl || fullUrl,
-            children: [{ type: 'text', value: part }],
-          }
-        } else {
-          return {
-            type: 'text',
-            value: part,
-          }
-        }
-      })
-    )
-
-    return {
-      tree: { type: 'root', children },
-      references: [],
-    }
-  }
 }

package/src/model/CapiResponse.test.ts

@@ -13,31 +13,6 @@ describe('CAPI response', () => {
     })
   })
 
-  describe('Author URL mapping', () => {
-    test('converts any straight quotes in the author name to curly quotes', () => {
-      const straighQuoteAuthorName = `Sinéad O'Connor`
-      const curlyQuoteAuthorName = `Sinéad O’Connor`
-
-      const article = cloneDeep(baseCapiObject)
-      delete article.annotations
-      article.annotations = [
-        {
-          id: 'http://api.ft.com/things/eeeeeeee-eeee-eeee-eeee-eeeeeeeeeeee',
-          predicate: 'http://www.ft.com/ontology/annotation/hasAuthor',
-          prefLabel: straighQuoteAuthorName,
-        },
-      ]
-      const capiResponse = new CapiResponse(article, context)
-
-      expect(
-        capiResponse.getAuthorUrlMapping().get(curlyQuoteAuthorName)
-      ).toBeDefined()
-      expect(
-        capiResponse.getAuthorUrlMapping().get(straighQuoteAuthorName)
-      ).toBeUndefined()
-    })
-  })
-
   describe('Content Type', () => {
     test(`uses first entry in types if there's no type`, () => {
       const article = cloneDeep(baseCapiObject)
@@ -84,6 +59,15 @@ describe('CAPI response', () => {
     })
   })
 
+  describe('Modified timestamp', () => {
+    test('generates a timestamp from the last modified date', () => {
+      const article = cloneDeep(baseCapiObject)
+      const capiResponse = new CapiResponse(article, context)
+
+      expect(capiResponse.modifiedTimestamp()).toEqual(1712061296789)
+    })
+  })
+
   describe('liveBlogPosts', () => {
     test('returns a resolved array of the articles contained sorted by most recently published', async () => {
       const liveBlogPackage = cloneDeep({