@financial-times/cp-content-pipeline-schema 2.4.1 → 2.5.0

package/src/model/CapiResponse.ts

@@ -30,6 +30,15 @@ import {
 
 import { ContentPackageContainsArgs, Media } from '../generated'
 import { Topper } from './Topper'
+import { z } from 'zod'
+
+function isPlainObject(object: unknown): object is Record<string, unknown> {
+  if (object && typeof object === 'object') {
+    return true
+  }
+
+  return false
+}
 
 type Design = {
   theme: LiteralUnionScalarValues<typeof PackageDesign>
@@ -56,7 +65,7 @@ type ZodInputObject = {
 
 function flattenFormattedZodIssues(
   issues: FormattedZodIssues,
-  object: ZodInputValue,
+  object: unknown,
   path: string[] = [],
   flattened: string[] = []
 ): string[] {
@@ -94,7 +103,7 @@ function flattenFormattedZodIssues(
       key !== '_errors' &&
       !Array.isArray(value) &&
       object &&
-      (Array.isArray(object) || typeof object === 'object')
+      (Array.isArray(object) || isPlainObject(object))
     ) {
       const isArrayKey = !Number.isNaN(parseInt(key))
       flattenFormattedZodIssues(
@@ -109,6 +118,36 @@ function flattenFormattedZodIssues(
   return flattened
 }
 
+function getContentType(
+  content: { type?: string; types: string[] },
+  validate?: true
+): LiteralUnionScalarValues<typeof ContentType>
+function getContentType(
+  content: { type?: string; types: string[] },
+  validate: false
+): LiteralUnionScalarValues<typeof ContentType> | string
+function getContentType(
+  content: { type?: string; types: string[] },
+  validate = true
+) {
+  const TYPE_REGEX = /https?:\/\/www.ft.com\/ontology\/content\//
+  const useType =
+    'type' in content && content.type ? content.type : content.types[0]
+  const type = useType.replace(TYPE_REGEX, '')
+
+  if (validate && !validLiteralUnionValue(type, ContentType.values)) {
+    throw new Error('Content type is invalid: ' + useType)
+  }
+
+  return type
+}
+
+const baselineContentSchema = z.object({
+  id: z.string(),
+  type: z.string().optional(),
+  types: z.array(z.string()),
+})
+
 export class CapiResponse {
   constructor(
     private capiData: ContentTypeSchemas,
@@ -117,11 +156,15 @@ export class CapiResponse {
   ) {}
 
   static fromJSON(
-    content: any,
+    content: unknown,
     context: QueryContext,
     packageContainer?: CapiResponse
   ): CapiResponse {
-    const model = new CapiResponse(content, context, packageContainer)
+    // if the content doesn't meet the baseline schema, something's gone
+    // unexpectedly very wrong. throwing an non-operational error is the
+    // right thing to do in that case, so we call .parse not .safeParse
+    const baseContent = baselineContentSchema.parse(content)
+    const type = getContentType(baseContent)
 
     /*
      * Check if the incoming data matches the types defined in our data source schema
@@ -129,7 +172,8 @@ export class CapiResponse {
      * As there is no agreed schema provided by CAPI there is a chance it will be out of date / incorrect at times
      * Manual updates will be required at times to keep it in sync with the responses we receive
      */
-    const schemaResponse = model.schema().safeParse(content)
+    const schemaResponse = schemas(type).safeParse(content)
+
     /*
      * Log an error if the response fails validation
      * Logs should be used to warn us of issues or response changes
@@ -146,32 +190,31 @@ export class CapiResponse {
             schemaResponse.error.format(),
             content
           ),
-          contentId: content.id,
-          contentType: model.type(false),
+          contentId: baseContent.id,
+          contentType: type,
         }),
       })
 
       context.metrics?.count(
-        `graphql.datasource.CapiDataSource.${model.type(
-          false
-        )}.validation.failure.count`,
+        `graphql.datasource.CapiDataSource.${type}.validation.failure.count`,
         1
       )
     } else {
       context.metrics?.count(
-        `graphql.datasource.CapiDataSource.${model.type(
-          false
-        )}.validation.success.count`,
+        `graphql.datasource.CapiDataSource.${type}.validation.success.count`,
         1
       )
     }
 
-    return model
+    // we can "safely" cast content to the schema return types here. if the type is
+    // incorrect, we've at least logged that error.
+    return new CapiResponse(
+      content as ContentTypeSchemas,
+      context,
+      packageContainer
+    )
   }
 
-  schema() {
-    return schemas(this.type())
-  }
   bodyXML(): string | null {
     if ('bodyXML' in this.capiData) {
       return this.capiData.bodyXML
@@ -281,7 +324,10 @@ export class CapiResponse {
     return this.capiData.title
   }
   id() {
-    return uuidFromUrl(this.capiData.id)
+    const uuid = uuidFromUrl(this.capiData.id)
+    this.context.addSurrogateKeys([uuid])
+
+    return uuid
   }
   standfirst(): string | null {
     if ('standfirst' in this.capiData && this.capiData.standfirst) {
@@ -410,18 +456,11 @@ export class CapiResponse {
   type(validate?: true): LiteralUnionScalarValues<typeof ContentType>
   type(validate: false): LiteralUnionScalarValues<typeof ContentType> | string
   type(validate = true) {
-    const TYPE_REGEX = /https?:\/\/www.ft.com\/ontology\/content\//
-    const useType =
-      'type' in this.capiData && this.capiData.type
-        ? this.capiData.type
-        : this.capiData.types[0]
-    const type = useType.replace(TYPE_REGEX, '')
-
-    if (validate && !validLiteralUnionValue(type, ContentType.values)) {
-      throw new Error('Content type is invalid: ' + useType)
+    if (validate) {
+      return getContentType(this.capiData, true)
+    } else {
+      return getContentType(this.capiData, false)
     }
-
-    return type
   }
 
   topperHasImage() {

package/src/resolvers/content-tree/bodyXMLToTree.ts

@@ -10,7 +10,7 @@ function isNode(node: ContentTree.Node | undefined): node is ContentTree.Node {
 }
 
 type ContentTreeTransform = (
-  $el: cheerio.Cheerio<any>,
+  $el: cheerio.Cheerio<cheerio.Element>,
   traverse: () => AnyNode[],
   context?: QueryContext
 ) => AnyNode | AnyNode[]

package/src/resolvers/content-tree/tagMappings.test.ts

@@ -1,5 +1,19 @@
+import { AnyNode, OldClip } from './Workarounds'
 import tagMappings, { getBooleanAttributeValue } from './tagMappings'
 import cheerio from 'cheerio'
+
+function expectNotArray<T>(thing: T | T[]): asserts thing is T {
+  expect(thing).not.toBeInstanceOf(Array)
+}
+
+function expectNodeType<T extends AnyNode>(
+  node: AnyNode | AnyNode[],
+  type: T['type']
+): asserts node is T {
+  expectNotArray(node)
+  expect(node.type).toBe(type)
+}
+
 describe('tagMappings test', () => {
   it('getBooleanAttributeValue attrubute', () => {
     let $el = cheerio.load('<ft-content autoplay></ft-content>')('ft-content')
@@ -20,7 +34,7 @@ describe('tagMappings test', () => {
     const bodyXML = `
     <body>
       <experimental>
-        <ft-content autoplay=\"false\" caption=\"caption data\" data-asset-type=\"clip\" data-copyright=\"Reuters\" 
+        <ft-content autoplay=\"false\" caption=\"caption data\" data-asset-type=\"clip\" data-copyright=\"Reuters\"
         data-layout=\"in-line\" description=\"description\" href=\"https://storytelling-clips.s3.amazonaws.com/World/Joe%20Biden%20on%20debt%20ceiling%20deal%20v1.mp4\"
         loop=\"true\" muted=\"false\" poster=\"https://d1e00ek4ebabms.cloudfront.net/production/65233b7d-5260-4acc-b01c-38ff470068e0.jpg\"
         poster-alt=\"\" poster-copyright=\"Reuters\" poster-id=\"db0a46c6-9e60-4f46-bc0b-c158cb8394d0\" type=\"http://www.ft.com/ontology/content/clip\">
@@ -30,8 +44,9 @@ describe('tagMappings test', () => {
     const selector =
       'ft-content[type="http://www.ft.com/ontology/content/clip"]'
     const $el = cheerio.load(bodyXML)(selector)
-    const mapping: any = tagMappings[selector]($el, () => [])
-    expect(mapping.type).toBe('clip')
+    const mapping = tagMappings[selector]($el, () => [])
+
+    expectNodeType<OldClip>(mapping, 'clip')
     expect(mapping.url).toBe(
       'https://storytelling-clips.s3.amazonaws.com/World/Joe%20Biden%20on%20debt%20ceiling%20deal%20v1.mp4'
     )

package/src/resolvers/content-tree/tagMappings.ts

@@ -20,6 +20,7 @@ import {
 import * as scrollytelling from '@financial-times/n-scrollytelling-image/server'
 import { LiteralToPrimitiveDeep } from 'type-fest'
 import { CapiResponse } from '../../model/CapiResponse'
+import type { Cheerio, Element } from 'cheerio'
 
 const tableResponsiveStyleMap: Record<string, Table['responsiveStyle']> = {
   stacked: 'flat',
@@ -623,8 +624,11 @@ const commonTagMappings: TagMappings = {
 // We can't trust cheerio.attr() method since with attributes like "autoplay"
 //doesn't return the value instead return "autoplay" and "autoplay" = "false"
 //should be interpreted as false
-export const getBooleanAttributeValue = ($el: any, attribute: string) => {
-  const value = $el.get(0).attribs[attribute]
+export const getBooleanAttributeValue = (
+  $el: Cheerio<Element>,
+  attribute: string
+) => {
+  const value = $el.get(0)?.attribs[attribute]
   return [null, undefined, 'false', false].includes(value) ? false : true
 }
 

package/src/types/cache.ts

@@ -11,7 +11,11 @@ interface ContextableCache<TContext extends QueryContext> {
 }
 
 export function isContextableCache<TContext extends QueryContext>(
-  cache: any
+  cache: unknown
 ): cache is ContextableCache<TContext> {
-  return 'withContext' in cache
+  if (cache && typeof cache === 'object' && 'withContext' in cache) {
+    return true
+  }
+
+  return false
 }