@financial-times/cp-content-pipeline-schema 2.3.0 → 2.3.1

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

@@ -1,8 +1,9 @@
-import type { ContentTree } from '@financial-times/content-tree'
 import * as cheerio from 'cheerio'
-import { isTag, isText } from 'domhandler'
 
+import type { ContentTree } from '@financial-times/content-tree'
 import { AnyNode } from './Workarounds'
+import { QueryContext } from '../..'
+import { isTag, isText } from 'domhandler'
 
 function isNode(node: ContentTree.Node | undefined): node is ContentTree.Node {
   return Boolean(node && 'type' in node)
@@ -10,7 +11,8 @@ function isNode(node: ContentTree.Node | undefined): node is ContentTree.Node {
 
 type ContentTreeTransform = (
   $el: cheerio.Cheerio<any>,
-  traverse: () => AnyNode[]
+  traverse: () => AnyNode[],
+  context?: QueryContext
 ) => AnyNode | AnyNode[]
 
 export type TagMappings = Record<string, ContentTreeTransform>
@@ -21,7 +23,8 @@ export type TagMappings = Record<string, ContentTreeTransform>
 */
 export default function bodyXMLToTree(
   xml: string,
-  tagMappings: TagMappings
+  tagMappings: TagMappings,
+  context?: QueryContext
 ): ContentTree.Body {
   const $ = cheerio.load(xml)
 
@@ -41,8 +44,10 @@ export default function bodyXMLToTree(
 
       if (matchedSelector) {
         const contentTreeTransform = tagMappings[matchedSelector]
-        return contentTreeTransform($(node), () =>
-          flattenAndTraverseChildren(node.children)
+        return contentTreeTransform(
+          $(node),
+          () => flattenAndTraverseChildren(node.children),
+          context
         )
       }
 

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

@@ -1,6 +1,6 @@
-import { parentPort } from 'worker_threads'
-
-import type { AnyNode } from './Workarounds'
+import { OperationalError } from '@dotcom-reliability-kit/errors'
+import { AnyNode } from './Workarounds'
+import { QueryContext } from '../..'
 
 type ValuesOfTuple<Tuple extends readonly string[]> = Tuple[number]
 
@@ -19,39 +19,24 @@ export const phrasingTypes = [
   'link',
 ] as const
 
-export interface PredicateError {
-  event: string
-  code: string
-  message: string
-  expected: AnyNode['type'] | readonly AnyNode['type'][]
-  actual: AnyNode['type'][]
-  error: Error
-}
-
-// reliability-kit's logger creates a worker thread under the hood but we can't
-// access this thread directly from these worker threads. Instead let's send
-// all our recoverable errors to the piscina pool managing us.
-const postLoggerError = (error: Omit<PredicateError, 'error'>) => {
-  // stacks will only copy over if they're in an Error object
-  const errorWithStack = { ...error, error: new Error() }
-  parentPort?.postMessage(errorWithStack)
-}
-
 export const findChildOftype = <NodeType extends AnyNode>(
   type: NodeType['type'],
   nodes: AnyNode[],
-  parentType: AnyNode['type']
+  parentType: AnyNode['type'],
+  context?: QueryContext
 ): NodeType | undefined => {
   const predicate = (node: AnyNode): node is NodeType => node.type === type
   const child = nodes.find(predicate)
 
   if (!child) {
-    postLoggerError({
+    context?.logger.error({
       event: 'RECOVERABLE_ERROR',
-      code: 'BODY_XML_UNEXPECTED_STRUCTURE',
-      message: `Didn't find expected child type in ${parentType}`,
-      expected: type,
-      actual: nodes.map((node) => node.type),
+      error: new OperationalError({
+        code: 'BODY_XML_UNEXPECTED_STRUCTURE',
+        message: `Didn't find expected child type in ${parentType}`,
+        expected: type,
+        actual: nodes.map((node) => node.type),
+      }),
     })
   }
 
@@ -61,18 +46,21 @@ export const findChildOftype = <NodeType extends AnyNode>(
 export const everyChildIsType = <NodeType extends AnyNode>(
   type: NodeType['type'],
   nodes: AnyNode[],
-  parentType: AnyNode['type']
+  parentType: AnyNode['type'],
+  context?: QueryContext
 ): NodeType[] => {
   const predicate = (node: AnyNode): node is NodeType => node.type === type
   const allChildrenAreType = nodes.every(predicate)
 
   if (!allChildrenAreType) {
-    postLoggerError({
+    context?.logger.error({
       event: 'RECOVERABLE_ERROR',
-      code: 'BODY_XML_UNEXPECTED_STRUCTURE',
-      message: `Unexpected children types for ${parentType}`,
-      expected: type,
-      actual: nodes.map((node) => node.type),
+      error: new OperationalError({
+        code: 'BODY_XML_UNEXPECTED_STRUCTURE',
+        message: `Unexpected children types for ${parentType}`,
+        expected: type,
+        actual: nodes.map((node) => node.type),
+      }),
     })
   }
 
@@ -82,19 +70,22 @@ export const everyChildIsType = <NodeType extends AnyNode>(
 export const childrenOfTypes = <Types extends readonly AnyNode['type'][]>(
   types: Types,
   nodes: AnyNode[],
-  parentType: AnyNode['type']
+  parentType: AnyNode['type'],
+  context?: QueryContext
 ): NodeOfType<ValuesOfTuple<Types>>[] => {
   const predicate = (node: AnyNode): node is NodeOfType<ValuesOfTuple<Types>> =>
     types.includes(node.type)
   const allChildrenAreType = nodes.every(predicate)
 
   if (!allChildrenAreType) {
-    postLoggerError({
+    context?.logger.error({
       event: 'RECOVERABLE_ERROR',
-      code: 'BODY_XML_UNEXPECTED_STRUCTURE',
-      message: `Unexpected ordered children types for ${parentType}`,
-      expected: types,
-      actual: nodes.map((node) => node.type),
+      error: new OperationalError({
+        code: 'BODY_XML_UNEXPECTED_STRUCTURE',
+        message: `Unexpected ordered children types for ${parentType}`,
+        expected: types,
+        actual: nodes.map((node) => node.type),
+      }),
     })
   }
 
@@ -119,18 +110,21 @@ export const childrenOfOrderedTypes = <
 >(
   types: Types,
   nodes: AnyNode[],
-  parentType: AnyNode['type']
+  parentType: AnyNode['type'],
+  context?: QueryContext
 ): NodesOfTypes<Types> => {
   if (nodesAreOrderedTypes(types, nodes)) {
     return nodes
   }
 
-  postLoggerError({
+  context?.logger.error({
     event: 'RECOVERABLE_ERROR',
-    code: 'BODY_XML_UNEXPECTED_STRUCTURE',
-    message: `Unexpected children types for ${parentType}`,
-    expected: types,
-    actual: nodes.map((node) => node.type),
+    error: new OperationalError({
+      code: 'BODY_XML_UNEXPECTED_STRUCTURE',
+      message: `Unexpected children types for ${parentType}`,
+      expected: types,
+      actual: nodes.map((node) => node.type),
+    }),
   })
 
   return nodes as unknown as NodesOfTypes<Types>

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

@@ -19,6 +19,7 @@ import {
 
 import * as scrollytelling from '@financial-times/n-scrollytelling-image/server'
 import { LiteralToPrimitiveDeep } from 'type-fest'
+import { CapiResponse } from '../../model/CapiResponse'
 
 const tableResponsiveStyleMap: Record<string, Table['responsiveStyle']> = {
   stacked: 'flat',
@@ -81,10 +82,10 @@ const validScrollytellingOption = <
     : false
 }
 
-const articleTagMappings = (topperHasImage: boolean): TagMappings => ({
+const articleTagMappings = (capiData?: CapiResponse): TagMappings => ({
   'body > ft-content[type="http://www.ft.com/ontology/content/ImageSet"]:first-child':
     ($el) => ({
-      type: topperHasImage ? 'image-set' : 'main-image',
+      type: capiData?.topperHasImage() ? 'image-set' : 'main-image',
       id: $el.attr('url') || '',
     }),
   'body > ft-content[type="http://www.ft.com/ontology/content/ImageSet"]:not(:first-child),:not(body) > ft-content[type="http://www.ft.com/ontology/content/ImageSet"]':
@@ -103,55 +104,60 @@ const liveBlogPostTagMappings: TagMappings = {
 
 const commonTagMappings: TagMappings = {
   body: ($el, traverse) => ({ type: 'body', version: 1, children: traverse() }),
-  'a:not([data-asset-type])': ($el, traverse) => ({
+  'a:not([data-asset-type])': ($el, traverse, context) => ({
     type: 'link',
     url: $el.attr('href') || '',
     title: '',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'link'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'link', context),
   }),
-  strong: ($el, traverse) => ({
+  strong: ($el, traverse, context) => ({
     type: 'strong',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'strong'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'strong', context),
   }),
-  em: ($el, traverse) => ({
+  em: ($el, traverse, context) => ({
     type: 'emphasis',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'emphasis'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'emphasis', context),
   }),
-  blockquote: ($el, traverse) => ({
+  blockquote: ($el, traverse, context) => ({
     type: 'blockquote',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'blockquote'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'blockquote', context),
   }),
   // strip any (redundant) line breaks in between two paragraphs
   'p + br': ($el) => ($el.next()[0]?.tagName === 'p' ? [] : { type: 'break' }),
   br: () => ({ type: 'break' }),
   hr: () => ({ type: 'thematic-break' }),
-  s: ($el, traverse) => ({
+  s: ($el, traverse, context) => ({
     type: 'strikethrough',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'strikethrough'),
+    children: childrenOfTypes(
+      phrasingTypes,
+      traverse(),
+      'strikethrough',
+      context
+    ),
   }),
-  ul: ($el, traverse) => ({
+  ul: ($el, traverse, context) => ({
     type: 'list',
-    children: everyChildIsType('list-item', traverse(), 'list'),
+    children: everyChildIsType('list-item', traverse(), 'list', context),
     ordered: false,
   }),
-  ol: ($el, traverse) => ({
+  ol: ($el, traverse, context) => ({
     type: 'list',
-    children: everyChildIsType('list-item', traverse(), 'list'),
+    children: everyChildIsType('list-item', traverse(), 'list', context),
     ordered: true,
   }),
-  li: ($el, traverse) => ({
+  li: ($el, traverse, context) => ({
     type: 'list-item',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'list-item'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'list-item', context),
   }),
   'a[data-asset-type="tweet"]': ($el) => ({
     type: 'tweet',
     id: $el.attr('href') || '',
   }),
 
-  'h4,h5,h6': ($el, traverse) => ({
+  'h4,h5,h6': ($el, traverse, context) => ({
     type: 'heading',
     level: 'label',
-    children: everyChildIsType('text', traverse(), 'heading'),
+    children: everyChildIsType('text', traverse(), 'heading', context),
   }),
   'pull-quote': ($el) => ({
     type: 'pullquote',
@@ -230,7 +236,7 @@ const commonTagMappings: TagMappings = {
       description: $description.text(),
     }
   },
-  '.n-content-layout': ($el, traverse) => {
+  '.n-content-layout': ($el, traverse, context) => {
     //TODO: this is a bit gross??
     const isValidWidth = (
       str: string
@@ -254,14 +260,16 @@ const commonTagMappings: TagMappings = {
           ...everyChildIsType<ContentTree.LayoutSlot>(
             'layout-slot',
             otherChildren,
-            'layout'
+            'layout',
+            context
           ),
         ]
       } else {
         return everyChildIsType<ContentTree.LayoutSlot>(
           'layout-slot',
           children,
-          'layout'
+          'layout',
+          context
         )
       }
     }
@@ -273,7 +281,7 @@ const commonTagMappings: TagMappings = {
       children: getChildren(traverse()),
     }
   },
-  '.n-content-layout__slot': ($el, traverse) => ({
+  '.n-content-layout__slot': ($el, traverse, context) => ({
     type: 'layout-slot',
     children: childrenOfTypes(
       ['heading', 'paragraph', 'layout-image'],
@@ -283,7 +291,8 @@ const commonTagMappings: TagMappings = {
         .flatMap((node) =>
           node.type === 'layout-slot' ? node.children : [node]
         ),
-      'layout-slot'
+      'layout-slot',
+      context
     ),
   }),
   '.n-content-layout img': ($el) => ({
@@ -294,7 +303,7 @@ const commonTagMappings: TagMappings = {
     credit: $el.attr('data-copyright') || '',
   }),
 
-  table: ($el, traverse): Table | ContentTree.Layout => {
+  table: ($el, traverse, context): Table | ContentTree.Layout => {
     const layoutSmallscreen = $el.attr('data-table-layout-smallscreen')
     const responsiveStyle: Table['responsiveStyle'] =
       layoutSmallscreen && layoutSmallscreen in tableResponsiveStyleMap
@@ -309,12 +318,14 @@ const commonTagMappings: TagMappings = {
     const caption = findChildOftype<TableCaption>(
       'table-caption',
       children,
-      'table'
+      'table',
+      context
     )
     const footer = findChildOftype<TableFooter>(
       'table-footer',
       children,
-      'table'
+      'table',
+      context
     )
 
     // HACK:KB:20230523 transform single-cell tables into infoboxes, for legacy articles
@@ -328,7 +339,8 @@ const commonTagMappings: TagMappings = {
         children: childrenOfTypes(
           ['paragraph', 'heading'],
           bodies[0].children[0].children[0].children,
-          'layout-slot'
+          'layout-slot',
+          context
         ),
       }
 
@@ -337,7 +349,12 @@ const commonTagMappings: TagMappings = {
             {
               type: 'heading',
               level: 'subheading',
-              children: everyChildIsType('text', caption.children, 'heading'),
+              children: everyChildIsType(
+                'text',
+                caption.children,
+                'heading',
+                context
+              ),
             },
             slot,
           ]
@@ -417,34 +434,44 @@ const commonTagMappings: TagMappings = {
   // HACK needs to come before tr. TODO sort by selector specificity
   'tfoot > tr': ($el, traverse) => traverse(),
   // HACK needs to come before td. TODO sort by selector specificity
-  'tfoot > tr > td': ($el, traverse) => ({
+  'tfoot > tr > td': ($el, traverse, context) => ({
     type: 'table-footer',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'table-footer'),
+    children: childrenOfTypes(
+      phrasingTypes,
+      traverse(),
+      'table-footer',
+      context
+    ),
   }),
-  tr: ($el, traverse) => ({
+  tr: ($el, traverse, context) => ({
     type: 'table-row',
-    children: everyChildIsType('table-cell', traverse(), 'table-row'),
+    children: everyChildIsType('table-cell', traverse(), 'table-row', context),
   }),
-  td: ($el, traverse) => ({
+  td: ($el, traverse, context) => ({
     type: 'table-cell',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'table-cell'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'table-cell', context),
   }),
-  th: ($el, traverse) => ({
+  th: ($el, traverse, context) => ({
     type: 'table-cell',
     heading: true,
-    children: childrenOfTypes(phrasingTypes, traverse(), 'table-cell'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'table-cell', context),
   }),
-  thead: ($el, traverse) => ({
+  thead: ($el, traverse, context) => ({
     type: 'table-body',
-    children: everyChildIsType('table-row', traverse(), 'table-body'),
+    children: everyChildIsType('table-row', traverse(), 'table-body', context),
   }),
-  tbody: ($el, traverse) => ({
+  tbody: ($el, traverse, context) => ({
     type: 'table-body',
-    children: everyChildIsType('table-row', traverse(), 'table-body'),
+    children: everyChildIsType('table-row', traverse(), 'table-body', context),
   }),
-  caption: ($el, traverse) => ({
+  caption: ($el, traverse, context) => ({
     type: 'table-caption',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'table-caption'),
+    children: childrenOfTypes(
+      phrasingTypes,
+      traverse(),
+      'table-caption',
+      context
+    ),
   }),
 
   // HACK needs to come after LayoutImage. TODO sort by selector specificity
@@ -468,14 +495,15 @@ const commonTagMappings: TagMappings = {
     format: 'standard',
   }),
 
-  'scrollable-block': ($el, traverse) => {
+  'scrollable-block': ($el, traverse, context) => {
     return {
       type: 'scrolly-block',
       theme: $el.attr('theme') === '2' ? 'serif' : 'sans',
       children: everyChildIsType(
         'scrolly-section',
         traverse(),
-        'scrolly-block'
+        'scrolly-block',
+        context
       ),
     }
   },
@@ -486,7 +514,7 @@ const commonTagMappings: TagMappings = {
       id: $el.attr('url') ?? '',
     }),
 
-  'scrollable-section': ($el, traverse) => {
+  'scrollable-section': ($el, traverse, context) => {
     const [firstChild, ...restChildren] = traverse()
 
     const display = $el.attr('theme-display')
@@ -510,29 +538,33 @@ const commonTagMappings: TagMappings = {
         ...everyChildIsType<ContentTree.ScrollyImage>(
           'scrolly-image',
           [firstChild],
-          'scrolly-section'
+          'scrolly-section',
+          context
         ),
         ...everyChildIsType<ContentTree.ScrollyCopy>(
           'scrolly-copy',
           restChildren,
-          'scrolly-section'
+          'scrolly-section',
+          context
         ),
       ] as [ContentTree.ScrollyImage, ...ContentTree.ScrollyCopy[]],
     }
   },
 
-  'scrollable-text': ($el, traverse) => ({
+  'scrollable-text': ($el, traverse, context) => ({
     type: 'scrolly-copy',
     children: childrenOfTypes(
       ['scrolly-heading', 'paragraph'],
       traverse(),
-      'scrolly-copy'
+      'scrolly-copy',
+      context
     ),
   }),
 
   'scrollable-text > p, scrollable-text > h1, scrollable-text > h2': (
     $el,
-    traverse
+    traverse,
+    context
   ) => {
     const textStyle = $el.attr('theme-style')
 
@@ -540,31 +572,41 @@ const commonTagMappings: TagMappings = {
       return {
         type: 'scrolly-heading',
         level: scrollytellingOptionsToContentTree.textStyle[textStyle],
-        children: everyChildIsType('text', traverse(), 'scrolly-heading'),
+        children: everyChildIsType(
+          'text',
+          traverse(),
+          'scrolly-heading',
+          context
+        ),
       }
     }
 
     return {
       type: 'paragraph',
-      children: childrenOfTypes(phrasingTypes, traverse(), 'paragraph'),
+      children: childrenOfTypes(
+        phrasingTypes,
+        traverse(),
+        'paragraph',
+        context
+      ),
     }
   },
 
   // HACK:KB:20230703 needs to come after `scrollable-text > X`. todo sort by specificity
-  p: ($el, traverse) => ({
+  p: ($el, traverse, context) => ({
     type: 'paragraph',
-    children: childrenOfTypes(phrasingTypes, traverse(), 'paragraph'),
+    children: childrenOfTypes(phrasingTypes, traverse(), 'paragraph', context),
   }),
 
-  h1: ($el, traverse) => ({
+  h1: ($el, traverse, context) => ({
     type: 'heading',
     level: 'chapter',
-    children: everyChildIsType('text', traverse(), 'heading'),
+    children: everyChildIsType('text', traverse(), 'heading', context),
   }),
-  'h2,h3': ($el, traverse) => ({
+  'h2,h3': ($el, traverse, context) => ({
     type: 'heading',
     level: 'subheading',
-    children: everyChildIsType('text', traverse(), 'heading'),
+    children: everyChildIsType('text', traverse(), 'heading', context),
   }),
   'ft-content[type="http://www.ft.com/ontology/content/ClipSet"]': ($el) => ({
     type: 'clip-set',