Merge pull request #361 from cofacts/choosing-media-articles

refactor: combine processMedia and processImage

src/graphql/cofacts-api.graphql

@@ -213,6 +213,9 @@ type Article implements Node {
   Automated reply from AI before human fact checkers compose an fact check
   """
   aiReplies: [AIReply!]!
+
+  """Automated transcript"""
+  aiTranscripts: [AITranscript!]!
   articleCategories(
     """
     Deprecated. Please use statuses instead. When specified, returns only article categories with the specified status
@@ -684,6 +687,9 @@ interface AIResponse implements Node {
 enum AIResponseTypeEnum {
   """The AI Response is an automated analysis / reply of an article."""
   AI_REPLY
+
+  """AI transcribed text of the specified article."""
+  TRANSCRIPT
 }
 
 enum AIResponseStatusEnum {
@@ -698,6 +704,30 @@ type OpenAICompletionUsage {
   totalTokens: Int!
 }
 
+"""
+Transcript from OCR or speech-to-text AI models for the specified MediaEntry ID as docId.
+"""
+type AITranscript implements Node & AIResponse {
+  id: ID!
+
+  """The id for the document that this AI response is for."""
+  docId: ID
+
+  """AI response type"""
+  type: AIResponseTypeEnum!
+
+  """The user triggered this AI response"""
+  user: User
+
+  """Processing status of AI"""
+  status: AIResponseStatusEnum!
+
+  """AI response text. Populated after status becomes SUCCESS."""
+  text: String
+  createdAt: String!
+  updatedAt: String
+}
+
 """The linkage between an Article and a Category"""
 type ArticleCategory implements Node {
   id: ID!
@@ -841,6 +871,12 @@ type ArticleConnectionEdge implements Edge {
   cursor: String!
   score: Float
   highlight: Highlights
+
+  """
+  The search hit's similarity with provided mediaUrl.
+            Ranges from 0 to 1. 0 if mediaUrl is not provided, or the hit is not matched by mediaUrl.
+  """
+  mediaSimilarity: Float!
 }
 
 type ArticleConnectionPageInfo implements PageInfo {

src/webhook/handlers/__fixtures__/processMedia.ts

@@ -0,0 +1,124 @@
+import { ListArticlesInProcessMediaQuery } from 'typegen/graphql';
+
+export const oneImageArticle: GqlResponse<ListArticlesInProcessMediaQuery> = {
+  data: {
+    ListArticles: {
+      edges: [
+        {
+          score: 87,
+          node: {
+            articleType: 'IMAGE',
+            attachmentUrl: 'http://foo/image.jpeg',
+            id: 'image-article-1',
+          },
+          mediaSimilarity: 0.87,
+          highlight: null,
+        },
+      ],
+    },
+  },
+};
+
+export const oneIdenticalImageArticle: GqlResponse<ListArticlesInProcessMediaQuery> =
+  {
+    data: {
+      ListArticles: {
+        edges: [
+          {
+            score: 100,
+            node: {
+              articleType: 'IMAGE',
+              attachmentUrl: 'http://foo/image.jpeg',
+              id: 'image-article-1',
+            },
+            mediaSimilarity: 1,
+            highlight: null,
+          },
+        ],
+      },
+    },
+  };
+
+export const identicalImageAndTextFound: GqlResponse<ListArticlesInProcessMediaQuery> =
+  {
+    data: {
+      ListArticles: {
+        edges: [
+          {
+            score: 100,
+            mediaSimilarity: 1,
+            node: {
+              articleType: 'IMAGE',
+              attachmentUrl: 'http://foo/image.jpeg',
+              id: 'image-article-1',
+            },
+            highlight: null,
+          },
+          {
+            score: 87.6656,
+            mediaSimilarity: 0,
+            node: {
+              articleType: 'TEXT',
+              attachmentUrl: null,
+              id: 'text-article-1',
+            },
+            highlight: {
+              text: 'Text matching <HIGHLIGHT>query transcript</HIGHLIGHT>',
+              hyperlinks: [
+                {
+                  title:
+                    'Hyperlink title matching <HIGHLIGHT>query transcript</HIGHLIGHT>',
+                  summary:
+                    'Hyperlink summary matching <HIGHLIGHT>query transcript</HIGHLIGHT>',
+                },
+              ],
+            },
+          },
+        ],
+      },
+    },
+  };
+
+export const twelveImageArticles: GqlResponse<ListArticlesInProcessMediaQuery> =
+  {
+    data: {
+      ListArticles: {
+        edges: Array.from(Array(6)).reduce<
+          NonNullable<ListArticlesInProcessMediaQuery['ListArticles']>['edges']
+        >(
+          (arr) =>
+            arr.concat([
+              {
+                score: 100,
+                mediaSimilarity: 1,
+                node: {
+                  articleType: 'IMAGE',
+                  attachmentUrl: 'http://foo/image.jpeg',
+                  id: 'image-article-1',
+                },
+                highlight: null,
+              },
+              {
+                score: 87.6656,
+                mediaSimilarity: 0.87,
+                node: {
+                  articleType: 'IMAGE',
+                  attachmentUrl: 'http://foo/image2.jpeg',
+                  id: 'image-article-2',
+                },
+                highlight: null,
+              },
+            ]),
+          []
+        ),
+      },
+    },
+  };
+
+export const notFound: GqlResponse<ListArticlesInProcessMediaQuery> = {
+  data: {
+    ListArticles: {
+      edges: [],
+    },
+  },
+};