refactor(webhook): extract text and media search

We will reuse them when proessing batch
cofacts · Dec 27, 2023 · 57e80ff · 57e80ff
1 parent bbbfee8
commit 57e80ff
Show file tree

Hide file tree

Showing 3 changed files with 92 additions and 78 deletions.
diff --git a/src/webhook/handlers/initState.ts b/src/webhook/handlers/initState.ts
@@ -12,22 +12,18 @@ import type {
   Context,
   CooccurredMessage,
 } from 'src/types/chatbotState';
-import gql from 'src/lib/gql';
 import {
   createPostbackAction,
   ellipsis,
   POSTBACK_NO_ARTICLE_FOUND,
   createHighlightContents,
   createTextMessage,
   createArticleSourceReply,
+  searchText,
 } from './utils';
 import ga from 'src/lib/ga';
 import detectDialogflowIntent from 'src/lib/detectDialogflowIntent';
 import choosingArticle from './choosingArticle';
-import {
-  ListArticlesInInitStateQuery,
-  ListArticlesInInitStateQueryVariables,
-} from 'typegen/graphql';
 
 const SIMILARITY_THRESHOLD = 0.95;
 
@@ -81,43 +77,16 @@ const initState = async ({
   }
 
   // Search for articles
-  const {
-    data: { ListArticles },
-  } = await gql`
-    query ListArticlesInInitState($text: String!) {
-      ListArticles(
-        filter: { moreLikeThis: { like: $text } }
-        orderBy: [{ _score: DESC }]
-        first: 4
-      ) {
-        edges {
-          node {
-            text
-            id
-            articleType
-          }
-          highlight {
-            text
-            hyperlinks {
-              title
-              summary
-            }
-          }
-        }
-      }
-    }
-  `<ListArticlesInInitStateQuery, ListArticlesInInitStateQueryVariables>({
-    text: input,
-  });
+  const result = await searchText(input);
 
   const inputSummary = ellipsis(input, 12);
 
-  if (ListArticles?.edges.length) {
+  if (result?.edges.length) {
     // Track if find similar Articles in DB.
     visitor.event({ ec: 'UserInput', ea: 'ArticleSearch', el: 'ArticleFound' });
 
     // Track which Article is searched. And set tracking event as non-interactionHit.
-    ListArticles.edges.forEach((edge) => {
+    result.edges.forEach((edge) => {
       visitor.event({
         ec: 'Article',
         ea: 'Search',
@@ -126,7 +95,7 @@ const initState = async ({
       });
     });
 
-    const edgesSortedWithSimilarity = ListArticles.edges
+    const edgesSortedWithSimilarity = result.edges
       .map((edge) => ({
         ...edge,
         similarity: stringSimilarity.compareTwoStrings(

diff --git a/src/webhook/handlers/processMedia.ts b/src/webhook/handlers/processMedia.ts
@@ -14,6 +14,7 @@ import {
   createTextMessage,
   createAskArticleSubmissionConsentReply,
   createHighlightContents,
+  searchMedia,
 } from './utils';
 import gql from 'src/lib/gql';
 import ga from 'src/lib/ga';
@@ -44,50 +45,14 @@ export default async function (message: CooccurredMessage, userId: string) {
     msgs: [message],
   };
 
-  const {
-    data: { ListArticles },
-  } = await gql`
-    query ListArticlesInProcessMedia($mediaUrl: String!) {
-      ListArticles(
-        filter: {
-          mediaUrl: $mediaUrl
-          articleTypes: [TEXT, IMAGE, AUDIO, VIDEO]
-          transcript: { shouldCreate: true }
-        }
-        orderBy: [{ _score: DESC }]
-        first: 9
-      ) {
-        edges {
-          score
-          mediaSimilarity
-          node {
-            id
-            articleType
-            attachmentUrl(variant: THUMBNAIL)
-          }
-          highlight {
-            text
-            hyperlinks {
-              title
-              summary
-            }
-          }
-        }
-      }
-    }
-  `<ListArticlesInProcessMediaQuery, ListArticlesInProcessMediaQueryVariables>(
-    {
-      mediaUrl: proxyUrl,
-    },
-    { userId }
-  );
+  const result = await searchMedia(proxyUrl, userId);
 
-  if (ListArticles && ListArticles.edges.length) {
+  if (result && result.edges.length) {
     // Track if find similar Articles in DB.
     visitor.event({ ec: 'UserInput', ea: 'ArticleSearch', el: 'ArticleFound' });
 
     // Track which Article is searched. And set tracking event as non-interactionHit.
-    ListArticles.edges.forEach((edge) => {
+    result.edges.forEach((edge) => {
       visitor.event({
         ec: 'Article',
         ea: 'Search',
@@ -96,14 +61,14 @@ export default async function (message: CooccurredMessage, userId: string) {
       });
     });
 
-    const edgesSortedWithSimilarity = [...ListArticles.edges].sort(
+    const edgesSortedWithSimilarity = [...result.edges].sort(
       (a, b) => b.mediaSimilarity - a.mediaSimilarity
     );
 
     const hasIdenticalDocs =
       edgesSortedWithSimilarity[0].mediaSimilarity >= SIMILARITY_THRESHOLD;
 
-    if (ListArticles.edges.length === 1 && hasIdenticalDocs) {
+    if (result.edges.length === 1 && hasIdenticalDocs) {
       visitor.send();
 
       return await choosingArticle({
@@ -118,7 +83,7 @@ export default async function (message: CooccurredMessage, userId: string) {
       });
     }
 
-    const articleOptions = ListArticles.edges
+    const articleOptions = result.edges
       .map(
         (
           {

diff --git a/src/webhook/handlers/utils.ts b/src/webhook/handlers/utils.ts
@@ -24,6 +24,10 @@ import type {
   CreateReferenceWordsReplyFragment,
   CreateAiReplyMutation,
   CreateAiReplyMutationVariables,
+  ListArticlesInInitStateQuery,
+  ListArticlesInInitStateQueryVariables,
+  ListArticlesInProcessMediaQuery,
+  ListArticlesInProcessMediaQueryVariables,
 } from 'typegen/graphql';
 
 import type { Input as ChoosingReplyInput } from './choosingReply';
@@ -851,3 +855,79 @@ export function getLineContentProxyURL(messageId: string) {
 
   return `${process.env.RUMORS_LINE_BOT_URL}/getcontent?token=${jwt}`;
 }
+
+export async function searchText(
+  text: string
+): Promise<ListArticlesInInitStateQuery['ListArticles']> {
+  const {
+    data: { ListArticles },
+  } = await gql`
+    query ListArticlesInInitState($text: String!) {
+      ListArticles(
+        filter: { moreLikeThis: { like: $text } }
+        orderBy: [{ _score: DESC }]
+        first: 4
+      ) {
+        edges {
+          node {
+            text
+            id
+            articleType
+          }
+          highlight {
+            text
+            hyperlinks {
+              title
+              summary
+            }
+          }
+        }
+      }
+    }
+  `<ListArticlesInInitStateQuery, ListArticlesInInitStateQueryVariables>({
+    text,
+  });
+  return ListArticles;
+}
+
+export async function searchMedia(
+  mediaUrl: string,
+  userId: string
+): Promise<ListArticlesInProcessMediaQuery['ListArticles']> {
+  const {
+    data: { ListArticles },
+  } = await gql`
+    query ListArticlesInProcessMedia($mediaUrl: String!) {
+      ListArticles(
+        filter: {
+          mediaUrl: $mediaUrl
+          articleTypes: [TEXT, IMAGE, AUDIO, VIDEO]
+          transcript: { shouldCreate: true }
+        }
+        orderBy: [{ _score: DESC }]
+        first: 9
+      ) {
+        edges {
+          score
+          mediaSimilarity
+          node {
+            id
+            articleType
+            attachmentUrl(variant: THUMBNAIL)
+          }
+          highlight {
+            text
+            hyperlinks {
+              title
+              summary
+            }
+          }
+        }
+      }
+    }
+  `<ListArticlesInProcessMediaQuery, ListArticlesInProcessMediaQueryVariables>(
+    { mediaUrl },
+    { userId }
+  );
+  return ListArticles;
+}