chore: remove const fn, update db fn, update findSimilarContent

README.md

@@ -13,6 +13,7 @@ This is a plugin for [Ubiquibot](https://github.com/ubiquity/ubiquibot-kernel).
 ## Usage
 
 - Add the following to your `.ubiquibot-config.yml` file with the appropriate URL:
+
 ```yaml
 - plugin: https://ubiquity-os-comment-vector-embeddings-main.ubiquity.workers.dev
   with:

manifest.json

@@ -1,5 +1,13 @@
 {
   "name": "Generate vector embeddings",
   "description": "Enables the storage, updating, and deletion of issue comment embeddings.",
-  "ubiquity:listeners": ["issue_comment.created", "issue_comment.edited", "issue_comment.deleted", "issues.opened", "issues.edited", "issues.deleted", "issues.labeled"]
+  "ubiquity:listeners": [
+    "issue_comment.created",
+    "issue_comment.edited",
+    "issue_comment.deleted",
+    "issues.opened",
+    "issues.edited",
+    "issues.deleted",
+    "issues.labeled"
+  ]
 }

src/adapters/supabase/helpers/embeddings.ts

@@ -112,6 +112,12 @@ export class Embeddings extends Super {
       throw new Error(this.context.logger.error("Markdown not found", { sourceId })?.logMessage.raw);
     }
 
+    const embeddingData = await this.getEmbedding(sourceId);
+
+    if (!embeddingData) {
+      return await this.createEmbedding(sourceId, type, body, metadata);
+    }
+
     const embedding = await this._embedWithVoyage(body);
 
     const toStore: Omit<CommentType, "created_at"> = {
@@ -123,12 +129,6 @@ export class Embeddings extends Super {
       modified_at: new Date().toISOString(),
     };
 
-    const embeddingData = await this.getEmbedding(sourceId);
-
-    if (!embeddingData) {
-      return await this.createEmbedding(sourceId, type, body, metadata);
-    }
-
     const { error } = await this.supabase.from("content").update(toStore).eq("source_id", sourceId);
 
     if (error) {
@@ -157,10 +157,10 @@ export class Embeddings extends Super {
 
   // Working with embeddings
 
-  async findSimilarIssues(markdown: string, threshold: number, currentId: string): Promise<IssueSimilaritySearchResult[]> {
+  async findSimilarContent(markdown: string, threshold: number, currentId: string): Promise<IssueSimilaritySearchResult[]> {
     const embedding = await this._embedWithVoyage(markdown);
-    const { data, error } = await this.supabase.rpc("find_similar_issues", {
-      current_id: currentId,
+    const { data, error } = await this.supabase.rpc("find_similar_content", {
+      curr_source_id: currentId,
       query_embedding: embedding,
       threshold: threshold,
     });

src/handlers/issue-matching.ts

@@ -24,7 +24,7 @@ export interface IssueGraphqlResponse {
   similarity: number;
 }
 
-const commentBuilder = (matchResultArray: Map<string, Array<string>>): string => {
+function commentBuilder(matchResultArray: Map<string, Array<string>>) {
   const commentLines: string[] = [">[!NOTE]", ">The following contributors may be suitable for this task:"];
   matchResultArray.forEach((issues, assignee) => {
     commentLines.push(`>### [${assignee}](https://www.github.com/${assignee})`);
@@ -50,7 +50,7 @@ export async function issueMatching(context: Context<"issues.opened" | "issues.e
   // create a new comment with users who completed task most similar to the issue
   // if the comment already exists, it should update the comment with the new users
   const matchResultArray: Map<string, Array<string>> = new Map();
-  const similarIssues = await supabase.embeddings.findSimilarIssues(issueContent, context.config.jobMatchingThreshold, issue.node_id);
+  const similarIssues = await supabase.embeddings.findSimilarContent(issueContent, context.config.jobMatchingThreshold, issue.node_id);
   if (similarIssues && similarIssues.length > 0) {
     // Find the most similar issue and the users who completed the task
     similarIssues.sort((a, b) => b.similarity - a.similarity);

src/handlers/task-deduplication.ts

@@ -15,7 +15,7 @@ export interface IssueGraphqlResponse {
  * @param context
  * @returns true if the issue is similar to an existing issue, false otherwise
  */
-export async function taskSimilaritySearch(context: Context<"issues.opened">): Promise<CallbackResult> {
+export async function taskSimilaritySearch(context: Context<"issues.opened" | "issues.edited">): Promise<CallbackResult> {
   const {
     logger,
     adapters: { supabase },
@@ -26,9 +26,9 @@ export async function taskSimilaritySearch(context: Context<"issues.opened">): P
   } = context;
   const similarIssues: IssueSimilaritySearchResult[] = [];
 
-  similarIssues.push(...(await supabase.embeddings.findSimilarIssues(issue.title, context.config.warningThreshold, issue.node_id)));
+  similarIssues.push(...(await supabase.embeddings.findSimilarContent(issue.title, context.config.warningThreshold, issue.node_id)));
   if (issue.body) {
-    similarIssues.push(...(await supabase.embeddings.findSimilarIssues(issue.body, context.config.warningThreshold, issue.node_id)));
+    similarIssues.push(...(await supabase.embeddings.findSimilarContent(issue.body, context.config.warningThreshold, issue.node_id)));
   }
 
   logger.info(`Found ${similarIssues.length} similar issues`);
@@ -53,7 +53,7 @@ export async function taskSimilaritySearch(context: Context<"issues.opened">): P
     if (similarIssues.length > 0) {
       logger.info(`Similar issue which matches more than ${context.config.warningThreshold} already exists`);
       await handleSimilarIssuesComment(context, issue.number, similarIssues);
-      return { statusCode: 200 }
+      return { statusCode: 200 };
     }
   }
 

src/proxy-callbacks.ts

@@ -48,7 +48,6 @@ const callbacks = {
   "issues.deleted": [deleteTaskEmbedding],
 } as ProxyCallbacks;
 
-
 /**
  * 
 

supabase/config.toml

@@ -8,7 +8,7 @@ enabled = true
 port = 54321
 # Schemas to expose in your API. Tables, views and stored procedures in this schema will get API
 # endpoints. public and storage are always included.
-schemas = ["public", "storage", "graphql_public"]
+schemas = ["public", "graphql_public"]
 # Extra schemas to add to the search_path of every request. public is always included.
 extra_search_path = ["public", "extensions"]
 # The maximum number of rows returns from a view, table, or stored procedure. Limits payload size

supabase/migrations/20240930150911_content.sql

@@ -1,3 +1,4 @@
+create extension if not exists vector;
 DROP TABLE IF EXISTS content;
 
 CREATE TABLE IF NOT EXISTS content (
@@ -13,16 +14,16 @@ CREATE TABLE IF NOT EXISTS content (
 
 ALTER TABLE "content" ENABLE ROW LEVEL SECURITY;
 
-CREATE OR REPLACE FUNCTION find_similar_content(current_source_id VARCHAR, query_embedding vector(1024), threshold float8)
+CREATE OR REPLACE FUNCTION find_similar_content(curr_source_id VARCHAR, query_embedding vector(1024), threshold float8)
 RETURNS TABLE(source_id VARCHAR, content_plaintext TEXT, similarity float8) AS $$
 BEGIN
   RETURN QUERY
-  SELECT source_id,
-         plaintext AS content_plaintext,
-         1 - (embedding <=> query_embedding) AS similarity
+  SELECT content.source_id AS source_id,
+         content.plaintext AS content_plaintext,
+         1 - (content.embedding <=> query_embedding) AS similarity
   FROM content
-  WHERE source_id <> current_source_id
-    AND 1 - (embedding <=> query_embedding) >= threshold
+  WHERE content.source_id <> curr_source_id
+    AND 1 - (content.embedding <=> query_embedding) >= threshold
   ORDER BY similarity DESC;
 END;
 $$ LANGUAGE plpgsql;