feat: Issue Deduplication

.cspell.json

@@ -26,7 +26,8 @@
     "voyageai",
     "vectordump",
     "payloadobject",
-    "markdownit"
+    "markdownit",
+    "plpgsql"
   ],
   "dictionaries": ["typescript", "node", "software-terms"],
   "import": ["@cspell/dict-typescript/cspell-ext.json", "@cspell/dict-node/cspell-ext.json", "@cspell/dict-software-terms"],

src/adapters/supabase/helpers/issues.ts

@@ -3,14 +3,10 @@ import { SuperSupabase } from "./supabase";
 import { Context } from "../../../types/context";
 import { markdownToPlainText } from "../../utils/markdown-to-plaintext";
 
-export interface IssueType {
-  id: string;
-  markdown?: string;
-  author_id: number;
-  created_at: string;
-  modified_at: string;
-  payloadObject: Record<string, unknown> | null;
-  embedding: number[];
+export interface IssueSimilaritySearchResult {
+  issue_id: string;
+  issue_plaintext: string;
+  similarity: number;
 }
 
 export class Issues extends SuperSupabase {
@@ -70,15 +66,13 @@ export class Issues extends SuperSupabase {
     }
   }
 
-  async findSimilarIssues(markdown: string, threshold: number): Promise<IssueType[] | null> {
+  async findSimilarIssues(markdown: string, threshold: number, currentId: string): Promise<IssueSimilaritySearchResult[] | null> {
     const embedding = await this.context.adapters.voyage.embedding.createEmbedding(markdown);
-    const { data, error } = await this.supabase
-      .from("issues")
-      .select("*")
-      .eq("type", "issue")
-      .textSearch("embedding", embedding.join(","))
-      .order("embedding", { foreignTable: "issues", ascending: false })
-      .lte("embedding", threshold);
+    const { data, error } = await this.supabase.rpc("find_similar_issues", {
+      current_id: currentId,
+      query_embedding: embedding,
+      threshold: threshold,
+    });
     if (error) {
       this.context.logger.error("Error finding similar issues", error);
       return [];

src/handlers/issue-deduplication.ts

@@ -0,0 +1,138 @@
+import { IssueSimilaritySearchResult } from "../adapters/supabase/helpers/issues";
+import { Context } from "../types";
+const MATCH_THRESHOLD = 0.95;
+const WARNING_THRESHOLD = 0.5;
+
+export interface IssueGraphqlResponse {
+  node: {
+    title: string;
+    url: string;
+  };
+}
+
+/**
+ * Check if an issue is similar to any existing issues in the database
+ * @param context
+ * @returns true if the issue is similar to an existing issue, false otherwise
+ */
+export async function issueChecker(context: Context): Promise<boolean> {
+  const {
+    logger,
+    payload,
+    adapters: { supabase },
+  } = context;
+
+  const issue = payload.issue;
+
+  //First Check if an issue with more than MATCH_THRESHOLD similarity exists (Very Similar)
+  const similarIssue = await supabase.issue.findSimilarIssues(issue.body + issue.title, MATCH_THRESHOLD, issue.node_id);
+  if (similarIssue && similarIssue?.length > 0) {
+    logger.info(`Similar issue which matches more than ${MATCH_THRESHOLD} already exists`);
+    //Close the issue as "unplanned"
+    await context.octokit.issues.update({
+      owner: payload.repository.owner.login,
+      repo: payload.repository.name,
+      issue_number: issue.number,
+      state: "closed",
+      labels: ["unplanned"],
+    });
+    return true;
+  }
+
+  //Second Check if an issue with more than WARNING_THRESHOLD similarity exists (Warning)
+  const warningIssue = await supabase.issue.findSimilarIssues(issue.body + issue.title, WARNING_THRESHOLD, issue.node_id);
+  if (warningIssue && warningIssue?.length > 0) {
+    logger.info(`Similar issue which matches more than ${WARNING_THRESHOLD} already exists`);
+    //Add a comment immediately next to the issue
+    //Build a list of similar issues url
+    const issueList: IssueGraphqlResponse[] = await Promise.all(
+      warningIssue.map(async (issue: IssueSimilaritySearchResult) => {
+        //fetch the issue url and title using globalNodeId
+        const issueUrl: IssueGraphqlResponse = await context.octokit.graphql(
+          `query($issueNodeId: ID!) {
+                    node(id: $issueNodeId) {
+                        ... on Issue {
+                        title
+                        url
+                        }
+                    }
+                    }`,
+          {
+            issueNodeId: issue.issue_id,
+          }
+        );
+        return issueUrl;
+      })
+    );
+
+    // Reopen the issue
+    await context.octokit.issues.update({
+      owner: payload.repository.owner.login,
+      repo: payload.repository.name,
+      issue_number: issue.number,
+      state: "open",
+    });
+    //Remove the "unplanned" label if it exists
+    if (issue.labels && issue.labels.find((label) => label.name === "unplanned")) {
+      await context.octokit.issues.removeLabel({
+        owner: payload.repository.owner.login,
+        repo: payload.repository.name,
+        issue_number: issue.number,
+        name: "unplanned",
+      });
+    }
+    // Check if there is already a comment on the issue
+    const existingComment = await context.octokit.issues.listComments({
+      owner: payload.repository.owner.login,
+      repo: payload.repository.name,
+      issue_number: issue.number,
+    });
+    if (existingComment.data.length > 0) {
+      // Find the comment that lists the similar issues
+      const commentToUpdate = existingComment.data.find(
+        (comment) => comment && comment.body && comment.body.includes("This issue seems to be similar to the following issue(s)")
+      );
+
+      if (commentToUpdate) {
+        // Update the comment with the latest list of similar issues
+        const body = issueList.map((issue) => `- [${issue.node.title}](${issue.node.url})`).join("\n");
+        const updatedBody = `This issue seems to be similar to the following issue(s):\n\n${body}`;
+        await context.octokit.issues.updateComment({
+          owner: payload.repository.owner.login,
+          repo: payload.repository.name,
+          comment_id: commentToUpdate.id,
+          body: updatedBody,
+        });
+      } else {
+        // Add a new comment to the issue
+        await createNewComment(context, issueList);
+      }
+    } else {
+      // Add a new comment to the issue
+      await createNewComment(context, issueList);
+    }
+    return true;
+  }
+
+  logger.info("No similar issue found");
+  return false;
+}
+
+/**
+ * Create a new comment on the issue with the list of similar issues
+ * @param context
+ * @param resolvedIssueList
+ */
+async function createNewComment(context: Context, resolvedIssueList: IssueGraphqlResponse[]) {
+  let body = "This issue seems to be similar to the following issue(s):\n\n";
+  resolvedIssueList.forEach((issue) => {
+    const issueLine = `- [${issue.node.title}](${issue.node.url})\n`;
+    body += issueLine;
+  });
+  await context.octokit.issues.createComment({
+    owner: context.payload.repository.owner.login,
+    repo: context.payload.repository.name,
+    issue_number: context.payload.issue.number,
+    body: body,
+  });
+}

src/plugin.ts

@@ -13,6 +13,7 @@ import { VoyageAIClient } from "voyageai";
 import { deleteIssues } from "./handlers/delete-issue";
 import { addIssue } from "./handlers/add-issue";
 import { updateIssue } from "./handlers/update-issue";
+import { issueChecker } from "./handlers/issue-deduplication";
 
 /**
  * The main plugin function. Split for easier testing.
@@ -31,11 +32,13 @@ export async function runPlugin(context: Context) {
   } else if (isIssueEvent(context)) {
     switch (eventName) {
       case "issues.opened":
+        await issueChecker(context);
         return await addIssue(context);
-      case "issues.deleted":
-        return await deleteIssues(context);
       case "issues.edited":
+        await issueChecker(context);
         return await updateIssue(context);
+      case "issues.deleted":
+        return await deleteIssues(context);
     }
   } else {
     logger.error(`Unsupported event: ${eventName}`);