feat: add similar issues endpoint

Author: danielroe

server/api/clusters/[owner]/[repo].get.ts

@@ -1,5 +1,5 @@
-import { clusterEmbeddings } from '../../../utils/cluster'
-import { getEmbeddingsForIssue } from '../../../utils/embeddings'
+import { clusterEmbeddings } from '~~/server/utils/cluster'
+import { getEmbeddingsForIssue } from '~~/server/utils/embeddings'
 
 import { isAllowedRepo, type AllowedRepo } from '#shared/repos'
 
@@ -62,7 +62,7 @@ export default defineCachedEventHandler(async (event) => {
 }, {
   swr: true,
   getKey(event) {
-    const [owner, repo] = getRouterParam(event, 'repo')?.split('/') || []
+    const { owner, repo } = getRouterParams(event)
     return `clusters:${owner}:${repo}`.toLowerCase()
   },
   maxAge: 60 * 60 * 1000,

server/api/issues/[owner]/[repo].get.ts

@@ -1,6 +1,6 @@
 import { Octokit } from '@octokit/rest'
 
-import { getLabels, type Issue } from '../../../utils/github'
+import { getLabels, type Issue } from '~~/server/utils/github'
 import { isAllowedRepo } from '#shared/repos'
 
 const labelsToExclude = ['documentation', 'invalid', 'enhancement']
@@ -41,7 +41,7 @@ export default defineCachedEventHandler(async (event) => {
 }, {
   swr: true,
   getKey(event) {
-    const [owner, repo] = getRouterParam(event, 'repo')?.split('/') || []
+    const { owner, repo } = getRouterParams(event)
     return `issues:${owner}:${repo}`.toLowerCase()
   },
   maxAge: 60 * 60 * 1000,

server/api/issues/[owner]/[repo]/[number].get.ts

@@ -0,0 +1,72 @@
+import { Octokit } from '@octokit/rest'
+
+import { isAllowedRepo } from '#shared/repos'
+import { getEmbeddingsForIssue, getStoredEmbeddingsForIssue } from '~~/server/utils/embeddings'
+
+export default defineCachedEventHandler(async (event) => {
+  const { owner, repo, number } = getRouterParams(event)
+
+  if (!owner || !repo || !number) {
+    throw createError({
+      status: 400,
+      message: 'Invalid repository',
+    })
+  }
+
+  const source = `${owner}/${repo}`
+
+  if (!isAllowedRepo(source)) {
+    throw createError({
+      status: 400,
+      message: 'Repository not allowed',
+    })
+  }
+
+  const vectorize = typeof hubVectorize !== 'undefined' ? hubVectorize('issues') : null
+
+  let issueEmbeddings = await getStoredEmbeddingsForIssue(event, owner, repo, number)
+  if (!issueEmbeddings) {
+    const options = owner !== 'unjs' ? { auth: useRuntimeConfig(event).github.token } : {}
+    const octokit = new Octokit(options)
+
+    const { data: issue } = await octokit.issues.get({
+      owner,
+      repo,
+      issue_number: parseInt(number),
+    })
+
+    issueEmbeddings = await getEmbeddingsForIssue(event, issue)
+  }
+
+  const results = await vectorize?.query(issueEmbeddings, {
+    // TODO: support similar repos
+    // filter: {
+    //   owner,
+    //   repository: repo,
+    // },
+  })
+
+  return results?.matches.map((m) => {
+    const groups = m.id.match(/^issue:(?<owner>[^:]+):(?<repo>[^:]+):(?<number>\d+)$/)?.groups
+    if (!groups) {
+      console.error('Invalid match', m.id)
+      return
+    }
+    return {
+      owner: groups.owner,
+      repo: groups.repo,
+      number: parseInt(groups.number!),
+      score: m.score,
+    }
+  }).filter(Boolean)
+}, {
+  swr: true,
+  getKey(event) {
+    const { owner, repo, number } = getRouterParams(event)
+    return `issue:${owner}:${repo}:${number}`.toLowerCase()
+  },
+  maxAge: 60 * 60 * 1000,
+  staleMaxAge: 60 * 60 * 1000,
+  shouldBypassCache: event => getHeader(event, 'force') === 'true',
+  shouldInvalidateCache: event => getHeader(event, 'force') === 'true',
+})

server/utils/embeddings.ts

@@ -3,6 +3,13 @@ import { hash } from 'ohash'
 
 import { getLabels, type Issue } from './github'
 
+export async function getStoredEmbeddingsForIssue(event: H3Event, owner: string, repo: string, number: number | string) {
+  const storage = hubKV()
+  const storageKey = `issue:${owner}:${repo}:${number}`
+  const res = await storage.getItem<StoredEmbeddings>(storageKey)
+  return res?.embeddings
+}
+
 export async function getEmbeddingsForIssue(event: H3Event, issue: Issue) {
   const storage = hubKV()
   const vectorize = typeof hubVectorize !== 'undefined' ? hubVectorize('issues') : null