From 9781ec023b421754254dddcbdd1f30f6fe8e6bff Mon Sep 17 00:00:00 2001
From: Kevin Lu <kevinlu1248@gmail.com>
Date: Wed, 5 Jun 2024 11:07:29 -0700
Subject: [PATCH 1/7] E2E PR Test

---
 sweep_chat/components/App.tsx     | 67 +++++++++++++++++--------------
 sweep_chat/cypress/e2e/spec.cy.ts | 10 ++++-
 2 files changed, 46 insertions(+), 31 deletions(-)

diff --git a/sweep_chat/components/App.tsx b/sweep_chat/components/App.tsx
index 47cf850409..91c4df76f0 100644
--- a/sweep_chat/components/App.tsx
+++ b/sweep_chat/components/App.tsx
@@ -522,37 +522,44 @@ const parsePullRequests = async (repoName: string, message: string, octokit: Oct
   const [orgName, repo] = repoName.split("/")
   const pulls = []
 
-  const prURLs = message.match(new RegExp(`https?:\/\/github.com\/${repoName}\/pull\/(?<prNumber>[0-9]+)`, 'gm'));
-  for (const prURL of prURLs || []) {
-    const prNumber = prURL.split("/").pop()
-    console.log(prNumber)
-    const pr = await octokit!.rest.pulls.get({
-      owner: orgName,
-      repo: repo,
-      pull_number: parseInt(prNumber!)
-    })
-    const title = pr.data.title
-    const body = pr.data.body
-    const labels = pr.data.labels.map((label) => label.name)
-    const status = pr.data.state === "open" ? "open" : pr.data.merged ? "merged" : "closed"
-    const file_diffs = (await octokit!.rest.pulls.listFiles({
-      owner: orgName,
-      repo: repo,
-      pull_number: parseInt(prNumber!)
-    })).data
-    // console.log(file_diffs)
-    pulls.push({
-      number: parseInt(prNumber!),
-      repo_name: repoName,
-      title,
-      body,
-      labels,
-      status,
-      file_diffs
-    } as PullRequest)
-  }
+  try {
+    const prURLs = message.match(new RegExp(`https?:\/\/github.com\/${repoName}\/pull\/(?<prNumber>[0-9]+)`, 'gm'));
+    for (const prURL of prURLs || []) {
+      const prNumber = prURL.split("/").pop()
+      const pr = await octokit!.rest.pulls.get({
+        owner: orgName,
+        repo: repo,
+        pull_number: parseInt(prNumber!)
+      })
+      const title = pr.data.title
+      const body = pr.data.body
+      const labels = pr.data.labels.map((label) => label.name)
+      const status = pr.data.state === "open" ? "open" : pr.data.merged ? "merged" : "closed"
+      const file_diffs = (await octokit!.rest.pulls.listFiles({
+        owner: orgName,
+        repo: repo,
+        pull_number: parseInt(prNumber!)
+      })).data
+      // console.log(file_diffs)
+      pulls.push({
+        number: parseInt(prNumber!),
+        repo_name: repoName,
+        title,
+        body,
+        labels,
+        status,
+        file_diffs
+      } as PullRequest)
+    }
 
-  return pulls
+    return pulls
+  } catch (e: any) {
+    toast({
+      title: "Failed to retrieve pull request",
+      description: `The following error has occurred: ${e.message}. Sometimes, logging out and logging back in can resolve this issue.`,
+      variant: "destructive"
+    });
+  } 
 }
 
 function App() {
diff --git a/sweep_chat/cypress/e2e/spec.cy.ts b/sweep_chat/cypress/e2e/spec.cy.ts
index 08d117064a..912117d166 100644
--- a/sweep_chat/cypress/e2e/spec.cy.ts
+++ b/sweep_chat/cypress/e2e/spec.cy.ts
@@ -1,4 +1,5 @@
 const testMessage = "In the vector search logic, how would I migrate the KNN to use HNSW instead?"
+const testPullRequestMessage = "Help me review this PR: https://github.com/sweepai/sweep/pull/3978"
 
 describe('sweep chat', () => {
   beforeEach(() => {
@@ -22,13 +23,20 @@ describe('sweep chat', () => {
     cy.get(':nth-child(5) > .flex').type(testMessage + "{enter}")
     cy.wait(1000)
     cy.get(':nth-child(5) > .inline-flex').click()
-    cy.wait(3000)
     cy.on('uncaught:exception', (err, runnable) => {
       expect(err.message).to.include('No snippets found');
       return false;
     })
   })
 
+  it("can preview pull requests", () => {
+    cy.get('.grow > .flex').type("sweepai/sweep").blur()
+    cy.get(':nth-child(5) > .flex', { timeout: 10000 }).should('have.attr', 'placeholder', 'Type a message...')
+
+    cy.get(':nth-child(5) > .flex').type(testPullRequestMessage + "{enter}")
+    cy.get('a > .bg-zinc-800').contains('Minor ticket utils fix')
+  })
+
   it("can send a message", () => {
     cy.get('.grow > .flex').type("sweepai/sweep").blur()
     cy.get(':nth-child(5) > .flex', { timeout: 10000 }).should('have.attr', 'placeholder', 'Type a message...')

From ff0632f0c54f092a276ffc51d587a3ffa76af6f3 Mon Sep 17 00:00:00 2001
From: Kevin Lu <kevinlu1248@gmail.com>
Date: Wed, 5 Jun 2024 12:07:22 -0700
Subject: [PATCH 2/7] Temporary work

---
 sweep_chat/components/App.tsx | 109 ++++++++++++++++++++++++++++------
 sweep_chat/package-lock.json  |  20 +++++++
 sweep_chat/package.json       |   2 +
 3 files changed, 112 insertions(+), 19 deletions(-)

diff --git a/sweep_chat/components/App.tsx b/sweep_chat/components/App.tsx
index 91c4df76f0..965e48f7e4 100644
--- a/sweep_chat/components/App.tsx
+++ b/sweep_chat/components/App.tsx
@@ -35,6 +35,7 @@ import PulsingLoader from "./shared/PulsingLoader";
 
 import { Octokit } from "octokit";
 import { Command, CommandEmpty, CommandGroup, CommandInput, CommandItem, CommandList, CommandSeparator } from "./ui/command";
+import clarinet from "clarinet";
 import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover";
 
 if (typeof window !== 'undefined') {
@@ -457,6 +458,57 @@ const MessageDisplay = ({ message, className, onEdit }: { message: Message, clas
   );
 };
 
+function getJSONPrefix(buffer: string): [any[], number] {
+  let stack: string[] = [];
+  const matchingBrackets: Record<string, string> = {
+    '[': ']',
+    '{': '}',
+    '(': ')'
+  };
+  var currentIndex = 0;
+  const results = [];
+  let inString = false;
+  let escapeNext = false;
+
+  for (let i = 0; i < buffer.length; i++) {
+    const char = buffer[i];
+
+    if (escapeNext) {
+      escapeNext = false;
+      continue;
+    }
+
+    if (char === '\\') {
+      escapeNext = true;
+      continue;
+    }
+
+    if (char === '"') {
+      inString = !inString;
+    }
+
+    if (!inString) {
+      if (matchingBrackets[char]) {
+        stack.push(char);
+      } else if (matchingBrackets[stack[stack.length - 1]] === char) {
+        stack.pop();
+        if (stack.length === 0) {
+          try {
+            results.push(JSON.parse(buffer.slice(currentIndex, i + 1)));
+            currentIndex = i + 1;
+          } catch (e) {
+            continue;
+          }
+        }
+      }
+    }
+  }
+  if (currentIndex == 0) {
+    console.log(results, currentIndex, buffer);
+  }
+  return [results, currentIndex];
+}
+
 async function* streamMessages(
   reader: ReadableStreamDefaultReader<Uint8Array>,
   isStream: React.MutableRefObject<boolean>,
@@ -482,31 +534,50 @@ async function* streamMessages(
         done = true;
         continue;
       }
-
+      
       if (value) {
         const decodedValue = new TextDecoder().decode(value);
         buffer += decodedValue;
-        buffer = buffer.replace("][{", "]\n[{"); // Ensure proper JSON formatting for split
-        buffer = buffer.replace("][[", "]\n[["); // Ensure proper JSON formatting for split
-        buffer = buffer.replace("]][", "]]\n["); // Ensure proper JSON formatting for split
-        const bufferLines = buffer.split("\n");
-
-        for (let line of bufferLines) { // Process all lines except the potentially incomplete last one
-          if (!line) {
-            continue
-          }
-          try {
-            const parsedLine = JSON.parse(line);
-            if (parsedLine) {
-              yield parsedLine
-            }
-          } catch (error) {
-            console.error("Failed to parse line:", line, error);
-          }
+
+        const [parsedObjects, currentIndex] = getJSONPrefix(buffer)
+        for (let parsedObject of parsedObjects) {
+          console.log(parsedObject)
+          yield parsedObject
         }
+        buffer = buffer.slice(currentIndex)
+
+        // const parsedLine = JSON.parse(buffer);
+        // if (parsedLine) {
+        //   yield parsedLine;
+        // }
 
-        buffer = bufferLines[bufferLines.length - 1]; // Keep the last line in the buffer
+        // buffer = "";
       }
+
+      // if (value) {
+      //   const decodedValue = new TextDecoder().decode(value);
+      //   buffer += decodedValue;
+      //   buffer = buffer.replace("][{", "]\n[{"); // Ensure proper JSON formatting for split
+      //   buffer = buffer.replace("][[", "]\n[["); // Ensure proper JSON formatting for split
+      //   buffer = buffer.replace("]][", "]]\n["); // Ensure proper JSON formatting for split
+      //   const bufferLines = buffer.split("\n");
+
+      //   for (let line of bufferLines) { // Process all lines except the potentially incomplete last one
+      //     if (!line) {
+      //       continue
+      //     }
+      //     try {
+      //       const parsedLine = JSON.parse(line);
+      //       if (parsedLine) {
+      //         yield parsedLine
+      //       }
+      //     } catch (error) {
+      //       console.error("Failed to parse line:", line, error);
+      //     }
+      //   }
+
+      //   buffer = bufferLines[bufferLines.length - 1]; // Keep the last line in the buffer
+      // }
     } catch (error) {
       console.error("Error during streaming:", error);
       throw error;
diff --git a/sweep_chat/package-lock.json b/sweep_chat/package-lock.json
index cdd4e07651..f3bc921c5e 100644
--- a/sweep_chat/package-lock.json
+++ b/sweep_chat/package-lock.json
@@ -19,6 +19,8 @@
         "@radix-ui/react-switch": "^1.0.3",
         "@radix-ui/react-toast": "^1.1.5",
         "@sentry/nextjs": "^8.2.1",
+        "@types/clarinet": "^0.12.3",
+        "clarinet": "^0.12.6",
         "class-variance-authority": "^0.7.0",
         "clsx": "^2.1.1",
         "cmdk": "^1.0.0",
@@ -3397,6 +3399,14 @@
         "@types/node": "*"
       }
     },
+    "node_modules/@types/clarinet": {
+      "version": "0.12.3",
+      "resolved": "https://registry.npmjs.org/@types/clarinet/-/clarinet-0.12.3.tgz",
+      "integrity": "sha512-7Fwiv6RNH6JW86U7Vvae7sFHrNmNWxMIx9iAH49F+5d9Kizew5GMYz6rixupn3PK4t+oqP7GZMqOQR77QbcrAQ==",
+      "dependencies": {
+        "@types/node": "*"
+      }
+    },
     "node_modules/@types/connect": {
       "version": "3.4.36",
       "resolved": "https://registry.npmjs.org/@types/connect/-/connect-3.4.36.tgz",
@@ -4882,6 +4892,16 @@
       "resolved": "https://registry.npmjs.org/cjs-module-lexer/-/cjs-module-lexer-1.3.1.tgz",
       "integrity": "sha512-a3KdPAANPbNE4ZUv9h6LckSl9zLsYOP4MBmhIPkRaeyybt+r4UghLvq+xw/YwUcC1gqylCkL4rdVs3Lwupjm4Q=="
     },
+    "node_modules/clarinet": {
+      "version": "0.12.6",
+      "resolved": "https://registry.npmjs.org/clarinet/-/clarinet-0.12.6.tgz",
+      "integrity": "sha512-0FR+TrvLbYHLjhzs9oeIbd3yfZmd4u2DzYQEjUTm2dNfh4Y/9RIRWPjsm3aBtrVEpjKI7+lWa4ouqEXoml84mQ==",
+      "engines": {
+        "chrome": ">=16.0.912",
+        "firefox": ">=0.8.0",
+        "node": ">=0.3.6"
+      }
+    },
     "node_modules/class-variance-authority": {
       "version": "0.7.0",
       "resolved": "https://registry.npmjs.org/class-variance-authority/-/class-variance-authority-0.7.0.tgz",
diff --git a/sweep_chat/package.json b/sweep_chat/package.json
index 58893cc2f1..d6279efcd0 100644
--- a/sweep_chat/package.json
+++ b/sweep_chat/package.json
@@ -23,6 +23,8 @@
     "@radix-ui/react-switch": "^1.0.3",
     "@radix-ui/react-toast": "^1.1.5",
     "@sentry/nextjs": "^8.2.1",
+    "@types/clarinet": "^0.12.3",
+    "clarinet": "^0.12.6",
     "class-variance-authority": "^0.7.0",
     "clsx": "^2.1.1",
     "cmdk": "^1.0.0",

From 39579cfa8a62ead050a84721e7b4369d35d0923d Mon Sep 17 00:00:00 2001
From: Kevin Lu <kevinlu1248@gmail.com>
Date: Wed, 5 Jun 2024 16:18:31 -0700
Subject: [PATCH 3/7] Loading bar fixes and autocomplete

---
 sweep_chat/components/App.tsx                 | 132 +++++++-------
 .../components/shared/PulsingLoader.tsx       |   7 +-
 sweep_chat/components/ui/autocomplete.tsx     | 165 ++++++++++++++++++
 sweep_chat/components/ui/skeleton.tsx         |  15 ++
 sweep_chat/package-lock.json                  |  55 ++++--
 sweep_chat/package.json                       |   2 +-
 6 files changed, 289 insertions(+), 87 deletions(-)
 create mode 100644 sweep_chat/components/ui/autocomplete.tsx
 create mode 100644 sweep_chat/components/ui/skeleton.tsx

diff --git a/sweep_chat/components/App.tsx b/sweep_chat/components/App.tsx
index 965e48f7e4..7c86f5baa7 100644
--- a/sweep_chat/components/App.tsx
+++ b/sweep_chat/components/App.tsx
@@ -17,6 +17,7 @@ import {
   AccordionItem,
   AccordionTrigger,
 } from "@/components/ui/accordion";
+import { AutoComplete } from "@/components/ui/autocomplete";
 import { Toaster } from "@/components/ui/toaster";
 import { toast } from "@/components/ui/use-toast";
 import { useSession, signIn, SessionProvider, signOut } from "next-auth/react";
@@ -35,7 +36,6 @@ import PulsingLoader from "./shared/PulsingLoader";
 
 import { Octokit } from "octokit";
 import { Command, CommandEmpty, CommandGroup, CommandInput, CommandItem, CommandList, CommandSeparator } from "./ui/command";
-import clarinet from "clarinet";
 import { Popover, PopoverContent, PopoverTrigger } from "./ui/popover";
 
 if (typeof window !== 'undefined') {
@@ -43,6 +43,8 @@ if (typeof window !== 'undefined') {
   posthog.debug(false)
 }
 
+type Repository = any;
+
 interface Snippet {
   content: string;
   start: number;
@@ -403,7 +405,7 @@ const MessageDisplay = ({ message, className, onEdit }: { message: Message, clas
               <AccordionTrigger className="border-none py-0 text-left">
                 <div className="text-xs text-gray-400 flex align-center">
                   {!message.function_call!.is_complete ? (
-                    <PulsingLoader size={4} />
+                    <PulsingLoader size={0.5} />
                   ) : (
                     <FaCheck
                       className="inline-block mr-2"
@@ -504,7 +506,7 @@ function getJSONPrefix(buffer: string): [any[], number] {
     }
   }
   if (currentIndex == 0) {
-    console.log(results, currentIndex, buffer);
+    console.log(buffer);
   }
   return [results, currentIndex];
 }
@@ -541,43 +543,10 @@ async function* streamMessages(
 
         const [parsedObjects, currentIndex] = getJSONPrefix(buffer)
         for (let parsedObject of parsedObjects) {
-          console.log(parsedObject)
           yield parsedObject
         }
         buffer = buffer.slice(currentIndex)
-
-        // const parsedLine = JSON.parse(buffer);
-        // if (parsedLine) {
-        //   yield parsedLine;
-        // }
-
-        // buffer = "";
       }
-
-      // if (value) {
-      //   const decodedValue = new TextDecoder().decode(value);
-      //   buffer += decodedValue;
-      //   buffer = buffer.replace("][{", "]\n[{"); // Ensure proper JSON formatting for split
-      //   buffer = buffer.replace("][[", "]\n[["); // Ensure proper JSON formatting for split
-      //   buffer = buffer.replace("]][", "]]\n["); // Ensure proper JSON formatting for split
-      //   const bufferLines = buffer.split("\n");
-
-      //   for (let line of bufferLines) { // Process all lines except the potentially incomplete last one
-      //     if (!line) {
-      //       continue
-      //     }
-      //     try {
-      //       const parsedLine = JSON.parse(line);
-      //       if (parsedLine) {
-      //         yield parsedLine
-      //       }
-      //     } catch (error) {
-      //       console.error("Failed to parse line:", line, error);
-      //     }
-      //   }
-
-      //   buffer = bufferLines[bufferLines.length - 1]; // Keep the last line in the buffer
-      // }
     } catch (error) {
       console.error("Error during streaming:", error);
       throw error;
@@ -587,6 +556,9 @@ async function* streamMessages(
       }
     }
   }
+  if (buffer) {
+    console.warn("Buffer:", buffer)
+  }
 }
 
 const parsePullRequests = async (repoName: string, message: string, octokit: Octokit): Promise<PullRequest[]> => {
@@ -653,6 +625,8 @@ function App() {
   const { data: session } = useSession()
 
   const posthog = usePostHog();
+  const [octokit, setOctokit] = useState<Octokit | null>(null)
+  const [repos, setRepos] = useState<Repository[]>([])
 
   if (session) {
     posthog.identify(
@@ -663,6 +637,19 @@ function App() {
         image: session.user!.image,
       }
     );
+  } else {
+    return (
+      <main className="flex h-screen items-center justify-center p-12">
+        <Toaster />
+        <Button onClick={() => signIn("github")} variant="secondary">
+          <FaGithub
+            className="inline-block mr-2"
+            style={{ marginTop: -2 }}
+          />
+          Sign in with GitHub
+        </Button>
+      </main>
+    )
   }
 
   useEffect(() => {
@@ -676,7 +663,7 @@ function App() {
 
   const lastAssistantMessageIndex = messages.findLastIndex((message) => message.role === "assistant" && message.content.trim().length > 0)
 
-  const startStream = async (message: string, newMessages: Message[]) => {
+  const startStream = async (message: string, newMessages: Message[], snippets: Snippet[]) => {
     setIsLoading(true);
     isStream.current = true;
 
@@ -737,6 +724,7 @@ function App() {
           variant: "destructive"
         });
         setIsLoading(false);
+        isStream.current = false;
         posthog.capture("chat errored", {
           repoName,
           snippets,
@@ -824,28 +812,29 @@ function App() {
     });
   }
 
-  const [octokit, setOctokit] = useState<Octokit | null>(null)
   useEffect(() => {
     if (session) {
       const octokit = new Octokit({auth: session.user!.accessToken})
-      setOctokit(octokit)
+      setOctokit(octokit);
+      (async () => {
+        const maxPages = 5;
+        let allRepositories: Repository[] = [];
+        let page = 1;
+        let response;
+        do {
+          response = await octokit.rest.repos.listForAuthenticatedUser({
+            visibility: "all",
+            sort: "pushed",
+            per_page: 100,
+            page: page,
+          });
+          allRepositories = allRepositories.concat(response.data);
+          setRepos(allRepositories)
+          page++;
+        } while (response.data.length !== 0 && page < maxPages);
+      })()
     }
-  }, [session])
-
-  if (!session) {
-    return (
-      <main className="flex h-screen items-center justify-center p-12">
-        <Toaster />
-        <Button onClick={() => signIn("github")} variant="secondary">
-          <FaGithub
-            className="inline-block mr-2"
-            style={{ marginTop: -2 }}
-          />
-          Sign in with GitHub
-        </Button>
-      </main>
-    )
-  }
+  }, [session.user!.accessToken])
 
   return (
     <main className="flex h-screen flex-col items-center justify-between p-12">
@@ -881,17 +870,17 @@ function App() {
         </div>
       </div>
       <div className={`mb-4 w-full flex items-center ${repoNameValid ? "" : "grow"}`}>
-        <Input
-          data-ph-capture-attribute-repo-name={repoName}
-          value={repoName}
-          onChange={(e)=>setRepoName(e.target.value.replace(/\s/g,''))}
-          onKeyDown={(e) => {
-            if (e.key === "Enter") {
-              e.currentTarget.blur();
-            }
-          }}
-          onBlur={async () => {
+        <AutoComplete
+          options={repos.map((repo) => ({label: repo.full_name, value: repo.full_name}))}
+          placeholder="Repository name"
+          emptyMessage="No repositories found"
+          value={{label: repoName, value: repoName}}
+          onValueChange={(option) => setRepoName(option.value)}
+          disabled={repoNameDisabled}
+          onBlur={async (repoName: string) => {
+            console.log(repoName)
             const cleanedRepoName = repoName.replace(/\s/g, '') // might be unsafe but we'll handle it once we get there
+            console.log(repoName)
             setRepoName(cleanedRepoName)
             if (cleanedRepoName === "") {
               setRepoNameValid(false)
@@ -943,10 +932,7 @@ function App() {
             }
             setRepoNameDisabled(false);
           }}
-          placeholder="Repository name"
-          disabled={repoNameDisabled}
         />
-        
         <Dialog>
           <DialogTrigger asChild>
             <Button variant="outline" className="ml-4">
@@ -1011,15 +997,17 @@ function App() {
               ]
               setMessages(newMessages)
               if (index == 0) {
-                setSnippets([])
+                setSnippets([]) 
+                startStream(content, newMessages, [])
+              } else {
+                startStream(content, newMessages, snippets)
               }
-              startStream(content, newMessages)
             }}
           />
         ))}
         {isLoading && (
           <div className="flex justify-around w-full py-2">
-            <PulsingLoader size={8} />
+            <PulsingLoader size={1.5} />
           </div>
         )}
       </div>
@@ -1065,7 +1053,7 @@ function App() {
                 const newMessages: Message[] = [...messages, { content: currentMessage, role: "user", annotations: { pulls } }];
                 setMessages(newMessages);
                 setCurrentMessage("");
-                startStream(currentMessage, newMessages)
+                startStream(currentMessage, newMessages, snippets)
               }
             }}
             onChange={(e) => setCurrentMessage(e.target.value)}
diff --git a/sweep_chat/components/shared/PulsingLoader.tsx b/sweep_chat/components/shared/PulsingLoader.tsx
index 7e8cfc074f..72e2286e89 100644
--- a/sweep_chat/components/shared/PulsingLoader.tsx
+++ b/sweep_chat/components/shared/PulsingLoader.tsx
@@ -1,9 +1,10 @@
 export default function PulsingLoader ({
-  size = 8
+  size = 2
 }: {
   size: number
 }) {
   return (
-    <div className={`animate-pulse rounded-full h-${size} w-${size} bg-zinc-700 mr-2`}></div>
+    <div className={`animate-pulse rounded-full bg-zinc-700 mr-2 p-4`} style={{ padding: `${size}rem` }}></div>
   )
-}
\ No newline at end of file
+}
+
diff --git a/sweep_chat/components/ui/autocomplete.tsx b/sweep_chat/components/ui/autocomplete.tsx
new file mode 100644
index 0000000000..9183890a1b
--- /dev/null
+++ b/sweep_chat/components/ui/autocomplete.tsx
@@ -0,0 +1,165 @@
+// Source: https://armand-salle.fr/post/autocomplete-select-shadcn-ui/
+
+import {
+    CommandGroup,
+    CommandItem,
+    CommandList,
+    CommandInput,
+  } from "./command"
+  import { Command as CommandPrimitive } from "cmdk"
+  import { useState, useRef, useCallback, type KeyboardEvent } from "react"
+  
+  import { Skeleton } from "./skeleton"
+  
+  import { Check } from "lucide-react"
+  import { cn } from "../../lib/utils"
+  
+  export type Option = Record<"value" | "label", string> & Record<string, string>
+  
+  type AutoCompleteProps = {
+    options: Option[]
+    emptyMessage: string
+    value?: Option
+    onValueChange?: (value: Option) => void
+    isLoading?: boolean
+    onBlur?: (value: string) => void
+    disabled?: boolean
+    placeholder?: string
+  }
+  
+  export const AutoComplete = ({
+    options,
+    placeholder,
+    emptyMessage,
+    value,
+    onValueChange,
+    onBlur,
+    disabled = false,
+    isLoading = false,
+  }: AutoCompleteProps) => {
+    const inputRef = useRef<HTMLInputElement>(null)
+  
+    const [isOpen, setOpen] = useState(false)
+    const [selected, setSelected] = useState<Option>(value as Option)
+    const [inputValue, setInputValue] = useState<string>(value?.label || "")
+  
+    const handleKeyDown = useCallback(
+      (event: KeyboardEvent<HTMLDivElement>) => {
+        const input = inputRef.current
+        if (!input) {
+          return
+        }
+  
+        // Keep the options displayed when the user is typing
+        if (!isOpen) {
+          setOpen(true)
+        }
+  
+        // This is not a default behaviour of the <input /> field
+        if (event.key === "Enter" && input.value !== "") {
+          const optionToSelect = options.find(
+            (option) => option.label === input.value,
+          )
+          if (optionToSelect) {
+            setSelected(optionToSelect)
+            onValueChange?.(optionToSelect)
+          }
+        }
+  
+        if (event.key === "Escape") {
+          input.blur()
+        }
+      },
+      [isOpen, options, onValueChange],
+    )
+  
+    const handleBlur = useCallback(() => {
+        setOpen(false)
+        onValueChange?.({
+            value: inputValue,
+            label: inputValue,
+        })
+        onBlur?.(inputValue)
+    }, [selected, inputValue])
+  
+    const handleSelectOption = useCallback(
+      (selectedOption: Option) => {
+        setInputValue(selectedOption.label)
+  
+        setSelected(selectedOption)
+        onValueChange?.(selectedOption)
+  
+        // This is a hack to prevent the input from being focused after the user selects an option
+        // We can call this hack: "The next tick"
+        setTimeout(() => {
+          inputRef?.current?.blur()
+        }, 0)
+      },
+      [onValueChange],
+    )
+  
+    return (
+      <CommandPrimitive onKeyDown={handleKeyDown} className="w-full">
+        <div>
+          <CommandInput
+            ref={inputRef}
+            value={inputValue}
+            onValueChange={isLoading ? undefined : setInputValue}
+            onBlur={handleBlur}
+            onFocus={() => setOpen(true)}
+            placeholder={placeholder}
+            disabled={disabled}
+            className="text-base"
+          />
+        </div>
+        <div className="relative mt-1">
+          <div
+            className={cn(
+              "animate-in fade-in-0 zoom-in-95 absolute top-0 z-10 w-full rounded-md bg-black outline-none border border-gray-700",
+              isOpen ? "block" : "hidden",
+            )}
+          >
+            <CommandList className="rounded-lg ring-1 ring-slate-200">
+              {isLoading ? (
+                <CommandPrimitive.Loading>
+                  <div className="p-1">
+                    <Skeleton className="h-8 w-full" />
+                  </div>
+                </CommandPrimitive.Loading>
+              ) : null}
+              {options.length > 0 && !isLoading ? (
+                <CommandGroup>
+                  {options.map((option) => {
+                    const isSelected = selected?.value === option.value
+                    return (
+                      <CommandItem
+                        key={option.value}
+                        value={option.label}
+                        onMouseDown={(event) => {
+                          event.preventDefault()
+                          event.stopPropagation()
+                        }}
+                        onSelect={() => handleSelectOption(option)}
+                        className={cn(
+                          "flex w-full items-center gap-2",
+                          !isSelected ? "pl-8" : null,
+                        )}
+                      >
+                        {isSelected ? <Check className="w-4" /> : null}
+                        {option.label}
+                      </CommandItem>
+                    )
+                  })}
+                </CommandGroup>
+              ) : null}
+              {!isLoading ? (
+                <CommandPrimitive.Empty className="select-none rounded-sm px-2 py-3 text-center text-sm border border-gray-700">
+                  {emptyMessage}
+                </CommandPrimitive.Empty>
+              ) : null}
+            </CommandList>
+          </div>
+        </div>
+      </CommandPrimitive>
+    )
+  }
\ No newline at end of file
diff --git a/sweep_chat/components/ui/skeleton.tsx b/sweep_chat/components/ui/skeleton.tsx
new file mode 100644
index 0000000000..01b8b6d4f7
--- /dev/null
+++ b/sweep_chat/components/ui/skeleton.tsx
@@ -0,0 +1,15 @@
+import { cn } from "@/lib/utils"
+
+function Skeleton({
+  className,
+  ...props
+}: React.HTMLAttributes<HTMLDivElement>) {
+  return (
+    <div
+      className={cn("animate-pulse rounded-md bg-muted", className)}
+      {...props}
+    />
+  )
+}
+
+export { Skeleton }
diff --git a/sweep_chat/package-lock.json b/sweep_chat/package-lock.json
index f3bc921c5e..b145295be6 100644
--- a/sweep_chat/package-lock.json
+++ b/sweep_chat/package-lock.json
@@ -20,7 +20,6 @@
         "@radix-ui/react-toast": "^1.1.5",
         "@sentry/nextjs": "^8.2.1",
         "@types/clarinet": "^0.12.3",
-        "clarinet": "^0.12.6",
         "class-variance-authority": "^0.7.0",
         "clsx": "^2.1.1",
         "cmdk": "^1.0.0",
@@ -37,6 +36,7 @@
         "react-markdown": "^9.0.1",
         "react-syntax-highlighter": "^15.5.0",
         "remark-gfm": "^4.0.0",
+        "skeleton": "^0.1.2",
         "tailwind-merge": "^2.3.0",
         "tailwindcss-animate": "^1.0.7",
         "usehooks-ts": "^3.1.0"
@@ -4892,16 +4892,6 @@
       "resolved": "https://registry.npmjs.org/cjs-module-lexer/-/cjs-module-lexer-1.3.1.tgz",
       "integrity": "sha512-a3KdPAANPbNE4ZUv9h6LckSl9zLsYOP4MBmhIPkRaeyybt+r4UghLvq+xw/YwUcC1gqylCkL4rdVs3Lwupjm4Q=="
     },
-    "node_modules/clarinet": {
-      "version": "0.12.6",
-      "resolved": "https://registry.npmjs.org/clarinet/-/clarinet-0.12.6.tgz",
-      "integrity": "sha512-0FR+TrvLbYHLjhzs9oeIbd3yfZmd4u2DzYQEjUTm2dNfh4Y/9RIRWPjsm3aBtrVEpjKI7+lWa4ouqEXoml84mQ==",
-      "engines": {
-        "chrome": ">=16.0.912",
-        "firefox": ">=0.8.0",
-        "node": ">=0.3.6"
-      }
-    },
     "node_modules/class-variance-authority": {
       "version": "0.7.0",
       "resolved": "https://registry.npmjs.org/class-variance-authority/-/class-variance-authority-0.7.0.tgz",
@@ -5039,6 +5029,19 @@
         "react-dom": "^18.0.0"
       }
     },
+    "node_modules/coffee-script": {
+      "version": "1.3.3",
+      "resolved": "https://registry.npmjs.org/coffee-script/-/coffee-script-1.3.3.tgz",
+      "integrity": "sha512-QjQ1T4BqyHv19k6XSfdhy/QLlIOhywz0ekBUCa9h71zYMJlfDTGan/Z1JXzYkZ6v8R+GhvL/p4FZPbPW8WNXlg==",
+      "deprecated": "CoffeeScript on NPM has moved to \"coffeescript\" (no hyphen)",
+      "bin": {
+        "cake": "bin/cake",
+        "coffee": "bin/coffee"
+      },
+      "engines": {
+        "node": ">=0.4.0"
+      }
+    },
     "node_modules/color-convert": {
       "version": "2.0.1",
       "resolved": "https://registry.npmjs.org/color-convert/-/color-convert-2.0.1.tgz",
@@ -5061,6 +5064,14 @@
       "integrity": "sha512-IfEDxwoWIjkeXL1eXcDiow4UbKjhLdq6/EuSVR9GMN7KVH3r9gQ83e73hsz1Nd1T3ijd5xv1wcWRYO+D6kCI2w==",
       "dev": true
     },
+    "node_modules/colors": {
+      "version": "0.6.2",
+      "resolved": "https://registry.npmjs.org/colors/-/colors-0.6.2.tgz",
+      "integrity": "sha512-OsSVtHK8Ir8r3+Fxw/b4jS1ZLPXkV6ZxDRJQzeD7qo0SqMXWrHDM71DgYzPMHY8SFJ0Ao+nNU2p1MmwdzKqPrw==",
+      "engines": {
+        "node": ">=0.1.90"
+      }
+    },
     "node_modules/combined-stream": {
       "version": "1.0.8",
       "resolved": "https://registry.npmjs.org/combined-stream/-/combined-stream-1.0.8.tgz",
@@ -11154,6 +11165,28 @@
         "url": "https://github.com/sponsors/isaacs"
       }
     },
+    "node_modules/skeleton": {
+      "version": "0.1.2",
+      "resolved": "https://registry.npmjs.org/skeleton/-/skeleton-0.1.2.tgz",
+      "integrity": "sha512-+7AFCihPnKe93c3qOlWbY3zTMYHe5IuQSA1azTTJ9Z6ILs9GTDx9QCd7zjaGKQtKS8QGQ3dQ+BDXDEhmOUm7vg==",
+      "dependencies": {
+        "coffee-script": "1.3.x",
+        "colors": "0.6.x",
+        "mkdirp": "0.3.x"
+      },
+      "bin": {
+        "skeleton": "bin/skeleton"
+      },
+      "engines": {
+        "node": "*"
+      }
+    },
+    "node_modules/skeleton/node_modules/mkdirp": {
+      "version": "0.3.5",
+      "resolved": "https://registry.npmjs.org/mkdirp/-/mkdirp-0.3.5.tgz",
+      "integrity": "sha512-8OCq0De/h9ZxseqzCH8Kw/Filf5pF/vMI6+BH7Lu0jXz2pqYCjTAQRolSxRIi+Ax+oCCjlxoJMP0YQ4XlrQNHg==",
+      "deprecated": "Legacy versions of mkdirp are no longer supported. Please update to mkdirp 1.x. (Note that the API surface has changed to use Promises in 1.x.)"
+    },
     "node_modules/slash": {
       "version": "3.0.0",
       "resolved": "https://registry.npmjs.org/slash/-/slash-3.0.0.tgz",
diff --git a/sweep_chat/package.json b/sweep_chat/package.json
index d6279efcd0..fcfb2733a6 100644
--- a/sweep_chat/package.json
+++ b/sweep_chat/package.json
@@ -24,7 +24,6 @@
     "@radix-ui/react-toast": "^1.1.5",
     "@sentry/nextjs": "^8.2.1",
     "@types/clarinet": "^0.12.3",
-    "clarinet": "^0.12.6",
     "class-variance-authority": "^0.7.0",
     "clsx": "^2.1.1",
     "cmdk": "^1.0.0",
@@ -41,6 +40,7 @@
     "react-markdown": "^9.0.1",
     "react-syntax-highlighter": "^15.5.0",
     "remark-gfm": "^4.0.0",
+    "skeleton": "^0.1.2",
     "tailwind-merge": "^2.3.0",
     "tailwindcss-animate": "^1.0.7",
     "usehooks-ts": "^3.1.0"

From 5c0df6e548d48fcc9df33c596bfb90c40faa6acc Mon Sep 17 00:00:00 2001
From: Kevin Lu <kevinlu1248@gmail.com>
Date: Wed, 5 Jun 2024 16:19:49 -0700
Subject: [PATCH 4/7] Border fix

---
 sweep_chat/components/ui/autocomplete.tsx | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/sweep_chat/components/ui/autocomplete.tsx b/sweep_chat/components/ui/autocomplete.tsx
index 9183890a1b..dde7e5aa03 100644
--- a/sweep_chat/components/ui/autocomplete.tsx
+++ b/sweep_chat/components/ui/autocomplete.tsx
@@ -115,11 +115,11 @@ import {
         <div className="relative mt-1">
           <div
             className={cn(
-              "animate-in fade-in-0 zoom-in-95 absolute top-0 z-10 w-full rounded-md bg-black outline-none border border-gray-700",
+              "animate-in fade-in-0 zoom-in-95 absolute top-0 z-10 w-full rounded-md bg-black outline-none border border-zinc-700",
               isOpen ? "block" : "hidden",
             )}
           >
-            <CommandList className="rounded-lg ring-1 ring-slate-200">
+            <CommandList className="rounded-lg ring-slate-200">
               {isLoading ? (
                 <CommandPrimitive.Loading>
                   <div className="p-1">

From 6fddf692b130adb247131f0095f02e81beeb3a5b Mon Sep 17 00:00:00 2001
From: MartinYe1234 <52641447+MartinYe1234@users.noreply.github.com>
Date: Wed, 5 Jun 2024 16:28:22 -0700
Subject: [PATCH 5/7] v1 (#3981)

Co-authored-by: Martin Ye <martin@sweep.dev>
---
 sweepai/core/on_comment_prompts.py      | 214 ++++++++++++++
 sweepai/core/planning_prompts.py        |  25 +-
 sweepai/core/review_utils.py            |  35 ++-
 sweepai/core/sweep_bot.py               | 363 +++++++++++++++++++++---
 sweepai/handlers/on_comment.py          |  61 ++--
 sweepai/handlers/on_ticket.py           |   5 +-
 sweepai/handlers/review_pr.py           |   2 +-
 sweepai/utils/concurrency_utils.py      |  14 +
 sweepai/utils/file_utils.py             |  35 ++-
 sweepai/utils/ticket_rendering_utils.py |  11 +-
 sweepai/utils/ticket_utils.py           |  18 --
 11 files changed, 652 insertions(+), 131 deletions(-)
 create mode 100644 sweepai/core/on_comment_prompts.py
 create mode 100644 sweepai/utils/concurrency_utils.py

diff --git a/sweepai/core/on_comment_prompts.py b/sweepai/core/on_comment_prompts.py
new file mode 100644
index 0000000000..1e5c7186d3
--- /dev/null
+++ b/sweepai/core/on_comment_prompts.py
@@ -0,0 +1,214 @@
+issue_sub_request_on_comment_system_prompt = """
+You are a talented software engineer who is responsible for addressing a comment let of a pull request. Identify every single one of the user's requests let in this comment. Be complete. The changes should be atomic.
+
+Guidelines:
+- For well-specified comments, where all required steps are already listed, simply break down the comment.
+- For less well-specified comments, where the user's requests are vague or incomplete, infer the user's intent and break down the request in the comment accordingly. This means you will need to analyze the existing changes already made in the pull request and list out all the extra changes that the user is asking for. The user may also be asking you to undo a certain change that you made.
+- A sub request should correspond to a code or test change.
+- A sub request should not be speculative, such as "catch any other errors", "abide by best practices" or "update any other code". Instead explicitly state the changes you would like to see.
+- Tests and error handling will be run automatically in the CI/CD pipeline, so do not mention them in the sub requests.
+- Topologically sort the sub requests, such that each sub request only depends on sub requests that come before it. For example, create helper functions before using them."""
+
+on_comment_pr_diffs_format = """
+Here are the changes that were made in the pull request:
+
+# PR Changes
+
+{pr_changes}
+"""
+
+# need to update to make it better at saying  things like "update any other code"
+issue_sub_request_on_comment_prompt = """\
+Break down the user comment in the pull request to identify every single one of the user's requests. Be complete. The changes should be atomic.
+
+Guidelines:
+- If it is very clear what the user comments wants, simply break down the comment into actionable steps.
+- For less well-specified user comments, where the user's requests are vague or incomplete, infer the user's intent and break down the comment into actionable steps accordingly.
+- A sub request should correspond to a code or test change.
+- A sub request should not be speculative, such as "catch any other errors", "abide by best practices" or "update any other code". Instead explicitly state the changes you would like to see.
+- Tests and error handling will be run automatically in the CI/CD pipeline, so do not mention them in the sub requests.
+- Topologically sort the sub requests, such that each sub request only depends on sub requests that come before it. For example, create helper functions before using them.
+- Make sure you are addressing the user's comment and not the original issue for the pull request
+
+First, identify the user's comment to ensure that you are not mistakenly addressing the original issue for the pull request. Respond below with the user comment:
+<user_comment>
+{{User's comment left of the pull request here. Repeat it verbatim.}}
+</user_comment>
+
+Once you have identified the user's comment you are then to create the issue sub requests. Respond in the following format:
+<issue_sub_requests>
+<issue_sub_request>
+A relevant, subtask based off of the user's comment.
+</issue_sub_request>
+<justification>
+1. Why this subtask is needed.
+2. A detailed explanation of the subtask, including the specific code entities that need to be changed.
+</justification>
+[additional sub requests as needed]
+</issue_sub_requests>"""
+
+proposed_plan_on_comment_system_prompt = """
+You are a diligent, meticulous AI assistant and will write COMPLETE code changes to resolve a user's pull request comment. They are requesting further changes to an already created pull request. Being complete means that there will be absolutely NO paraphrasing, abbreviating, or placeholder comments like "# rest of code here" or "# rest of test cases", since that is lazy. Code files, the changes already made in the pull request, the user's comment requesting further changes, and relevant parts of the codebase will be provided.
+Your role is to carefully analyze the issue and codebase, then to make the necessary code changes to resolve the issue. Reference specific files, functions, variables and code files in your plan. Organize the steps logically and break them into small, manageable tasks.
+Prioritize using existing code and functions to make efficient and maintainable changes. Ensure your suggestions fully resolve the issue.
+
+You must accomplish this task:
+# Issue Analysis: Analyze the user's comment and codebase to understand the problem. This section will vary in verbosity depending on the complexity of the user comment, but each section should be at least 1 paragraph long. In the issue analysis you provide you will detail a series of proposed changes to resolve the issue.
+"""
+
+plan_generation_steps_on_comment_system_prompt = """
+You are a diligent, meticulous AI assistant and will be responsible for creating a high level overview of how to satisfy a user's request for further changes that has been left on a pull request. Code files, the changes that have already been made in the pull request, the user's comment which you must resolve and relevant parts of the codebase will be provided.
+You will also be given a series of individual proposed changes to the pull request that are meant to tackle the user's comment. It is your responsability to now put these proposed changes into one cohesive plan. This plan should be detailed and in depth, detailing the order in which files are created and modified.
+You must strive to create the simplest plan possible with the least steps possible and lowest chance of error.
+
+You must accomplish this task:
+# Plan Generation: Given a series of proposed changes it is now your job to generate a detailed plan with all the required code changes. You will need to detail in depth the order in which files are created and modified as well as what the exact code modifications are.
+"""
+
+# anthropic prompt
+proposed_plan_on_comment_prompt = """Your job is to write a high quality, detailed, set of proposed changes in order resolve the pull request comment left by the user.
+
+You will analyze the provided code files, repository, and pull request comment to understand the requested change that the user wants to the already existing pull request. Detail a series code changes to fully resolve the user's request. The proposed changes should utilize the relevant code files and utility modules provided.
+
+Guidelines:
+<guidelines>
+- Always include the full file path and reference the provided files 
+- Prioritize using existing code and utility methods to minimize writing new code, make sure to explicitly reference any class or function names
+- Break the each task into small steps that are easy to complete
+- Make sure to only accomplish what the user is asking and nothing more, if they wanted more they can ask for it later
+<guidelines>
+
+Please use the following XML format for your response, replacing the placeholders with the appropriate information:
+
+# 1. Issue Analysis:
+<issue_analysis>
+a. Identify potential root causes and fixes for the user comment left on the pull request by referencing specific code entities in the relevant files. Then, select which of the root causes will most likely resolve the user's comment based on the current state of the pull request and codebase. (write at least 1 paragraph)
+
+b. Detail ALL of the changes that need to be made to the pull request (excluding tests) to resolve the user's comment. For each of the sub requests here write a detailed set of code changes spanning at least one change, possibly more. Be specific and direct, using the phrases "add", "replace", and "remove". Be complete and precise. You must address all the following subrequests and you must cover ALL changes that are required per sub request.
+
+# Issue Sub Requests
+
+{issue_sub_requests}
+
+Reference the provided code files, summaries, entity names, and necessary files/directories. The format should be:
+<issue_and_proposed_changes>
+<issue_sub_request>
+...
+</issue_sub_request>
+<proposed_changes>
+For each of the sub requests here, pinpoint the exact places to make changes. Describe exactly what to do, referencing specific code entities in the relevant files.
+Break the above steps up into seperate actionable steps and number them.
+Example:
+1. Step 1
+   - Step 1a
+   - Step 1b
+2. Step 2
+   - Step 2a
+   ...
+...
+</proposed_changes>
+</issue_and_proposed_changes>
+
+c. Detail ALL changes that do not correspond to an sub request from the user's issue. These changes should be necessary to resolve the issue but are not explicitly mentioned in the user's request. This code change should describe exactly what to do, referencing specific code entities in the relevant files. (optional)
+
+d. Sort the proposed changes topologically. This means that each proposed change should only depend on proposed changes that come before it. You are essentially making a TODO list.
+</issue_analysis>
+"""
+
+plan_generation_steps_on_comment_prompt = """Your job is to write a high quality, detailed, plan of how to resolve a user's comment that they have left on a pull request requesting further changes. 
+
+You will analyze the provided code files, repository, user comment and changes that have already been made to the pull request to understand the user's requested change. You have been given a series of proposed changes that are meant to implement the user's comment. 
+It is now your job to put these proposed changes together into a comprehensive and cohesive plan. This plan that you create will be given to an intern who has 0 prior knowledge of coding or the codebase. The intern will follow your plan to the letter and is incapable of thinking for themselves, so you must be extremely detailed and precise in your instructions.
+
+Below is the issue analysis and proposed changes that you will need to use to create the plan. This will tell you the thinking behind the person who made the proposed changes and what they are aiming for.
+
+# Issue Analysis and Proposed Changes
+
+{issue_analysis_and_proposed_changes}
+
+Questions to answer:
+<questions>
+1. What kind of task is this? Is it related to a new feature, a bug fix, unit tests?
+</questions>
+Respond with the following answers xml block, do this before you create the plan:
+<answers>
+{{Answers to the above question here}}
+</answers>
+
+Based on the answers to the above question go through each of the guidelines below and see which are applicable in your case. To show that you have correctly identified the correct guidelines to follow you will
+respond back with the selected guidelines.
+Guidelines to follow:
+<guidelines>
+- If the task is a new feature, the plan should not impact existing features whenever possible and should not introduce new bugs. The plan should split the required code modifications into small easy changes.
+- If the task is a bug fix, the plan should avoid making large changes and aim to keep existing logic flows intact. The plan should split the required code modifications into small easy changes.
+- If the task is related the creation/addition of new unit tests, the plan should attempt to one shot the changes. That is, generate all unit tests in one go instead of splitting the changes into multiple steps.
+- Use multiple <modify> blocks for the same file if there are multiple distinct changes to make in that file, such as for imports.
+- A <modify> block must contain exactly one change in one <new_code> tag.
+- To remove code, replace it with empty <new_code> tags.
+- Never leave todo or placeholder comments like "# rest of code" in the code. Always fully implement the changes.
+</guidelines>
+Respond with the xml format for applicable guidelines below, do this before you create the plan:
+<selected_guidelines_to_follow>
+{{List out the guidelines you will follow here}}
+</selected_guidelines_to_follow>
+
+# Plan Creation
+Now based on the selected guidelines you are to return the final plan.
+Plan: Write all necessary code changes to resolve the issue, indicating which code sections to modify and how to modify it.
+    - When modifying code you MUST do the following:
+        - First, copy the original code in <original_code> tags, copying them VERBATIM from the file. Do NOT paraphrase or abbreviate the source code. Placeholder comments like "# existing code" are not permitted. The <original_code> block must NOT be empty. Start from the last header like a function or class definition and include the entire block of code that needs to be modified.
+        - Next, write the new code in <new_code> tags, specifying necessary imports and referencing relevant type definitions, interfaces, and schemas. BE EXACT and COMPLETE as this code will replace the mentioned <original_code>.
+    - When creating files you MUST do the following:
+        - First, describe in detail EVERYTHING you will need in this file. Skip writing <original_code> tags.
+        - Next, write the new file in <new_code> tags, specifying necessary imports and referencing relevant type definitions, interfaces, and schemas. BE EXACT and COMPLETE as this file will be created in the mentioned <file_path>.
+    - In both cases, paraphrasing, abbreviating the source code, or placeholder comments such as "# rest of code" are NEVER PERMITTED. NEVER LEAVE COMMENTS DESCRIBING WHAT YOU WILL DO, JUST DO IT.
+    - Remember that when you are creating your plan, the changes will be applied IN ORDER meaning that if you create a file in the first <modify> block, you must consider the file created in second <modify> block meaning that if you are modifying the newly created file, you must now include the correct <original_code> block.
+
+Respond with the following xml format for the plan:
+<plan>  
+<modify file="file_path"> 
+Describe ALL changes to be made. Implement all the changes FULLY. This means that under no circumstances should you leave any placeholder comments like "# rest of code" or "# rest of test cases". If you find yourself doing this you MUST rewrite the plan so that this doesn't happen.
+
+1. If you are creating a file, you may skip this step. Otherwise, copy the original code into <original_code></original_code> tags, copying them VERBATIM from the file. Do NOT paraphrase or abbreviate the source code. Placeholder comments like "# existing code" are not permitted. The referenced original code span should be just enough to cover the change, with 5 extra lines above and below for context. Start from the last header like a function or class definition and include the entire block of code that needs to be modified.
+
+2. Write the new code in <new_code></new_code> tags, specifying necessary imports and referencing relevant type definitions, interfaces, and schemas. BE COMPLETE as this code will replace the mentioned <original_code></original_code>. Paraphrasing, abbreviating the source code, or placeholder comments such as "# rest of code" are NEVER PERMITTED.
+</modify>
+
+[additional modifies as needed until all proposed changes are handled, for the same file or different files]
+</plan>"""
+
+
+rename_on_comment_system_prompt = """
+You are a talented software engineer who is responsible for addressing a comment let of a pull request.
+Your only task is to identify if the user is requesting you to rename or delete a file in the comment that they left.
+"""
+
+rename_on_comment_prompt = """Your job is to handle all renames and deletions in the codebase based on the comment the user left on the pull request. Renames will be needed for any form of file or directory movement, translations or migrations.
+
+Identify all renames that would need to occur in the codebase to resolve the user's issue. Respond in the following format:
+
+<renames>
+<rename>
+<old_name>Current full file path of the file.</old_name>
+<new_name>New full file path of the file. Set to empty to delete the file.</new_name>
+</rename>
+[additional renames as needed]
+</renames>
+
+If no renames are needed, respond with an empty <renames> tag."""
+
+"""
+Respond in this format:
+
+# 1. Comprehensive error analysis:
+<thinking>
+First, for each error message in the logs:
+a. Identify the root cause.
+b. Then, spell out all lines of code that may have the same failure as the erroring lines of code.
+</thinking>
+
+# 2. Root cause analysis:
+<root_cause_analysis>
+Final response, including all areas requiring changes.
+</root_cause_analysis>
+"""
\ No newline at end of file
diff --git a/sweepai/core/planning_prompts.py b/sweepai/core/planning_prompts.py
index c7c3ed7337..054d550cb0 100644
--- a/sweepai/core/planning_prompts.py
+++ b/sweepai/core/planning_prompts.py
@@ -42,7 +42,7 @@
 """
 
 plan_generation_steps_system_prompt = """
-You are a diligent, meticulous AI assistant and will be responsible for creating a high level overview of how to solve resolve a Github Issue. Code files, a description of the issue, and relevant parts of the codebase will be provided.
+You are a diligent, meticulous AI assistant and will be responsible for creating a high level overview of how to resolve a Github Issue. Code files, a description of the issue, and relevant parts of the codebase will be provided.
 You will also be given a series of individual proposed changes to the codebase that are meant to tackle the Github Issue. It is your responsability to now put these proposed changes into one cohesive plan. This plan should be detailed and in depth, detailing the order in which files are created and modified.
 You must strive to create the simplest plan possible with the least steps possible and lowest chance of error.
 
@@ -50,16 +50,6 @@
 # Plan Generation: Given a series of proposed changes it is now your job to generate a detailed plan with all the required code changes. You will need to detail in depth the order in which files are created and modified as well as what the exact code modifications are.
 """
 
-test = """
-2. Plan: Write all necessary code changes to resolve the issue, indicating which code sections to modify and how to modify it.
-    - When modifying code you MUST do the following:
-        - First, copy the original code in <original_code> tags, copying them VERBATIM from the file. Do NOT paraphrase or abbreviate the source code. Placeholder comments like "# existing code" are not permitted. The <original_code> block must NOT be empty. Start from the last header like a function or class definition and include the entire block of code that needs to be modified.
-        - Next, write the new code in <new_code> tags, specifying necessary imports and referencing relevant type definitions, interfaces, and schemas. BE EXACT and COMPLETE as this code will replace the mentioned <original_code>.
-    - When creating files you MUST do the following:
-        - First, describe in detail EVERYTHING you will need in this file. Skip writing <original_code> tags.
-        - Next, write the new file in <new_code> tags, specifying necessary imports and referencing relevant type definitions, interfaces, and schemas. BE EXACT and COMPLETE as this file will be created in the mentioned <file_path>.
-    - In both cases, paraphrasing, abbreviating the source code, or placeholder comments such as "# rest of code" are NEVER PERMITTED. NEVER LEAVE COMMENTS DESCRIBING WHAT YOU WILL DO, JUST DO IT."""
-
 # anthropic prompt
 proposed_plan_prompt = """Your job is to write a high quality, detailed, set of proposed changes in order resolve a GitHub issue.
 
@@ -171,19 +161,6 @@
 [additional modifies as needed until all proposed changes are handled, for the same file or different files]
 </plan>"""
 
-plan_creation_prompt = """
-# 2. Plan:
-<plan>  
-<modify file="file_path"> 
-Describe ALL changes to be made.
-
-1. If you are creating a file, you may skip this step. Otherwise, copy the original code into <original_code></original_code> tags, copying them VERBATIM from the file. Do NOT paraphrase or abbreviate the source code. Placeholder comments like "# existing code" are not permitted. The referenced original code span should be just enough to cover the change, with 5 extra lines above and below for context. Start from the last header like a function or class definition and include the entire block of code that needs to be modified.
-
-2. Write the new code in <new_code></new_code> tags, specifying necessary imports and referencing relevant type definitions, interfaces, and schemas. BE COMPLETE as this code will replace the mentioned <original_code></original_code>. Paraphrasing, abbreviating the source code, or placeholder comments such as "# rest of code" are NEVER PERMITTED.
-</modify>
-
-[additional modifies as needed until all proposed changes are handled, for the same file or different files]
-</plan>"""
 
 anthropic_rename_prompt = """Your job is to handle all renames and deletions in the codebase to resolve a user's issue. Renames will be needed for any form of file or directory movement, translations or migrations.
 
diff --git a/sweepai/core/review_utils.py b/sweepai/core/review_utils.py
index 8000c18032..b51fe2b1a4 100644
--- a/sweepai/core/review_utils.py
+++ b/sweepai/core/review_utils.py
@@ -10,12 +10,10 @@
 import numpy as np
 from sklearn.cluster import DBSCAN
 from tqdm import tqdm
-from sweepai.chat.api import posthog_trace
 from sweepai.config.client import SweepConfig
 from sweepai.core.chat import ChatGPT
 from sweepai.core.entities import Message, UnsuitableFileException
 from sweepai.core.review_annotations import get_diff_annotations
-from sweepai.core.sweep_bot import safe_decode
 from sweepai.core.vector_db import cosine_similarity, embed_text_array
 from sweepai.dataclasses.codereview import CodeReview, CodeReviewByGroup, CodeReviewIssue, FunctionDef, GroupedFilesForReview, PRChange, Patch
 from sweepai.logn.cache import file_cache
@@ -24,7 +22,7 @@
 from github.Repository import Repository
 from github.PullRequest import PullRequest
 
-from sweepai.utils.file_utils import read_file_with_fallback_encodings
+from sweepai.utils.file_utils import read_file_with_fallback_encodings, safe_decode
 from sweepai.utils.github_utils import ClonedRepo, MockClonedRepo, update_file
 from sweepai.utils.hash import hash_sha256
 from sweepai.utils.str_utils import add_line_numbers, extract_object_fields_from_string, extract_objects_from_string, object_to_xml, objects_to_xml, remove_lines_from_text
@@ -327,15 +325,26 @@ def format_pr_changes(
     return formatted_changes, all_formatted_patches, all_formatted_source_code
 
 # render a group of pr changes for review
-def render_pr_changes(pr_changes: dict[str, PRChange]) -> str:
-    formatted_changes, formatted_patches, formatted_source_code = format_pr_changes(pr_changes)
+def render_pr_changes(
+    pr_changes: dict[str, PRChange],
+    truncate: bool = False,
+    include_annotations: bool = True
+) -> str:
+    formatted_changes, formatted_patches, formatted_source_code = format_pr_changes(
+        pr_changes, truncate=truncate, include_annotations=include_annotations
+    )
     if len(formatted_changes) >= MAX_CHAR_BUDGET:
+        truncate = True
         # go again with pruned code
-        formatted_changes, formatted_patches, formatted_source_code = format_pr_changes(pr_changes, truncate=True)
+        formatted_changes, formatted_patches, formatted_source_code = format_pr_changes(
+            pr_changes, truncate=truncate, include_annotations=include_annotations
+        )
     if len(formatted_changes) >= MAX_CHAR_BUDGET:
+        truncate = True
+        include_annotations = False
         # go again without annotations and prune code
         formatted_changes, formatted_patches, formatted_source_code = format_pr_changes(
-            pr_changes, truncate=True, include_annotations=False
+            pr_changes, truncate=truncate, include_annotations=include_annotations
         )
     if len(formatted_changes) >= MAX_CHAR_BUDGET:
         # simply too many changes to handle, split them up
@@ -345,14 +354,18 @@ def render_pr_changes(pr_changes: dict[str, PRChange]) -> str:
 # render all changes for all groups, handle cases where grouped files are too large
 def format_all_pr_changes_by_groups(
     grouped_files: dict[str, list[str]],
-    pr_changes: dict[str, PRChange]
+    pr_changes: dict[str, PRChange],
+    truncate: bool = False,
+    include_annotations: bool = True
 ):
     formatted_pr_changes_by_groups: dict[str, GroupedFilesForReview] = {}
     for _, files_to_review in grouped_files.items():
         # get all relevant PRChange objects
         changes_to_review = {file_name: pr_changes[file_name] for file_name in files_to_review}
         # build change overview
-        formatted_changes, formatted_patches, formatted_source_code = render_pr_changes(changes_to_review)
+        formatted_changes, formatted_patches, formatted_source_code = render_pr_changes(
+            changes_to_review, truncate=truncate, include_annotations=include_annotations
+        )
         if formatted_changes:
             group = GroupedFilesForReview(
                 file_names=files_to_review,
@@ -1216,7 +1229,6 @@ def sort_code_issues_by_severity(
         
 
 # get the best issue to return based on group vote
-@posthog_trace
 def get_group_voted_best_issue_index(
     username: str, 
     file_name: str, 
@@ -1265,7 +1277,6 @@ def get_code_reviews_for_file(
     return code_review_by_group
 
 # run 5 seperate instances of review_pr and then group the resulting issues and only take the issues that appear the majority of the time (> 3)
-@posthog_trace
 def group_vote_review_pr(
     username: str, 
     pr_changes: dict[str, PRChange], 
@@ -1416,7 +1427,6 @@ def group_vote_review_pr(
         majority_code_review_by_file[group_name] = copy.deepcopy(final_code_review)
     return majority_code_review_by_file
 
-@posthog_trace
 def review_pr_detailed_checks(
     username: str, 
     cloned_repo: ClonedRepo,
@@ -1453,7 +1463,6 @@ def get_pr_summary_from_patches(
     pr_summary = review_bot.get_pr_summary(formatted_pr_patches, chat_logger=chat_logger)
     return pr_summary
     
-@posthog_trace
 def sort_code_issues_by_severity(
     username: str, 
     code_review_by_file: dict[str, CodeReview], 
diff --git a/sweepai/core/sweep_bot.py b/sweepai/core/sweep_bot.py
index 0b6d0959a1..9e01ae7c12 100644
--- a/sweepai/core/sweep_bot.py
+++ b/sweepai/core/sweep_bot.py
@@ -1,11 +1,7 @@
-import base64
 import os
 import re
 from typing import Iterator
 
-import chardet
-
-from github.Repository import Repository
 from loguru import logger
 from networkx import Graph
 
@@ -43,6 +39,17 @@
     issue_sub_request_system_prompt,
     anthropic_rename_prompt,
 )
+from sweepai.core.on_comment_prompts import (
+    issue_sub_request_on_comment_system_prompt,
+    on_comment_pr_diffs_format,
+    rename_on_comment_prompt,
+    rename_on_comment_system_prompt,
+    issue_sub_request_on_comment_prompt,
+    proposed_plan_on_comment_system_prompt,
+    proposed_plan_on_comment_prompt,
+    plan_generation_steps_on_comment_system_prompt,
+    plan_generation_steps_on_comment_prompt
+)
 from sweepai.utils.chat_logger import ChatLogger
 # from sweepai.utils.previous_diff_utils import get_relevant_commits
 from sweepai.utils.diff import generate_diff
@@ -158,34 +165,6 @@ def parse_renames(renames_string: str):
         rename_dict[old_name.strip()] = new_name.strip()
     return rename_dict
 
-def safe_decode(
-    repo: Repository,
-    path: str,
-    *args,
-    **kwargs
-):
-    """
-    By default, this function will decode the file contents from the repo.
-    But if the file > 1MB, we will fetch the raw content and then decode it manually ourselves.
-    It's a strange bug that occurs when the file is too large and the GitHub API doesn't decode it properly and returns encoding="none".
-    Reference: https://docs.github.com/en/rest/repos/contents?apiVersion=2022-11-28#get-repository-content
-    """
-    try:
-        contents = repo.get_contents(path, *args, **kwargs)
-        if contents.encoding == "none":
-            blob = repo.get_git_blob(contents.sha)
-            detected_encoding = chardet.detect(base64.b64decode(blob.content))['encoding']
-            if detected_encoding is None:
-                return None
-            else:
-                try:
-                    return base64.b64decode(blob.content).decode(detected_encoding)
-                except UnicodeDecodeError as e:
-                    raise e
-        return contents.decoded_content.decode("utf-8")
-    except Exception as e:
-        raise e
-
 def remove_line_numbers(s: str) -> str:
     # Check if more than 50% of lines have line numbers
     # Remove line numbers with spaces after (e.g. "1: {code}")
@@ -799,6 +778,326 @@ def get_files_to_change(
 
     return [], ""
 
+
+@streamable
+def get_files_to_change_for_on_comment(
+    relevant_snippets: list[Snippet],
+    read_only_snippets: list[Snippet],
+    problem_statement: str,
+    repo_name: str,
+    cloned_repo: ClonedRepo,
+    additional_context: str = "",
+    import_graph: Graph | None = None,
+    pr_info: str = "",
+    pr_diffs: str = "",
+    chat_logger: ChatLogger = None,
+    seed: int = 0,
+    images: list[tuple[str, str, str]] | None = None
+) -> Iterator[tuple[dict[str, str], list[FileChangeRequest], str]]:
+    use_openai = False
+    problem_statement = problem_statement.strip("\n")
+    file_change_requests: list[FileChangeRequest] = []
+    messages: list[Message] = []
+    user_facing_message = ""
+    messages.append(
+        Message(role="system", content=issue_sub_request_on_comment_system_prompt, key="system")
+    )
+
+    new_relevant_snippets = []
+    new_read_only_snippets = []
+    
+    for snippet in relevant_snippets:
+        if snippet in new_relevant_snippets or snippet in new_read_only_snippets:
+            continue
+        if "test" in snippet.file_path:
+            new_read_only_snippets.append(snippet)
+        else:
+            new_relevant_snippets.append(snippet)
+    
+    relevant_snippets = new_relevant_snippets
+    read_only_snippets = new_read_only_snippets + read_only_snippets
+
+    interleaved_snippets = []
+    for i in range(max(len(relevant_snippets), len(read_only_snippets))):
+        if i < len(relevant_snippets):
+            interleaved_snippets.append(relevant_snippets[i])
+        if i < len(read_only_snippets):
+            interleaved_snippets.append(read_only_snippets[i])
+
+    interleaved_snippets = partition_snippets_if_test(interleaved_snippets, include_tests=False)
+    max_snippets = get_max_snippets(interleaved_snippets)
+    relevant_snippets = [snippet for snippet in max_snippets if any(snippet.file_path == relevant_snippet.file_path for relevant_snippet in relevant_snippets)]
+    read_only_snippets = [snippet for snippet in max_snippets if not any(snippet.file_path == relevant_snippet.file_path for relevant_snippet in relevant_snippets)]
+
+    messages.append(
+        Message(
+            role="user",
+            content=format_snippets(
+                relevant_snippets,
+                read_only_snippets,
+                problem_statement,
+            ),
+            key="relevant_snippets",
+        )
+    )
+    if additional_context:
+        messages.append(
+            Message(
+                role="user",
+                content=additional_context,
+            )
+        )
+    if pr_info:
+        messages.append(
+            Message(
+                role="user",
+                content=f"# Pr Description - use this to get an idea of what the original pull request was about:\n\n{pr_info}",
+            )
+        )
+    messages.append(
+        Message(
+            role="user",
+            content=f"# Comment left on the Pull Request - THIS IS THE USER REQUEST YOU ARE TO RESOLVE:\n<user_comment>\n{problem_statement}\n</user_comment>",
+        )
+    )
+    formatted_pr_diffs = on_comment_pr_diffs_format.format(pr_changes=pr_diffs)
+    messages.append(
+        Message(role="user", content=formatted_pr_diffs, key="pr_diffs")
+    )
+    print("messages")
+    for message in messages:
+        print(message.content + "\n\n")
+    joint_message = "\n\n".join(message.content for message in messages[1:])
+    print("messages", joint_message)
+    issue_sub_request_chat_gpt = ChatGPT(
+        messages=[
+            Message(
+                role="system",
+                content=issue_sub_request_on_comment_system_prompt,
+            ),
+        ],
+    )
+    renames_chat_gpt = ChatGPT(
+        messages=[
+            Message(
+                role="system",
+                content=rename_on_comment_system_prompt,
+            ),
+        ],
+    )
+    MODEL = "claude-3-opus-20240229"
+
+    renames_response = renames_chat_gpt.chat_anthropic(
+        content=joint_message + "\n\n" + rename_on_comment_prompt,
+        temperature=0.1,
+        images=images,
+        use_openai=use_openai,
+        seed=seed + 1,
+        stop_sequences=["</renames>"],
+        model=MODEL # haiku can troll this reponse
+    )
+    renames_dict = parse_renames(renames_response)
+    # need better validation
+    if renames_dict:
+        relevant_snippets = [Snippet(
+            file_path=renames_dict.get(snippet.file_path, snippet.file_path),
+            start=snippet.start,
+            end=snippet.end,
+            content=snippet.content,
+            score=snippet.score,
+            type_name=snippet.type_name,
+        ) for snippet in relevant_snippets]
+        read_only_snippets = [Snippet(
+            file_path=renames_dict.get(snippet.file_path, snippet.file_path),
+            start=snippet.start,
+            end=snippet.end,
+            content=snippet.content,
+            score=snippet.score,
+            type_name=snippet.type_name,
+        ) for snippet in read_only_snippets]
+        messages = [
+            message if message.key != "relevant_snippets" else Message(
+                role="user",
+                content=format_snippets(
+                    relevant_snippets,
+                    read_only_snippets,
+                    problem_statement,
+                ),
+                key="relevant_snippets",
+            ) for message in messages
+        ]
+        renames_string = "# Warning\n<warnings>\nIMPORTANT:" + "\n".join(
+            f"`{old_name}` has already been renamed to `{new_name}`" for old_name, new_name in renames_dict.items()
+        ) + "\nDo NOT include any renaming steps in your plan.\n</warnings>"
+        messages.append(
+            Message(
+                role="user",
+                content=renames_string,
+                key="renames"
+            )
+        )
+        joint_message = "\n\n".join(message.content for message in messages[1:])
+        user_facing_message += "We decided to make the following renames to help you solve the GitHub issue:\n"  +"\n".join(
+            f"Rename `{old_name}` to `{new_name}`" for old_name, new_name in renames_dict.items()
+        ) + "\n\n"
+        yield renames_dict, user_facing_message, []
+
+    issue_sub_requests = ""
+    if not use_openai:
+        issue_sub_request_response = continuous_llm_calls(
+            issue_sub_request_chat_gpt,
+            content=joint_message + "\n\n" + issue_sub_request_on_comment_prompt,
+            model=MODEL,
+            temperature=0.1,
+            images=images,
+            use_openai=use_openai,
+            seed=seed,
+            stop_sequences=["</issue_sub_requests>"],
+            response_cleanup=cleanup_fcrs,
+            MAX_CALLS=10
+        )
+        issue_sub_request_pattern = re.compile(r"<issue_sub_requests>(.*?)</issue_sub_requests>", re.DOTALL)
+        issue_sub_request_match = issue_sub_request_pattern.search(issue_sub_request_response)
+        if not issue_sub_request_match:
+            raise Exception("Failed to match issue excerpts")
+        issue_sub_requests = issue_sub_request_match.group(1)
+        issue_sub_requests = issue_sub_requests.strip("\n")
+        issue_sub_requests = re.sub(r"<justification>\n(.*?)\n</justification>\n*", "\n", issue_sub_requests, flags=re.DOTALL).strip("\n")
+
+        user_facing_message += "I'm going to follow the following steps to help you solve the GitHub issue:\n" 
+        single_issue_sub_request_pattern = re.compile(r"<issue_sub_request>(.*?)</issue_sub_request>", re.DOTALL)
+        for i, single_issue_sub_request_match in enumerate(single_issue_sub_request_pattern.finditer(issue_sub_requests)):
+            user_facing_message += f"{i + 1}. {single_issue_sub_request_match.group(1).strip()}\n"
+        user_facing_message += "\n\n"
+        yield renames_dict, user_facing_message, []
+
+    open("msg.txt", "w").write(joint_message + "\n\n" + proposed_plan_on_comment_prompt.format(issue_sub_requests=issue_sub_requests))
+    
+    chat_gpt = ChatGPT(
+        messages=[
+            Message(
+                role="system",
+                content=proposed_plan_on_comment_system_prompt,
+            ),
+        ],
+    )
+    # handle stop sequences better for multiple chained calls
+    proposed_plan_response: str = continuous_llm_calls(
+        chat_gpt,
+        content=joint_message + "\n\n" + proposed_plan_on_comment_prompt.format(issue_sub_requests=issue_sub_requests),
+        model=MODEL,
+        temperature=0.1,
+        images=images,
+        use_openai=use_openai,
+        seed=seed,
+        stop_sequences=["</issue_analysis>"],
+        response_cleanup=cleanup_fcrs,
+        MAX_CALLS=10
+    )
+    # get the issue analysis from the proposed plan response
+    issue_analysis_and_proposed_changes, failed , _ = extract_object_fields_from_string(proposed_plan_response, ["issue_analysis"])
+
+    # TODO: add error case for no issue_analysis
+
+    chat_gpt.messages= [
+        Message(
+            role="system",
+            content=plan_generation_steps_on_comment_system_prompt,
+        ),
+    ]
+    # handle stop sequences better for multiple chained calls
+    files_to_change_response: str = continuous_llm_calls(
+        chat_gpt,
+        content=joint_message + "\n\n" + plan_generation_steps_on_comment_prompt.format(
+            issue_analysis_and_proposed_changes=issue_analysis_and_proposed_changes
+        ),
+        model=MODEL,
+        temperature=0.1,
+        images=images,
+        use_openai=use_openai,
+        seed=seed,
+        stop_sequences=["</plan>"],
+        response_cleanup=cleanup_fcrs,
+        MAX_CALLS=10
+    )
+    # breakpoint()
+    relevant_modules = []
+    pattern = re.compile(r"<relevant_modules>(.*?)</relevant_modules>", re.DOTALL)
+    relevant_modules_match = pattern.search(files_to_change_response)
+    if relevant_modules_match:
+        relevant_modules = [relevant_module.strip() for relevant_module in relevant_modules_match.group(1).split("\n") if relevant_module.strip()]
+    print("relevant_modules", relevant_modules)
+    file_change_requests = []
+    try:
+        for re_match in re.finditer(
+            FileChangeRequest._regex, files_to_change_response, re.DOTALL
+        ):
+            file_change_request = FileChangeRequest.from_string(re_match.group(0))
+            file_change_request.raw_relevant_files = " ".join(relevant_modules)
+            file_change_request.filename = renames_dict.get(file_change_request.filename, file_change_request.filename)
+            file_change_requests.append(file_change_request)
+        
+        yield renames_dict, user_facing_message + "Here are the changes we decided to make. I'm currently just making some edits:\n", file_change_requests
+        
+        error_message, error_indices = get_error_message(
+            file_change_requests,
+            cloned_repo,
+            renames_dict=renames_dict,
+        )
+
+        for error_resolution_count in range(3):
+            if not error_message:
+                break
+            # todo: segment these into smaller calls to handle different edge cases
+            # delete the error messages
+            chat_gpt.messages = [message if message.role != "system" else Message(
+                content=fix_files_to_change_system_prompt,
+                role="system"
+            ) for message in chat_gpt.messages]
+            fix_attempt = continuous_llm_calls(
+                chat_gpt,
+                content=fix_files_to_change_prompt.format(
+                    error_message=error_message,
+                    allowed_indices=english_join([str(index) for index in range(len(error_indices))]),
+                ),
+                model=MODEL,
+                temperature=0.1,
+                images=images,
+                seed=seed,
+                stop_sequences=["</error_resolutions"],
+                response_cleanup=cleanup_fcrs,
+                use_openai=error_resolution_count < 2,
+            )
+            drops, matches = parse_patch_fcrs(fix_attempt)
+            for index, new_fcr in matches:
+                if index >= len(error_indices):
+                    logger.warning(f"Index {index} not in error indices")
+                    continue
+                if "COPIED_FROM_PREVIOUS_MODIFY" in new_fcr.instructions:
+                    # if COPIED_FROM_PREVIOUS_CREATE, we just need to override the filename
+                    file_change_requests[error_indices[index]].filename = renames_dict.get(new_fcr.filename, new_fcr.filename)
+                    continue
+                file_change_requests[error_indices[index]] = new_fcr
+            for drop in sorted(drops, reverse=True):
+                if drop >= len(error_indices):
+                    logger.warning(f"Index {drop} not in error indices")
+                    continue
+                file_change_requests.pop(error_indices[drop])
+            logger.debug("Old indices", error_indices)
+            error_message, error_indices = get_error_message(file_change_requests, cloned_repo, renames_dict=renames_dict)
+            logger.debug("New indices", error_indices)
+            yield renames_dict, user_facing_message + "Here are the changes we decided to make. I'm currently just making some edits:\n", file_change_requests
+            # breakpoint()
+        # breakpoint()
+
+        validate_file_change_requests(file_change_requests, cloned_repo, renames_dict=renames_dict)
+        yield renames_dict, user_facing_message + "Here are the changes we decided to make. I'm done making edits and now I'm just validating the changes using a linter to catch any mistakes like syntax errors or undefined variables:\n", file_change_requests
+        return renames_dict, file_change_requests, files_to_change_response
+    except RegexMatchError as e:
+        print("RegexMatchError", e)
+
+    return [], ""
+
 def context_get_files_to_change(
     relevant_snippets: list[Snippet],
     read_only_snippets: list[Snippet],
diff --git a/sweepai/handlers/on_comment.py b/sweepai/handlers/on_comment.py
index dccd1ef82d..30b4bbc33f 100644
--- a/sweepai/handlers/on_comment.py
+++ b/sweepai/handlers/on_comment.py
@@ -19,16 +19,20 @@
 )
 from sweepai.core.entities import MockPR, NoFilesException, Snippet, render_fcrs
 from sweepai.core.pull_request_bot import PRSummaryBot
-from sweepai.core.sweep_bot import get_files_to_change, validate_file_change_requests
+from sweepai.core.sweep_bot import get_files_to_change_for_on_comment, validate_file_change_requests
 from sweepai.handlers.create_pr import handle_file_change_requests
-from sweepai.core.review_utils import get_pr_changes
+from sweepai.core.review_utils import format_pr_info, get_pr_changes, smart_prune_file_based_on_patches
 from sweepai.utils.chat_logger import ChatLogger
+from sweepai.utils.concurrency_utils import fire_and_forget_wrapper
 from sweepai.utils.diff import generate_diff
 from sweepai.utils.event_logger import posthog
 from sweepai.utils.github_utils import ClonedRepo, commit_multi_file_changes, get_github_client, sanitize_string_for_github, validate_and_sanitize_multi_file_changes
-from sweepai.utils.str_utils import BOT_SUFFIX, FASTER_MODEL_MESSAGE, blockquote
-from sweepai.utils.ticket_rendering_utils import sweeping_gif
-from sweepai.utils.ticket_utils import center, fire_and_forget_wrapper, prep_snippets
+from sweepai.utils.str_utils import BOT_SUFFIX, FASTER_MODEL_MESSAGE, add_line_numbers, blockquote
+from sweepai.utils.ticket_rendering_utils import center, sweeping_gif
+from sweepai.utils.ticket_utils import prep_snippets
+
+from github.Repository import Repository
+from github.PullRequest import PullRequest
 
 num_of_snippets_to_query = 30
 total_number_of_snippet_tokens = 15_000
@@ -67,11 +71,10 @@ def on_comment(
         )
         organization, repo_name = repo_full_name.split("/")
         start_time = time.time()
-
         _token, g = get_github_client(installation_id)
-        repo = g.get_repo(repo_full_name)
+        repo: Repository = g.get_repo(repo_full_name)
         if pr is None:
-            pr = repo.get_pull(pr_number)
+            pr: PullRequest = repo.get_pull(pr_number)
         pr_title = pr.title
         pr_body = (
             pr.body.split("<details>\n<summary><b>🎉 Latest improvements to Sweep:")[0]
@@ -164,16 +167,6 @@ def on_comment(
 
         logger.bind(**metadata)
 
-        elapsed_time = time.time() - start_time
-        posthog.capture(
-            username,
-            "started",
-            properties={
-                **metadata,
-                "duration": elapsed_time,
-                "tracking_id": tracking_id,
-            },
-        )
         logger.info(f"Getting repo {repo_full_name}")
         # Telemetry logic end
 
@@ -223,24 +216,28 @@ def edit_comment(new_comment: str) -> None:
             )
 
             # Generate diffs for this PR
-            pr_diff_string = None
-            pr_files_modified = None
+            pr_diff_string = ""
             if pr_number:
+                pr_changes, dropped_files, unsuitable_files = get_pr_changes(repo, pr)
                 patches = []
-                pr_files_modified = {}
-                files = pr.get_files()
-                for file in files:
-                    if file.status == "modified":
+                source_codes = []
+                for file_name, pr_change in pr_changes.items():
+                    if pr_change.status == "modified":
                         # Get the entire file contents, not just the patch
-                        pr_files_modified[file.filename] = repo.get_contents(
-                            file.filename, ref=branch_name
-                        ).decoded_content.decode("utf-8")
-
+                        numbered_source_code = add_line_numbers(pr_change.new_code, start=1)
+                        pruned_source_code = smart_prune_file_based_on_patches(numbered_source_code, pr_change.patches)
+                        source_codes.append(f'<file_with_patches_applied file_name="{file_name}">\n{pruned_source_code}\n</file>')
+                        patch_changes = [patch.changes for patch in pr_change.patches]
+                        patches_string = "\n".join(patch_changes)
                         patches.append(
-                            f'<file file_path="{file.filename}">\n{file.patch}\n</file>'
+                            f'<patches file_name="{file_name}">\n{patches_string}\n</patches>'
                         )
+                # create source code string
+                source_code_string = (
+                    "<code_files_with_patches_applied>\n" + "\n".join(source_codes) + "\n</code_files_with_patches_applied>"
+                )
                 pr_diff_string = (
-                    "<files_changed>\n" + "\n".join(patches) + "\n</files_changed>"
+                    "<pr_changes>\n" + "\n".join(patches) + "\n\n# Here are all the code files with the above patches applied:\n\n" + source_code_string + "</pr_changes>"
                 )
 
             # This means it's a comment on a file
@@ -302,11 +299,13 @@ def edit_comment(new_comment: str) -> None:
             edit_comment(SWEEPING_GIF + "I just completed searching for relevant files, now I'm making changes...")
             if file_comment:
                 formatted_query = f"The user left this GitHub PR Review comment in `{pr_path}`:\n<comment>\n{comment}\n</comment>\nThis was where they left their comment on the PR:\n<review_code_chunk>\n{formatted_pr_chunk}\n</review_code_chunk>.\n\nResolve their comment."
-            renames_dict, file_change_requests, plan = get_files_to_change(
+            pull_request_info = format_pr_info(pr)
+            renames_dict, file_change_requests, plan = get_files_to_change_for_on_comment(
                 relevant_snippets=snippets,
                 read_only_snippets=[],
                 problem_statement=formatted_query,
                 repo_name=repo_name,
+                pr_info=pull_request_info,
                 pr_diffs=pr_diff_string,
                 cloned_repo=cloned_repo,
             )
diff --git a/sweepai/handlers/on_ticket.py b/sweepai/handlers/on_ticket.py
index 79b6b70708..fc60fba6c5 100644
--- a/sweepai/handlers/on_ticket.py
+++ b/sweepai/handlers/on_ticket.py
@@ -39,10 +39,11 @@
 from sweepai.handlers.create_pr import (
     handle_file_change_requests,
 )
+from sweepai.utils.concurrency_utils import fire_and_forget_wrapper
 from sweepai.utils.image_utils import get_image_contents_from_urls, get_image_urls_from_issue
 from sweepai.utils.issue_validator import validate_issue
 from sweepai.utils.prompt_constructor import get_issue_request
-from sweepai.utils.ticket_rendering_utils import add_emoji, process_summary, remove_emoji, get_payment_messages, get_comment_header, send_email_to_user, raise_on_no_file_change_requests, handle_empty_repository, delete_old_prs
+from sweepai.utils.ticket_rendering_utils import add_emoji, center, process_summary, remove_emoji, get_payment_messages, get_comment_header, send_email_to_user, raise_on_no_file_change_requests, handle_empty_repository, delete_old_prs
 from sweepai.utils.validate_license import validate_license
 from sweepai.utils.buttons import Button, ButtonList
 from sweepai.utils.chat_logger import ChatLogger
@@ -73,9 +74,7 @@
     to_branch_name
 )
 from sweepai.utils.ticket_utils import (
-    center,
     fetch_relevant_files,
-    fire_and_forget_wrapper,
 )
 
 @posthog_trace
diff --git a/sweepai/handlers/review_pr.py b/sweepai/handlers/review_pr.py
index 065bafb4d8..c361c45a01 100644
--- a/sweepai/handlers/review_pr.py
+++ b/sweepai/handlers/review_pr.py
@@ -19,9 +19,9 @@
     review_pr_detailed_checks,
     sort_code_issues_by_severity,
 )
+from sweepai.utils.concurrency_utils import fire_and_forget_wrapper
 from sweepai.utils.github_utils import ClonedRepo, get_github_client, refresh_token
 from sweepai.utils.ticket_rendering_utils import create_update_review_pr_comment
-from sweepai.utils.ticket_utils import fire_and_forget_wrapper
 from sweepai.utils.validate_license import validate_license
 from sweepai.utils.chat_logger import ChatLogger
 from sweepai.utils.event_logger import posthog
diff --git a/sweepai/utils/concurrency_utils.py b/sweepai/utils/concurrency_utils.py
new file mode 100644
index 0000000000..e241d7c6c0
--- /dev/null
+++ b/sweepai/utils/concurrency_utils.py
@@ -0,0 +1,14 @@
+def fire_and_forget_wrapper(call):
+    """
+    This decorator is used to run a function in a separate thread.
+    It does not return anything and does not wait for the function to finish.
+    It fails silently.
+    """
+
+    def wrapper(*args, **kwargs):
+        try:
+            return call(*args, **kwargs)
+        except Exception:
+            pass
+
+    return wrapper
\ No newline at end of file
diff --git a/sweepai/utils/file_utils.py b/sweepai/utils/file_utils.py
index 3336644b3a..19cd23f17f 100644
--- a/sweepai/utils/file_utils.py
+++ b/sweepai/utils/file_utils.py
@@ -1,3 +1,7 @@
+import base64
+import chardet
+from github.Repository import Repository
+
 # attempts to decode a file with the following encodings
 def read_file_with_fallback_encodings(
     file_path: str, encodings=["utf-8", "windows-1252", "iso-8859-1"]
@@ -31,4 +35,33 @@ def encode_file_with_fallback_encodings(
     encodings_string = ", ".join(encodings)
     raise UnicodeEncodeError(
         f"Could not encode the file with any of the specified encodings: {encodings_string}"
-    )
\ No newline at end of file
+    )
+
+
+def safe_decode(
+    repo: Repository,
+    path: str,
+    *args,
+    **kwargs
+):
+    """
+    By default, this function will decode the file contents from the repo.
+    But if the file > 1MB, we will fetch the raw content and then decode it manually ourselves.
+    It's a strange bug that occurs when the file is too large and the GitHub API doesn't decode it properly and returns encoding="none".
+    Reference: https://docs.github.com/en/rest/repos/contents?apiVersion=2022-11-28#get-repository-content
+    """
+    try:
+        contents = repo.get_contents(path, *args, **kwargs)
+        if contents.encoding == "none":
+            blob = repo.get_git_blob(contents.sha)
+            detected_encoding = chardet.detect(base64.b64decode(blob.content))['encoding']
+            if detected_encoding is None:
+                return None
+            else:
+                try:
+                    return base64.b64decode(blob.content).decode(detected_encoding)
+                except UnicodeDecodeError as e:
+                    raise e
+        return contents.decoded_content.decode("utf-8")
+    except Exception as e:
+        raise e
\ No newline at end of file
diff --git a/sweepai/utils/ticket_rendering_utils.py b/sweepai/utils/ticket_rendering_utils.py
index 7a6531cc0d..300fbf2a06 100644
--- a/sweepai/utils/ticket_rendering_utils.py
+++ b/sweepai/utils/ticket_rendering_utils.py
@@ -19,7 +19,6 @@
 
 
 from sweepai.agents.pr_description_bot import PRDescriptionBot
-from sweepai.chat.api import posthog_trace
 from sweepai.config.client import (
     RESTART_SWEEP_BUTTON,
     SweepConfig,
@@ -34,6 +33,7 @@
 from sweepai.handlers.on_check_suite import clean_gh_logs
 from sweepai.utils.buttons import create_action_buttons
 from sweepai.utils.chat_logger import ChatLogger
+from sweepai.utils.concurrency_utils import fire_and_forget_wrapper
 from sweepai.utils.github_utils import (
     CURRENT_USERNAME,
     get_github_client,
@@ -50,10 +50,6 @@
     sep,
     stars_suffix,
 )
-from sweepai.utils.ticket_utils import (
-    center,
-    fire_and_forget_wrapper,
-)
 from sweepai.utils.user_settings import UserSettings
 
 
@@ -105,6 +101,8 @@
 
 SWEEP_PR_REVIEW_HEADER = "# Sweep: PR Review"
 
+def center(text: str) -> str:
+    return f"<div align='center'>{text}</div>"
 
 # Add :eyes: emoji to ticket
 def add_emoji(issue: Issue, comment_id: int = None, reaction_content="eyes"):
@@ -571,7 +569,6 @@ def parse_issues_from_code_review(issue_string: str):
 
 
 # converts the list of issues inside a code_review into markdown text to display in a github comment
-@posthog_trace
 def render_code_review_issues(
     username: str,
     pr: PullRequest,
@@ -652,7 +649,6 @@ def format_code_sections(text: str) -> str:
 
 
 # turns code_review_by_file into markdown string
-@posthog_trace
 def render_pr_review_by_file(
     username: str,
     pr: PullRequest,
@@ -729,7 +725,6 @@ def render_pr_review_by_file(
 
 # handles the creation or update of the Sweep comment letting the user know that Sweep is reviewing a pr
 # returns the comment_id
-@posthog_trace
 def create_update_review_pr_comment(
     username: str,
     pr: PullRequest,
diff --git a/sweepai/utils/ticket_utils.py b/sweepai/utils/ticket_utils.py
index cf56480345..ff9459d0be 100644
--- a/sweepai/utils/ticket_utils.py
+++ b/sweepai/utils/ticket_utils.py
@@ -554,24 +554,6 @@ def fetch_relevant_files(
 SLOW_MODE = False
 SLOW_MODE = True
 
-def center(text: str) -> str:
-    return f"<div align='center'>{text}</div>"
-
-
-def fire_and_forget_wrapper(call):
-    """
-    This decorator is used to run a function in a separate thread.
-    It does not return anything and does not wait for the function to finish.
-    It fails silently.
-    """
-
-    def wrapper(*args, **kwargs):
-        try:
-            return call(*args, **kwargs)
-        except Exception:
-            pass
-
-    return wrapper
 
 if __name__ == "__main__":
     from sweepai.utils.github_utils import MockClonedRepo

From ce41a64d1090fa64a1b1f9016dbd491149940e26 Mon Sep 17 00:00:00 2001
From: Kevin Lu <kevinlu1248@gmail.com>
Date: Wed, 5 Jun 2024 16:36:32 -0700
Subject: [PATCH 6/7] Build fixes

---
 sweep_chat/components/App.tsx | 67 ++++++++++++++++++-----------------
 1 file changed, 34 insertions(+), 33 deletions(-)

diff --git a/sweep_chat/components/App.tsx b/sweep_chat/components/App.tsx
index 7c86f5baa7..9fdb5060e4 100644
--- a/sweep_chat/components/App.tsx
+++ b/sweep_chat/components/App.tsx
@@ -602,6 +602,7 @@ const parsePullRequests = async (repoName: string, message: string, octokit: Oct
       description: `The following error has occurred: ${e.message}. Sometimes, logging out and logging back in can resolve this issue.`,
       variant: "destructive"
     });
+    return []
   } 
 }
 
@@ -628,6 +629,39 @@ function App() {
   const [octokit, setOctokit] = useState<Octokit | null>(null)
   const [repos, setRepos] = useState<Repository[]>([])
 
+  useEffect(() => {
+    if (messagesContainerRef.current) {
+      const { scrollTop, scrollHeight, clientHeight } = messagesContainerRef.current;
+      if (scrollHeight - scrollTop - clientHeight < 100) {
+        messagesContainerRef.current.scrollTop = messagesContainerRef.current.scrollHeight;
+      }
+    }
+  }, [messages]);
+
+  useEffect(() => {
+    if (session) {
+      const octokit = new Octokit({auth: session.user!.accessToken})
+      setOctokit(octokit);
+      (async () => {
+        const maxPages = 5;
+        let allRepositories: Repository[] = [];
+        let page = 1;
+        let response;
+        do {
+          response = await octokit.rest.repos.listForAuthenticatedUser({
+            visibility: "all",
+            sort: "pushed",
+            per_page: 100,
+            page: page,
+          });
+          allRepositories = allRepositories.concat(response.data);
+          setRepos(allRepositories)
+          page++;
+        } while (response.data.length !== 0 && page < maxPages);
+      })()
+    }
+  }, [session?.user!.accessToken])
+
   if (session) {
     posthog.identify(
       session.user!.email!,
@@ -652,15 +686,6 @@ function App() {
     )
   }
 
-  useEffect(() => {
-    if (messagesContainerRef.current) {
-      const { scrollTop, scrollHeight, clientHeight } = messagesContainerRef.current;
-      if (scrollHeight - scrollTop - clientHeight < 100) {
-        messagesContainerRef.current.scrollTop = messagesContainerRef.current.scrollHeight;
-      }
-    }
-  }, [messages]);
-
   const lastAssistantMessageIndex = messages.findLastIndex((message) => message.role === "assistant" && message.content.trim().length > 0)
 
   const startStream = async (message: string, newMessages: Message[], snippets: Snippet[]) => {
@@ -812,30 +837,6 @@ function App() {
     });
   }
 
-  useEffect(() => {
-    if (session) {
-      const octokit = new Octokit({auth: session.user!.accessToken})
-      setOctokit(octokit);
-      (async () => {
-        const maxPages = 5;
-        let allRepositories: Repository[] = [];
-        let page = 1;
-        let response;
-        do {
-          response = await octokit.rest.repos.listForAuthenticatedUser({
-            visibility: "all",
-            sort: "pushed",
-            per_page: 100,
-            page: page,
-          });
-          allRepositories = allRepositories.concat(response.data);
-          setRepos(allRepositories)
-          page++;
-        } while (response.data.length !== 0 && page < maxPages);
-      })()
-    }
-  }, [session.user!.accessToken])
-
   return (
     <main className="flex h-screen flex-col items-center justify-between p-12">
       <Toaster />

From ef749fe1605d8471deaaaa236bcaed1ecdc4a573 Mon Sep 17 00:00:00 2001
From: Kevin Lu <kevinlu1248@gmail.com>
Date: Wed, 5 Jun 2024 16:42:33 -0700
Subject: [PATCH 7/7] Cypress fixes

---
 sweep_chat/cypress/e2e/spec.cy.ts | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/sweep_chat/cypress/e2e/spec.cy.ts b/sweep_chat/cypress/e2e/spec.cy.ts
index 912117d166..7a82a5f604 100644
--- a/sweep_chat/cypress/e2e/spec.cy.ts
+++ b/sweep_chat/cypress/e2e/spec.cy.ts
@@ -12,12 +12,12 @@ describe('sweep chat', () => {
   });
 
   it("can set a repo", () => {
-    cy.get('.grow > .flex').type("sweepai/sweep").blur()
+    cy.get("[tabindex=\"-1\"] > :nth-child(2) > .items-center > input").type("sweepai/sweep").blur()
     cy.get(':nth-child(5) > .flex', { timeout: 10000 }).should('have.attr', 'placeholder', 'Type a message...')
   })
 
   it("can stop the chat", () => {
-    cy.get('.grow > .flex').type("sweepai/sweep").blur()
+    cy.get("[tabindex=\"-1\"] > :nth-child(2) > .items-center > input").type("sweepai/sweep").blur()
     cy.get(':nth-child(5) > .flex', { timeout: 10000 }).should('have.attr', 'placeholder', 'Type a message...')
 
     cy.get(':nth-child(5) > .flex').type(testMessage + "{enter}")
@@ -30,7 +30,7 @@ describe('sweep chat', () => {
   })
 
   it("can preview pull requests", () => {
-    cy.get('.grow > .flex').type("sweepai/sweep").blur()
+    cy.get("[tabindex=\"-1\"] > :nth-child(2) > .items-center > input").type("sweepai/sweep").blur()
     cy.get(':nth-child(5) > .flex', { timeout: 10000 }).should('have.attr', 'placeholder', 'Type a message...')
 
     cy.get(':nth-child(5) > .flex').type(testPullRequestMessage + "{enter}")
@@ -38,7 +38,7 @@ describe('sweep chat', () => {
   })
 
   it("can send a message", () => {
-    cy.get('.grow > .flex').type("sweepai/sweep").blur()
+    cy.get("[tabindex=\"-1\"] > :nth-child(2) > .items-center > input").type("sweepai/sweep").blur()
     cy.get(':nth-child(5) > .flex', { timeout: 10000 }).should('have.attr', 'placeholder', 'Type a message...')
 
     cy.get(':nth-child(5) > .flex').type(testMessage + "{enter}")