upgrade to v0.2.1

CHANGELOG.md

@@ -5,6 +5,13 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.2.1] - 2024-02-22
+
+- Added configurable timeout setting and retry mechanism to accommodate Azure OpenAI / cloud service performance variation.
+- Minor enhancement to code indentation in completion.
+- Chat function can now take into account custom API parameters configured in settings.
+- Update LICENSE template.
+
 ## [0.2.0] - 2024-02-07
 
 - Support `2023-12-01-preview` API version and enable `response_format: {"type": "json_object"}` by default in code completion

LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright [yyyy] [name of copyright owner]
+   Copyright 2024 carusyte
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

package.json

@@ -3,7 +3,7 @@
   "displayName": "GAI Choy",
   "description": "G̲enerative A̲I̲ empowered, C̲ode H̲elper O̲n Y̲our side. Yet another Copilot for coding, with built-in integration capability with Azure OpenAI models or, local LLM such as CodeShell.",
   "publisher": "carusyte",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "icon": "assets/logo.png",
   "pricing": "Free",
   "keywords": [
@@ -63,7 +63,6 @@
             "gpt-4",
             "gpt-4-32k"
           ],
-          "when": "config.GAIChoy.RunEnvForLLMs == 'Azure OpenAI'",
           "type": "string",
           "order": 4
         },
@@ -76,7 +75,6 @@
             "gpt-4",
             "gpt-4-32k"
           ],
-          "when": "config.GAIChoy.RunEnvForLLMs == 'Azure OpenAI'",
           "type": "string",
           "order": 5
         },
@@ -85,7 +83,6 @@
           "type": "null",
           "scope": "application",
           "markdownDescription": "[Set API Key](command:gaichoy.set_api_key)",
-          "when": "config.GAIChoy.RunEnvForLLMs == 'Azure OpenAI'",
           "order": 6
         },
         "GAIChoy.ApiVersion": {
@@ -99,22 +96,26 @@
             "2023-09-01-preview",
             "2023-12-01-preview"
           ],
-          "when": "config.GAIChoy.RunEnvForLLMs == 'Azure OpenAI'",
           "type": "string",
           "order": 7
         },
         "GAIChoy.ApiParameters": {
           "description": "The API parameters for Azure OpenAI. Format: key=value pairs delimited by semicolons.",
           "default": "temperature=0.2",
-          "when": "config.GAIChoy.RunEnvForLLMs == 'Azure OpenAI'",
           "type": "string",
           "order": 8
         },
+        "GAIChoy.ApiTimeout": {
+          "description": "The timeout in seconds before API call stops waiting or retrying.",
+          "type": "number",
+          "default": 60,
+          "order": 9
+        },
         "GAIChoy.AutoTriggerCompletion": {
           "description": "Whether or not to automatically trigger completion when typing.",
           "default": false,
           "type": "boolean",
-          "order": 9
+          "order": 10
         },
         "GAIChoy.AutoCompletionDelay": {
           "description": "The delay in seconds before automatic code completion triggers.",
@@ -125,7 +126,7 @@
             3
           ],
           "default": 2,
-          "order": 10
+          "order": 11
         },
         "GAIChoy.CompletionMaxTokens": {
           "description": "Maximum number of tokens for which suggestions will be displayed",
@@ -140,7 +141,7 @@
             4096
           ],
           "default": 64,
-          "order": 11
+          "order": 12
         },
         "GAIChoy.ChatMaxTokens": {
           "description": "Maximum number of tokens for which chat messages will be displayed",
@@ -154,20 +155,20 @@
             32768
           ],
           "default": 2048,
-          "order": 12
+          "order": 13
         },
         "GAIChoy.EnableDebugMessage": {
           "description": "Prints debug message to extension output.",
           "type": "boolean",
           "default": false,
-          "order": 13
+          "order": 14
         },
         "GAIChoy.ClearChatHistory": {
           "description": "Clear the chat history",
           "type": "null",
           "scope": "application",
           "markdownDescription": "[Clear chat history](command:gaichoy.clear_chat_history)",
-          "order": 14
+          "order": 15
         }
       }
     },

src/CodeShellCompletionProvider.ts

@@ -76,6 +76,13 @@ export class CodeShellCompletionProvider implements InlineCompletionItemProvider
             if (index === 0){
                 return line;
             }
+            // if the line contains leading whitespace with length equal or greater than indentation,
+            // that probably indicates the line has already been indented properly, and we can skip prepending
+            // excessive indentation in this case.
+            let leadingWhitespace = line.match(/^\s*/);
+            if (leadingWhitespace && leadingWhitespace[0].length >= indentation.length) {
+                return line;
+            }
             return indentation + line;
         });
         return indentedLines.join("\n");

src/FetchStream.ts

@@ -6,6 +6,7 @@ import ExtensionResource from "./ExtensionResource";
 export interface FetchStreamOptions {
   url: string;
   requestInit: RequestInit;
+  timeout: number;
   onmessage: (data: string) => void;
   ondone?: (data: Array<string>) => void;
   onerror?: (response: Response) => void;
@@ -14,13 +15,15 @@ export interface FetchStreamOptions {
 export class FetchStream {
   url: string;
   requestInit: RequestInit;
+  timeout: number;
   onmessage: FetchStreamOptions["onmessage"];
   ondone: FetchStreamOptions["ondone"];
   onerror: FetchStreamOptions["onerror"];
 
   constructor(options: FetchStreamOptions) {
     this.url = options.url;
     this.requestInit = options.requestInit;
+    this.timeout = options.timeout;
     this.onmessage = options.onmessage;
     this.ondone = options.ondone;
     this.onerror = options.onerror;
@@ -35,28 +38,56 @@ export class FetchStream {
     });
     const xres = ExtensionResource.instance;
     var responseMessage:Array<string> = [];
-    fetch(this.url, this.requestInit)
-      .then(response => {
-        if (response.status === 200) {
-          return response.body!;
-        } else {
-          return Promise.reject(response);
-        }
-      }).then(async (readableStream) => {
-        // read strings from readableStream chunk by chunk and feed into parser, push into responseMessage
-        for await (const chunk of readableStream) {
-          const chunkString = chunk.toString();
-          parser.feed(chunkString);
-          responseMessage.push(chunkString);
-        }
-      }).then(() => {
-        this.ondone?.(responseMessage);
-      }).catch(error => {
-        window.showErrorMessage(`${error}`);
-        window.setStatusBarMessage(`${error}`, 10000);
-        xres.errorMessage("caught error trying to fetch from " + this.url);
-        xres.errorMessage(error.message);
-        this.onerror?.(error);
-      });
+    // retry the following fetch action until success or this.timeout (seconds) is exceeded. 
+    // each attempt to fetch should not exceed the remaining wait time, which equals this.timeout - total elapsed time.
+    const startTime = Date.now();
+    let elapsedTime = 0;
+
+    const fetchWithTimeout = () => {
+      const remainingTime = this.timeout * 1000 - elapsedTime;
+      if (remainingTime <= 0) {
+        this.onerror?.(new Response(null, { status: 408, statusText: 'FetchStream timeout exceeded.' }));
+        return;
+      }
+
+      const abortController = new AbortController();
+      const id = setTimeout(() => abortController.abort(), remainingTime);
+      this.requestInit.signal = abortController.signal;
+
+      fetch(this.url, this.requestInit)
+        .then(response => {
+          clearTimeout(id);
+          if (response.status === 200) {
+            return response.body!;
+          } else {
+            throw new Error(`HTTP error! status: ${response.status}`);
+            // return Promise.reject(response);
+          }
+        }).then(async (readableStream) => {
+          // read strings from readableStream chunk by chunk and feed into parser, push into responseMessage
+          for await (const chunk of readableStream) {
+            const chunkString = chunk.toString();
+            parser.feed(chunkString);
+            responseMessage.push(chunkString);
+          }
+        }).then(() => {
+          this.ondone?.(responseMessage);
+        }).catch(error => {
+          elapsedTime = Date.now() - startTime;
+          if (elapsedTime < this.timeout * 1000) {
+            xres.errorMessage("retrying error calling " + this.url);
+            xres.errorMessage(error.message);
+            fetchWithTimeout();
+          } else {
+            window.showErrorMessage(`${error}`);
+            window.setStatusBarMessage(`${error}`, 10000);
+            xres.errorMessage("caught error trying to fetch from " + this.url);
+            xres.errorMessage(error.message);
+            this.onerror?.(error);
+          }
+        });
+    };
+
+    fetchWithTimeout();
   }
 }

src/RequestEventStream.ts

@@ -3,6 +3,7 @@ import { Agent } from "https";
 import { workspace } from "vscode";
 import { FetchStream } from "./FetchStream";
 import { ChatItem } from "./ChatMemory";
+import { AzureOAI } from "./llm/AzureOAI";
 import AbortController from "abort-controller";
 import ExtensionResource from "./ExtensionResource";
 
@@ -19,6 +20,8 @@ export async function postEventStream(prompt: string, chatList: Array<ChatItem>,
     const maxtokens = workspace.getConfiguration("GAIChoy").get("ChatMaxTokens") as number;
     const modelEnv = workspace.getConfiguration("GAIChoy").get("RunEnvForLLMs") as string;
     const allowSelfSignedCert = workspace.getConfiguration("GAIChoy").get("AllowSelfSignedCert") as boolean
+    const parameters = workspace.getConfiguration("GAIChoy").get("ApiParameters") as string;
+    const timeout = workspace.getConfiguration("GAIChoy").get("ApiTimeout") as number;
 
     // get API key from secret storage
     let api_key = await ExtensionResource.instance.getApiKey();
@@ -88,6 +91,9 @@ export async function postEventStream(prompt: string, chatList: Array<ChatItem>,
             // "max_tokens": maxtokens,
             // "stop": ["|<end>|", "|end|", "<|endoftext|>", "## human"],
         };
+
+        AzureOAI.mergeParameters(body, parameters)
+
         for (let item of chatList) {
             if (item.humanMessage.content.length > 0) {
                 // @ts-ignore
@@ -133,6 +139,7 @@ export async function postEventStream(prompt: string, chatList: Array<ChatItem>,
     new FetchStream({
         url: serverAddress + uri,
         requestInit: requestInit,
+        timeout: timeout,
         onmessage: msgCallback,
         ondone: doneCallback,
         onerror: errorCallback

src/llm/AzureOAI.ts

@@ -31,6 +31,7 @@ export class AzureOAI {
         const model = workspace.getConfiguration("GAIChoy").get("ChatModel") as string;
         const api_version = workspace.getConfiguration("GAIChoy").get("ApiVersion") as string;
         const parameters = workspace.getConfiguration("GAIChoy").get("ApiParameters") as string;
+        const timeout = workspace.getConfiguration("GAIChoy").get("ApiTimeout") as number;
 
         // get API key from secret storage
         let api_key = await ExtensionResource.instance.getApiKey();
@@ -120,12 +121,32 @@ Expected response in JSON format:
 
         ExtensionResource.instance.debugMessage("request.data: " + data)
         const uri = "/openai/deployments/" + model + "/chat/completions?api-version=" + api_version
-        const response = await axiosInstance.post<OpenAiCompletionResponse>(serverAddress + uri, data, { headers: headers });
-        ExtensionResource.instance.debugMessage("response.data: " + response.data)
-        const content = response.data.choices[0].message.content;
-        ExtensionResource.instance.debugMessage("response.data.choices[0].message.content: " + content)
-        let contentJSON = JSON.parse(this.trimTripleBackticks(content));
-        return contentJSON.generated_code;
+
+        try {
+            let startTime = Date.now();
+            let elapsedTime = 0;
+            do {
+                try {
+                    // adjust the timeout parameter at each round of retry. deduct by the elapsedTime.
+                    let postTimeout = Math.max(timeout * 1000 - elapsedTime, 0);
+                    const response = await axiosInstance.post<OpenAiCompletionResponse>(serverAddress + uri, data, { headers: headers, timeout: postTimeout });
+                    ExtensionResource.instance.debugMessage("response.data: " + response.data);
+                    const content = response.data.choices[0].message.content;
+                    ExtensionResource.instance.debugMessage("response.data.choices[0].message.content: " + content);
+                    let contentJSON = JSON.parse(this.trimTripleBackticks(content));
+                    return contentJSON.generated_code;
+                } catch (error) {
+                    ExtensionResource.instance.debugMessage("Error in code completion: " + error);
+                    if (Date.now() - startTime > timeout * 1000) {
+                        throw new Error('API timeout exceeded');
+                    }
+                }
+                elapsedTime = Date.now() - startTime;
+            } while (elapsedTime < timeout * 1000);
+        } catch (error) {
+            ExtensionResource.instance.debugMessage("Final Error: " + error);
+            throw error;
+        }
     }
 
     static trimTripleBackticks(str: string){