Merge branch 'release/v0.8.0'

u1and0 · Dec 30, 2024 · c8f3986 · c8f3986
2 parents a32264e + bf4e304
commit c8f3986
Show file tree

Hide file tree

Showing 12 changed files with 535 additions and 203 deletions.
diff --git a/.github/workflows/deno.yml b/.github/workflows/deno.yml
@@ -38,10 +38,10 @@ jobs:
         uses: actions/checkout@v4
 
       - name: Setup Deno
-        uses: denoland/setup-deno@v1
+        uses: denoland/setup-deno@v2
         # uses: denoland/setup-deno@61fe2df320078202e33d7d5ad347e7dcfa0e8f31  # v1.1.2
         with:
-          deno-version: v1.x
+          deno-version: v2.x.x
 
       # Uncomment this step to verify the use of 'deno fmt' on each commit.
       - name: Verify formatting
@@ -63,25 +63,35 @@ jobs:
     strategy:
       matrix:
         include:
-          - name: linux
+          - name: linux-x64
             vm: ubuntu-latest
-          - name: macosx
+            target: x86_64-unknown-linux-gnu
+          - name: linux-arm64
+            vm: ubuntu-latest
+            target: aarch64-unknown-linux-gnu
+          - name: macosx-x64
+            vm: macos-latest
+            target: x86_64-apple-darwin
+          - name: macosx-arm64
             vm: macos-latest
-          - name: windows
+            target: aarch64-apple-darwin
+          - name: windows-x64
             vm: windows-latest
+            target: x86_64-pc-windows-msvc
+
 
     steps:
       - name: Setup repo
         uses: actions/checkout@v4
 
       - name: Setup Deno
-        uses: denoland/setup-deno@v1
+        uses: denoland/setup-deno@v2
         # uses: denoland/setup-deno@61fe2df320078202e33d7d5ad347e7dcfa0e8f31  # v1.1.2
         with:
-          deno-version: v1.x
+          deno-version: v2.x.x
 
       - name: Build binary
-        run: deno compile --allow-net --allow-env --output gpt index.ts
+        run: deno compile --allow-net --allow-env --target ${{ matrix.target }} --output gpt index.ts
 
       - name: Binary upload (Unix, MacOS)
         # Artifact upload only occurs when tag matches

diff --git a/README.md b/README.md
@@ -1,4 +1,4 @@
-<img src="https://img.shields.io/badge/version-v0.7.0-FF7777.svg"></img>
+<img src="https://img.shields.io/badge/version-v0.8.0-FF7777.svg"></img>
 <img src="https://img.shields.io/badge/LICENSE-MIT-3388FF.svg"></img>
 <img src="https://shield.deno.dev/deno/%5E1.39"></img>
 <img src="https://github.com/u1and0/gpt-cli/actions/workflows/deno.yml/badge.svg"></img>
@@ -10,10 +10,16 @@ Command-line interface  that enables interactive chat with LLMs.
 # Quick start
 
 ```
-$ curl -LO https://github.com/u1and0/gpt-cli/releases/download/v0.7.0/gpt-cli-linux.zip
+# Install binary
+$ curl -LO https://github.com/u1and0/gpt-cli/releases/download/v0.8.0/gpt-cli-linux.zip
 $ unzip gpt-cli.zip
 $ chmod 755 gpt
 $ sudo ln -s ./gpt /usr/bin
+
+# Setup API key
+$ export OPENAI_API_KEY='sk-*****'
+
+# Run
 $ gpt
 Ctrl-D to confirm input, q or exit to end chat
 You: hi
@@ -46,7 +52,7 @@ You have 3 options.
 The simplest way.
 
 ```
-$ curl -LO https://github.com/u1and0/gpt-cli/releases/download/v0.7.0/gpt-cli-linux.zip
+$ curl -LO https://github.com/u1and0/gpt-cli/releases/download/v0.8.0/gpt-cli-linux.zip
 $ unzip gpt-cli-linux.zip
 $ chmod 755 gpt
 $ sudo ln -s ./gpt /usr/bin
@@ -104,14 +110,30 @@ export OPENAI_API_KEY='sk-*****'
 export ANTHROPIC_API_KEY='sk-ant-*****'
 ```
 
-### Goolgle API (for Gemini)
+### Google API (for Gemini)
 
 [Get Google API key](https://aistudio.google.com/app/apikey), then set environment argument.
 
 ```
 export GOOGLE_API_KEY='*****'
 ```
 
+### Groq API
+
+[Get Groq API key](https://console.groq.com/keys), then set environment argument.
+
+```
+export GROQ_API_KEY='*****'
+```
+
+### Together AI API
+
+[Get Together AI API key](https://api.together.xyz/settings/api-keys), then set environment argument.
+
+```
+export TOGETHER_AI_API_KEY='*****'
+```
+
 ### Replicate API (for Open Models)
 
 [Get Replicate API token](https://replicate.com/account/api-tokens), then set environment argument.
@@ -138,7 +160,7 @@ $ gpt -m gpt-4o-mini -x 1000 -t 1.0 [OPTIONS] PROMPT
 |--------------|-------------|------|----|
 | -v | --version | boolean | Show version |
 | -h | --help | boolean | Show this message |
-| -m | --model | string | OpenAI, Anthropic, Google, Replicate, Ollama model (default gpt-4o-mini) |
+| -m | --model | string | LLM model (default gpt-4o-mini) |
 | -x | --max\_tokens | number | Number of AI answer tokens (default 1000) |
 | -t | --temperature | number | Higher number means more creative answers, lower number means more exact answers (default 1.0) |
 | -u | --url | string | URL and port number for ollama server |
@@ -153,38 +175,54 @@ A Questions for Model
 ## Models
 - [OpenAI](https://platform.openai.com/docs/models)
     - gpt-4o-mini
-    - gpt-4o...
+    - gpt-4o
+    - o1
+    - o1-preview
+    - o1-mini...
 - [Anthropic](https://docs.anthropic.com/claude/docs/models-overview)
     - claude-3-opus-20240229
-    - claude-3-haiku-20240307
-    - claude-instant-1.2...
+    - claude-3-5-sonnet-latest
+    - claude-3-5-haiku-latest
 - [Gemini](https://ai.google.dev/gemini-api/docs/models/gemini)
     - gemini-1.5-pro-latest
-    - gemini-pro
+    - gemini-2.0-flash-exp...
+- [Groq](https://console.groq.com/docs/models)
+    - groq/llama3-groq-70b-8192-tool-use-preview
+    - groq/llama-3.3-70b-specdec
+    - groq/llama3.1-70b-specdec
+    - groq/llama-3.2-1b-preview
+    - groq/llama-3.2-3b-preview...
+- [TogetherAI](https://api.together.ai/models)
+    - togetherai/meta-llama/Llama-3.3-70B-Instruct-Turbo
+    - togetherai/Qwen/QwQ-32B-Preview
+    - togetherai/meta-llama/Llama-3.1-405B-Instruct-Turbo
+    - togetherai/google/gemma-2-27b-it
+    - togetherai/mistralai/Mistral-7B-Instruct-v0.3...
 - [Replicate](https://replicate.com/models)
-    - meta/meta-llama-3-70b-instruct
-    - meta/llama-2-7b-chat
-    - mistralai/mistral-7b-instruct-v0.2
-    - mistralai/mixtral-8x7b-instruct-v0.1
-    - snowflake/snowflake-arctic-instruct
-    - replicate/flan-t5-xl...
+    - replicate/meta/meta-llama-3-70b-instruct
+    - replicate/meta/llama-2-7b-chat
+    - replicate/mistralai/mistral-7b-instruct-v0.2
+    - replicate/mistralai/mixtral-8x7b-instruct-v0.1
+    - replicate/snowflake/snowflake-arctic-instruct
+    - replicate/replicate/flan-t5-xl...
 - [Ollama](https://ollama.com/library)  ** Using before "$ ollama serve" locally **
-    - phi3
-    - llama3:70b
-    - mixtral:8x7b-text-v0.1-q5\_K\_M...
+    - ollama/phi3
+    - ollama/llama3:70b
+    - ollama/mixtral:8x7b-text-v0.1-q5\_K\_M...
 
 ## / command
 Help (/commands):
 
-- /?, /help       Help for a command
-- /clear          Clear session context
-- /bye            Exit
+- /?, /help         Help for a command
+- /clear            Clear session context
+- /modelStack       Show model's history
+- /bye,/exit,/quit  Exit
 
 ## @ command
 Help (@commands): Change model while asking.
 
-- @{ModelName}    Change LLM model
-- ex)   @gemini-1.5-pro any prompt...
+- @ModelName    Change LLM model
+- ex)   @gemini-1.5-pro your question...
 
 ## Test
 
@@ -215,12 +253,12 @@ hook_add = '''
     " Create test code
     command! -nargs=0 -range GPTGenerateTest <line1>,<line2>call gptcli#GPT('You are the best code tester. Please write test code that covers all cases to try the given code.', { "temperature": 0.5, "model": "claude-3-haiku-20240307" })
     command! -nargs=0 -range GPTErrorBustor <line1>,<line2>call gptcli#GPT('Your task is to analyze the provided code snippet, identify any bugs or errors present, and provide a corrected version of the code that resolves these issues. Explain the problems you found in the original code and how your fixes address them. The corrected code should be functional, efficient, and adhere to best practices in programming.', {"temperature": 0.5, "model": "claude-3-sonnet-20240229"})
-    command! -nargs=0 -range GPTCodeOptimizer <line1>,<line2>call gptcli#GPT("Your task is to analyze the provided code snippet and suggest improvements to optimize its performance. Identify areas where the code can be made more efficient, faster, or less resource-intensive. Provide specific suggestions for optimization, along with explanations of how these changes can enhance the code performance. The optimized code should maintain the same functionality as the original code while demonstrating improved efficiency.", { "model": "meta/meta-llama-3-70b-instruct" })
+    command! -nargs=0 -range GPTCodeOptimizer <line1>,<line2>call gptcli#GPT("Your task is to analyze the provided code snippet and suggest improvements to optimize its performance. Identify areas where the code can be made more efficient, faster, or less resource-intensive. Provide specific suggestions for optimization, along with explanations of how these changes can enhance the code performance. The optimized code should maintain the same functionality as the original code while demonstrating improved efficiency.", { "model": "replicate/meta/meta-llama-3-70b-instruct" })
 
     " Any system prompt
     command! -nargs=? -range GPTComplete <line1>,<line2>call gptcli#GPT(<q-args>, { "model": "claude-3-haiku-20240307" })
     " Chat with GPT
-    command! -nargs=? GPTChat call gptcli#GPTWindow(<q-args>, { "model": "phi3:instruct", "url": "http://localhost:11434"})
+    command! -nargs=? GPTChat call gptcli#GPTWindow(<q-args>, { "model": "ollama/phi3:instruct", "url": "http://localhost:11434"})
 ```
 
 ![Peek 2024-04-01 03-35.gif](https://qiita-image-store.s3.ap-northeast-1.amazonaws.com/0/113494/f243f19b-ee47-9821-5899-7ed2acc17320.gif)

diff --git a/deno.json b/deno.json
@@ -4,5 +4,10 @@
   },
   "test": {
     "exclude": ["node_modules/"]
-  }
+  },
+  "dependencies": {
+      "@langchain/groq": "^0.1.2"
+  },
+  "nodeModulesDir": "auto"
 }
+
diff --git a/index.ts b/index.ts
@@ -25,9 +25,21 @@ import { commandMessage, helpMessage } from "./lib/help.ts";
 import { LLM, Message } from "./lib/llm.ts";
 import { getUserInputInMessage, readStdin } from "./lib/input.ts";
 import { Params, parseArgs } from "./lib/params.ts";
-import { Command, extractAtModel, isCommand } from "./lib/command.ts";
-
-const VERSION = "v0.7.0";
+import {
+  Command,
+  extractAtModel,
+  handleSlashCommand,
+  isAtCommand,
+  isSlashCommand,
+  modelStack,
+} from "./lib/command.ts";
+
+const VERSION = "v0.8.0";
+
+/** 灰色のテキストで表示 */
+function consoleInfoWithGrayText(s: string): void {
+  console.info(`\x1b[90m${s}\x1b[0m`);
+}
 
 const llmAsk = async (params: Params) => {
   params.debug && console.debug(params);
@@ -48,51 +60,46 @@ const llmAsk = async (params: Params) => {
     }
 
     // 対話的回答
+    consoleInfoWithGrayText(commandMessage);
     while (true) {
       // ユーザーからの入力待ち
-      let humanMessage = await getUserInputInMessage(messages);
-
-      /** /commandを実行する
-       * Help: ヘルプメッセージを出力する
-       * Clear: systemp promptを残してコンテキストを削除する
-       * Bye: コマンドを終了する
-       */
-      if (isCommand(humanMessage)) {
-        switch (humanMessage) {
-          case Command.Help: {
-            console.log(commandMessage);
-            continue; // Slashコマンドを処理したら次のループへ
-          }
-          case Command.Clear: {
-            // system promptが設定されていれば、それを残してコンテキストクリア
-            console.log("Context clear successful");
-            messages = params.systemPrompt
-              ? [new SystemMessage(params.systemPrompt)]
-              : [];
-            continue; // Slashコマンドを処理したら次のループへ
-          }
-          case Command.Bye: {
-            Deno.exit(0);
-          }
-        }
-      } else if (humanMessage?.content.toString().startsWith("@")) {
+      let humanMessage: HumanMessage | Command = await getUserInputInMessage(
+        messages,
+      );
+
+      // /commandを実行する
+      if (isSlashCommand(humanMessage)) {
+        messages = handleSlashCommand(humanMessage, messages);
+        continue;
+      } else if (isAtCommand(humanMessage)) {
         // @Model名で始まるinput はllmモデルを再指定する
         const { model, message } = extractAtModel(
           humanMessage.content.toString(),
         );
         // モデル名指定以外のプロンプトがなければ前のプロンプトを引き継ぐ。
-        humanMessage = message ? message : messages.at(-2);
+        // 前のプロンプトもなければ空のHumanMessageを渡す
+        humanMessage = message || messages.at(-2) || new HumanMessage("");
+
+        // @コマンドで指定したモデルのパースに成功したら
+        // モデルスタックに追加して新しいモデルで会話を始める。
+        // パースに失敗したら、以前のモデルを復元してエラー表示して
+        // 前のモデルに戻して会話を継続。
         if (model) {
+          const modelBackup = params.model;
           params.model = model;
-          llm = new LLM(params);
+          try {
+            llm = new LLM(params);
+          } catch (error: unknown) {
+            console.error(error);
+            params.model = modelBackup;
+            continue;
+          }
+          modelStack.push(model);
         }
       }
 
-      // 最後のメッセージがHumanMessageではない場合
       // ユーザーからの問いを追加
-      if (humanMessage) {
-        messages.push(humanMessage);
-      }
+      messages.push(humanMessage);
       // console.debug(messages);
       // AIからの回答を追加
       const aiMessage = await llm.ask(messages);
@@ -119,6 +126,8 @@ const main = async () => {
     Deno.exit(0);
   }
 
+  // modelStackに使用した最初のモデルを追加
+  modelStack.push(params.model);
   // 標準入力をチェック
   const stdinContent: string | null = await readStdin();
   if (stdinContent) {