@marizmelo/llm-cli 0.0.3 → 0.0.5

package/README.md

@@ -51,10 +51,18 @@ llm-cli
 
 #### Ollama (Local)
 ```bash
-# Install Ollama first, then:
+# 1. Install Ollama (https://ollama.com)
+# 2. Pull a model
+ollama pull llama3.2:latest
+
+# 3. Start llm-cli and switch to Ollama
 llm-cli
 /provider switch ollama
+
+# 4. (Optional) Switch to a different model
+/provider model <model-name>
 ```
+The default model is `llama3.2:latest`. Use `/provider model` to list available models and switch between them.
 
 #### Traditional Setup (Environment Variables)
 ```bash

package/bundle/gemini.js

@@ -139658,8 +139658,8 @@ var GIT_COMMIT_INFO, CLI_VERSION;
 var init_git_commit = __esm({
   "packages/core/dist/src/generated/git-commit.js"() {
     "use strict";
-    GIT_COMMIT_INFO = "30280aba";
-    CLI_VERSION = "0.0.2";
+    GIT_COMMIT_INFO = "65cf1981";
+    CLI_VERSION = "0.0.5";
   }
 });
 
@@ -141931,18 +141931,16 @@ var init_ollama_provider = __esm({
         this.model = config.model;
       }
       async generateContent(request3, userPromptId) {
-        const prompt = this.convertToOllamaPrompt(request3);
+        const messages = this.convertToOllamaMessages(request3);
+        const tools = this.convertToOllamaTools(request3);
         const ollamaRequest = {
           model: this.model,
-          prompt,
+          messages,
           stream: false,
-          options: {
-            temperature: request3.generationConfig?.temperature || 0,
-            top_p: request3.generationConfig?.topP || 1,
-            num_predict: request3.generationConfig?.maxOutputTokens
-          }
+          ...tools.length > 0 ? { tools } : {},
+          options: this.buildOptions(request3)
         };
-        const response = await fetch(`${this.baseUrl}/api/generate`, {
+        const response = await fetch(`${this.baseUrl}/api/chat`, {
           method: "POST",
           headers: {
             "Content-Type": "application/json"
@@ -141950,24 +141948,23 @@ var init_ollama_provider = __esm({
           body: JSON.stringify(ollamaRequest)
         });
         if (!response.ok) {
-          throw new Error(`Ollama API error: ${response.status} ${response.statusText}`);
+          const body = await response.text().catch(() => "");
+          throw new Error(`Ollama API error: ${response.status} ${response.statusText} - ${body}`);
         }
         const ollamaResponse = await response.json();
-        return this.convertFromOllamaResponse(ollamaResponse);
+        return this.convertFromOllamaChatResponse(ollamaResponse);
       }
       async generateContentStream(request3, userPromptId) {
-        const prompt = this.convertToOllamaPrompt(request3);
+        const messages = this.convertToOllamaMessages(request3);
+        const tools = this.convertToOllamaTools(request3);
         const ollamaRequest = {
           model: this.model,
-          prompt,
+          messages,
           stream: true,
-          options: {
-            temperature: request3.generationConfig?.temperature || 0,
-            top_p: request3.generationConfig?.topP || 1,
-            num_predict: request3.generationConfig?.maxOutputTokens
-          }
+          ...tools.length > 0 ? { tools } : {},
+          options: this.buildOptions(request3)
         };
-        const response = await fetch(`${this.baseUrl}/api/generate`, {
+        const response = await fetch(`${this.baseUrl}/api/chat`, {
           method: "POST",
           headers: {
             "Content-Type": "application/json"
@@ -141975,7 +141972,8 @@ var init_ollama_provider = __esm({
           body: JSON.stringify(ollamaRequest)
         });
         if (!response.ok) {
-          throw new Error(`Ollama API error: ${response.status} ${response.statusText}`);
+          const body = await response.text().catch(() => "");
+          throw new Error(`Ollama API error: ${response.status} ${response.statusText} - ${body}`);
         }
         const reader = response.body?.getReader();
         if (!reader) {
@@ -141984,17 +141982,19 @@ var init_ollama_provider = __esm({
         return this.createStreamGenerator(reader);
       }
       async countTokens(request3) {
-        const text = this.convertToOllamaPrompt(request3);
+        const messages = this.convertToOllamaMessages(request3);
+        const text = messages.map((msg) => msg.content).join("");
         const estimatedTokens = Math.ceil(text.length / 4);
         return {
           totalTokens: estimatedTokens
         };
       }
       async embedContent(request3) {
-        const prompt = this.convertToOllamaPrompt(request3);
+        const messages = this.convertToOllamaMessages(request3);
+        const text = messages.map((msg) => msg.content).join("");
         const ollamaRequest = {
           model: this.model,
-          prompt
+          prompt: text
         };
         const response = await fetch(`${this.baseUrl}/api/embeddings`, {
           method: "POST",
@@ -142021,58 +142021,231 @@ var init_ollama_provider = __esm({
       validateConfig(config) {
         return config.provider === "ollama" && !!config.model;
       }
-      convertToOllamaPrompt(request3) {
+      buildOptions(request3) {
+        const options2 = {};
+        const temp = request3.generationConfig?.temperature ?? request3.config?.temperature;
+        if (temp !== void 0)
+          options2.temperature = temp;
+        const topP = request3.generationConfig?.topP ?? request3.config?.topP;
+        if (topP !== void 0)
+          options2.top_p = topP;
+        const maxTokens = request3.generationConfig?.maxOutputTokens ?? request3.config?.maxOutputTokens;
+        if (maxTokens !== void 0)
+          options2.num_predict = maxTokens;
+        return options2;
+      }
+      convertToOllamaMessages(request3) {
+        const messages = [];
+        if (request3.systemInstruction) {
+          const systemText = request3.systemInstruction.parts?.map((part) => part.text || "").join("") || "";
+          if (systemText.trim()) {
+            messages.push({ role: "system", content: systemText });
+          }
+        }
         if (request3.contents && Array.isArray(request3.contents)) {
-          return request3.contents.map((content) => {
-            if (content.parts && Array.isArray(content.parts)) {
-              return content.parts.map((part) => {
-                if (part.text)
-                  return part.text;
-                if (part.inlineData?.data) {
-                  return "";
+          for (const content of request3.contents) {
+            if (!content.parts || !Array.isArray(content.parts))
+              continue;
+            const role = content.role || "user";
+            const functionCallParts = content.parts.filter((p) => p.functionCall);
+            const functionResponseParts = content.parts.filter((p) => p.functionResponse);
+            const textParts = content.parts.filter((p) => p.text !== void 0);
+            if (functionCallParts.length > 0) {
+              const textContent2 = textParts.map((p) => p.text).join("");
+              const toolCalls = functionCallParts.map((p) => ({
+                function: {
+                  name: p.functionCall.name,
+                  arguments: p.functionCall.args || {}
                 }
-                return "";
-              }).join("");
+              }));
+              messages.push({
+                role: "assistant",
+                content: textContent2,
+                tool_calls: toolCalls
+              });
+            } else if (functionResponseParts.length > 0) {
+              for (const part of functionResponseParts) {
+                const output = part.functionResponse.response?.output ?? JSON.stringify(part.functionResponse.response ?? {});
+                messages.push({
+                  role: "tool",
+                  content: typeof output === "string" ? output : JSON.stringify(output),
+                  name: part.functionResponse.name
+                });
+              }
+            } else {
+              const textContent2 = textParts.map((p) => p.text).join("");
+              if (textContent2.trim()) {
+                messages.push({
+                  role: role === "model" ? "assistant" : "user",
+                  content: textContent2
+                });
+              }
             }
-            return "";
-          }).join("\n");
+          }
         }
-        return "";
+        return messages;
+      }
+      convertToOllamaTools(request3) {
+        const tools = [];
+        if (request3.config?.tools) {
+          for (const tool of request3.config.tools) {
+            if (tool.functionDeclarations) {
+              for (const func of tool.functionDeclarations) {
+                tools.push({
+                  type: "function",
+                  function: {
+                    name: func.name,
+                    description: func.description,
+                    parameters: func.parameters || {}
+                  }
+                });
+              }
+            }
+          }
+        }
+        return tools;
       }
-      convertFromOllamaResponse(ollamaResponse) {
+      convertFromOllamaChatResponse(ollamaResponse) {
+        const message = ollamaResponse.message;
+        const text = message?.content || "";
+        const parts = [];
+        if (text) {
+          parts.push({ text });
+        }
+        if (message?.tool_calls) {
+          for (const toolCall of message.tool_calls) {
+            if (toolCall.function) {
+              const args = typeof toolCall.function.arguments === "string" ? JSON.parse(toolCall.function.arguments) : toolCall.function.arguments || {};
+              parts.push({
+                functionCall: {
+                  name: toolCall.function.name,
+                  args
+                }
+              });
+            }
+          }
+        }
         return {
           candidates: [
             {
               content: {
-                parts: [
-                  {
-                    text: ollamaResponse.response
-                  }
-                ]
+                parts
               },
-              // Ollama doesn't have token limits like Gemini - it just stops when complete
               finishReason: ollamaResponse.done ? "STOP" : "STOP"
             }
-          ]
+          ],
+          text,
+          functionCalls: parts.filter((p) => p.functionCall).map((p) => p.functionCall)
         };
       }
       async *createStreamGenerator(reader) {
         const decoder = new TextDecoder();
+        let buffer = "";
+        let accumulatedToolCalls = [];
+        let accumulatedText = "";
         try {
           while (true) {
             const { done, value } = await reader.read();
             if (done)
               break;
-            const chunk = decoder.decode(value);
-            const lines = chunk.split("\n").filter((line) => line.trim());
+            buffer += decoder.decode(value, { stream: true });
+            const lines = buffer.split("\n");
+            buffer = lines.pop() || "";
             for (const line of lines) {
+              const trimmedLine = line.trim();
+              if (!trimmedLine)
+                continue;
               try {
-                const ollamaResponse = JSON.parse(line);
-                yield this.convertFromOllamaResponse(ollamaResponse);
+                const chunk = JSON.parse(trimmedLine);
+                if (chunk.message?.content) {
+                  accumulatedText += chunk.message.content;
+                  yield {
+                    candidates: [
+                      {
+                        content: {
+                          parts: [{ text: chunk.message.content }]
+                        },
+                        finishReason: void 0
+                      }
+                    ],
+                    text: chunk.message.content
+                  };
+                }
+                if (chunk.message?.tool_calls) {
+                  for (const toolCall of chunk.message.tool_calls) {
+                    if (toolCall.function) {
+                      const args = typeof toolCall.function.arguments === "string" ? JSON.parse(toolCall.function.arguments) : toolCall.function.arguments || {};
+                      accumulatedToolCalls.push({
+                        function: { name: toolCall.function.name, arguments: args }
+                      });
+                    }
+                  }
+                }
+                if (chunk.done && accumulatedToolCalls.length > 0) {
+                  const parts = [];
+                  for (const tc of accumulatedToolCalls) {
+                    parts.push({
+                      functionCall: {
+                        name: tc.function.name,
+                        args: tc.function.arguments
+                      }
+                    });
+                  }
+                  yield {
+                    candidates: [
+                      {
+                        content: { parts },
+                        finishReason: "STOP"
+                      }
+                    ],
+                    text: "",
+                    functionCalls: parts.map((p) => p.functionCall)
+                  };
+                }
               } catch (e2) {
               }
             }
           }
+          if (buffer.trim()) {
+            try {
+              const chunk = JSON.parse(buffer.trim());
+              if (chunk.message?.content) {
+                yield {
+                  candidates: [
+                    {
+                      content: {
+                        parts: [{ text: chunk.message.content }]
+                      },
+                      finishReason: chunk.done ? "STOP" : void 0
+                    }
+                  ],
+                  text: chunk.message.content
+                };
+              }
+              if (chunk.done && accumulatedToolCalls.length > 0) {
+                const parts = [];
+                for (const tc of accumulatedToolCalls) {
+                  parts.push({
+                    functionCall: {
+                      name: tc.function.name,
+                      args: tc.function.arguments
+                    }
+                  });
+                }
+                yield {
+                  candidates: [
+                    {
+                      content: { parts },
+                      finishReason: "STOP"
+                    }
+                  ],
+                  text: "",
+                  functionCalls: parts.map((p) => p.functionCall)
+                };
+              }
+            } catch (e2) {
+            }
+          }
         } finally {
           reader.releaseLock();
         }
@@ -142199,7 +142372,7 @@ function createContentGeneratorConfig(config, authType) {
   return contentGeneratorConfig;
 }
 async function createContentGenerator(config, gcConfig, sessionId2) {
-  const version2 = "0.0.3";
+  const version2 = "0.0.5";
   const userAgent2 = `GeminiCLI/${version2} (${process.platform}; ${process.arch})`;
   const baseHeaders = {
     "User-Agent": userAgent2
@@ -274561,7 +274734,7 @@ async function getPackageJson() {
 // packages/cli/src/utils/version.ts
 async function getCliVersion() {
   const pkgJson = await getPackageJson();
-  return "0.0.3";
+  return "0.0.5";
 }
 
 // packages/cli/src/ui/commands/aboutCommand.ts
@@ -274613,7 +274786,7 @@ init_open();
 import process30 from "node:process";
 
 // packages/cli/src/generated/git-commit.ts
-var GIT_COMMIT_INFO2 = "03ffabb5";
+var GIT_COMMIT_INFO2 = "37f235d7";
 
 // packages/cli/src/ui/commands/bugCommand.ts
 init_dist3();
@@ -316757,6 +316930,11 @@ main().catch((error) => {
  * Copyright 2025 Google LLC
  * SPDX-License-Identifier: Apache-2.0
  */
+/**
+ * @license
+ * Copyright 2026 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
 /*! Bundled license information:
 
 react/cjs/react.production.js:

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@marizmelo/llm-cli",
-  "version": "0.0.3",
+  "version": "0.0.5",
   "engines": {
     "node": ">=20.0.0"
   },