diff options
Diffstat (limited to 'src/ollama-client.ts')
| -rw-r--r-- | src/ollama-client.ts | 499 |
1 files changed, 214 insertions, 285 deletions
diff --git a/src/ollama-client.ts b/src/ollama-client.ts index 30e4d41..be255d3 100644 --- a/src/ollama-client.ts +++ b/src/ollama-client.ts @@ -28,6 +28,152 @@ export interface ToolCallEvent { result: string; } +/** + * Approval request event for tools that require user confirmation. + */ +export interface ApprovalRequestEvent { + toolName: string; + friendlyName: string; + message: string; + args: Record<string, unknown>; +} + +export interface ModelOptions { + temperature?: number; + num_ctx?: number; + num_predict?: number; +} + +/** + * Result returned by a chat request strategy. + */ +interface ChatRequestResult { + content: string; + toolCalls: ToolCallResponse[]; +} + +/** + * A strategy function that performs a single HTTP request to the Ollama chat API. + * Different implementations handle non-streaming, mobile fallback, and desktop streaming. + */ +type ChatRequestStrategy = ( + workingMessages: ChatMessage[], +) => Promise<ChatRequestResult>; + +/** + * Options for the shared agent loop. + */ +interface AgentLoopOptions { + messages: ChatMessage[]; + tools?: OllamaToolDefinition[]; + app?: App; + onToolCall?: (event: ToolCallEvent) => void; + onApprovalRequest?: (event: ApprovalRequestEvent) => Promise<boolean>; + sendRequest: ChatRequestStrategy; +} + +/** + * System prompt injected when tools are available. + */ +const TOOL_SYSTEM_PROMPT = + "You are a helpful assistant with access to tools for interacting with an Obsidian vault. " + + "When you use the search_files tool, the results contain exact file paths. " + + "You MUST use these exact paths when calling read_file or referencing files. " + + "NEVER guess or modify file paths \u2014 always use the paths returned by search_files verbatim. " + + "Some tools (such as delete_file) require user approval before they execute. " + + "If the user declines an action, ask them why so you can better assist them."; + +/** + * Shared agent loop: injects the system prompt, calls the strategy for each + * iteration, executes tool calls, and loops until the model returns a final + * text response or the iteration cap is reached. + */ +async function chatAgentLoop(opts: AgentLoopOptions): Promise<string> { + const { messages, tools, app, onToolCall, onApprovalRequest, sendRequest } = opts; + const maxIterations = 10; + let iterations = 0; + + const workingMessages = messages.map((m) => ({ ...m })); + + // Inject system prompt when tools are available + if (tools !== undefined && tools.length > 0) { + workingMessages.unshift({ role: "system", content: TOOL_SYSTEM_PROMPT }); + } + + while (iterations < maxIterations) { + iterations++; + + const { content, toolCalls } = await sendRequest(workingMessages); + + // No tool calls — return the final content + if (toolCalls.length === 0) { + return content; + } + + // Append assistant message with tool_calls to working history + workingMessages.push({ + role: "assistant", + content, + tool_calls: toolCalls, + }); + + if (app === undefined) { + throw new Error("App reference required for tool execution."); + } + + // Execute each tool call and append results + for (const tc of toolCalls) { + const fnName = tc.function.name; + const fnArgs = tc.function.arguments; + const toolEntry = findToolByName(fnName); + + let result: string; + if (toolEntry === undefined) { + result = `Error: Unknown tool "${fnName}".`; + } else if (toolEntry.requiresApproval) { + let approved = false; + if (onApprovalRequest !== undefined) { + const message = toolEntry.approvalMessage !== undefined + ? toolEntry.approvalMessage(fnArgs) + : `Allow ${toolEntry.friendlyName}?`; + approved = await onApprovalRequest({ + toolName: fnName, + friendlyName: toolEntry.friendlyName, + message, + args: fnArgs, + }); + } + result = approved + ? await toolEntry.execute(app, fnArgs) + : `Action declined by user: ${toolEntry.friendlyName} was not approved.`; + } else { + result = await toolEntry.execute(app, fnArgs); + } + + if (onToolCall !== undefined) { + const friendlyName = toolEntry !== undefined ? toolEntry.friendlyName : fnName; + const summary = toolEntry !== undefined ? toolEntry.summarize(fnArgs) : `Called ${fnName}`; + const resultSummary = toolEntry !== undefined ? toolEntry.summarizeResult(result) : ""; + onToolCall({ toolName: fnName, friendlyName, summary, resultSummary, args: fnArgs, result }); + } + + workingMessages.push({ + role: "tool", + tool_name: fnName, + content: result, + }); + } + + // Loop continues — model sees tool results + } + + throw new Error("Tool calling loop exceeded maximum iterations."); +} + +// --------------------------------------------------------------------------- +// Utility functions +// --------------------------------------------------------------------------- + export async function testConnection(ollamaUrl: string): Promise<string> { try { const response = await requestUrl({ @@ -119,8 +265,6 @@ export async function showModel(ollamaUrl: string, model: string): Promise<Model const modelInfo = json.model_info as Record<string, unknown> | undefined; if (modelInfo !== undefined && modelInfo !== null) { - // Look for context_length in model_info - // Keys are typically "<family>.context_length" e.g. "llama.context_length" for (const key of Object.keys(modelInfo)) { if (key.endsWith(".context_length") || key === "context_length") { const val = modelInfo[key]; @@ -141,6 +285,10 @@ export async function showModel(ollamaUrl: string, model: string): Promise<Model } } +// --------------------------------------------------------------------------- +// Non-streaming chat (requestUrl, no UI callbacks) +// --------------------------------------------------------------------------- + /** * Send a chat message with optional tool-calling agent loop. * When tools are provided, the function handles the multi-turn tool @@ -155,38 +303,18 @@ export async function sendChatMessage( onToolCall?: (event: ToolCallEvent) => void, onApprovalRequest?: (event: ApprovalRequestEvent) => Promise<boolean>, ): Promise<string> { - const maxIterations = 10; - let iterations = 0; - - const workingMessages = messages.map((m) => ({ ...m })); - - // Inject a system prompt when tools are available to guide the model - if (tools !== undefined && tools.length > 0) { - const systemPrompt: ChatMessage = { - role: "system", - content: - "You are a helpful assistant with access to tools for interacting with an Obsidian vault. " + - "When you use the search_files tool, the results contain exact file paths. " + - "You MUST use these exact paths when calling read_file or referencing files. " + - "NEVER guess or modify file paths — always use the paths returned by search_files verbatim.", + const sendRequest: ChatRequestStrategy = async (workingMessages) => { + const body: Record<string, unknown> = { + model, + messages: workingMessages, + stream: false, }; - workingMessages.unshift(systemPrompt); - } - while (iterations < maxIterations) { - iterations++; + if (tools !== undefined && tools.length > 0) { + body.tools = tools; + } try { - const body: Record<string, unknown> = { - model, - messages: workingMessages, - stream: false, - }; - - if (tools !== undefined && tools.length > 0) { - body.tools = tools; - } - const response = await requestUrl({ url: `${ollamaUrl}/api/chat`, method: "POST", @@ -203,93 +331,28 @@ export async function sendChatMessage( const content = typeof msg.content === "string" ? msg.content : ""; const toolCalls = Array.isArray(msg.tool_calls) ? msg.tool_calls as ToolCallResponse[] : []; - // If no tool calls, return the final content - if (toolCalls.length === 0) { - return content; - } - - // Append assistant message with tool_calls to working history - const assistantMsg: ChatMessage = { - role: "assistant", - content, - tool_calls: toolCalls, - }; - workingMessages.push(assistantMsg); - - // Execute each tool call and append results - if (app === undefined) { - throw new Error("App reference required for tool execution."); - } - - for (const tc of toolCalls) { - const fnName = tc.function.name; - const fnArgs = tc.function.arguments; - const toolEntry = findToolByName(fnName); - - let result: string; - if (toolEntry === undefined) { - result = `Error: Unknown tool "${fnName}".`; - } else if (toolEntry.requiresApproval) { - let approved = false; - if (onApprovalRequest !== undefined) { - const message = toolEntry.approvalMessage !== undefined - ? toolEntry.approvalMessage(fnArgs) - : `Allow ${toolEntry.friendlyName}?`; - approved = await onApprovalRequest({ - toolName: fnName, - friendlyName: toolEntry.friendlyName, - message, - args: fnArgs, - }); - } - result = approved - ? await toolEntry.execute(app, fnArgs) - : `Action declined by user: ${toolEntry.friendlyName} was not approved.`; - } else { - result = await toolEntry.execute(app, fnArgs); - } - - if (onToolCall !== undefined) { - const friendlyName = toolEntry !== undefined ? toolEntry.friendlyName : fnName; - const summary = toolEntry !== undefined ? toolEntry.summarize(fnArgs) : `Called ${fnName}`; - const resultSummary = toolEntry !== undefined ? toolEntry.summarizeResult(result) : ""; - onToolCall({ toolName: fnName, friendlyName, summary, resultSummary, args: fnArgs, result }); - } - - workingMessages.push({ - role: "tool", - tool_name: fnName, - content: result, - }); - } - - // Loop continues — model sees tool results + return { content, toolCalls }; } catch (err: unknown) { if (err instanceof Error) { throw new Error(`Chat request failed: ${err.message}`); } throw new Error("Chat request failed: unknown error."); } - } + }; - throw new Error("Tool calling loop exceeded maximum iterations."); + return chatAgentLoop({ + messages, + tools, + app, + onToolCall, + onApprovalRequest, + sendRequest, + }); } -/** - * Approval request event for tools that require user confirmation. - */ -export interface ApprovalRequestEvent { - toolName: string; - friendlyName: string; - message: string; - args: Record<string, unknown>; -} - -export interface ModelOptions { - temperature?: number; - num_ctx?: number; - num_predict?: number; -} +// --------------------------------------------------------------------------- +// Streaming chat +// --------------------------------------------------------------------------- /** * Streaming chat options. @@ -343,8 +406,7 @@ async function* readNdjsonStream( /** * Send a chat message with streaming. - * Streams text chunks via onChunk callback. Supports tool-calling agent loop: - * tool execution rounds are non-streamed, only the final text response streams. + * Streams text chunks via onChunk callback. Supports tool-calling agent loop. * Returns the full accumulated response text. * * On mobile platforms, falls back to non-streaming via Obsidian's requestUrl() @@ -354,40 +416,36 @@ async function* readNdjsonStream( export async function sendChatMessageStreaming( opts: StreamingChatOptions, ): Promise<string> { - if (Platform.isMobile) { - return sendChatMessageStreamingMobile(opts); - } - return sendChatMessageStreamingDesktop(opts); + const { ollamaUrl, model, tools, app, options, onChunk, onToolCall, onApprovalRequest, onCreateBubble, abortSignal } = opts; + + const sendRequest: ChatRequestStrategy = Platform.isMobile + ? buildMobileStrategy(ollamaUrl, model, tools, options, onChunk, onCreateBubble) + : buildDesktopStreamingStrategy(ollamaUrl, model, tools, options, onChunk, onCreateBubble, abortSignal); + + return chatAgentLoop({ + messages: opts.messages, + tools, + app, + onToolCall, + onApprovalRequest, + sendRequest, + }); } /** - * Mobile fallback: uses Obsidian's requestUrl() (non-streaming) so the request + * Mobile strategy: uses Obsidian's requestUrl() (non-streaming) so the request * goes through the native networking layer and can reach localhost / LAN. + * Delivers the full response as a single chunk. */ -async function sendChatMessageStreamingMobile( - opts: StreamingChatOptions, -): Promise<string> { - const { ollamaUrl, model, messages, tools, app, options, onChunk, onToolCall, onApprovalRequest, onCreateBubble } = opts; - const maxIterations = 10; - let iterations = 0; - - const workingMessages = messages.map((m) => ({ ...m })); - - if (tools !== undefined && tools.length > 0) { - const systemPrompt: ChatMessage = { - role: "system", - content: - "You are a helpful assistant with access to tools for interacting with an Obsidian vault. " + - "When you use the search_files tool, the results contain exact file paths. " + - "You MUST use these exact paths when calling read_file or referencing files. " + - "NEVER guess or modify file paths — always use the paths returned by search_files verbatim.", - }; - workingMessages.unshift(systemPrompt); - } - - while (iterations < maxIterations) { - iterations++; - +function buildMobileStrategy( + ollamaUrl: string, + model: string, + tools: OllamaToolDefinition[] | undefined, + options: ModelOptions | undefined, + onChunk: (text: string) => void, + onCreateBubble: () => void, +): ChatRequestStrategy { + return async (workingMessages) => { onCreateBubble(); const body: Record<string, unknown> = { @@ -421,67 +479,11 @@ async function sendChatMessageStreamingMobile( const content = typeof msg.content === "string" ? msg.content : ""; const toolCalls = Array.isArray(msg.tool_calls) ? msg.tool_calls as ToolCallResponse[] : []; - // Deliver the full content as a single chunk to the UI if (content !== "") { onChunk(content); } - if (toolCalls.length === 0) { - return content; - } - - const assistantMsg: ChatMessage = { - role: "assistant", - content, - tool_calls: toolCalls, - }; - workingMessages.push(assistantMsg); - - if (app === undefined) { - throw new Error("App reference required for tool execution."); - } - - for (const tc of toolCalls) { - const fnName = tc.function.name; - const fnArgs = tc.function.arguments; - const toolEntry = findToolByName(fnName); - - let result: string; - if (toolEntry === undefined) { - result = `Error: Unknown tool "${fnName}".`; - } else if (toolEntry.requiresApproval) { - let approved = false; - if (onApprovalRequest !== undefined) { - const message = toolEntry.approvalMessage !== undefined - ? toolEntry.approvalMessage(fnArgs) - : `Allow ${toolEntry.friendlyName}?`; - approved = await onApprovalRequest({ - toolName: fnName, - friendlyName: toolEntry.friendlyName, - message, - args: fnArgs, - }); - } - result = approved - ? await toolEntry.execute(app, fnArgs) - : `Action declined by user: ${toolEntry.friendlyName} was not approved.`; - } else { - result = await toolEntry.execute(app, fnArgs); - } - - if (onToolCall !== undefined) { - const friendlyName = toolEntry !== undefined ? toolEntry.friendlyName : fnName; - const summary = toolEntry !== undefined ? toolEntry.summarize(fnArgs) : `Called ${fnName}`; - const resultSummary = toolEntry !== undefined ? toolEntry.summarizeResult(result) : ""; - onToolCall({ toolName: fnName, friendlyName, summary, resultSummary, args: fnArgs, result }); - } - - workingMessages.push({ - role: "tool", - tool_name: fnName, - content: result, - }); - } + return { content, toolCalls }; } catch (err: unknown) { if (err instanceof Error) { const msg = err.message.toLowerCase(); @@ -496,38 +498,22 @@ async function sendChatMessageStreamingMobile( } throw new Error("Chat request failed: unknown error."); } - } - - throw new Error("Tool calling loop exceeded maximum iterations."); + }; } /** - * Desktop streaming: uses native fetch() for real token-by-token streaming. + * Desktop streaming strategy: uses native fetch() for real token-by-token streaming. */ -async function sendChatMessageStreamingDesktop( - opts: StreamingChatOptions, -): Promise<string> { - const { ollamaUrl, model, messages, tools, app, options, onChunk, onToolCall, onApprovalRequest, onCreateBubble, abortSignal } = opts; - const maxIterations = 10; - let iterations = 0; - - const workingMessages = messages.map((m) => ({ ...m })); - - if (tools !== undefined && tools.length > 0) { - const systemPrompt: ChatMessage = { - role: "system", - content: - "You are a helpful assistant with access to tools for interacting with an Obsidian vault. " + - "When you use the search_files tool, the results contain exact file paths. " + - "You MUST use these exact paths when calling read_file or referencing files. " + - "NEVER guess or modify file paths — always use the paths returned by search_files verbatim.", - }; - workingMessages.unshift(systemPrompt); - } - - while (iterations < maxIterations) { - iterations++; - +function buildDesktopStreamingStrategy( + ollamaUrl: string, + model: string, + tools: OllamaToolDefinition[] | undefined, + options: ModelOptions | undefined, + onChunk: (text: string) => void, + onCreateBubble: () => void, + abortSignal?: AbortSignal, +): ChatRequestStrategy { + return async (workingMessages) => { onCreateBubble(); const body: Record<string, unknown> = { @@ -556,7 +542,7 @@ async function sendChatMessageStreamingDesktop( } if (response.body === null) { - throw new Error("Response body is null — streaming not supported."); + throw new Error("Response body is null \u2014 streaming not supported."); } const reader = response.body.getReader(); @@ -580,68 +566,11 @@ async function sendChatMessageStreamingDesktop( } } catch (err: unknown) { if (err instanceof DOMException && err.name === "AbortError") { - return content; + return { content, toolCalls: [] }; } throw err; } - if (toolCalls.length === 0) { - return content; - } - - const assistantMsg: ChatMessage = { - role: "assistant", - content, - tool_calls: toolCalls, - }; - workingMessages.push(assistantMsg); - - if (app === undefined) { - throw new Error("App reference required for tool execution."); - } - - for (const tc of toolCalls) { - const fnName = tc.function.name; - const fnArgs = tc.function.arguments; - const toolEntry = findToolByName(fnName); - - let result: string; - if (toolEntry === undefined) { - result = `Error: Unknown tool "${fnName}".`; - } else if (toolEntry.requiresApproval) { - let approved = false; - if (onApprovalRequest !== undefined) { - const message = toolEntry.approvalMessage !== undefined - ? toolEntry.approvalMessage(fnArgs) - : `Allow ${toolEntry.friendlyName}?`; - approved = await onApprovalRequest({ - toolName: fnName, - friendlyName: toolEntry.friendlyName, - message, - args: fnArgs, - }); - } - result = approved - ? await toolEntry.execute(app, fnArgs) - : `Action declined by user: ${toolEntry.friendlyName} was not approved.`; - } else { - result = await toolEntry.execute(app, fnArgs); - } - - if (onToolCall !== undefined) { - const friendlyName = toolEntry !== undefined ? toolEntry.friendlyName : fnName; - const summary = toolEntry !== undefined ? toolEntry.summarize(fnArgs) : `Called ${fnName}`; - const resultSummary = toolEntry !== undefined ? toolEntry.summarizeResult(result) : ""; - onToolCall({ toolName: fnName, friendlyName, summary, resultSummary, args: fnArgs, result }); - } - - workingMessages.push({ - role: "tool", - tool_name: fnName, - content: result, - }); - } - } - - throw new Error("Tool calling loop exceeded maximum iterations."); + return { content, toolCalls }; + }; } |