Package restructure and API changes, several fixes

- More emphasis on `api` package. It now holds database model structs
  from `lmcli/models` (which is now gone) as well as the tool spec,
  call, and result types. `tools.Tool` is now `api.ToolSpec`.
  `api.ChatCompletionClient` was renamed to
  `api.ChatCompletionProvider`.

- Change ChatCompletion interface and implementations to no longer do
  automatic tool call recursion - they simply return a ToolCall message
  which the caller can decide what to do with (e.g. prompt for user
  confirmation before executing)

- `api.ChatCompletionProvider` functions have had their ReplyCallback
  parameter removed, as now they only return a single reply.

- Added a top-level `agent` package, moved the current built-in tools
  implementations under `agent/toolbox`. `tools.ExecuteToolCalls` is now
  `agent.ExecuteToolCalls`.

- Fixed request context handling in openai, google, ollama (use
  `NewRequestWithContext`), cleaned up request cancellation in TUI

- Fix tool call tui persistence bug (we were skipping message with empty
  content)

- Now handle tool calling from TUI layer

TODO:
- Prompt users before executing tool calls
- Automatically send tool results to the model (or make this toggleable)

pkg/api/provider/ollama/ollama.go

@@ -11,7 +11,6 @@ import (
 	"strings"
 
 	"git.mlow.ca/mlow/lmcli/pkg/api"
-	"git.mlow.ca/mlow/lmcli/pkg/lmcli/model"
 )
 
 type OllamaClient struct {
@@ -43,8 +42,8 @@ type OllamaResponse struct {
 }
 
 func createOllamaRequest(
-	params model.RequestParameters,
-	messages []model.Message,
+	params api.RequestParameters,
+	messages []api.Message,
 ) OllamaRequest {
 	requestMessages := make([]OllamaMessage, 0, len(messages))
 
@@ -64,11 +63,11 @@ func createOllamaRequest(
 	return request
 }
 
-func (c *OllamaClient) sendRequest(ctx context.Context, req *http.Request) (*http.Response, error) {
+func (c *OllamaClient) sendRequest(req *http.Request) (*http.Response, error) {
 	req.Header.Set("Content-Type", "application/json")
 
 	client := &http.Client{}
-	resp, err := client.Do(req.WithContext(ctx))
+	resp, err := client.Do(req)
 	if err != nil {
 		return nil, err
 	}
@@ -83,12 +82,11 @@ func (c *OllamaClient) sendRequest(ctx context.Context, req *http.Request) (*htt
 
 func (c *OllamaClient) CreateChatCompletion(
 	ctx context.Context,
-	params model.RequestParameters,
-	messages []model.Message,
-	callback api.ReplyCallback,
-) (string, error) {
+	params api.RequestParameters,
+	messages []api.Message,
+) (*api.Message, error) {
 	if len(messages) == 0 {
-		return "", fmt.Errorf("Can't create completion from no messages")
+		return nil, fmt.Errorf("Can't create completion from no messages")
 	}
 
 	req := createOllamaRequest(params, messages)
@@ -96,46 +94,40 @@ func (c *OllamaClient) CreateChatCompletion(
 
 	jsonData, err := json.Marshal(req)
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 
-	httpReq, err := http.NewRequest("POST", c.BaseURL+"/chat", bytes.NewBuffer(jsonData))
+	httpReq, err := http.NewRequestWithContext(ctx, "POST", c.BaseURL+"/chat", bytes.NewBuffer(jsonData))
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 
-	resp, err := c.sendRequest(ctx, httpReq)
+	resp, err := c.sendRequest(httpReq)
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 	defer resp.Body.Close()
 
 	var completionResp OllamaResponse
 	err = json.NewDecoder(resp.Body).Decode(&completionResp)
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 
-	content := completionResp.Message.Content
-	if callback != nil {
-		callback(model.Message{
-			Role:    model.MessageRoleAssistant,
-			Content: content,
-		})
-	}
-
-	return content, nil
+	return &api.Message{
+		Role:    api.MessageRoleAssistant,
+		Content: completionResp.Message.Content,
+	}, nil
 }
 
 func (c *OllamaClient) CreateChatCompletionStream(
 	ctx context.Context,
-	params model.RequestParameters,
-	messages []model.Message,
-	callback api.ReplyCallback,
+	params api.RequestParameters,
+	messages []api.Message,
 	output chan<- api.Chunk,
-) (string, error) {
+) (*api.Message, error) {
 	if len(messages) == 0 {
-		return "", fmt.Errorf("Can't create completion from no messages")
+		return nil, fmt.Errorf("Can't create completion from no messages")
 	}
 
 	req := createOllamaRequest(params, messages)
@@ -143,17 +135,17 @@ func (c *OllamaClient) CreateChatCompletionStream(
 
 	jsonData, err := json.Marshal(req)
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 
-	httpReq, err := http.NewRequest("POST", c.BaseURL+"/chat", bytes.NewBuffer(jsonData))
+	httpReq, err := http.NewRequestWithContext(ctx, "POST", c.BaseURL+"/chat", bytes.NewBuffer(jsonData))
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 
-	resp, err := c.sendRequest(ctx, httpReq)
+	resp, err := c.sendRequest(httpReq)
 	if err != nil {
-		return "", err
+		return nil, err
 	}
 	defer resp.Body.Close()
 
@@ -166,7 +158,7 @@ func (c *OllamaClient) CreateChatCompletionStream(
 			if err == io.EOF {
 				break
 			}
-			return "", err
+			return nil, err
 		}
 
 		line = bytes.TrimSpace(line)
@@ -177,7 +169,7 @@ func (c *OllamaClient) CreateChatCompletionStream(
 		var streamResp OllamaResponse
 		err = json.Unmarshal(line, &streamResp)
 		if err != nil {
-			return "", err
+			return nil, err
 		}
 
 		if len(streamResp.Message.Content) > 0 {
@@ -189,12 +181,8 @@ func (c *OllamaClient) CreateChatCompletionStream(
 		}
 	}
 
-	if callback != nil {
-		callback(model.Message{
-			Role:    model.MessageRoleAssistant,
-			Content: content.String(),
-		})
-	}
-
-	return content.String(), nil
+	return &api.Message{
+		Role:    api.MessageRoleAssistant,
+		Content: content.String(),
+	}, nil
 }