X-Git-Url: https://git.giacomolli.uk/onechat/blobdiff_plain/d9fddc0f506eb4ead7049f5638674b2b8773fcaa..496569ab5b179932c60b2924bf4c61176c8be004:/chain.go?ds=inline

diff --git a/chain.go b/chain.go
index 0211787..f201812 100644
--- a/chain.go
+++ b/chain.go
@@ -22,21 +22,36 @@ type ChainClient interface {
 // answer. The replies, the tool calls, and the results go to the log of tools
 // as they happen. The thoughts never do.
 //
-// Each later request begins with the whole input of the request before it,
-// then its output items, then the tool results. The output items hold the
-// reasoning with its encrypted content, because the calls do not store it. So
-// each request begins with all of the previous one and reads it from the prompt
+// Each later request begins with the whole input of the request before it, then
+// its output items, then the tool results. The output items hold the reasoning
+// with its encrypted content, because the calls do not store it. So each
+// request begins with all of the previous one and reads it from the prompt
 // cache. The messages of take join each later request after the tool results.
 type Chain struct {
 	client  ChainClient
 	tools   *Toolbox
 	request ModelRequest
 	take    func() []Message
+	images  *ImageStore
+}
+
+// ChainFunctionCall holds the fields of a function call output item.
+type ChainFunctionCall struct {
+	Type      string `json:"type"`
+	Name      string `json:"name"`
+	Arguments string `json:"arguments"`
+	CallID    string `json:"call_id"`
+}
+
+// ChainFunctionCallOutput is the input item with the result of a function call.
+type ChainFunctionCallOutput struct {
+	Type   string `json:"type"`
+	CallID string `json:"call_id"`
+	Output any    `json:"output"`
 }
 
 // Run runs the chain from input, the input of its first request, and returns
-// the text of the final answer. An error of a log write wraps
-// errChainLogWrite.
+// the text of the final answer. An error of a log write wraps errChainLogWrite.
 func (c *Chain) Run(ctx context.Context, input []any) (string, error) {
 	request := c.request
 	for request.Turn = "first"; ; request.Turn = "next" {
@@ -47,6 +62,7 @@ func (c *Chain) Run(ctx context.Context, input []any) (string, error) {
 		}
 
 		var text string
+
 		input, text, err = c.handle(ctx, input, response)
 		if err != nil {
 			return "", err
@@ -59,12 +75,15 @@ func (c *Chain) Run(ctx context.Context, input []any) (string, error) {
 }
 
 // handle logs the replies of response and runs its tool calls. It returns the
-// input of the next request: input, the output items, the tool results, and
-// the messages of take. After the final answer it returns no input, and the
-// replies of the answer as its text.
+// input of the next request: input, the output items, the tool results, and the
+// messages of take. After the final answer it returns no input, and the replies
+// of the answer as its text.
 func (c *Chain) handle(ctx context.Context, input []any, response ModelResponse) ([]any, string, error) {
-	var outputs []any
-	var replies []string
+	var (
+		outputs []any
+		replies []string
+	)
+
 	for _, raw := range response.Output {
 		item, err := chainInputItem(raw)
 		if err != nil {
@@ -74,6 +93,7 @@ func (c *Chain) handle(ctx context.Context, input []any, response ModelResponse)
 		input = append(input, item)
 
 		var call ChainFunctionCall
+
 		err = json.Unmarshal(raw, &call)
 		if err != nil {
 			return nil, "", fmt.Errorf("decode output item: %w", err)
@@ -117,33 +137,23 @@ func (c *Chain) handle(ctx context.Context, input []any, response ModelResponse)
 
 	input = append(input, outputs...)
 	for _, message := range c.take() {
-		input = append(input, modelUserMessage(message.Text))
+		item, err := imageUserMessage(c.images, message.Text)
+		if err != nil {
+			return nil, "", fmt.Errorf("read queued images: %w", err)
+		}
+
+		input = append(input, item)
 	}
 
 	return input, "", nil
 }
 
-// ChainFunctionCall holds the fields of a function call output item.
-type ChainFunctionCall struct {
-	Type      string `json:"type"`
-	Name      string `json:"name"`
-	Arguments string `json:"arguments"`
-	CallID    string `json:"call_id"`
-}
-
-// ChainFunctionCallOutput is the input item with the result of a function
-// call.
-type ChainFunctionCallOutput struct {
-	Type   string `json:"type"`
-	CallID string `json:"call_id"`
-	Output string `json:"output"`
-}
-
 // chainInputItem returns an output item as an input item for the next request.
 // The calls do not store their items, so the item goes back whole, without its
 // id.
 func chainInputItem(raw json.RawMessage) (map[string]any, error) {
 	var item map[string]any
+
 	err := json.Unmarshal(raw, &item)
 	if err != nil {
 		return nil, fmt.Errorf("decode output item: %w", err)