X-Git-Url: https://git.giacomolli.uk/onechat/blobdiff_plain/cb514b662bcf9d90b2cf02789c989f777d774d3f..59948b44c0fe2165b90e3aa0d18646f6c3d74af3:/model.go?ds=inline

diff --git a/model.go b/model.go
index 61dbc58..cc77f9a 100644
--- a/model.go
+++ b/model.go
@@ -41,9 +41,9 @@ const (
 var errModelCallLimit = errors.New("call limit reached")
 
 // modelEndpoint returns the URL and the authorizer of the endpoint in config.
-// The ChatGPT backend uses the token file in dir. The OpenAI API uses the key in
-// OPENAI_API_KEY, and there is no fallback from one to the other. All clients
-// share the authorizer, so one refresh runs at a time.
+// The ChatGPT backend uses the token file in dir. The OpenAI API uses the key
+// in OPENAI_API_KEY, and there is no fallback from one to the other. All
+// clients share the authorizer, so one refresh runs at a time.
 func modelEndpoint(config Config, dir string) (string, ModelAuthorizer, error) {
 	if config.Endpoint == configEndpointAPI {
 		key := os.Getenv("OPENAI_API_KEY")
@@ -73,22 +73,22 @@ func newModelClient(url string, auth ModelAuthorizer, cacheKey, usagePath string
 
 // modelUserMessage returns a user input item with one text part for each text.
 func modelUserMessage(texts ...string) any {
-	type part struct {
-		Type string `json:"type"`
-		Text string `json:"text"`
-	}
-
-	content := make([]part, len(texts))
+	content := make([]ModelInputPart, len(texts))
 	for n, text := range texts {
-		content[n] = part{
+		content[n] = ModelInputPart{
 			Type: "input_text",
 			Text: text,
 		}
 	}
 
+	return modelUserParts(content)
+}
+
+// modelUserParts keeps text and image parts in one user message.
+func modelUserParts(content []ModelInputPart) any {
 	return struct {
-		Role    string `json:"role"`
-		Content []part `json:"content"`
+		Role    string           `json:"role"`
+		Content []ModelInputPart `json:"content"`
 	}{
 		Role:    "user",
 		Content: content,
@@ -98,8 +98,10 @@ func modelUserMessage(texts ...string) any {
 // modelOutputText returns the text of the message items of output.
 func modelOutputText(output []json.RawMessage) (string, error) {
 	var text strings.Builder
+
 	for _, raw := range output {
 		var item ModelOutputItem
+
 		err := json.Unmarshal(raw, &item)
 		if err != nil {
 			return "", fmt.Errorf("decode output item: %w", err)
@@ -131,6 +133,7 @@ type ModelAPIKey string
 // Authorize sets the key as the bearer token.
 func (k ModelAPIKey) Authorize(ctx context.Context, header http.Header) error {
 	header.Set("Authorization", "Bearer "+string(k))
+
 	return ctx.Err()
 }
 
@@ -168,6 +171,16 @@ func (c *ModelClient) LimitReached() bool {
 	return c.limit > 0 && c.calls.Load() >= c.limit
 }
 
+// CallLimit returns the configured call limit, or zero for no limit.
+func (c *ModelClient) CallLimit() int64 {
+	return c.limit
+}
+
+// UsagePath returns the file that records the calls of this client.
+func (c *ModelClient) UsagePath() string {
+	return c.usage.path
+}
+
 // Create sends the call and reads its event stream to the end. Over the call
 // limit it sends nothing and returns errModelCallLimit.
 func (c *ModelClient) Create(ctx context.Context, request ModelRequest) (ModelResponse, error) {
@@ -204,6 +217,7 @@ func (c *ModelClient) Create(ctx context.Context, request ModelRequest) (ModelRe
 
 	httpRequest.Header.Set("Content-Type", "application/json")
 	httpRequest.Header.Set("Accept", "text/event-stream")
+
 	err = c.auth.Authorize(ctx, httpRequest.Header)
 	if err != nil {
 		return ModelResponse{}, fmt.Errorf("authorize model call: %w", err)
@@ -307,6 +321,13 @@ type ModelOutputItem struct {
 	} `json:"content"`
 }
 
+// ModelInputPart is a text or image part in a user message or tool result.
+type ModelInputPart struct {
+	Type     string `json:"type"`
+	Text     string `json:"text,omitempty"`
+	ImageURL string `json:"image_url,omitempty"`
+}
+
 // ModelRequest is one model call. Input holds the input items in order, such as
 // modelUserMessage values or output items of an earlier response. Node names
 // the node that a compaction builds, or is nil. Turn names the request of a
@@ -424,6 +445,7 @@ type ModelWireError struct {
 // without the turn state.
 func modelQuotaHeaders(header http.Header) map[string]string {
 	var quota map[string]string
+
 	for name, values := range header {
 		if !strings.HasPrefix(name, modelQuotaPrefix) || name == modelTurnStateHeader || len(values) == 0 {
 			continue
@@ -444,8 +466,11 @@ func modelQuotaHeaders(header http.Header) map[string]string {
 // the completed response when the stream sent none.
 func modelReadStream(stream io.Reader) (ModelResponse, error) {
 	var response ModelResponse
+
 	reader := bufio.NewReader(stream)
+
 	var data bytes.Buffer
+
 	for {
 		line, err := reader.ReadBytes('\n')
 		if err != nil && !errors.Is(err, io.EOF) {
@@ -462,6 +487,7 @@ func modelReadStream(stream io.Reader) (ModelResponse, error) {
 		} else if data.Len() > 0 {
 			done, eventErr := modelReadEvent(&response, data.Bytes())
 			data.Reset()
+
 			if eventErr != nil || done {
 				return response, eventErr
 			}
@@ -477,6 +503,7 @@ func modelReadStream(stream io.Reader) (ModelResponse, error) {
 // stream.
 func modelReadEvent(response *ModelResponse, data []byte) (bool, error) {
 	var event ModelStreamEvent
+
 	err := json.Unmarshal(data, &event)
 	if err != nil {
 		return true, fmt.Errorf("decode stream event: %w", err)