fix: file payload

cmd/client-telegram/main.go

@@ -188,30 +188,24 @@ func (l *Logic) Handler(ctx context.Context, b *bot.Bot, update *models.Update)
 		fmt.Println("photo file ID:", highResImg.FileID)
 		fmt.Printf("photo info: W: %d, H: %d, Size: %d\n", highResImg.Width, highResImg.Height, highResImg.FileSize)
 		fmt.Println("caption:", update.Message.Caption)
-		f, err := b.GetFile(ctx, &bot.GetFileParams{
-			FileID: highResImg.FileID,
-		})
+		data, err := downloadFile(ctx, b, highResImg.FileID)
 		if err != nil {
-			fmt.Println("error getting file:", err)
+			fmt.Println("error downloading photo:", err)
 			return
 		}
-
-		// Download the file
-		dlURL := b.FileDownloadLink(f)
-		resp, err := http.Get(dlURL)
-		if err != nil {
-			fmt.Println("error downloading file:", err)
-			return
+		payloads = append(payloads, data)
+		if update.Message.Caption != "" {
+			msg = update.Message.Caption
 		}
-		defer func() { _ = resp.Body.Close() }()
+	}
 
-		data, err := io.ReadAll(resp.Body)
+	if update.Message.Document != nil {
+		data, err := downloadFile(ctx, b, update.Message.Document.FileID)
 		if err != nil {
-			fmt.Println("error reading file:", err)
+			fmt.Println("error downloading document:", err)
 			return
 		}
 		payloads = append(payloads, data)
-
 		if update.Message.Caption != "" {
 			msg = update.Message.Caption
 		}
@@ -282,6 +276,21 @@ func (l *Logic) Handler(ctx context.Context, b *bot.Bot, update *models.Update)
 
 }
 
+func downloadFile(ctx context.Context, b *bot.Bot, fileID string) ([]byte, error) {
+	f, err := b.GetFile(ctx, &bot.GetFileParams{FileID: fileID})
+	if err != nil {
+		return nil, err
+	}
+
+	resp, err := http.Get(b.FileDownloadLink(f))
+	if err != nil {
+		return nil, err
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	return io.ReadAll(resp.Body)
+}
+
 func biggestImage(photos []models.PhotoSize) models.PhotoSize {
 	if len(photos) == 0 {
 		return models.PhotoSize{}

go.mod

@@ -9,6 +9,7 @@ require (
 	github.com/go-telegram/bot v1.17.0
 	github.com/google/uuid v1.6.0
 	github.com/mark3labs/mcp-go v0.29.1-0.20250521213157-f99e5472f312
+	github.com/openai/openai-go v1.8.2
 	github.com/philippgille/chromem-go v0.7.0
 	golang.org/x/oauth2 v0.30.0
 	google.golang.org/api v0.236.0
@@ -41,7 +42,6 @@ require (
 	github.com/mattn/go-colorable v0.1.2 // indirect
 	github.com/mattn/go-isatty v0.0.8 // indirect
 	github.com/mbleigh/raymond v0.0.0-20250414171441-6b3a58ab9e0a // indirect
-	github.com/openai/openai-go v1.8.2 // indirect
 	github.com/spf13/cast v1.7.1 // indirect
 	github.com/tidwall/gjson v1.18.0 // indirect
 	github.com/tidwall/match v1.1.1 // indirect

internal/ai/agent/logic.go

@@ -3,6 +3,7 @@ package agent
 
 import (
 	"context"
+	"encoding/base64"
 	"errors"
 	"fmt"
 	"log/slog"
@@ -207,11 +208,13 @@ func (l *Logic) HandleMessage(ctx context.Context, sessionID string, req domain.
 	userPromptParts := make([]*ai.Part, 0, len(req.InlineData)+1)
 	for _, inlineData := range req.InlineData {
 		// If we have some inline data convert them to prompt parts
-		userPromptParts = append(userPromptParts, ai.NewMediaPart(inlineData.MimeType, string(inlineData.Data)))
+		userPromptParts = append(userPromptParts, ai.NewMediaPart(inlineData.MimeType, "data:"+inlineData.MimeType+";base64,"+base64.StdEncoding.EncodeToString(inlineData.Data)))
 	}
 
-	// Add the user's request at the end too
-	userPromptParts = append(userPromptParts, ai.NewTextPart(req.Message))
+	// Add the user's request at the end too (skip if empty, e.g. file sent without caption)
+	if req.Message != "" {
+		userPromptParts = append(userPromptParts, ai.NewTextPart(req.Message))
+	}
 	hist = append(hist, ai.NewUserMessage(userPromptParts...))
 
 	// Inject static and dynamic information into the system prompt

internal/history/history.go

@@ -77,6 +77,8 @@ func (l *Logic) Save(ctx context.Context, sessionID string, history []*ai.Messag
 		return errors.New("invalid sessionID")
 	}
 
+	history = stripMedia(history)
+
 	var b []byte
 	var err error
 	if l.config.HistorySummary > 0 && len(history) >= l.config.HistorySummary {
@@ -121,6 +123,24 @@ func (l *Logic) summarize(ctx context.Context, history []*ai.Message) (*ai.Messa
 	return ai.NewModelTextMessage(fmt.Sprintf("Summarized history:\n\n%s", summary)), nil
 }
 
+// stripMedia replaces PartMedia parts with a text placeholder so binary file
+// data is not re-sent on every subsequent turn.
+func stripMedia(msgs []*ai.Message) []*ai.Message {
+	out := make([]*ai.Message, len(msgs))
+	for i, msg := range msgs {
+		stripped := make([]*ai.Part, 0, len(msg.Content))
+		for _, p := range msg.Content {
+			if p.Kind == ai.PartMedia {
+				stripped = append(stripped, ai.NewTextPart("[attached: "+p.ContentType+"]"))
+			} else {
+				stripped = append(stripped, p)
+			}
+		}
+		out[i] = &ai.Message{Role: msg.Role, Content: stripped, Metadata: msg.Metadata}
+	}
+	return out
+}
+
 func contentToString(history []*ai.Message) string {
 	var sb strings.Builder
 	for _, h := range history {