p7: make image prompt truncation UTF-8 safe

4 files changed, 120 insertions, 3 deletions

diff --git a/internal/image/gemini_prompt.go b/internal/image/gemini_prompt.go
index 5932174..6d84ee9 100644
--- a/internal/image/gemini_prompt.go
+++ b/internal/image/gemini_prompt.go
@@ -9,7 +9,7 @@ import (
 func normalizeCustomPrompt(prompt string) string {
 	prompt = strings.TrimSpace(prompt)
 	if len(prompt) > maxCustomPrompt {
-		prompt = prompt[:maxCustomPrompt-3] + "..."
+		prompt = truncateToByteLimit(prompt, maxCustomPrompt)
 	}
 	return prompt
 }
diff --git a/internal/image/gemini_test.go b/internal/image/gemini_test.go
index aae8014..91e0ce4 100644
--- a/internal/image/gemini_test.go
+++ b/internal/image/gemini_test.go
@@ -9,6 +9,7 @@ import (
 	"net/http/httptest"
 	"strings"
 	"testing"
+	"unicode/utf8"
 
 	"google.golang.org/genai"
 )
@@ -189,6 +190,55 @@ func TestGeminiProvider_Search_CustomPromptIsTruncated(t *testing.T) {
 	}
 }
 
+func TestGeminiProvider_Search_CustomPromptTruncatesUTF8Safely(t *testing.T) {
+	originalText := geminiGenerateText
+	originalImage := geminiGenerateImage
+	t.Cleanup(func() {
+		geminiGenerateText = originalText
+		geminiGenerateImage = originalImage
+	})
+
+	geminiGenerateText = func(context.Context, *GeminiProvider, string, string, string, float32, int32) (string, error) {
+		t.Fatal("unexpected text generation for truncated custom prompt")
+		return "", nil
+	}
+
+	var gotPrompt string
+	geminiGenerateImage = func(_ context.Context, _ *GeminiProvider, prompt, _ string) ([]byte, string, error) {
+		gotPrompt = prompt
+		return mustJPEGBytes(t), "image/jpeg", nil
+	}
+
+	client := NewGeminiProvider(&GeminiConfig{APIKey: "test-key"})
+	longPrompt := strings.Repeat("яa", 2500)
+	if len(longPrompt) <= maxCustomPrompt {
+		t.Fatal("test prompt did not exceed truncation limit")
+	}
+
+	results, err := client.Search(context.Background(), &SearchOptions{
+		Query:        "ябълка",
+		CustomPrompt: longPrompt,
+	})
+	if err != nil {
+		t.Fatalf("Search() unexpected error: %v", err)
+	}
+	if !utf8.ValidString(gotPrompt) {
+		t.Fatalf("Search() returned invalid UTF-8 prompt: %q", gotPrompt)
+	}
+	if len(gotPrompt) > maxCustomPrompt {
+		t.Fatalf("prompt length = %d, want <= %d", len(gotPrompt), maxCustomPrompt)
+	}
+	if !strings.HasSuffix(gotPrompt, "...") {
+		t.Fatalf("prompt = %q, want ellipsis suffix", gotPrompt)
+	}
+	if client.LastPrompt() != gotPrompt {
+		t.Fatalf("LastPrompt() = %q, want %q", client.LastPrompt(), gotPrompt)
+	}
+	if len(results) != 1 {
+		t.Fatalf("expected 1 result, got %d", len(results))
+	}
+}
+
 func TestGeminiProvider_GenerateImage_UsesCustomPrompt(t *testing.T) {
 	originalText := geminiGenerateText
 	originalImage := geminiGenerateImage
diff --git a/internal/image/prompt.go b/internal/image/prompt.go
index addf936..c67c8b3 100644
--- a/internal/image/prompt.go
+++ b/internal/image/prompt.go
@@ -6,6 +6,7 @@ import (
 )
 
 const maxImagePromptChars = 1000
+const promptEllipsis = "..."
 
 func promptSubject(englishTranslation, fallback string) string {
 	subject := normalizePromptText(englishTranslation)
@@ -95,6 +96,38 @@ func withTerminalPunctuation(text string) string {
 	}
 }
 
+func truncateToByteLimit(text string, maxBytes int) string {
+	text = strings.TrimSpace(text)
+	if text == "" || maxBytes <= 0 {
+		return ""
+	}
+	if len(text) <= maxBytes {
+		return text
+	}
+
+	if maxBytes <= len(promptEllipsis) {
+		return truncateToRuneBoundary(text, maxBytes)
+	}
+
+	return truncateToRuneBoundary(text, maxBytes-len(promptEllipsis)) + promptEllipsis
+}
+
+func truncateToRuneBoundary(text string, maxBytes int) string {
+	if maxBytes <= 0 {
+		return ""
+	}
+
+	end := 0
+	for i := range text {
+		if i > maxBytes {
+			break
+		}
+		end = i
+	}
+
+	return text[:end]
+}
+
 // buildEducationalPrompt assembles the final image-generation prompt from a
 // pre-chosen artistic style, an optional scene description, and the word subject.
 func buildEducationalPrompt(style, scene, subject string) string {
@@ -128,7 +161,7 @@ func buildEducationalPrompt(style, scene, subject string) string {
 				)
 				maxSceneLen := maxImagePromptChars - len(template)
 				if maxSceneLen > 3 && len(scene) > maxSceneLen {
-					scene = scene[:maxSceneLen] + "..."
+					scene = truncateToByteLimit(scene, maxSceneLen)
 				}
 				prompt = fmt.Sprintf(
 					"Generate a %s flashcard image illustrating \"%s\". Scene: %s "+
@@ -149,7 +182,7 @@ func buildEducationalPrompt(style, scene, subject string) string {
 	}
 
 	if len(prompt) > maxImagePromptChars {
-		prompt = prompt[:997] + "..."
+		prompt = truncateToByteLimit(prompt, maxImagePromptChars)
 	}
 
 	return prompt
diff --git a/internal/image/prompt_test.go b/internal/image/prompt_test.go
index 73eb30e..957c84c 100644
--- a/internal/image/prompt_test.go
+++ b/internal/image/prompt_test.go
@@ -1,8 +1,10 @@
 package image
 
 import (
+	"fmt"
 	"strings"
 	"testing"
+	"unicode/utf8"
 )
 
 func TestPromptSubjectUsesTranslationFirst(t *testing.T) {
@@ -58,3 +60,35 @@ func TestFallbackVisualDirectionForSingleWord(t *testing.T) {
 		t.Fatalf("fallbackVisualDirection() = %q", got)
 	}
 }
+
+func TestBuildEducationalPromptTruncatesLongSceneWithoutBreakingUTF8(t *testing.T) {
+	t.Parallel()
+
+	style := "Photorealism"
+	subject := "ябълка"
+	for {
+		template := fmt.Sprintf(
+			"Generate a %s flashcard image illustrating \"%s\". Scene:  "+
+				"The image should be educational and suitable for language learning flashcards. "+
+				"Requirements: The main subject or concept must be clearly visible, centered, well lit, and easy to identify.",
+			style, subject,
+		)
+		if (maxImagePromptChars-len(template))%2 == 1 {
+			break
+		}
+		subject += "x"
+	}
+
+	scene := strings.Repeat("Ярка сцена с ябълки и хора, ", 80)
+	got := buildEducationalPrompt(style, scene, subject)
+
+	if !utf8.ValidString(got) {
+		t.Fatalf("buildEducationalPrompt() returned invalid UTF-8: %q", got)
+	}
+	if len(got) > maxImagePromptChars {
+		t.Fatalf("buildEducationalPrompt() length = %d, want <= %d", len(got), maxImagePromptChars)
+	}
+	if !strings.Contains(got, "...") {
+		t.Fatalf("buildEducationalPrompt() = %q, want truncated scene marker", got)
+	}
+}