Skip to content

Drop in a screenshot and convert it to clean code (HTML/Tailwind/React/Vue)

Python

Support multiple screenshot uploads

Abi Raja974ddb0a9532
diff --git a/backend/evals/core.py b/backend/evals/core.py
index 38b38afc..e34b349b 100644
--- a/backend/evals/core.py
+++ b/backend/evals/core.py
@@ -11,7 +11,7 @@ from openai.types.chat import ChatCompletionMessageParam
async def generate_code_for_image(image_url: str, stack: Stack, model: Llm) -> str:
- prompt_messages = assemble_prompt(image_url, stack)
+ prompt_messages = assemble_prompt([image_url], stack)
return await generate_code_core(prompt_messages, model)
diff --git a/backend/prompts/__init__.py b/backend/prompts/__init__.py
index fa381fc3..2fb085f5 100644
--- a/backend/prompts/__init__.py
+++ b/backend/prompts/__init__.py
@@ -40,9 +40,9 @@ async def create_prompt(
else:
# Assemble the prompt for non-imported code
if input_mode == "image":
- image_url = prompt["images"][0]
+ image_urls = prompt.get("images", [])
text_prompt = prompt.get("text", "")
- prompt_messages = assemble_prompt(image_url, stack, text_prompt)
+ prompt_messages = assemble_prompt(image_urls, stack, text_prompt)
elif input_mode == "text":
prompt_messages = assemble_text_prompt(prompt["text"], stack)
elif input_mode == "video":
@@ -62,9 +62,9 @@ async def create_prompt(
]
else:
# Default to image mode for backward compatibility
- image_url = prompt["images"][0]
+ image_urls = prompt.get("images", [])
text_prompt = prompt.get("text", "")
- prompt_messages = assemble_prompt(image_url, stack, text_prompt)
+ prompt_messages = assemble_prompt(image_urls, stack, text_prompt)
if generation_type == "update":
# Transform the history tree into message format
@@ -145,7 +145,7 @@ def assemble_imported_code_prompt(
def assemble_prompt(
- image_data_url: str,
+ image_data_urls: list[str],
stack: Stack,
text_prompt: str = "",
) -> list[ChatCompletionMessageParam]:
@@ -156,16 +156,20 @@ def assemble_prompt(
if text_prompt.strip():
user_prompt = user_prompt.strip() + "\n\nAdditional instructions: " + text_prompt
- user_content: list[ChatCompletionContentPartParam] = [
- {
- "type": "image_url",
- "image_url": {"url": image_data_url, "detail": "high"},
- },
+ user_content: list[ChatCompletionContentPartParam] = []
+ for image_data_url in image_data_urls:
+ user_content.append(
+ {
+ "type": "image_url",
+ "image_url": {"url": image_data_url, "detail": "high"},
+ }
+ )
+ user_content.append(
{
"type": "text",
"text": user_prompt,
- },
- ]
+ }
+ )
return [
{
"role": "system",
diff --git a/frontend/src/App.tsx b/frontend/src/App.tsx
index b2c122ff..9af6b0a3 100644
--- a/frontend/src/App.tsx
+++ b/frontend/src/App.tsx
@@ -255,10 +255,12 @@ function App() {
// Kick off the code generation
if (referenceImages.length > 0) {
+ const images =
+ inputMode === "video" ? [referenceImages[0]] : referenceImages;
doGenerateCode({
generationType: "create",
inputMode,
- prompt: { text: textPrompt, images: [referenceImages[0]] },
+ prompt: { text: textPrompt, images },
});
}
}
diff --git a/frontend/src/components/sidebar/Sidebar.tsx b/frontend/src/components/sidebar/Sidebar.tsx
index b3d1b338..3cf2003b 100644
--- a/frontend/src/components/sidebar/Sidebar.tsx
+++ b/frontend/src/components/sidebar/Sidebar.tsx
@@ -231,11 +231,26 @@ function Sidebar({
})}
>
{inputMode === "image" && (
- <img
- className="w-[340px] border border-gray-200 rounded-md"
- src={referenceImages[0]}
- alt="Reference"
- />
+ <>
+ {referenceImages.length === 1 ? (
+ <img
+ className="w-[340px] border border-gray-200 rounded-md"
+ src={referenceImages[0]}
+ alt="Reference"
+ />
+ ) : (
+ <div className="grid grid-cols-2 gap-2 w-[340px] max-h-[360px] overflow-y-auto border border-gray-200 rounded-md p-2">
+ {referenceImages.map((image, index) => (
+ <img
+ key={`${image}-${index}`}
+ className="w-full h-24 object-cover rounded"
+ src={image}
+ alt={`Reference ${index + 1}`}
+ />
+ ))}
+ </div>
+ )}
+ </>
)}
{inputMode === "video" && (
<video
@@ -248,7 +263,11 @@ function Sidebar({
)}
</div>
<div className="text-gray-400 uppercase text-sm text-center mt-1">
- {inputMode === "video" ? "Original Video" : "Original Screenshot"}
+ {inputMode === "video"
+ ? "Original Video"
+ : referenceImages.length > 1
+ ? `Original Screenshots (${referenceImages.length})`
+ : "Original Screenshot"}
</div>
</div>
)}
diff --git a/frontend/src/components/unified-input/tabs/UploadTab.tsx b/frontend/src/components/unified-input/tabs/UploadTab.tsx
index f967b8c2..70a58210 100644
--- a/frontend/src/components/unified-input/tabs/UploadTab.tsx
+++ b/frontend/src/components/unified-input/tabs/UploadTab.tsx
@@ -77,6 +77,7 @@ function UploadTab({ doCreate }: Props) {
setFiles([]);
setTextPrompt("");
setShowTextPrompt(false);
+ setUploadedInputMode("image");
};
const handleKeyDown = (e: React.KeyboardEvent<HTMLTextAreaElement>) => {
@@ -88,7 +89,7 @@ function UploadTab({ doCreate }: Props) {
const { getRootProps, getInputProps, isFocused, isDragAccept, isDragReject } =
useDropzone({
- maxFiles: 1,
+ maxFiles: 10,
maxSize: 1024 * 1024 * 20,
accept: {
"image/png": [".png"],
@@ -99,6 +100,21 @@ function UploadTab({ doCreate }: Props) {
"video/webm": [".webm"],
},
onDrop: (acceptedFiles) => {
+ if (acceptedFiles.length === 0) return;
+
+ const hasVideo = acceptedFiles.some(
+ (file) =>
+ file.type.startsWith("video/") ||
+ [".mp4", ".mov", ".webm"].some((ext) =>
+ file.name.toLowerCase().endsWith(ext)
+ )
+ );
+
+ if (hasVideo && acceptedFiles.length > 1) {
+ toast.error("Please upload a single video or multiple images.");
+ return;
+ }
+
setFiles(
acceptedFiles.map((file: File) =>
Object.assign(file, {
@@ -107,18 +123,11 @@ function UploadTab({ doCreate }: Props) {
) as FileWithPreview[]
);
- const firstFile = acceptedFiles[0];
- const isVideo =
- firstFile?.type?.startsWith("video/") ||
- [".mp4", ".mov", ".webm"].some((ext) =>
- firstFile?.name?.toLowerCase().endsWith(ext)
- );
-
Promise.all(acceptedFiles.map((file) => fileToDataURL(file)))
.then((dataUrls) => {
if (dataUrls.length > 0) {
const inputMode =
- isVideo || (dataUrls[0] as string).startsWith("data:video")
+ hasVideo || (dataUrls[0] as string).startsWith("data:video")
? "video"
: "image";
setUploadedDataUrls(dataUrls as string[]);
@@ -206,11 +215,11 @@ function UploadTab({ doCreate }: Props) {
</div>
<div className="text-center">
<p className="text-gray-700 font-medium">
- Drop a screenshot or video here
+ Drop screenshots or a video here
</p>
</div>
<p className="text-xs text-gray-400 mt-2">
- Supports PNG, JPG, MP4, MOV, WebM (max 20MB, 30s video)
+ Supports PNG, JPG, MP4, MOV, WebM (max 20MB each, 30s video)
</p>
</div>
</div>
@@ -226,16 +235,21 @@ function UploadTab({ doCreate }: Props) {
controls
/>
) : (
- <img
- src={files[0]?.preview}
- alt="Uploaded screenshot"
- className="w-full h-auto max-h-[400px] object-contain rounded-lg border border-gray-200"
- />
+ <div className="grid grid-cols-2 sm:grid-cols-3 gap-3 max-h-[420px] overflow-y-auto rounded-lg border border-gray-200 p-3">
+ {files.map((file, index) => (
+ <img
+ key={`${file.name}-${index}`}
+ src={file.preview}
+ alt={`Uploaded screenshot ${index + 1}`}
+ className="w-full h-32 object-cover rounded-md border border-gray-100"
+ />
+ ))}
+ </div>
)}
<button
onClick={handleClear}
className="absolute top-2 right-2 bg-white rounded-full p-1.5 shadow-md hover:bg-gray-100 transition-colors"
- aria-label="Remove file"
+ aria-label="Remove files"
>
<svg
xmlns="http://www.w3.org/2000/svg"