Drop in a screenshot and convert it to clean code (HTML/Tailwind/React/Vue)
Python
extract only html content
Abi Rajaedfd16ef1db5
diff --git a/backend/codegen/__init__.py b/backend/codegen/__init__.pynew file mode 100644index 00000000..e69de29bdiff --git a/backend/codegen/test_utils.py b/backend/codegen/test_utils.pynew file mode 100644index 00000000..cd663f81--- /dev/null+++ b/backend/codegen/test_utils.py@@ -0,0 +1,57 @@+import unittest+from codegen.utils import extract_html_content+++class TestUtils(unittest.TestCase):++ def test_extract_html_content_with_html_tags(self):+ text = "<html><body><p>Hello, World!</p></body></html>"+ expected = "<html><body><p>Hello, World!</p></body></html>"+ result = extract_html_content(text)+ self.assertEqual(result, expected)++ def test_extract_html_content_without_html_tags(self):+ text = "No HTML content here."+ expected = "No HTML content here."+ result = extract_html_content(text)+ self.assertEqual(result, expected)++ def test_extract_html_content_with_partial_html_tags(self):+ text = "<html><body><p>Hello, World!</p></body>"+ expected = "<html><body><p>Hello, World!</p></body>"+ result = extract_html_content(text)+ self.assertEqual(result, expected)++ def test_extract_html_content_with_multiple_html_tags(self):+ text = "<html><body><p>First</p></body></html> Some text <html><body><p>Second</p></body></html>"+ expected = "<html><body><p>First</p></body></html>"+ result = extract_html_content(text)+ self.assertEqual(result, expected)++ ## The following are tests based on actual LLM outputs++ def test_extract_html_content_some_explanation_before(self):+ text = """Got it! You want the song list to be displayed horizontally. I'll update the code to ensure that the song list is displayed in a horizontal layout.++ Here's the updated code:++ <html lang="en"><head></head><body class="bg-black text-white"></body></html>"""+ expected = '<html lang="en"><head></head><body class="bg-black text-white"></body></html>'+ result = extract_html_content(text)+ self.assertEqual(result, expected)++ def test_markdown_tags(self):+ text = "```html<head></head>```"+ expected = "```html<head></head>```"+ result = extract_html_content(text)+ self.assertEqual(result, expected)++ def test_doctype_text(self):+ text = '<!DOCTYPE html><html lang="en"><head></head><body></body></html>'+ expected = '<html lang="en"><head></head><body></body></html>'+ result = extract_html_content(text)+ self.assertEqual(result, expected)+++if __name__ == "__main__":+ unittest.main()diff --git a/backend/codegen/utils.py b/backend/codegen/utils.pynew file mode 100644index 00000000..107579b8--- /dev/null+++ b/backend/codegen/utils.py@@ -0,0 +1,14 @@+import re+++def extract_html_content(text: str):+ # Use regex to find content within <html> tags and include the tags themselves+ match = re.search(r"(<html.*?>.*?</html>)", text, re.DOTALL)+ if match:+ return match.group(1)+ else:+ # Otherwise, we just send the previous HTML over+ print(+ "[HTML Extraction] No <html> tags found in the generated content: " + text+ )+ return textdiff --git a/backend/routes/generate_code.py b/backend/routes/generate_code.pyindex d280830a..6426f202 100644--- a/backend/routes/generate_code.py+++ b/backend/routes/generate_code.py@@ -2,6 +2,7 @@ import osimport tracebackfrom fastapi import APIRouter, WebSocketimport openai+from codegen.utils import extract_html_contentfrom config import ANTHROPIC_API_KEY, IS_PROD, SHOULD_MOCK_AI_RESPONSEfrom custom_types import InputModefrom llm import (@@ -312,6 +313,9 @@ async def stream_code(websocket: WebSocket):print("Exact used model for generation: ", exact_llm_version)+ # Strip the completion of everything except the HTML content+ completion = extract_html_content(completion)+# Write the messages dict into a log so that we can debug laterwrite_logs(prompt_messages, completion) # type: ignore