Refactor HTML extraction logic to streamline response handling and improve error messaging

app.py

@@ -94,89 +94,18 @@ Focus on creating clean, spacious, and well-proportioned designs that feel premi
 
 def extract_html_from_response(response_text):
     """Extract HTML code from model response"""
-
-    # First, try to find HTML within code blocks
+    
+    # Try to find HTML within code blocks first
     html_pattern = r"```(?:html)?\s*(.*?)```"
     matches = re.findall(html_pattern, response_text, re.DOTALL | re.IGNORECASE)
-
+    
     if matches:
         return matches[0].strip()
-
-    # If no code blocks, look for HTML tags in the response
+    
+    # If no code blocks but response contains HTML tags, return the whole response
     if re.search(r"<[^>]+>", response_text):
-        # Find the first HTML tag
-        html_start = re.search(r"<", response_text)
-        if not html_start:
-            raise ValueError("No HTML content found in the response")
-
-        # Start from the first HTML tag
-        content_from_first_tag = response_text[html_start.start() :]
-
-        # Split into lines and process
-        lines = content_from_first_tag.split("\n")
-        html_lines = []
-
-        for line in lines:
-            line_stripped = line.strip()
-
-            # Skip empty lines at the beginning
-            if not line_stripped and not html_lines:
-                continue
-
-            # Stop if we hit obvious explanation text after we have HTML
-            if html_lines and line_stripped:
-                # Check if this line looks like explanation rather than HTML
-                if not re.search(r"<[^>]*>", line_stripped):
-                    # If it's a long explanatory sentence, stop here
-                    if len(line_stripped) > 80 or any(
-                        phrase in line_stripped.lower()
-                        for phrase in [
-                            "this creates",
-                            "this provides",
-                            "the form",
-                            "this design",
-                            "here's",
-                            "this code",
-                            "explanation:",
-                            "note:",
-                            "features:",
-                            "improvements:",
-                        ]
-                    ):
-                        break
-
-            html_lines.append(line.rstrip())
-
-            # If we have HTML and hit a line that ends with a period and looks like explanation, stop
-            if html_lines and line_stripped.endswith(".") and len(line_stripped) > 40:
-                if not re.search(r"<[^>]*>", line_stripped):
-                    html_lines.pop()  # Remove the explanatory line
-                    break
-
-        # Join and clean up
-        html_content = "\n".join(html_lines).strip()
-
-        # Remove any trailing text after the last complete HTML tag
-        # Find the last complete tag (either closing tag or self-closing)
-        last_complete_tag = None
-        for match in re.finditer(r"<[^>]+>", html_content):
-            tag = match.group()
-            if (
-                tag.startswith("</")
-                or tag.endswith("/>")
-                or any(
-                    void in tag.lower()
-                    for void in ["<input", "<img", "<br", "<hr", "<meta", "<link"]
-                )
-            ):
-                last_complete_tag = match
-
-        if last_complete_tag:
-            # Keep everything up to and including the last complete tag
-            html_content = html_content[: last_complete_tag.end()]
-
-        return html_content
-
+        return response_text.strip()
+    
     # If no HTML found, return error
     raise ValueError("No HTML content found in the response")