Create app.py

app.py

@@ -0,0 +1,55 @@
+import gradio as gr
+from bs4 import BeautifulSoup
+from markdownify import MarkdownConverter
+
+
+def md(soup, **options):
+    return MarkdownConverter(**options).convert_soup(soup)
+
+
+def main_fn(html: str, check: list[int]):
+    soup = BeautifulSoup(html, features="html.parser")
+
+    # Remove <script> e <style>
+    for tag in ["script", "style"]:
+        for t in soup.find_all(tag):
+            t.decompose()
+
+    body = soup.find("body")
+    main = soup.find("main")
+
+    strip_tags = []  # Futuramente pode-se usar `check` para filtrar tags
+
+    if main:
+        markdown = md(main, strip=strip_tags)
+    else:
+        markdown = md(body, strip=strip_tags)
+
+    title_tag = soup.find("title")
+    title = title_tag.get_text(strip=True) if title_tag else "Sem título"
+
+    return f"{title}\n======\n\n{markdown}"
+
+
+demo = gr.Interface(
+    fn=main_fn,
+    title="HTML para Markdown",
+    description="""
+    <div style="width: fit-content; margin: 0 auto;">
+        Cole aqui seu HTML bruto e o app vai converter para Markdown.
+    </div>""",
+    inputs=[
+        gr.TextArea(label="HTML", placeholder="Cole seu código HTML aqui...", lines=20),
+        gr.CheckboxGroup(
+            label="Ignorar tags (sem efeito - tudo será extraído)",
+            choices=["a", "img", "noscript"],
+            value=[],
+        ),
+    ],
+    outputs=[
+        gr.TextArea(label="Markdown gerado", show_copy_button=True)
+    ],
+    allow_flagging="never",
+)
+
+demo.launch(server_name="0.0.0.0")