Create app.py

app.py

@@ -0,0 +1,79 @@
+import os
+import gradio as gr
+import asyncio
+from telethon import TelegramClient
+from huggingface_hub import HfApi
+from dotenv import load_dotenv
+
+load_dotenv()
+
+api_id = int(os.getenv("API_ID"))
+api_hash = os.getenv("API_HASH")
+hf_token = os.getenv("HF_TOKEN")
+channel = os.getenv("CHANNEL_USERNAME")
+dataset_repo = os.getenv("DATASET_REPO")
+
+client = TelegramClient("my_session", api_id, api_hash)
+hf_api = HfApi()
+hf_api.set_access_token(hf_token)
+
+os.makedirs("downloads", exist_ok=True)
+
+async def process_filenames(name_input, progress=None):
+    await client.start()
+
+    filenames = [name.strip().lower() for name in name_input.replace(",", "\n").splitlines() if name.strip()]
+    results = []
+    found = set()
+
+    messages = [msg async for msg in client.iter_messages(channel, limit=300)]
+    total = len(messages)
+
+    for i, msg in enumerate(messages):
+        if progress:
+            progress(i / total, desc=f"Scanning message {i+1}/{total}...")
+
+        if msg.media and msg.file:
+            fname = msg.file.name or f"file_{msg.id}{msg.file.ext}"
+
+            for search in filenames:
+                if search in fname.lower() and fname not in found:
+                    found.add(fname)
+                    path = f"downloads/{fname}"
+
+                    if not os.path.exists(path):
+                        await msg.download_media(file=path)
+                        hf_api.upload_file(
+                            path_or_fileobj=path,
+                            path_in_repo=os.path.basename(path),
+                            repo_id=dataset_repo,
+                            repo_type="dataset"
+                        )
+                        results.append(f"✅ Uploaded: {fname}")
+                    else:
+                        results.append(f"⏩ Already exists: {fname}")
+                    break
+
+    matched_names = [r.split(":")[-1].strip().lower() for r in results if r.startswith("✅") or r.startswith("⏩")]
+    for name in filenames:
+        if not any(name in matched for matched in matched_names):
+            results.append(f"❌ Not found: {name}")
+
+    return "\n".join(results) if results else "❌ No files matched."
+
+def handle_upload(name_input):
+    with gr.Progress() as p:
+        return asyncio.run(process_filenames(name_input, p))
+
+def launch_ui():
+    with gr.Blocks() as app:
+        gr.Markdown("## 📥 Telegram File Fetcher → 📤 HF Dataset Uploader")
+        input_box = gr.Textbox(label="Enter filenames (comma or newline separated)", lines=6)
+        result_box = gr.Textbox(label="Results", lines=10)
+        go = gr.Button("Start Upload")
+        go.click(fn=handle_upload, inputs=input_box, outputs=result_box)
+    return app
+
+if __name__ == "__main__":
+    asyncio.run(client.connect())
+    launch_ui().launch(server_name="0.0.0.0", server_port=7860)