Update app.py

app.py

@@ -1,6 +1,6 @@
 import os
 os.system("playwright install")
-# app.py (Final, Working Async Version with Bugfixes)
+# app.py (Final, Working Async Version with All Bugs Fixed)
 
 import gradio as gr
 from playwright.async_api import async_playwright, Error as PlaywrightError
@@ -18,32 +18,49 @@ REVOLVER = None
 LIVE_CONTEXTS = {}  # { tab_id: { "context": PlaywrightContext, "page": PlaywrightPage } }
 APP_STARTED = False
 
-# --- 2. PLAIN DATA STATE CLASSES (Unchanged) ---
+# --- 2. PLAIN DATA STATE CLASSES (Copyable) ---
 class TabState:
+    """A plain data class representing a tab's state. Fully copyable."""
     def __init__(self, tab_id, proxy_used="Direct Connection"):
-        self.id, self.url, self.title = tab_id, "about:blank", "New Tab"
-        self.parsed_text, self.links, self.proxy_used = "Welcome!", [], proxy_used
+        self.id = tab_id
+        self.url = "about:blank"
+        self.title = "New Tab"
+        self.parsed_text = "Welcome! Navigate to a URL or search to get started."
+        self.links = []
+        self.proxy_used = proxy_used
 
 class BrowserState:
-    def __init__(self): self.tabs, self.active_tab_id = [], None
-    def get_active_tab(self): return next((t for t in self.tabs if t.id == self.active_tab_id), None)
+    """A plain data class representing the browser's overall state."""
+    def __init__(self):
+        self.tabs = []
+        self.active_tab_id = None
+    def get_active_tab(self):
+        if not self.active_tab_id: return None
+        return next((t for t in self.tabs if t.id == self.active_tab_id), None)
 
 class CredentialRevolver:
+    """Manages a rotating list of proxies from an environment variable."""
     def __init__(self, proxy_string: str):
         self.proxies = self._parse_proxies(proxy_string)
         if self.proxies: self.proxy_cycler = cycle(self.proxies)
         else: self.proxy_cycler = None
     def _parse_proxies(self, proxy_string: str):
-        proxies = [];
+        proxies = []
         for line in proxy_string.strip().splitlines():
-            try: parsed = urllib.parse.urlparse(f"//{line.strip()}"); server = f"{parsed.scheme or 'http'}://{parsed.hostname}:{parsed.port}"; proxies.append({"server": server, "username": parsed.username, "password": parsed.password})
+            try:
+                parsed = urllib.parse.urlparse(f"//{line.strip()}")
+                server = f"{parsed.scheme or 'http'}://{parsed.hostname}:{parsed.port}"
+                proxies.append({"server": server, "username": parsed.username, "password": parsed.password})
             except: pass
         return proxies
-    def get_next(self): return next(self.proxy_cycler) if self.proxy_cycler else None
-    def count(self): return len(self.proxies)
+    def get_next(self):
+        return next(self.proxy_cycler) if self.proxy_cycler else None
+    def count(self):
+        return len(self.proxies)
 
-# --- 3. ASYNC LOGIC (Unchanged) ---
+# --- 3. CORE ASYNC LOGIC ---
 async def _fetch_and_update_tab_state(tab_state: TabState, url: str):
+    """Uses Playwright to navigate and BeautifulSoup to parse, updating the TabState object."""
     log = f"▶️ Navigating to {url}..."; live_page = LIVE_CONTEXTS[tab_state.id]["page"]
     try:
         await live_page.goto(url, wait_until='domcontentloaded', timeout=30000)
@@ -61,6 +78,7 @@ async def _fetch_and_update_tab_state(tab_state: TabState, url: str):
     return log
 
 async def handle_action(browser_state: BrowserState, action: str, value=None):
+    """Modifies the state based on user actions."""
     log = ""; active_tab_state = browser_state.get_active_tab()
     if action == "new_tab":
         tab_id, proxy_config = str(uuid.uuid4()), REVOLVER.get_next()
@@ -90,7 +108,21 @@ async def handle_action(browser_state: BrowserState, action: str, value=None):
         browser_state.active_tab_id = value; log = f"Switched to tab."
     return browser_state, log
 
-# --- 4. GRADIO UI AND EVENT HANDLING (WITH FIXES) ---
+# **BUG FIX 1: NameError Fix**
+# This function is now defined before it is called by the master_handler.
+def update_ui_components(browser_state: BrowserState):
+    """Generates a dictionary of UI updates from the state. Not async."""
+    active_tab = browser_state.get_active_tab()
+    if not active_tab: return {page_content: gr.Markdown("No active tabs."), url_textbox: "", links_display: "", tab_selector: gr.Radio(choices=[])}
+    tab_choices = [(f"Tab {i}: {t.title[:25]}... (via {t.proxy_used})", t.id) for i, t in enumerate(browser_state.tabs)]
+    links_md = "### 🔗 Links on Page\n" + ('\n'.join(f"{i}. [{link['text'][:80]}]({link['url']})" for i, link in enumerate(active_tab.links[:25])) if active_tab.links else "_No links found._")
+    return {
+        page_content: gr.Markdown(f"# {active_tab.title}\n**URL:** {active_tab.url}\n\n---\n\n{active_tab.parsed_text[:2000]}..."),
+        url_textbox: gr.Textbox(value=active_tab.url), links_display: gr.Markdown(links_md),
+        tab_selector: gr.Radio(choices=tab_choices, value=active_tab.id, label="Active Tabs"),
+    }
+
+# --- 4. GRADIO UI AND EVENT HANDLING ---
 with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
     browser_state = gr.State(BrowserState())
     gr.Markdown("# 🛰️ Real Browser Demo (Final Working Version)")
@@ -107,7 +139,6 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
                 links_display = gr.Markdown("...");
                 with gr.Row(): click_num_box = gr.Number(label="Link #", scale=1, minimum=0, step=1); click_btn = gr.Button("Click Link", scale=2)
     
-    # This order must be consistent
     all_outputs = [page_content, url_textbox, links_display, tab_selector, log_display]
     
     async def master_handler(current_state, action, value=None):
@@ -121,7 +152,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
         new_state, log = await handle_action(current_state, action, value)
         ui_updates = update_ui_components(new_state)
         
-        # **THE CRITICAL FIX IS HERE:**
+        # **BUG FIX 2: ValueError Fix**
         # We must return a tuple with a value for EACH output component, in the correct order.
         return (
             new_state, # 1. For the browser_state output
@@ -129,29 +160,16 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
             ui_updates[url_textbox], # 3.
             ui_updates[links_display], # 4.
             ui_updates[tab_selector], # 5.
-            ui_updates[log_display] # 6.
+            log # 6. The log textbox is now updated correctly
         )
     
-    # **THE SECOND CRITICAL FIX IS HERE:**
-    # Each event listener is a simple, separate async function that Gradio can correctly await.
-    
-    async def on_load(state):
-        return await master_handler(state, "new_tab", None)
-    
-    async def on_go_click(state, value):
-        return await master_handler(state, "go", value)
-        
-    async def on_click_link(state, value):
-        return await master_handler(state, "click", value)
-        
-    async def on_new_tab(state):
-        return await master_handler(state, "new_tab", None)
-        
-    async def on_close_tab(state):
-        return await master_handler(state, "close_tab", None)
-        
-    async def on_switch_tab(state, value):
-        return await master_handler(state, "switch_tab", value)
+    # Define simple async lambdas for each event to ensure clean wiring
+    on_load = lambda s: master_handler(s, "new_tab", None)
+    on_go_click = lambda s, v: master_handler(s, "go", v)
+    on_click_link = lambda s, v: master_handler(s, "click", v)
+    on_new_tab = lambda s: master_handler(s, "new_tab", None)
+    on_close_tab = lambda s: master_handler(s, "close_tab", None)
+    on_switch_tab = lambda s, v: master_handler(s, "switch_tab", v)
         
     # Wire up the new, clean event handlers
     demo.load(on_load, inputs=[browser_state], outputs=[browser_state, *all_outputs])