Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
import os
|
2 |
os.system("playwright install")
|
3 |
-
# app.py (
|
4 |
|
5 |
import gradio as gr
|
6 |
-
from playwright.
|
7 |
from bs4 import BeautifulSoup
|
8 |
import urllib.parse
|
9 |
import datetime
|
@@ -13,14 +13,26 @@ import os
|
|
13 |
from itertools import cycle
|
14 |
import uuid
|
15 |
|
16 |
-
# --- 1.
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
|
|
22 |
|
23 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
class TabState:
|
25 |
def __init__(self, tab_id, proxy_used="Direct Connection"):
|
26 |
self.id, self.url, self.title = tab_id, "about:blank", "New Tab"
|
@@ -31,7 +43,9 @@ class BrowserState:
|
|
31 |
def __init__(self): self.tabs, self.active_tab_id = [], None
|
32 |
def get_active_tab(self): return next((t for t in self.tabs if t.id == self.active_tab_id), None)
|
33 |
|
34 |
-
|
|
|
|
|
35 |
def __init__(self, proxy_string: str):
|
36 |
self.proxies = self._parse_proxies(proxy_string)
|
37 |
if self.proxies: self.proxy_cycler = cycle(self.proxies)
|
@@ -45,18 +59,16 @@ class CredentialRevolver: # Unchanged
|
|
45 |
def get_next(self): return next(self.proxy_cycler) if self.proxy_cycler else None
|
46 |
def count(self): return len(self.proxies)
|
47 |
|
48 |
-
|
49 |
-
|
50 |
|
51 |
-
|
52 |
-
log = f"▶️ Navigating to {url}..."
|
53 |
-
live_page = LIVE_CONTEXTS[tab_state.id]["page"]
|
54 |
try:
|
55 |
-
|
56 |
-
tab_state.url = live_page.url
|
57 |
-
tab_state.title = await live_page.title() or "No Title"
|
58 |
log += f"\n✅ Arrived at: {tab_state.url}"
|
59 |
-
html_content =
|
60 |
soup = BeautifulSoup(html_content, 'lxml')
|
61 |
for script in soup(["script", "style", "nav", "footer"]): script.extract()
|
62 |
tab_state.parsed_text = soup.get_text(separator='\n', strip=True)
|
@@ -68,26 +80,26 @@ async def _fetch_and_update_tab_state(tab_state: TabState, url: str):
|
|
68 |
tab_state.links = []; log += f"\n❌ {error_message}"
|
69 |
return log
|
70 |
|
71 |
-
|
72 |
log = ""; active_tab_state = browser_state.get_active_tab()
|
73 |
if action == "new_tab":
|
74 |
-
tab_id = str(uuid.uuid4())
|
75 |
-
|
76 |
-
|
77 |
-
page = await context.new_page()
|
78 |
LIVE_CONTEXTS[tab_id] = {"context": context, "page": page}
|
79 |
new_tab = TabState(tab_id, proxy_used=proxy_config['server'] if proxy_config else "Direct")
|
80 |
browser_state.tabs.append(new_tab)
|
81 |
browser_state.active_tab_id = tab_id
|
82 |
-
log =
|
83 |
elif action == "go" and active_tab_state:
|
84 |
url = value if (urllib.parse.urlparse(value).scheme and urllib.parse.urlparse(value).netloc) else f"https://duckduckgo.com/html/?q={urllib.parse.quote_plus(value)}"
|
85 |
-
log =
|
86 |
elif action == "click" and active_tab_state and value is not None:
|
87 |
try:
|
88 |
link_index = int(value)
|
89 |
if 0 <= link_index < len(active_tab_state.links):
|
90 |
-
|
|
|
91 |
else: log = "Invalid link number."
|
92 |
except: log = "Please enter a valid number to click."
|
93 |
elif action == "close_tab" and active_tab_state:
|
@@ -98,14 +110,14 @@ async def handle_action(browser_state: BrowserState, action: str, value=None):
|
|
98 |
new_index = tab_index - 1 if tab_index > 0 else 0
|
99 |
browser_state.active_tab_id = browser_state.tabs[new_index].id
|
100 |
resources = LIVE_CONTEXTS.pop(tab_to_close_id)
|
101 |
-
|
102 |
log = f"💣 Tab closed."
|
103 |
else: log = "Cannot close the last tab."
|
104 |
elif action == "switch_tab" and value is not None:
|
105 |
browser_state.active_tab_id = value; log = f"Switched to tab."
|
106 |
return browser_state, log
|
107 |
|
108 |
-
def update_ui_components(browser_state: BrowserState):
|
109 |
active_tab = browser_state.get_active_tab()
|
110 |
if not active_tab: return {page_content: gr.Markdown("No active tabs."), url_textbox: "", links_display: "", tab_selector: gr.Radio(choices=[])}
|
111 |
tab_choices = [(f"Tab {i}: {t.title[:25]}... (via {t.proxy_used})", t.id) for i, t in enumerate(browser_state.tabs)]
|
@@ -116,26 +128,11 @@ def update_ui_components(browser_state: BrowserState): # This function is not as
|
|
116 |
tab_selector: gr.Radio(choices=tab_choices, value=active_tab.id, label="Active Tabs"),
|
117 |
}
|
118 |
|
119 |
-
# ---
|
120 |
-
async def startup():
|
121 |
-
global P, BROWSER, REVOLVER
|
122 |
-
print("🚀 App starting up...")
|
123 |
-
P = await async_playwright().start()
|
124 |
-
BROWSER = await P.firefox.launch(headless=True)
|
125 |
-
proxy_list_str = os.getenv("PROXY_LIST", "")
|
126 |
-
REVOLVER = CredentialRevolver(proxy_list_str)
|
127 |
-
print(f"✅ Playwright started. {REVOLVER.count()} proxies loaded.")
|
128 |
-
|
129 |
-
async def shutdown():
|
130 |
-
print("🧹 App shutting down...")
|
131 |
-
if BROWSER: await BROWSER.close()
|
132 |
-
if P: await P.stop()
|
133 |
-
print("✅ Playwright stopped.")
|
134 |
-
|
135 |
with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
|
136 |
browser_state = gr.State(BrowserState())
|
137 |
-
gr.Markdown("# 🛰️ Real Browser Demo (
|
138 |
-
|
139 |
with gr.Row():
|
140 |
with gr.Column(scale=3):
|
141 |
url_textbox = gr.Textbox(label="URL or Search Term", interactive=True)
|
@@ -146,44 +143,29 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
|
|
146 |
with gr.Row(): new_tab_btn = gr.Button("➕ New Tab"); close_tab_btn = gr.Button("❌ Close Tab")
|
147 |
tab_selector = gr.Radio(choices=[], label="Active Tabs", interactive=True)
|
148 |
with gr.Accordion("Clickable Links", open=True):
|
149 |
-
links_display = gr.Markdown("...")
|
150 |
with gr.Row(): click_num_box = gr.Number(label="Link #", scale=1, minimum=0, step=1); click_btn = gr.Button("Click Link", scale=2)
|
151 |
-
|
152 |
all_outputs = [page_content, url_textbox, links_display, tab_selector, log_display]
|
153 |
|
154 |
-
|
155 |
-
new_state, log =
|
156 |
-
ui_updates = update_ui_components(new_state)
|
157 |
-
ui_updates[log_display] = log
|
158 |
return new_state, ui_updates
|
159 |
|
160 |
-
|
161 |
-
|
162 |
-
|
|
|
163 |
return ui_updates
|
164 |
-
|
165 |
demo.load(initial_load, inputs=[browser_state], outputs=list(all_outputs))
|
166 |
|
167 |
-
# Event listeners -
|
168 |
-
go_btn.click(master_handler, [browser_state, url_textbox], [browser_state, *all_outputs],
|
169 |
-
url_textbox.submit(master_handler, [browser_state, url_textbox], [browser_state, *all_outputs],
|
170 |
-
click_btn.click(master_handler, [browser_state, click_num_box], [browser_state, *all_outputs],
|
171 |
-
new_tab_btn.click(master_handler, [browser_state], [browser_state, *all_outputs],
|
172 |
-
close_tab_btn.click(master_handler, [browser_state], [browser_state, *all_outputs]
|
173 |
-
tab_selector.input(master_handler, [browser_state, tab_selector], [browser_state, *all_outputs]
|
174 |
-
|
175 |
-
# The `startup` and `shutdown` events are not available in standard `Blocks`.
|
176 |
-
# We manage this by doing the startup inside the first `load` event.
|
177 |
-
# The global setup is the best way for Spaces.
|
178 |
-
# A small tweak to make it work without official startup events:
|
179 |
-
# The startup logic is moved to the global scope but needs an async context.
|
180 |
-
# We will use the existing startup/shutdown logic and assume the Gradio version supports it or handles it gracefully.
|
181 |
-
# The best practice would be to use a framework that has explicit startup/shutdown events like FastAPI.
|
182 |
-
# For Gradio Spaces, the singleton pattern with async calls in handlers is the way to go.
|
183 |
-
|
184 |
-
# Let's adjust for standard Gradio deployment.
|
185 |
-
# We'll run startup manually before launching.
|
186 |
-
import asyncio
|
187 |
-
asyncio.run(startup())
|
188 |
|
189 |
demo.launch()
|
|
|
1 |
import os
|
2 |
os.system("playwright install")
|
3 |
+
# app.py (Synchronous, state-separated, and Gradio API fix)
|
4 |
|
5 |
import gradio as gr
|
6 |
+
from playwright.sync_api import sync_playwright, Error as PlaywrightError
|
7 |
from bs4 import BeautifulSoup
|
8 |
import urllib.parse
|
9 |
import datetime
|
|
|
13 |
from itertools import cycle
|
14 |
import uuid
|
15 |
|
16 |
+
# --- 1. LIVE RESOURCES (Global, Non-copyable) ---
|
17 |
+
try:
|
18 |
+
p = sync_playwright().start()
|
19 |
+
browser = p.firefox.launch(headless=True, timeout=60000)
|
20 |
+
print("✅ Playwright browser launched successfully.")
|
21 |
+
except Exception as e:
|
22 |
+
print(f"❌ Could not launch Playwright browser. Original error: {e}"); exit()
|
23 |
|
24 |
+
LIVE_CONTEXTS = {} # { tab_id: { "context": PlaywrightContext, "page": PlaywrightPage } }
|
25 |
+
|
26 |
+
def cleanup():
|
27 |
+
print(f"🧹 Cleaning up: Closing {len(LIVE_CONTEXTS)} browser contexts...")
|
28 |
+
for tab_id, resources in LIVE_CONTEXTS.items():
|
29 |
+
if not resources["context"].is_closed():
|
30 |
+
resources["context"].close()
|
31 |
+
browser.close(); p.stop()
|
32 |
+
atexit.register(cleanup)
|
33 |
+
|
34 |
+
|
35 |
+
# --- 2. STATE DATA (Plain, Copyable Classes) ---
|
36 |
class TabState:
|
37 |
def __init__(self, tab_id, proxy_used="Direct Connection"):
|
38 |
self.id, self.url, self.title = tab_id, "about:blank", "New Tab"
|
|
|
43 |
def __init__(self): self.tabs, self.active_tab_id = [], None
|
44 |
def get_active_tab(self): return next((t for t in self.tabs if t.id == self.active_tab_id), None)
|
45 |
|
46 |
+
|
47 |
+
# --- 3. LOGIC ---
|
48 |
+
class CredentialRevolver:
|
49 |
def __init__(self, proxy_string: str):
|
50 |
self.proxies = self._parse_proxies(proxy_string)
|
51 |
if self.proxies: self.proxy_cycler = cycle(self.proxies)
|
|
|
59 |
def get_next(self): return next(self.proxy_cycler) if self.proxy_cycler else None
|
60 |
def count(self): return len(self.proxies)
|
61 |
|
62 |
+
proxy_list_str = os.getenv("PROXY_LIST", "")
|
63 |
+
revolver = CredentialRevolver(proxy_list_str)
|
64 |
|
65 |
+
def _fetch_and_update_tab_state(tab_state: TabState, url: str):
|
66 |
+
log = f"▶️ Navigating to {url}..."; live_page = LIVE_CONTEXTS[tab_state.id]["page"]
|
|
|
67 |
try:
|
68 |
+
live_page.goto(url, wait_until='domcontentloaded', timeout=30000)
|
69 |
+
tab_state.url, tab_state.title = live_page.url, live_page.title() or "No Title"
|
|
|
70 |
log += f"\n✅ Arrived at: {tab_state.url}"
|
71 |
+
html_content = live_page.content()
|
72 |
soup = BeautifulSoup(html_content, 'lxml')
|
73 |
for script in soup(["script", "style", "nav", "footer"]): script.extract()
|
74 |
tab_state.parsed_text = soup.get_text(separator='\n', strip=True)
|
|
|
80 |
tab_state.links = []; log += f"\n❌ {error_message}"
|
81 |
return log
|
82 |
|
83 |
+
def handle_action(browser_state: BrowserState, action: str, value=None):
|
84 |
log = ""; active_tab_state = browser_state.get_active_tab()
|
85 |
if action == "new_tab":
|
86 |
+
tab_id, proxy_config = str(uuid.uuid4()), revolver.get_next()
|
87 |
+
context = browser.new_context(proxy=proxy_config)
|
88 |
+
page = context.new_page()
|
|
|
89 |
LIVE_CONTEXTS[tab_id] = {"context": context, "page": page}
|
90 |
new_tab = TabState(tab_id, proxy_used=proxy_config['server'] if proxy_config else "Direct")
|
91 |
browser_state.tabs.append(new_tab)
|
92 |
browser_state.active_tab_id = tab_id
|
93 |
+
log = _fetch_and_update_tab_state(new_tab, "https://www.whatsmyip.org/")
|
94 |
elif action == "go" and active_tab_state:
|
95 |
url = value if (urllib.parse.urlparse(value).scheme and urllib.parse.urlparse(value).netloc) else f"https://duckduckgo.com/html/?q={urllib.parse.quote_plus(value)}"
|
96 |
+
log = _fetch_and_update_tab_state(active_tab_state, url)
|
97 |
elif action == "click" and active_tab_state and value is not None:
|
98 |
try:
|
99 |
link_index = int(value)
|
100 |
if 0 <= link_index < len(active_tab_state.links):
|
101 |
+
link_url = active_tab_state.links[link_index]['url']
|
102 |
+
log = _fetch_and_update_tab_state(active_tab_state, link_url)
|
103 |
else: log = "Invalid link number."
|
104 |
except: log = "Please enter a valid number to click."
|
105 |
elif action == "close_tab" and active_tab_state:
|
|
|
110 |
new_index = tab_index - 1 if tab_index > 0 else 0
|
111 |
browser_state.active_tab_id = browser_state.tabs[new_index].id
|
112 |
resources = LIVE_CONTEXTS.pop(tab_to_close_id)
|
113 |
+
if not resources['context'].is_closed(): resources['context'].close()
|
114 |
log = f"💣 Tab closed."
|
115 |
else: log = "Cannot close the last tab."
|
116 |
elif action == "switch_tab" and value is not None:
|
117 |
browser_state.active_tab_id = value; log = f"Switched to tab."
|
118 |
return browser_state, log
|
119 |
|
120 |
+
def update_ui_components(browser_state: BrowserState):
|
121 |
active_tab = browser_state.get_active_tab()
|
122 |
if not active_tab: return {page_content: gr.Markdown("No active tabs."), url_textbox: "", links_display: "", tab_selector: gr.Radio(choices=[])}
|
123 |
tab_choices = [(f"Tab {i}: {t.title[:25]}... (via {t.proxy_used})", t.id) for i, t in enumerate(browser_state.tabs)]
|
|
|
128 |
tab_selector: gr.Radio(choices=tab_choices, value=active_tab.id, label="Active Tabs"),
|
129 |
}
|
130 |
|
131 |
+
# --- Gradio UI Layout ---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
132 |
with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
|
133 |
browser_state = gr.State(BrowserState())
|
134 |
+
gr.Markdown("# 🛰️ Real Browser Demo (Synchronous Fix)")
|
135 |
+
gr.Markdown(f"This demo runs a real headless browser. **{revolver.count()} proxies loaded**.")
|
136 |
with gr.Row():
|
137 |
with gr.Column(scale=3):
|
138 |
url_textbox = gr.Textbox(label="URL or Search Term", interactive=True)
|
|
|
143 |
with gr.Row(): new_tab_btn = gr.Button("➕ New Tab"); close_tab_btn = gr.Button("❌ Close Tab")
|
144 |
tab_selector = gr.Radio(choices=[], label="Active Tabs", interactive=True)
|
145 |
with gr.Accordion("Clickable Links", open=True):
|
146 |
+
links_display = gr.Markdown("...");
|
147 |
with gr.Row(): click_num_box = gr.Number(label="Link #", scale=1, minimum=0, step=1); click_btn = gr.Button("Click Link", scale=2)
|
148 |
+
|
149 |
all_outputs = [page_content, url_textbox, links_display, tab_selector, log_display]
|
150 |
|
151 |
+
def master_handler(current_state, action, value):
|
152 |
+
new_state, log = handle_action(current_state, action, value)
|
153 |
+
ui_updates = update_ui_components(new_state); ui_updates[log_display] = log
|
|
|
154 |
return new_state, ui_updates
|
155 |
|
156 |
+
def initial_load(s):
|
157 |
+
# A small hack to initialize the browser state with one tab on load
|
158 |
+
# We call the handler directly but only return the UI update part.
|
159 |
+
_, ui_updates = master_handler(s, "new_tab", None)
|
160 |
return ui_updates
|
|
|
161 |
demo.load(initial_load, inputs=[browser_state], outputs=list(all_outputs))
|
162 |
|
163 |
+
# Event listeners - FIXED by removing the `fn_name` argument.
|
164 |
+
go_btn.click(master_handler, [browser_state, url_textbox], [browser_state, *all_outputs], show_progress="full")
|
165 |
+
url_textbox.submit(master_handler, [browser_state, url_textbox], [browser_state, *all_outputs], show_progress="full")
|
166 |
+
click_btn.click(master_handler, [browser_state, click_num_box], [browser_state, *all_outputs], show_progress="full")
|
167 |
+
new_tab_btn.click(master_handler, [browser_state], [browser_state, *all_outputs], show_progress="full")
|
168 |
+
close_tab_btn.click(master_handler, [browser_state], [browser_state, *all_outputs])
|
169 |
+
tab_selector.input(master_handler, [browser_state, tab_selector], [browser_state, *all_outputs])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
170 |
|
171 |
demo.launch()
|