Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,228 +1,246 @@
|
|
1 |
# app.py
|
2 |
|
3 |
import gradio as gr
|
4 |
-
import
|
|
|
|
|
5 |
import datetime
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
#
|
10 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
class Tab:
|
13 |
-
"""Represents a single browser tab
|
14 |
-
def __init__(self,
|
15 |
-
self.
|
16 |
-
self.
|
17 |
-
self.
|
18 |
-
self.
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
def go_back(self):
|
29 |
-
if self.current_index > 0:
|
30 |
-
self.current_index -= 1
|
31 |
-
return f"Going back to {self.current_url()}"
|
32 |
-
return "No more history to go back to."
|
33 |
-
|
34 |
-
def go_forward(self):
|
35 |
-
if self.current_index < len(self.history) - 1:
|
36 |
-
self.current_index += 1
|
37 |
-
return f"Going forward to {self.current_url()}"
|
38 |
-
return "No more history to go forward to."
|
39 |
-
|
40 |
-
def current_url(self):
|
41 |
-
return self.history[self.current_index] if self.current_index != -1 else "about:blank"
|
42 |
-
|
43 |
-
class PseudoBrowser:
|
44 |
-
"""Simulates a web browser's core functionality."""
|
45 |
def __init__(self):
|
46 |
-
self.
|
47 |
-
self.
|
48 |
-
self.active_tab_index = 0
|
49 |
self.bookmarks = set()
|
50 |
self.global_history = []
|
51 |
-
self.
|
52 |
|
53 |
def _get_active_tab(self):
|
|
|
|
|
54 |
return self.tabs[self.active_tab_index]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
|
56 |
-
def
|
57 |
-
"""
|
58 |
-
log = ""
|
59 |
-
if url in self.cache and not refresh:
|
60 |
-
log += f"⚡️ Loading '{url}' from cache.\n"
|
61 |
-
content = self.cache[url]
|
62 |
-
else:
|
63 |
-
log += f"☁️ Fetching '{url}' from network...\n"
|
64 |
-
content = f"<html><body><h1>Welcome to {url}</h1><p><i>Content 'rendered' at {datetime.datetime.now().strftime('%H:%M:%S')}</i></p></body></html>"
|
65 |
-
self.cache[url] = content
|
66 |
-
return content, log
|
67 |
-
|
68 |
-
def open_url(self, url):
|
69 |
tab = self._get_active_tab()
|
70 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
self.global_history.append((datetime.datetime.now(), url))
|
72 |
-
|
73 |
-
return content, log + "\n" + fetch_log
|
74 |
|
75 |
def back(self):
|
76 |
tab = self._get_active_tab()
|
77 |
-
|
78 |
-
|
79 |
-
|
|
|
|
|
80 |
|
81 |
def forward(self):
|
82 |
tab = self._get_active_tab()
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
|
|
87 |
def refresh(self):
|
88 |
tab = self._get_active_tab()
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
return
|
93 |
|
94 |
def new_tab(self):
|
95 |
-
|
96 |
-
|
|
|
|
|
97 |
self.active_tab_index = len(self.tabs) - 1
|
98 |
-
|
99 |
-
return content, f"✨ New tab opened.\n" + fetch_log
|
100 |
|
101 |
def close_tab(self):
|
102 |
-
if len(self.tabs)
|
103 |
-
return
|
104 |
|
105 |
-
|
106 |
-
|
|
|
107 |
if self.active_tab_index >= len(self.tabs):
|
108 |
self.active_tab_index = len(self.tabs) - 1
|
109 |
|
110 |
-
|
111 |
-
|
112 |
-
return content, log + "\n" + fetch_log
|
113 |
-
|
114 |
-
def switch_tab(self, tab_label):
|
115 |
-
# The label is "Tab 0: https://..."
|
116 |
-
index = int(tab_label.split(":")[0].replace("Tab", "").strip())
|
117 |
-
if 0 <= index < len(self.tabs):
|
118 |
-
self.active_tab_index = index
|
119 |
-
tab = self._get_active_tab()
|
120 |
-
content, fetch_log = self._fetch_content(tab.current_url())
|
121 |
-
return content, f"Switched to Tab {index}.\n" + fetch_log
|
122 |
-
return None, "❌ Invalid tab index."
|
123 |
-
|
124 |
-
def add_bookmark(self):
|
125 |
-
url = self._get_active_tab().current_url()
|
126 |
-
if url not in self.bookmarks:
|
127 |
-
self.bookmarks.add(url)
|
128 |
-
return f"⭐ Bookmarked: {url}"
|
129 |
-
return f"ℹ️ Already bookmarked: {url}"
|
130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
# --- Gradio UI and Event Handlers ---
|
132 |
|
133 |
-
def update_ui_components(
|
134 |
"""Generates all UI component values from the browser state."""
|
135 |
-
|
136 |
-
|
137 |
-
|
138 |
-
|
139 |
-
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
-
|
144 |
-
# History
|
145 |
-
history_md = "### 📜 Global History\n"
|
146 |
-
if not browser.global_history:
|
147 |
-
history_md += "_(empty)_"
|
148 |
-
else:
|
149 |
-
for ts, url in reversed(browser.global_history[-10:]): # Show last 10
|
150 |
-
history_md += f"- `{ts.strftime('%H:%M:%S')}`: {url}\n"
|
151 |
|
152 |
-
#
|
153 |
-
|
154 |
-
|
155 |
-
|
|
|
|
|
|
|
|
|
|
|
156 |
else:
|
157 |
-
|
158 |
-
|
159 |
-
|
160 |
return {
|
161 |
-
|
162 |
-
url_textbox: gr.Textbox(value=
|
|
|
163 |
tab_selector: gr.Radio(choices=tab_choices, value=active_tab_label, label="Active Tabs"),
|
164 |
-
history_display: gr.Markdown(history_md),
|
165 |
-
bookmarks_display: gr.Markdown(bookmarks_md)
|
166 |
}
|
167 |
|
168 |
# --- Event Handlers ---
|
169 |
-
|
170 |
-
|
171 |
-
|
172 |
-
|
173 |
-
|
174 |
-
|
175 |
-
|
176 |
-
|
177 |
-
|
178 |
-
|
179 |
-
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
|
184 |
-
|
185 |
-
|
186 |
-
|
187 |
-
|
188 |
-
|
189 |
-
|
190 |
-
|
191 |
-
|
192 |
-
|
193 |
-
|
194 |
-
|
195 |
-
|
196 |
-
|
197 |
-
|
198 |
-
def handle_new_tab(browser_state):
|
199 |
-
content, log = browser_state.new_tab()
|
200 |
-
return {
|
201 |
-
**update_ui_components(browser_state),
|
202 |
-
log_display: gr.Textbox(log)
|
203 |
-
}
|
204 |
-
|
205 |
-
def handle_close_tab(browser_state):
|
206 |
-
content, log = browser_state.close_tab()
|
207 |
-
# Update UI components. If content is None, it means the last tab wasn't closed.
|
208 |
-
updated_components = update_ui_components(browser_state)
|
209 |
-
if content is not None:
|
210 |
-
updated_components[html_view] = gr.HTML(value=content)
|
211 |
|
212 |
-
|
213 |
-
**updated_components,
|
214 |
-
log_display: gr.Textbox(log)
|
215 |
-
}
|
216 |
-
|
217 |
-
def handle_switch_tab(browser_state, tab_label):
|
218 |
-
content, log = browser_state.switch_tab(tab_label)
|
219 |
-
return {
|
220 |
-
**update_ui_components(browser_state),
|
221 |
-
log_display: gr.Textbox(log)
|
222 |
-
}
|
223 |
-
|
224 |
-
def handle_add_bookmark(browser_state):
|
225 |
-
log = browser_state.add_bookmark()
|
226 |
return {
|
227 |
**update_ui_components(browser_state),
|
228 |
log_display: gr.Textbox(log)
|
@@ -230,65 +248,62 @@ def handle_add_bookmark(browser_state):
|
|
230 |
|
231 |
# --- Gradio Interface Layout ---
|
232 |
|
233 |
-
with gr.Blocks(theme=gr.themes.Soft(), title="
|
234 |
-
|
|
|
235 |
|
236 |
-
gr.Markdown("# 🌐
|
237 |
-
gr.Markdown("
|
238 |
|
239 |
with gr.Row():
|
240 |
with gr.Column(scale=3):
|
241 |
-
# URL and Navigation Controls
|
242 |
with gr.Row():
|
243 |
-
back_btn = gr.Button("◀")
|
244 |
-
forward_btn = gr.Button("▶")
|
245 |
-
refresh_btn = gr.Button("🔄")
|
246 |
-
|
247 |
-
|
248 |
-
|
249 |
|
250 |
-
|
251 |
-
|
252 |
|
253 |
-
|
254 |
-
log_display = gr.Textbox(label="Log", interactive=False)
|
255 |
|
256 |
with gr.Column(scale=1):
|
257 |
-
# Tab Management
|
258 |
with gr.Row():
|
259 |
new_tab_btn = gr.Button("➕ New Tab")
|
260 |
-
close_tab_btn = gr.Button("❌ Close
|
261 |
tab_selector = gr.Radio(choices=[], label="Active Tabs", interactive=True)
|
262 |
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
|
|
267 |
|
268 |
# --- Component Wiring ---
|
269 |
-
|
270 |
-
# Define all outputs that can be updated
|
271 |
-
outputs = [html_view, url_textbox, tab_selector, history_display, bookmarks_display, log_display]
|
272 |
|
273 |
# Initial load
|
274 |
demo.load(
|
275 |
-
|
276 |
inputs=[browser_state],
|
277 |
-
outputs=
|
278 |
)
|
279 |
|
280 |
-
#
|
281 |
-
go_btn.click(
|
282 |
-
url_textbox.submit(
|
283 |
-
|
284 |
-
forward_btn.click(handle_forward, [browser_state], outputs)
|
285 |
-
refresh_btn.click(handle_refresh, [browser_state], outputs)
|
286 |
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
|
291 |
-
|
|
|
|
|
292 |
|
293 |
|
294 |
demo.launch()
|
|
|
1 |
# app.py
|
2 |
|
3 |
import gradio as gr
|
4 |
+
from playwright.sync_api import sync_playwright, Error as PlaywrightError
|
5 |
+
from bs4 import BeautifulSoup
|
6 |
+
import urllib.parse
|
7 |
import datetime
|
8 |
+
import atexit
|
9 |
+
import re
|
10 |
+
|
11 |
+
# --- GLOBAL PLAYWRIGHT SETUP ---
|
12 |
+
# Launch Playwright and a browser instance once when the app starts.
|
13 |
+
# This is crucial for performance and state management.
|
14 |
+
try:
|
15 |
+
p = sync_playwright().start()
|
16 |
+
# Using Firefox can sometimes be less prone to bot detection than Chromium.
|
17 |
+
# headless=True is essential for running on a server like Hugging Face Spaces.
|
18 |
+
browser = p.firefox.launch(headless=True, timeout=60000)
|
19 |
+
print("✅ Playwright browser launched successfully.")
|
20 |
+
except Exception as e:
|
21 |
+
print(f"❌ Could not launch Playwright browser: {e}")
|
22 |
+
# You might want to handle this more gracefully, but for a demo, exiting is fine.
|
23 |
+
exit()
|
24 |
+
|
25 |
+
# Ensure the browser is closed gracefully when the app exits.
|
26 |
+
def cleanup():
|
27 |
+
print("🧹 Cleaning up: Closing Playwright browser...")
|
28 |
+
browser.close()
|
29 |
+
p.stop()
|
30 |
+
atexit.register(cleanup)
|
31 |
+
|
32 |
+
|
33 |
+
# --- Core Browser Logic (Powered by Playwright) ---
|
34 |
|
35 |
class Tab:
|
36 |
+
"""Represents a single browser tab, now backed by a Playwright Page."""
|
37 |
+
def __init__(self, playwright_page):
|
38 |
+
self.page = playwright_page # The actual Playwright page object
|
39 |
+
self.title = "New Tab"
|
40 |
+
self.url = "about:blank"
|
41 |
+
self.parsed_text = "Welcome! Navigate to a URL or search to get started."
|
42 |
+
self.links = [] # A list of {'text': str, 'url': str}
|
43 |
+
|
44 |
+
def close(self):
|
45 |
+
"""Closes the underlying Playwright page."""
|
46 |
+
if not self.page.is_closed():
|
47 |
+
self.page.close()
|
48 |
+
|
49 |
+
class RealBrowser:
|
50 |
+
"""Manages multiple tabs and browser-level state."""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
def __init__(self):
|
52 |
+
self.tabs = []
|
53 |
+
self.active_tab_index = -1
|
|
|
54 |
self.bookmarks = set()
|
55 |
self.global_history = []
|
56 |
+
self.new_tab() # Start with one tab
|
57 |
|
58 |
def _get_active_tab(self):
|
59 |
+
if self.active_tab_index == -1 or self.active_tab_index >= len(self.tabs):
|
60 |
+
return None
|
61 |
return self.tabs[self.active_tab_index]
|
62 |
+
|
63 |
+
def _fetch_and_parse(self, tab, url):
|
64 |
+
"""Uses Playwright to navigate and BeautifulSoup to parse."""
|
65 |
+
log = f"▶️ Navigating to {url}..."
|
66 |
+
try:
|
67 |
+
# Navigate the page, waiting until the page is fully loaded.
|
68 |
+
# wait_until='domcontentloaded' is a good balance of speed and completeness.
|
69 |
+
tab.page.goto(url, wait_until='domcontentloaded', timeout=30000)
|
70 |
+
|
71 |
+
# Update tab state with the final URL after any redirects
|
72 |
+
tab.url = tab.page.url
|
73 |
+
tab.title = tab.page.title() or "No Title"
|
74 |
+
log += f"\n✅ Arrived at: {tab.url}"
|
75 |
+
log += f"\n📄 Title: {tab.title}"
|
76 |
+
|
77 |
+
# Get the fully-rendered HTML and parse it
|
78 |
+
html_content = tab.page.content()
|
79 |
+
soup = BeautifulSoup(html_content, 'lxml')
|
80 |
+
|
81 |
+
# Extract and clean text
|
82 |
+
for script in soup(["script", "style", "nav", "footer"]):
|
83 |
+
script.extract()
|
84 |
+
text = soup.get_text(separator='\n', strip=True)
|
85 |
+
tab.parsed_text = text
|
86 |
+
|
87 |
+
# Extract links
|
88 |
+
tab.links = []
|
89 |
+
for link in soup.find_all('a', href=True):
|
90 |
+
href = link['href']
|
91 |
+
absolute_url = urllib.parse.urljoin(tab.url, href)
|
92 |
+
# Filter out useless links
|
93 |
+
if absolute_url.startswith('http') and not re.match(r'javascript:|mailto:', absolute_url):
|
94 |
+
link_text = link.get_text(strip=True) or "[No Link Text]"
|
95 |
+
tab.links.append({'text': link_text, 'url': absolute_url})
|
96 |
+
log += f"\n🔗 Found {len(tab.links)} links."
|
97 |
+
|
98 |
+
except PlaywrightError as e:
|
99 |
+
error_message = str(e)
|
100 |
+
if "net::ERR" in error_message:
|
101 |
+
error_message = "Network error: Could not resolve host or connect."
|
102 |
+
elif "Timeout" in error_message:
|
103 |
+
error_message = f"Timeout: The page took too long to load."
|
104 |
+
|
105 |
+
tab.title = "Error"
|
106 |
+
tab.url = url
|
107 |
+
tab.parsed_text = f"❌ Failed to load page.\n\nError: {error_message}"
|
108 |
+
tab.links = []
|
109 |
+
log += f"\n❌ {error_message}"
|
110 |
+
|
111 |
+
return log
|
112 |
|
113 |
+
def go(self, term_or_url):
|
114 |
+
"""Opens a URL or performs a search in the active tab."""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
tab = self._get_active_tab()
|
116 |
+
if not tab: return "No active tab."
|
117 |
+
|
118 |
+
# Check if it's a URL or a search term
|
119 |
+
parsed_url = urllib.parse.urlparse(term_or_url)
|
120 |
+
if parsed_url.scheme and parsed_url.netloc:
|
121 |
+
url = term_or_url
|
122 |
+
else:
|
123 |
+
url = f"https://duckduckgo.com/html/?q={urllib.parse.quote_plus(term_or_url)}"
|
124 |
+
|
125 |
self.global_history.append((datetime.datetime.now(), url))
|
126 |
+
return self._fetch_and_parse(tab, url)
|
|
|
127 |
|
128 |
def back(self):
|
129 |
tab = self._get_active_tab()
|
130 |
+
if tab and tab.page.can_go_back():
|
131 |
+
# Playwright's go_back is async-like, we need to re-parse
|
132 |
+
tab.page.go_back(wait_until='domcontentloaded')
|
133 |
+
return self._fetch_and_parse(tab, tab.page.url)
|
134 |
+
return "Cannot go back."
|
135 |
|
136 |
def forward(self):
|
137 |
tab = self._get_active_tab()
|
138 |
+
if tab and tab.page.can_go_forward():
|
139 |
+
tab.page.go_forward(wait_until='domcontentloaded')
|
140 |
+
return self._fetch_and_parse(tab, tab.page.url)
|
141 |
+
return "Cannot go forward."
|
142 |
+
|
143 |
def refresh(self):
|
144 |
tab = self._get_active_tab()
|
145 |
+
if tab:
|
146 |
+
tab.page.reload(wait_until='domcontentloaded')
|
147 |
+
return self._fetch_and_parse(tab, tab.page.url)
|
148 |
+
return "No active tab."
|
149 |
|
150 |
def new_tab(self):
|
151 |
+
# Create a new page in the persistent browser context
|
152 |
+
page = browser.new_page()
|
153 |
+
tab = Tab(page)
|
154 |
+
self.tabs.append(tab)
|
155 |
self.active_tab_index = len(self.tabs) - 1
|
156 |
+
return self.go("https://duckduckgo.com/html/?q=news") # Navigate new tab to a default search
|
|
|
157 |
|
158 |
def close_tab(self):
|
159 |
+
if len(self.tabs) <= 1:
|
160 |
+
return "Cannot close the last tab."
|
161 |
|
162 |
+
tab_to_close = self.tabs.pop(self.active_tab_index)
|
163 |
+
tab_to_close.close()
|
164 |
+
|
165 |
if self.active_tab_index >= len(self.tabs):
|
166 |
self.active_tab_index = len(self.tabs) - 1
|
167 |
|
168 |
+
# No need to re-fetch, just update the UI state
|
169 |
+
return f"Tab closed. Switched to Tab {self.active_tab_index}."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
170 |
|
171 |
+
def switch_tab(self, tab_label):
|
172 |
+
try:
|
173 |
+
index = int(tab_label.split(":")[0].replace("Tab", "").strip())
|
174 |
+
if 0 <= index < len(self.tabs):
|
175 |
+
self.active_tab_index = index
|
176 |
+
return f"Switched to Tab {index}."
|
177 |
+
return "Invalid tab index."
|
178 |
+
except (ValueError, IndexError):
|
179 |
+
return "Invalid tab format."
|
180 |
+
|
181 |
# --- Gradio UI and Event Handlers ---
|
182 |
|
183 |
+
def update_ui_components(browser_state: RealBrowser):
|
184 |
"""Generates all UI component values from the browser state."""
|
185 |
+
active_tab = browser_state._get_active_tab()
|
186 |
+
if not active_tab:
|
187 |
+
return {
|
188 |
+
page_content: gr.Markdown("No active tabs. Please create a new one."),
|
189 |
+
url_textbox: "",
|
190 |
+
links_display: "",
|
191 |
+
tab_selector: gr.Radio(choices=[], label="Active Tabs"),
|
192 |
+
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
193 |
|
194 |
+
# Tab Selector
|
195 |
+
tab_choices = [f"Tab {i}: {tab.title[:40]}..." for i, tab in enumerate(browser_state.tabs)]
|
196 |
+
active_tab_label = f"Tab {browser_state.active_tab_index}: {active_tab.title[:40]}..."
|
197 |
+
|
198 |
+
# Links Display
|
199 |
+
links_md = "### 🔗 Links on Page\n"
|
200 |
+
if active_tab.links:
|
201 |
+
for i, link in enumerate(active_tab.links[:25]): # Show first 25 links
|
202 |
+
links_md += f"{i}. [{link['text'][:80]}]({link['url']})\n"
|
203 |
else:
|
204 |
+
links_md += "_No links found or page failed to load._"
|
205 |
+
|
|
|
206 |
return {
|
207 |
+
page_content: gr.Markdown(f"# {active_tab.title}\n**URL:** {active_tab.url}\n\n---\n\n{active_tab.parsed_text[:2000]}..."),
|
208 |
+
url_textbox: gr.Textbox(value=active_tab.url),
|
209 |
+
links_display: gr.Markdown(links_md),
|
210 |
tab_selector: gr.Radio(choices=tab_choices, value=active_tab_label, label="Active Tabs"),
|
|
|
|
|
211 |
}
|
212 |
|
213 |
# --- Event Handlers ---
|
214 |
+
def handle_action(browser_state, action, value=None):
|
215 |
+
if action == "go":
|
216 |
+
log = browser_state.go(value)
|
217 |
+
elif action == "click":
|
218 |
+
tab = browser_state._get_active_tab()
|
219 |
+
try:
|
220 |
+
link_index = int(value)
|
221 |
+
if tab and 0 <= link_index < len(tab.links):
|
222 |
+
link_url = tab.links[link_index]['url']
|
223 |
+
log = browser_state.go(link_url)
|
224 |
+
else:
|
225 |
+
log = "Invalid link number."
|
226 |
+
except (ValueError, TypeError):
|
227 |
+
log = "Please enter a valid number to click."
|
228 |
+
elif action == "back":
|
229 |
+
log = browser_state.back()
|
230 |
+
elif action == "forward":
|
231 |
+
log = browser_state.forward()
|
232 |
+
elif action == "refresh":
|
233 |
+
log = browser_state.refresh()
|
234 |
+
elif action == "new_tab":
|
235 |
+
log = browser_state.new_tab()
|
236 |
+
elif action == "close_tab":
|
237 |
+
log = browser_state.close_tab()
|
238 |
+
elif action == "switch_tab":
|
239 |
+
log = browser_state.switch_tab(value)
|
240 |
+
else:
|
241 |
+
log = "Unknown action."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
242 |
|
243 |
+
# After any action, update the entire UI based on the new state
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
244 |
return {
|
245 |
**update_ui_components(browser_state),
|
246 |
log_display: gr.Textbox(log)
|
|
|
248 |
|
249 |
# --- Gradio Interface Layout ---
|
250 |
|
251 |
+
with gr.Blocks(theme=gr.themes.Soft(), title="Real Browser Demo") as demo:
|
252 |
+
# The gr.State holds our Python class instance, persisting it across calls.
|
253 |
+
browser_state = gr.State(RealBrowser())
|
254 |
|
255 |
+
gr.Markdown("# 🌐 Real Browser Demo (Powered by Playwright)")
|
256 |
+
gr.Markdown("Type a URL or search term. This demo runs a real headless browser on the server to fetch and parse live websites.")
|
257 |
|
258 |
with gr.Row():
|
259 |
with gr.Column(scale=3):
|
|
|
260 |
with gr.Row():
|
261 |
+
back_btn = gr.Button("◀ Back")
|
262 |
+
forward_btn = gr.Button("▶ Forward")
|
263 |
+
refresh_btn = gr.Button("🔄 Refresh")
|
264 |
+
|
265 |
+
url_textbox = gr.Textbox(label="URL or Search Term", placeholder="https://news.ycombinator.com or 'best python libraries'", interactive=True)
|
266 |
+
go_btn = gr.Button("Go", variant="primary")
|
267 |
|
268 |
+
with gr.Accordion("Page Content (Text Only)", open=True):
|
269 |
+
page_content = gr.Markdown("Loading...")
|
270 |
|
271 |
+
log_display = gr.Textbox(label="Status Log", interactive=False)
|
|
|
272 |
|
273 |
with gr.Column(scale=1):
|
|
|
274 |
with gr.Row():
|
275 |
new_tab_btn = gr.Button("➕ New Tab")
|
276 |
+
close_tab_btn = gr.Button("❌ Close Tab")
|
277 |
tab_selector = gr.Radio(choices=[], label="Active Tabs", interactive=True)
|
278 |
|
279 |
+
with gr.Accordion("Clickable Links", open=True):
|
280 |
+
links_display = gr.Markdown("...")
|
281 |
+
with gr.Row():
|
282 |
+
click_num_box = gr.Number(label="Link #", scale=1, minimum=0, step=1)
|
283 |
+
click_btn = gr.Button("Click Link", scale=2)
|
284 |
|
285 |
# --- Component Wiring ---
|
286 |
+
all_outputs = [page_content, url_textbox, links_display, tab_selector, log_display]
|
|
|
|
|
287 |
|
288 |
# Initial load
|
289 |
demo.load(
|
290 |
+
lambda state: {**update_ui_components(state), log_display: "🚀 Browser Initialized! Ready to navigate."},
|
291 |
inputs=[browser_state],
|
292 |
+
outputs=all_outputs
|
293 |
)
|
294 |
|
295 |
+
# Event listeners
|
296 |
+
go_btn.click(lambda s, v: handle_action(s, "go", v), [browser_state, url_textbox], all_outputs, show_progress="full")
|
297 |
+
url_textbox.submit(lambda s, v: handle_action(s, "go", v), [browser_state, url_textbox], all_outputs, show_progress="full")
|
298 |
+
click_btn.click(lambda s, v: handle_action(s, "click", v), [browser_state, click_num_box], all_outputs, show_progress="full")
|
|
|
|
|
299 |
|
300 |
+
back_btn.click(lambda s: handle_action(s, "back"), [browser_state], all_outputs, show_progress="full")
|
301 |
+
forward_btn.click(lambda s: handle_action(s, "forward"), [browser_state], all_outputs, show_progress="full")
|
302 |
+
refresh_btn.click(lambda s: handle_action(s, "refresh"), [browser_state], all_outputs, show_progress="full")
|
303 |
|
304 |
+
new_tab_btn.click(lambda s: handle_action(s, "new_tab"), [browser_state], all_outputs, show_progress="full")
|
305 |
+
close_tab_btn.click(lambda s: handle_action(s, "close_tab"), [browser_state], all_outputs)
|
306 |
+
tab_selector.input(lambda s, v: handle_action(s, "switch_tab", v), [browser_state, tab_selector], all_outputs)
|
307 |
|
308 |
|
309 |
demo.launch()
|