Spaces:
Running
Running
Trisha Tomy
commited on
Commit
·
c24ca69
1
Parent(s):
cbcd3e0
trying fixes for loading
Browse files- src/proxy_lite/browser/browser.py +45 -101
src/proxy_lite/browser/browser.py
CHANGED
@@ -179,116 +179,60 @@ class BrowserSession:
|
|
179 |
# re-run for cases of mid-run redirects
|
180 |
@retry(
|
181 |
wait=wait_exponential(multiplier=1, min=1, max=10),
|
182 |
-
stop=stop_after_delay(5),
|
183 |
reraise=True,
|
184 |
before_sleep=before_sleep_log(logger, logging.ERROR),
|
185 |
)
|
186 |
async def update_poi(self) -> None:
|
187 |
try:
|
188 |
-
#
|
189 |
-
|
|
|
190 |
logger.debug("wait_for_load_state('networkidle') completed.")
|
191 |
|
192 |
-
#
|
193 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
194 |
try:
|
195 |
-
|
196 |
-
|
197 |
-
|
198 |
-
|
199 |
-
|
200 |
-
logger.error(f"DEBUGGING: Playwright Timeout (180s) on body selector at URL: {current_url}")
|
201 |
-
|
202 |
-
html_content = None
|
203 |
-
try:
|
204 |
-
if self.current_page:
|
205 |
-
html_content = await self.current_page.content()
|
206 |
-
# Log only a snippet of HTML to avoid excessively large logs
|
207 |
-
logger.error(f"DEBUGGING: HTML Content (first 1000 chars) when timeout occurred:\n{html_content[:1000]}...")
|
208 |
-
except Exception as html_e:
|
209 |
-
logger.error(f"DEBUGGING: Could not get HTML content: {html_e}")
|
210 |
-
|
211 |
-
screenshot_b64 = "N/A"
|
212 |
-
try:
|
213 |
-
if self.current_page:
|
214 |
-
# Capture screenshot at lower quality to keep log size manageable
|
215 |
-
screenshot_bytes = await self.current_page.screenshot(type="jpeg", quality=50)
|
216 |
-
screenshot_b64 = base64.b64encode(screenshot_bytes).decode("utf-8")
|
217 |
-
# Log only a very short snippet of base64 string
|
218 |
-
logger.error(f"DEBUGGING: Base64 Screenshot (truncated) when timeout occurred:\ndata:image/jpeg;base64,{screenshot_b64[:100]}... (full string is much longer)")
|
219 |
-
except Exception as ss_e:
|
220 |
-
logger.error(f"DEBUGGING: Could not take screenshot: {ss_e}")
|
221 |
-
|
222 |
-
# Re-raise the original exception to ensure the task still fails,
|
223 |
-
# but now with crucial debugging information in the logs.
|
224 |
-
raise e
|
225 |
-
# --- END TEMPORARY DEBUGGING CODE ---
|
226 |
-
|
227 |
-
except PlaywrightTimeoutError: # This outer catch is for the wait_for_load_state timeout
|
228 |
-
logger.error(f"Timeout waiting for website load state (networkidle): {self.current_url}")
|
229 |
-
raise # Re-raise if load_state itself times out
|
230 |
-
|
231 |
-
except Exception as e:
|
232 |
-
logger.error(f"An unexpected error occurred during page readiness check: {e}")
|
233 |
-
raise
|
234 |
-
|
235 |
-
# Run the bounding box javascript code to highlight the points of interest on the page
|
236 |
-
page_info = await self.current_page.evaluate(
|
237 |
-
"""() => {
|
238 |
-
overwriteDefaultSelectConvergence();
|
239 |
-
return findPOIsConvergence();
|
240 |
-
}""",
|
241 |
-
)
|
242 |
-
# Get the points of interest on the page
|
243 |
-
self.poi_elements = page_info["element_descriptions"]
|
244 |
-
element_centroids = page_info["element_centroids"]
|
245 |
-
try:
|
246 |
-
# Select all iframes on the page
|
247 |
-
iframes = await self.current_page.query_selector_all("iframe")
|
248 |
-
|
249 |
-
max_iframes = 10
|
250 |
-
|
251 |
-
# Define an asynchronous function to process and filter each iframe
|
252 |
-
tasks = [asyncio.create_task(self.process_iframe(iframe)) for iframe in iframes[:max_iframes]]
|
253 |
|
254 |
-
|
255 |
-
|
256 |
-
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
-
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
for centroid in iframe_poi["element_centroids"]:
|
269 |
-
centroid["x"] += iframe_offsets[index]["x"]
|
270 |
-
centroid["y"] += iframe_offsets[index]["y"]
|
271 |
-
centroid["left"] += iframe_offsets[index]["x"]
|
272 |
-
centroid["top"] += iframe_offsets[index]["y"]
|
273 |
-
centroid["right"] += iframe_offsets[index]["x"]
|
274 |
-
centroid["bottom"] += iframe_offsets[index]["y"]
|
275 |
-
element_centroids.extend(iframe_poi["element_centroids"])
|
276 |
-
|
277 |
-
except Exception as e:
|
278 |
-
logger.error(f"Error in finding iframes: {e}")
|
279 |
-
|
280 |
-
# Get the centroids of the points of interest
|
281 |
-
self.poi_centroids = [Point(x=xy["x"], y=xy["y"]) for xy in element_centroids]
|
282 |
-
self.bounding_boxes = [BoundingBox(**xy, label=str(i)) for i, xy in enumerate(element_centroids)]
|
283 |
-
self.pois = [
|
284 |
-
POI(info=info, element_centroid=centroid, bounding_box=bbox)
|
285 |
-
for info, centroid, bbox in zip(
|
286 |
-
self.poi_elements,
|
287 |
-
self.poi_centroids,
|
288 |
-
self.bounding_boxes,
|
289 |
-
strict=False,
|
290 |
-
)
|
291 |
-
]
|
292 |
|
293 |
@property
|
294 |
def poi_text(self) -> str:
|
|
|
179 |
# re-run for cases of mid-run redirects
|
180 |
@retry(
|
181 |
wait=wait_exponential(multiplier=1, min=1, max=10),
|
182 |
+
stop=stop_after_delay(5), # This retry is for the entire update_poi method, in case of intermittent issues
|
183 |
reraise=True,
|
184 |
before_sleep=before_sleep_log(logger, logging.ERROR),
|
185 |
)
|
186 |
async def update_poi(self) -> None:
|
187 |
try:
|
188 |
+
# Step 1: Wait for network to be idle. This indicates that initial requests have settled.
|
189 |
+
logger.debug("Attempting wait_for_load_state('networkidle')...")
|
190 |
+
await self.current_page.wait_for_load_state("networkidle", timeout=180000) # Increased timeout
|
191 |
logger.debug("wait_for_load_state('networkidle') completed.")
|
192 |
|
193 |
+
# Step 2: Wait for the 'loading' class to disappear from the body.
|
194 |
+
# This is a common and effective way to detect when SPAs like Salesforce are visually ready.
|
195 |
+
logger.debug("Attempting wait_for_selector('body:not(.loading)')...")
|
196 |
+
# Removed state="visible" as it's often too strict for 'body' in SPAs,
|
197 |
+
# and 'not(.loading)' implies it should become visible eventually.
|
198 |
+
await self.current_page.wait_for_selector("body:not(.loading)", timeout=180000)
|
199 |
+
logger.debug("wait_for_selector('body:not(.loading)') completed.")
|
200 |
+
|
201 |
+
# Optional Step 3 (Highly Recommended): If the above still times out,
|
202 |
+
# uncomment and replace with a reliable selector for an interactive element
|
203 |
+
# that only appears after the Salesforce UI is fully loaded and ready for user input.
|
204 |
+
# Example: await self.current_page.wait_for_selector("#some_salesforce_specific_id", timeout=180000, state="visible")
|
205 |
+
# Example: await self.current_page.wait_for_selector("text=App Launcher", timeout=180000, state="visible")
|
206 |
+
# For now, we'll rely on the 'body:not(.loading)' as the primary indicator.
|
207 |
+
|
208 |
+
except PlaywrightTimeoutError as e:
|
209 |
+
# --- START TEMPORARY DEBUGGING CODE ---
|
210 |
+
# This block captures state specifically when a Playwright timeout occurs
|
211 |
+
current_url = self.current_page.url if self.current_page else "N/A"
|
212 |
+
logger.error(f"DEBUGGING: Playwright Timeout (180s) during page readiness check at URL: {current_url}")
|
213 |
+
|
214 |
+
html_content = None
|
215 |
try:
|
216 |
+
if self.current_page:
|
217 |
+
html_content = await self.current_page.content()
|
218 |
+
logger.error(f"DEBUGGING: HTML Content (first 1000 chars) when timeout occurred:\n{html_content[:1000]}...")
|
219 |
+
except Exception as html_e:
|
220 |
+
logger.error(f"DEBUGGING: Could not get HTML content for debug: {html_e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
221 |
|
222 |
+
screenshot_b64 = "N/A"
|
223 |
+
try:
|
224 |
+
if self.current_page:
|
225 |
+
# Capture screenshot at lower quality (e.g., 50) to keep log size manageable.
|
226 |
+
# Higher quality might make logs too large for some platforms.
|
227 |
+
screenshot_bytes = await self.current_page.screenshot(type="jpeg", quality=50)
|
228 |
+
screenshot_b64 = base64.b64encode(screenshot_bytes).decode("utf-8")
|
229 |
+
# Log only a very short snippet of base64 string to confirm it's there
|
230 |
+
logger.error(f"DEBUGGING: Base64 Screenshot (truncated) when timeout occurred:\ndata:image/jpeg;base64,{screenshot_b64[:100]}... (full string is much longer)")
|
231 |
+
|
232 |
+
# If you want to view the full screenshot locally during development, you can save it:
|
233 |
+
# with open("debug_timeout_full_screenshot.jpeg", "wb") as f:
|
234 |
+
# f.write(screenshot_bytes)
|
235 |
+
# logger.error("DEBUGGING: Full screenshot saved to debug_timeout
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
236 |
|
237 |
@property
|
238 |
def poi_text(self) -> str:
|