ayrisdev/crawl4ai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix: use CDP to find context by browserContextId for concurrent sessions

Browse Source
This commit is contained in:
unclecode
2025-12-13 07:02:23 +00:00
parent c1e485e0b0
commit 8014805c17
1 changed files with 60 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

78
crawl4ai/browser_manager.py
View File

@@ -677,34 +677,76 @@ class BrowserManager:
tag="BROWSER" tag="BROWSER"
) )
# Find the specific context by matching target_id to pages # Find the specific context by matching browserContextId
# This is critical for concurrent crawls sharing a browser # Each Playwright context has pages, and we need to find which one
# corresponds to our CDP browserContextId
found_context = None found_context = None
if self.config.target_id and contexts: if contexts and len(contexts) == 1:
for ctx in contexts: # Only one context - use it directly
for page in ctx.pages: found_context = contexts[0]
# Playwright stores target ID in internal implementation elif contexts and len(contexts) > 1:
page_impl = getattr(page, '_impl_obj', None) # Multiple contexts - need to find the right one
page_target_id = getattr(page_impl, '_target_id', None) if page_impl else None # Use CDP to query which context owns our target
if page_target_id == self.config.target_id: try:
found_context = ctx # Get first page from any context to create CDP session
if self.logger: any_page = None
self.logger.debug( for ctx in contexts:
f"Found context by target_id: {self.config.target_id}", if ctx.pages:
tag="BROWSER" any_page = ctx.pages[0]
)
break break
if found_context:
break if any_page:
cdp = await any_page.context.new_cdp_session(any_page)
try:
result = await cdp.send("Target.getTargets")
targets = result.get("targetInfos", [])
# Find our target and its browserContextId
for target in targets:
if target.get("targetId") == self.config.target_id:
target_browser_context_id = target.get("browserContextId")
if target_browser_context_id == self.config.browser_context_id:
# Found it - now find which Playwright context has a page matching
for ctx in contexts:
for page in ctx.pages:
# Check if this page's context matches
# by checking if the page is in the right context
page_cdp = await ctx.new_cdp_session(page)
try:
page_targets = await page_cdp.send("Target.getTargets")
for pt in page_targets.get("targetInfos", []):
if pt.get("browserContextId") == self.config.browser_context_id:
found_context = ctx
break
finally:
await page_cdp.detach()
if found_context:
break
if found_context:
break
break
finally:
await cdp.detach()
except Exception as e:
if self.logger:
self.logger.warning(
f"Failed to find context via CDP: {e}",
tag="BROWSER"
)
if found_context: if found_context:
self.default_context = found_context self.default_context = found_context
if self.logger:
self.logger.debug(
f"Found context for browserContextId: {self.config.browser_context_id}",
tag="BROWSER"
)
elif contexts: elif contexts:
# Fallback to first context if we can't find the specific one # Fallback to first context if we can't find the specific one
self.default_context = contexts[0] self.default_context = contexts[0]
if self.logger: if self.logger:
self.logger.warning( self.logger.warning(
f"Could not find context for target_id {self.config.target_id}, " f"Could not find context for browserContextId {self.config.browser_context_id}, "
f"using first of {len(contexts)} context(s)", f"using first of {len(contexts)} context(s)",
tag="BROWSER" tag="BROWSER"
) )