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

Revert context matching attempts - Playwright cannot see CDP-created contexts

Browse Source
This commit is contained in:
unclecode
2025-12-13 07:57:29 +00:00
parent 8014805c17
commit 6185d3cb32
1 changed files with 24 additions and 94 deletions
Download Patch File Download Diff File Expand all files Collapse all files

118
crawl4ai/browser_manager.py
View File

@@ -676,78 +676,12 @@ class BrowserManager:
f"Using pre-existing browser context: {self.config.browser_context_id}", f"Using pre-existing browser context: {self.config.browser_context_id}",
tag="BROWSER" tag="BROWSER"
) )
# When connecting to a pre-created context, it should be in contexts
# Find the specific context by matching browserContextId if contexts:
# Each Playwright context has pages, and we need to find which one
# corresponds to our CDP browserContextId
found_context = None
if contexts and len(contexts) == 1:
# Only one context - use it directly
found_context = contexts[0]
elif contexts and len(contexts) > 1:
# Multiple contexts - need to find the right one
# Use CDP to query which context owns our target
try:
# Get first page from any context to create CDP session
any_page = None
for ctx in contexts:
if ctx.pages:
any_page = ctx.pages[0]
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:
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:
# 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.debug(
f"Could not find context for browserContextId {self.config.browser_context_id}, " f"Found {len(contexts)} existing context(s), using first one",
f"using first of {len(contexts)} context(s)",
tag="BROWSER" tag="BROWSER"
) )
else: else:
@@ -1226,33 +1160,29 @@ class BrowserManager:
await self._apply_stealth_to_page(page) await self._apply_stealth_to_page(page)
else: else:
context = self.default_context context = self.default_context
pages = context.pages
# When target_id is provided, use it to find the specific page page = next((p for p in pages if p.url == crawlerRunConfig.url), None)
# This is critical for concurrent requests sharing a browser if not page:
if self.config.browser_context_id and self.config.target_id: if pages:
page = await self._get_page_by_target_id(context, self.config.target_id) page = pages[0]
if not page: else:
# Fallback: create new page in existing context # Double-check under lock to avoid TOCTOU and ensure only
# one task calls new_page when pages=[] concurrently
async with self._page_lock: async with self._page_lock:
page = await context.new_page() pages = context.pages
await self._apply_stealth_to_page(page) if pages:
else: page = pages[0]
# Original logic for cases without pre-created target elif self.config.browser_context_id and self.config.target_id:
pages = context.pages # Pre-existing context/target provided - use CDP to get the page
page = next((p for p in pages if p.url == crawlerRunConfig.url), None) # This handles the case where Playwright doesn't see the target yet
if not page: page = await self._get_page_by_target_id(context, self.config.target_id)
if pages: if not page:
page = pages[0] # Fallback: create new page in existing context
else:
# Double-check under lock to avoid TOCTOU and ensure only
# one task calls new_page when pages=[] concurrently
async with self._page_lock:
pages = context.pages
if pages:
page = pages[0]
else:
page = await context.new_page() page = await context.new_page()
await self._apply_stealth_to_page(page) await self._apply_stealth_to_page(page)
else:
page = await context.new_page()
await self._apply_stealth_to_page(page)
else: else:
# Otherwise, check if we have an existing context for this config # Otherwise, check if we have an existing context for this config
config_signature = self._make_config_signature(crawlerRunConfig) config_signature = self._make_config_signature(crawlerRunConfig)