The agent that turns websites into APIs!
fix: har recording
kalil032127c8b95cec28
diff --git a/src/reverse_api/browser.py b/src/reverse_api/browser.pyindex f614b74..8ac7cf4 100644--- a/src/reverse_api/browser.py+++ b/src/reverse_api/browser.py@@ -241,8 +241,6 @@ class ManualBrowser:def _start_with_real_chrome(self, start_url: Optional[str] = None) -> Path:"""Start using the real Chrome browser with user's profile."""- # We need to use a COPY of the profile to avoid locking issues- # Chrome locks its profile when running, so we can't use it directlyimport shutilimport tempfile@@ -257,12 +255,12 @@ class ManualBrowser:temp_profile_dir = Path(tempfile.mkdtemp(prefix="chrome_profile_"))console.print(f" [dim]using real chrome (profile copy)[/dim]")- console.print(f" [dim]note: close chrome if you have it open[/dim]")+ console.print(f" [yellow]⚠️ please browse in the FIRST tab only[/yellow]")+ console.print(f" [yellow] (new tabs may not be recorded)[/yellow]")console.print()try:# Use launch_persistent_context with channel="chrome" to use real Chrome binary- # This gives us the real Chrome with a fresh profile that has all extensions/settingsself._context = self._playwright.chromium.launch_persistent_context(user_data_dir=str(temp_profile_dir),channel="chrome", # Use real Chrome binary@@ -278,19 +276,24 @@ class ManualBrowser:)self._using_persistent = True- # Get or create page- if self._context.pages:- page = self._context.pages[0]- else:- page = self._context.new_page()+ for existing_page in self._context.pages:+ try:+ existing_page.close()+ except Exception:+ pass++ # For HAR recording & context+ page = self._context.new_page()if start_url:page.goto(start_url, wait_until="domcontentloaded")+ else:+ page.goto("https://www.google.com", wait_until="domcontentloaded")# Wait for browser to closetry:while self._context.pages:- self._context.pages[0].wait_for_timeout(100) # Faster polling+ self._context.pages[0].wait_for_timeout(100)except Exception:pass@@ -376,11 +379,14 @@ class ManualBrowser:if start_url:page.goto(start_url, wait_until="domcontentloaded")+ else:+ # For HAR recording & context+ page.goto("about:blank")# Wait for browser to closetry:while self._context.pages:- self._context.pages[0].wait_for_timeout(100) # Faster polling+ self._context.pages[0].wait_for_timeout(100)except Exception:pass@@ -423,10 +429,26 @@ class ManualBrowser:spinner="dots",) as status:try:+ status.update(" [dim]flushing network traffic...[/dim]")+ import time++ time.sleep(1)+status.update(" [dim]saving har file...[/dim]")- self._context.close() # This saves the HAR file- except Exception:- pass+ self._context.close()++ if self.har_path.exists():+ har_size = self.har_path.stat().st_size+ status.update(f" [dim]har saved: {har_size:,} bytes[/dim]")+ else:+ console.print(+ " [yellow]warning: har file was not created[/yellow]"+ )++ except Exception as e:+ console.print(f" [yellow]warning: error saving har: {e}[/yellow]")+ if self.har_path.exists():+ console.print(f" [dim]har file exists despite error[/dim]")self._context = None# Only close browser if not using persistent context