Better context manager and cleanup of old browser instances

olmocr/bench/katex/render.py

@@ -19,6 +19,7 @@ import re
 import sqlite3
 import threading
 import unittest
+from contextlib import contextmanager
 from dataclasses import dataclass
 from typing import List, Optional
 
@@ -125,10 +126,6 @@ equation_cache = EquationCache()
 # --- End SQLite Cache Implementation ---
 
 
-# Thread-local storage for Playwright and browser instances
-_thread_local = threading.local()
-
-
 @dataclass
 class BoundingBox:
     x: float
@@ -158,21 +155,63 @@ def get_equation_hash(equation, bg_color="white", text_color="black", font_size=
     return hashlib.sha1(params_str.encode("utf-8")).hexdigest()
 
 
-def init_browser():
+# Thread-local storage for browser contexts
+_thread_local = threading.local()
+
+
+@contextmanager
+def browser_context():
     """
-    Initialize the Playwright and browser instance for the current thread if not already done.
+    Context manager for Playwright browser instances.
+    Returns a browser context that can be used for rendering equations.
+    Automatically handles initialization and cleanup.
+    Cleans up and recreates the browser every 100 uses to prevent memory leaks.
     """
+    # Initialize usage counter if not present
+    if not hasattr(_thread_local, "usage_count"):
+        _thread_local.usage_count = 0
+
+    # Check if we already have a browser for this thread
     if not hasattr(_thread_local, "playwright"):
         _thread_local.playwright = sync_playwright().start()
         _thread_local.browser = _thread_local.playwright.chromium.launch()
+        _thread_local.usage_count = 0
 
+    # Increment usage counter
+    _thread_local.usage_count += 1
 
-def get_browser():
-    """
-    Return the browser instance for the current thread.
-    """
-    init_browser()
-    return _thread_local.browser
+    # Check if we need to clean up the browser (every 100 uses)
+    if _thread_local.usage_count > 100:
+        # Clean up the old browser completely
+        print("Cleanup up old playwright instance to prevent memory leaks...")
+        
+        try:
+            _thread_local.browser.close()
+        except Exception:
+            pass  # Ignore errors during cleanup
+
+        try:
+            _thread_local.playwright.stop()
+        except Exception:
+            pass  # Ignore errors during cleanup
+
+        # Remove the attributes to force re-initialization
+        delattr(_thread_local, "playwright")
+        delattr(_thread_local, "browser")
+
+        # Re-initialize with a fresh browser
+        _thread_local.playwright = sync_playwright().start()
+        _thread_local.browser = _thread_local.playwright.chromium.launch()
+        _thread_local.usage_count = 1  # Reset to 1 since we just used it
+
+    # Create a new context for this operation
+    context = _thread_local.browser.new_context(viewport={"width": 800, "height": 400})
+
+    try:
+        yield context
+    finally:
+        # Clean up the context after use
+        context.close()
 
 
 def render_equation(
@@ -207,11 +246,10 @@ def render_equation(
     if not os.path.exists(katex_css_path) or not os.path.exists(katex_js_path):
         raise FileNotFoundError(f"KaTeX files not found. Please ensure katex.min.css and katex.min.js are in {script_dir}")
 
-    # Get the browser instance for the current thread.
-    browser = get_browser()
-
+    # Use the browser context manager
+    with browser_context() as context:
         # Create a new page.
-    page = browser.new_page(viewport={"width": 800, "height": 400})
+        page = context.new_page()
 
         # Basic HTML structure for rendering.
         page_html = f"""