patchtroy 0.4.5__tar.gz → 0.4.6__tar.gz

{patchtroy-0.4.5 → patchtroy-0.4.6}/Dockerfile

@@ -8,7 +8,7 @@ FROM python:3.11-slim
 LABEL maintainer="Marcus Zou <marcus.zou@icloud.com>"
 LABEL org.opencontainers.image.title="Patchtroy"
 LABEL org.opencontainers.image.description="Undetected stealth web scraper & markdown extractor microservice"
-LABEL org.opencontainers.image.version="0.4.5"
+LABEL org.opencontainers.image.version="0.4.6"
 
 # Python and environment settings
 ENV PYTHONUNBUFFERED=1 \

{patchtroy-0.4.5 → patchtroy-0.4.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: patchtroy
-Version: 0.4.5
+Version: 0.4.6
 Summary: Undetected stealth web scraper & markdown extractor for LLMs, pairing Patchright with Trafilatura.
 Project-URL: Homepage, https://github.com/marcuz-apl/patchtroy
 Project-URL: Repository, https://github.com/marcuz-apl/patchtroy

{patchtroy-0.4.5 → patchtroy-0.4.6}/PRD.md

@@ -152,7 +152,7 @@ Patchtroy occupies the unoccupied **"Lightweight Stealth Sweet Spot"**:
 - [x] Token counter & LLM chunking utility (`result.chunk(max_tokens=2048)`).
 - [x] Full test suite (32 tests passing).
 
-### Phase 4 (v0.4.5 — Current)
+### Phase 4 (v0.4.6 — Current)
 - [x] PyPI public release workflows & multi-OS matrix CI (`.github/workflows/`).
 - [x] Comprehensive benchmark publication vs Crawl4AI and Firecrawl (`benchmarks/BENCHMARKS.md`).
 - [x] Official documentation portal (Material for MkDocs).

patchtroy-0.4.6/VERSION

@@ -0,0 +1 @@
+v0.4.7-260902d

{patchtroy-0.4.5 → patchtroy-0.4.6}/benchmarks/BENCHMARKS.md

@@ -19,7 +19,7 @@ Existing open-source solutions force developers to choose between:
 
 ## 2. Feature & Architecture Matrix
 
-| Dimension | **Patchtroy** (v0.4.5) | **Crawl4AI** (v0.9.3) | **Firecrawl** (Self-Hosted) |
+| Dimension | **Patchtroy** (v0.4.6) | **Crawl4AI** (v0.9.3) | **Firecrawl** (Self-Hosted) |
 | :--- | :---: | :---: | :---: |
 | **Stealth Engine** | **✅ Native Patchright (C++ CDP masked)** | ❌ Standard Playwright (CDP leaked) | ⚠️ Standard Playwright / Puppeteer |
 | **Dynamic Challenge Resilience** | **✅ Native driver protection** | ❌ Frequently flagged on automated checks | ❌ Blocked unless using paid SaaS proxy |

{patchtroy-0.4.5 → patchtroy-0.4.6}/docs/benchmarks.md

@@ -1,6 +1,6 @@
 # 📊 Benchmarks vs Crawl4AI and Firecrawl
 
-A quantitative, reproducible benchmark comparing **Patchtroy (v0.4.5)** against **Crawl4AI (v0.9.3)** and **Firecrawl (Self-Hosted)**.
+A quantitative, reproducible benchmark comparing **Patchtroy (v0.4.6)** against **Crawl4AI (v0.9.3)** and **Firecrawl (Self-Hosted)**.
 
 ---
 

{patchtroy-0.4.5 → patchtroy-0.4.6}/docs/rest-api.md

@@ -53,7 +53,7 @@ curl http://localhost:4013/health
 ```json
 {
   "status": "healthy",
-  "version": "0.4.5",
+  "version": "0.4.6",
   "engine": "patchright",
   "active": true
 }

{patchtroy-0.4.5 → patchtroy-0.4.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "patchtroy"
-version = "0.4.5"
+version = "0.4.6"
 description = "Undetected stealth web scraper & markdown extractor for LLMs, pairing Patchright with Trafilatura."
 readme = "README.md"
 license = "Apache-2.0"

{patchtroy-0.4.5 → patchtroy-0.4.6}/src/patchtroy/__init__.py

@@ -6,7 +6,7 @@ from patchtroy.models import LinkItem, PatchtroyConfig, ScrapeResult
 from patchtroy.pool import BrowserContextPool
 from patchtroy.proxy import ProxyItem, ProxyManager
 
-__version__ = "0.4.5"
+__version__ = "0.4.6"
 __author__ = "Marcus Zou"
 __license__ = "Apache-2.0"
 __copyright__ = "Copyright 2026 Alfazen Inc."

{patchtroy-0.4.5 → patchtroy-0.4.6}/src/patchtroy/cli.py

@@ -11,7 +11,7 @@ from patchtroy.crawler import Patchtroy
 from patchtroy.models import PatchtroyConfig
 from patchtroy.utils import silence_windows_proactor_bug
 
-__version__ = "0.4.5"
+__version__ = "0.4.6"
 
 
 def main(argv: list[str] | None = None) -> int:

{patchtroy-0.4.5 → patchtroy-0.4.6}/src/patchtroy/pool.py

@@ -4,7 +4,6 @@ from __future__ import annotations
 
 import asyncio
 import logging
-import sys
 from collections.abc import AsyncIterator
 from contextlib import asynccontextmanager
 from typing import Any
@@ -90,8 +89,8 @@ class BrowserContextPool:
                     pass
                 self._playwright = None
 
-            if sys.platform == "win32":
-                await asyncio.sleep(0.05)
+            # Allow event loop to process subprocess termination callbacks before shutdown
+            await asyncio.sleep(0.05)
 
     @asynccontextmanager
     async def acquire_context(

{patchtroy-0.4.5 → patchtroy-0.4.6}/src/patchtroy/server.py

@@ -90,7 +90,7 @@ def create_app() -> FastAPI:
     app = FastAPI(
         title="Patchtroy REST Microservice",
         description="Undetected stealth web scraper & clean Markdown extractor for LLMs.",
-        version="0.4.5",
+        version="0.4.6",
         lifespan=lifespan,
     )
 
@@ -99,7 +99,7 @@ def create_app() -> FastAPI:
         """Microservice health and status check."""
         return {
             "status": "healthy",
-            "version": "0.4.5",
+            "version": "0.4.6",
             "engine": "patchright",
             "active": crawler_instance is not None and crawler_instance._pool.is_running,
         }

{patchtroy-0.4.5 → patchtroy-0.4.6}/src/patchtroy/utils.py

@@ -3,7 +3,6 @@
 from __future__ import annotations
 
 import random
-import sys
 from typing import Any
 from urllib.parse import urlparse
 
@@ -56,17 +55,33 @@ def is_valid_url(url: str) -> bool:
         return False
 
 
-def silence_windows_proactor_bug() -> None:
-    """Silence known Python Windows asyncio proactor pipe transport bug.
+def silence_subprocess_transport_bug() -> None:
+    """Silence Python asyncio BaseSubprocessTransport and pipe transport destructor errors.
 
-    On Windows, when an asyncio subprocess transport is garbage collected after
-    loop shutdown, _ProactorBasePipeTransport.__del__ and BaseSubprocessTransport.__del__
-    call __repr__, which calls fileno() on a closed pipe handle and raises
-    ValueError: I/O operation on closed pipe. This wraps both destructors to cleanly
-    catch and ignore closed pipe errors during interpreter teardown.
+    When an event loop finishes and closes (via asyncio.run), underlying child process
+    transports may remain in the garbage collector until interpreter exit. During GC,
+    their __del__ destructors attempt post-loop cleanup:
+      - On Unix/Linux: calls self.close() -> loop.call_soon() -> RuntimeError('Event loop is closed')
+      - On Windows: calls _warn() -> __repr__() -> fileno() -> ValueError('I/O operation on closed pipe')
+    Wrapping both destructors safely catches and suppresses these post-loop cleanup errors across
+    all platforms (Linux, macOS, Windows).
     """
-    if sys.platform != "win32":
-        return
+    try:
+        from asyncio.base_subprocess import BaseSubprocessTransport
+
+        orig_sub_del = getattr(BaseSubprocessTransport, "__del__", None)
+        if orig_sub_del and not getattr(BaseSubprocessTransport, "_patchtroy_safe", False):
+
+            def _safe_sub_del(self: Any, *args: Any, **kwargs: Any) -> None:
+                try:
+                    orig_sub_del(self, *args, **kwargs)
+                except (RuntimeError, ValueError, OSError):
+                    pass
+
+            BaseSubprocessTransport.__del__ = _safe_sub_del  # type: ignore[method-assign]
+            BaseSubprocessTransport._patchtroy_safe = True  # type: ignore[attr-defined]
+    except Exception:
+        pass
 
     try:
         from asyncio.proactor_events import _ProactorBasePipeTransport
@@ -77,7 +92,7 @@ def silence_windows_proactor_bug() -> None:
             def _safe_pipe_del(self: Any, *args: Any, **kwargs: Any) -> None:
                 try:
                     orig_pipe_del(self, *args, **kwargs)
-                except (ValueError, OSError):
+                except (RuntimeError, ValueError, OSError):
                     pass
 
             _ProactorBasePipeTransport.__del__ = _safe_pipe_del  # type: ignore[method-assign]
@@ -85,19 +100,6 @@ def silence_windows_proactor_bug() -> None:
     except Exception:
         pass
 
-    try:
-        from asyncio.base_subprocess import BaseSubprocessTransport
-
-        orig_sub_del = getattr(BaseSubprocessTransport, "__del__", None)
-        if orig_sub_del and not getattr(BaseSubprocessTransport, "_patchtroy_safe", False):
-
-            def _safe_sub_del(self: Any, *args: Any, **kwargs: Any) -> None:
-                try:
-                    orig_sub_del(self, *args, **kwargs)
-                except (ValueError, OSError):
-                    pass
 
-            BaseSubprocessTransport.__del__ = _safe_sub_del  # type: ignore[method-assign]
-            BaseSubprocessTransport._patchtroy_safe = True  # type: ignore[attr-defined]
-    except Exception:
-        pass
+# Backwards-compatible alias
+silence_windows_proactor_bug = silence_subprocess_transport_bug

{patchtroy-0.4.5 → patchtroy-0.4.6}/tests/test_server.py

@@ -17,7 +17,7 @@ async def test_health_check_endpoint():
         assert resp.status_code == 200
         data = resp.json()
         assert data["status"] == "healthy"
-        assert data["version"] == "0.4.5"
+        assert data["version"] == "0.4.6"
         assert data["engine"] == "patchright"
 
 

{patchtroy-0.4.5 → patchtroy-0.4.6}/uv.lock

@@ -900,7 +900,7 @@ wheels = [
 
 [[package]]
 name = "patchtroy"
-version = "0.4.5"
+version = "0.4.6"
 source = { editable = "." }
 dependencies = [
     { name = "beautifulsoup4" },

patchtroy-0.4.5/VERSION

@@ -1 +0,0 @@
-v0.4.5-260902b