browser-control-and-automation-cli 0.1.1.dev3__tar.gz → 0.1.3__tar.gz

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/AGENTS.md

@@ -199,6 +199,9 @@ the implementation, and where should a change land first.
   `Tab <id> is not debuggable: unknown` during extension-mode `read` or `open-tab`
   -> extension tried to attach Chrome debugger before the new tab reported a debuggable URL
   -> inspect `browser-cli-extension/src/background.js::openTab` and `browser-cli-extension/src/debugger.js::ensureAttached`
+  `read` fails in extension mode with `Separator is found, but chunk is longer than limit` or `chunk exceed the limit`
+  -> extension-side page extraction hit a large-result chunking limit on the current page
+  -> inspect `src/browser_cli/daemon/browser_service.py::read_page`, `src/browser_cli/drivers/_extension/page_actions.py`, and `browser-cli-extension/src/background/page_actions.js`; safe fix is a one-shot read fallback to Playwright rather than changing CLI contracts
 - If the user reports popup/runtime observer drift:
   start at `src/browser_cli/daemon/runtime_presentation.py`, then `src/browser_cli/extension/session.py`, then `browser-cli-extension/src/background.js`, `browser-cli-extension/src/popup_view.js`, and `browser-cli-extension/src/popup.js`.
 - If the user reports long-run stability drift across repeated reloads, reconnects, or artifact runs:

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: browser-control-and-automation-cli
-Version: 0.1.1.dev3
+Version: 0.1.3
 Summary: CLI-first browser reader for rendered HTML and page snapshots
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

browser_control_and_automation_cli-0.1.3/src/browser_cli/__init__.py

@@ -0,0 +1,21 @@
+"""browser-cli package."""
+
+from importlib.metadata import PackageNotFoundError, version
+
+__all__ = ["__version__"]
+
+
+def _resolve_version() -> str:
+    for distribution_name in (
+        "browser-control-and-automation-cli",
+        "browserctl",
+        "browser-cli",
+    ):
+        try:
+            return version(distribution_name)
+        except PackageNotFoundError:
+            continue
+    return "0+unknown"
+
+
+__version__ = _resolve_version()

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/src/browser_cli/daemon/browser_service.py

@@ -58,6 +58,12 @@ class BrowserService:
     READ_SCROLL_PAUSE_MS = 450
     READ_SCROLL_MAX_ROUNDS = 8
     READ_SCROLL_STABLE_ROUNDS = 2
+    _READ_EXTENSION_FALLBACK_PATTERNS = (
+        "Separator is found, but chunk is longer than limit",
+        "Separator is not found, and chunk exceed the limit",
+        "chunk is longer than limit",
+        "chunk exceed the limit",
+    )
 
     def __init__(
         self,
@@ -333,6 +339,40 @@ class BrowserService:
     ) -> dict[str, Any]:
         if output_mode not in {"html", "snapshot"}:
             raise InvalidInputError(f"Unsupported read output mode: {output_mode}")
+        try:
+            return await self._read_page_once(
+                url=url,
+                output_mode=output_mode,
+                scroll_bottom=scroll_bottom,
+            )
+        except Exception as exc:
+            if not self._should_retry_read_on_playwright(exc):
+                raise
+            logger.warning(
+                "Extension read failed with chunking error; retrying with playwright: %s",
+                exc,
+            )
+            await self._activate_driver("playwright", reason="extension-read-fallback")
+            payload = await self._read_page_once(
+                url=url,
+                output_mode=output_mode,
+                scroll_bottom=scroll_bottom,
+            )
+            payload["driver_fallback"] = {
+                "from": "extension",
+                "to": "playwright",
+                "reason": "extension-read-fallback",
+                "error": str(exc),
+            }
+            return payload
+
+    async def _read_page_once(
+        self,
+        *,
+        url: str,
+        output_mode: str,
+        scroll_bottom: bool,
+    ) -> dict[str, Any]:
         page = await self.new_tab(
             url=url, wait_until="load", timeout_seconds=self.READ_NAVIGATION_TIMEOUT_SECONDS
         )
@@ -372,6 +412,12 @@ class BrowserService:
             with contextlib.suppress(Exception):
                 await self.close_tab(page_id)
 
+    def _should_retry_read_on_playwright(self, exc: Exception) -> bool:
+        if self._driver_name != "extension":
+            return False
+        message = str(exc)
+        return any(pattern in message for pattern in self._READ_EXTENSION_FALLBACK_PATTERNS)
+
     async def close_tab(self, page_id: str) -> dict[str, Any]:
         self._snapshot_registry.clear_page(page_id)
         return await self._active_driver().close_tab(page_id)

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/src/browser_cli/daemon/client.py

@@ -48,6 +48,7 @@ from .transport import (
 STARTUP_TIMEOUT_SECONDS = 15.0
 STARTUP_PROBE_INTERVAL_SECONDS = 0.1
 TERMINATE_GRACE_SECONDS = 3.0
+_STREAM_LIMIT = 32 * 1024 * 1024
 
 
 def send_command(
@@ -72,7 +73,9 @@ def send_command(
 async def _send_command_async(action: str, args: dict[str, Any]) -> dict[str, Any]:
     app_paths = get_app_paths()
     try:
-        reader, writer = await asyncio.open_unix_connection(str(app_paths.socket_path))
+        reader, writer = await asyncio.open_unix_connection(
+            str(app_paths.socket_path), limit=_STREAM_LIMIT
+        )
     except OSError as exc:
         raise DaemonNotAvailableError(str(exc)) from exc
     request_payload = {

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/src/browser_cli/daemon/server.py

@@ -24,6 +24,7 @@ from .transport import (
 )
 
 logger = logging.getLogger(__name__)
+_STREAM_LIMIT = 32 * 1024 * 1024
 
 
 class BrowserDaemonServer:
@@ -50,6 +51,7 @@ class BrowserDaemonServer:
             self._server = await asyncio.start_unix_server(
                 self._handle_client,
                 path=str(app_paths.socket_path),
+                limit=_STREAM_LIMIT,
             )
             with contextlib.suppress(OSError):
                 os.chmod(app_paths.socket_path, 0o600)

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/src/browser_control_and_automation_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: browser-control-and-automation-cli
-Version: 0.1.1.dev3
+Version: 0.1.3
 Summary: CLI-first browser reader for rendered HTML and page snapshots
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/src/browser_control_and_automation_cli.egg-info/SOURCES.txt

@@ -254,6 +254,7 @@ tests/unit/test_guard_scripts.py
 tests/unit/test_install_skills_command.py
 tests/unit/test_lifecycle_commands.py
 tests/unit/test_network.py
+tests/unit/test_package_version.py
 tests/unit/test_paths_command.py
 tests/unit/test_profile_discovery.py
 tests/unit/test_python_compat_guard.py

{browser_control_and_automation_cli-0.1.1.dev3 → browser_control_and_automation_cli-0.1.3}/tests/unit/test_daemon_browser_service.py

@@ -108,6 +108,21 @@ class _FakeDriverBase:
         self.switched_to = page_id
         return await self.get_page_summary(page_id)
 
+    async def wait(
+        self,
+        page_id: str,
+        *,
+        seconds: float | None = None,
+        text: str | None = None,
+        gone: bool = False,
+        exact: bool = False,
+    ) -> dict:
+        _ = seconds
+        _ = text
+        _ = gone
+        _ = exact
+        return {"page_id": page_id, "waited": True}
+
 
 class _FakePlaywrightDriver(_FakeDriverBase):
     def __init__(self, chrome_environment=None, *, headless=None) -> None:
@@ -519,6 +534,39 @@ def test_browser_service_downgrades_at_safe_point_and_reports_state_reset(
     asyncio.run(_scenario())
 
 
+def test_read_page_falls_back_to_playwright_when_extension_hits_chunk_limit(
+    _patched_browser_service: _FakeExtensionHub,
+) -> None:
+    async def _scenario() -> None:
+        _patched_browser_service.connect()
+        service = browser_service_module.BrowserService(TabRegistry())
+        await service.ensure_started()
+        assert service.active_driver_name == "extension"
+
+        async def _extension_capture_html(page_id: str) -> dict[str, str]:
+            raise RuntimeError("Separator is found, but chunk is longer than limit")
+
+        async def _playwright_capture_html(page_id: str) -> dict[str, str]:
+            return {"page_id": page_id, "html": "<html>ok</html>"}
+
+        service._extension.capture_html = _extension_capture_html  # type: ignore[method-assign]  # noqa: SLF001
+        service._playwright.capture_html = _playwright_capture_html  # type: ignore[method-assign]  # noqa: SLF001
+
+        payload = await service.read_page(url="https://example.com/long-doc", output_mode="html")
+
+        assert payload["body"] == "<html>ok</html>"
+        assert payload["driver_fallback"] == {
+            "from": "extension",
+            "to": "playwright",
+            "reason": "extension-read-fallback",
+            "error": "Separator is found, but chunk is longer than limit",
+        }
+        assert service.active_driver_name == "playwright"
+        await service.stop()
+
+    asyncio.run(_scenario())
+
+
 def test_browser_service_downgrade_survives_previous_driver_stop_failure(
     _patched_browser_service: _FakeExtensionHub,
     monkeypatch: pytest.MonkeyPatch,

browser_control_and_automation_cli-0.1.3/tests/unit/test_package_version.py

@@ -0,0 +1,44 @@
+from __future__ import annotations
+
+from importlib.metadata import PackageNotFoundError
+
+import browser_cli
+
+
+def test_package_version_prefers_current_distribution_name(monkeypatch) -> None:
+    calls: list[str] = []
+
+    def _fake_version(name: str) -> str:
+        calls.append(name)
+        if name == "browser-control-and-automation-cli":
+            return "1.2.3"
+        raise PackageNotFoundError(name)
+
+    monkeypatch.setattr(browser_cli, "version", _fake_version)
+
+    assert browser_cli._resolve_version() == "1.2.3"
+    assert calls == ["browser-control-and-automation-cli"]
+
+
+def test_package_version_falls_back_to_legacy_distribution_names(monkeypatch) -> None:
+    calls: list[str] = []
+
+    def _fake_version(name: str) -> str:
+        calls.append(name)
+        if name == "browserctl":
+            return "2.0.0"
+        raise PackageNotFoundError(name)
+
+    monkeypatch.setattr(browser_cli, "version", _fake_version)
+
+    assert browser_cli._resolve_version() == "2.0.0"
+    assert calls == ["browser-control-and-automation-cli", "browserctl"]
+
+
+def test_package_version_uses_unknown_when_no_distribution_metadata_exists(monkeypatch) -> None:
+    def _fake_version(_name: str) -> str:
+        raise PackageNotFoundError("missing")
+
+    monkeypatch.setattr(browser_cli, "version", _fake_version)
+
+    assert browser_cli._resolve_version() == "0+unknown"

browser_control_and_automation_cli-0.1.1.dev3/src/browser_cli/__init__.py

@@ -1,7 +0,0 @@
-"""browser-cli package."""
-
-from importlib.metadata import version
-
-__all__ = ["__version__"]
-
-__version__ = version("browser-cli")