Synchronous Browser (#3745)

Split out sync methods in playwright

langchain/agents/agent_toolkits/playwright/toolkit.py

@@ -1,13 +1,16 @@
 """Playwright web browser toolkit."""
 from __future__ import annotations
 
-from typing import TYPE_CHECKING, List, Type
+from typing import TYPE_CHECKING, List, Optional, Type, cast
 
-from pydantic import Extra, Field, root_validator
+from pydantic import Extra, root_validator
 
 from langchain.agents.agent_toolkits.base import BaseToolkit
 from langchain.tools.base import BaseTool
-from langchain.tools.playwright.base import BaseBrowserTool
+from langchain.tools.playwright.base import (
+    BaseBrowserTool,
+    lazy_import_playwright_browsers,
+)
 from langchain.tools.playwright.click import ClickTool
 from langchain.tools.playwright.current_page import CurrentWebPageTool
 from langchain.tools.playwright.extract_hyperlinks import ExtractHyperlinksTool
@@ -15,16 +18,24 @@ from langchain.tools.playwright.extract_text import ExtractTextTool
 from langchain.tools.playwright.get_elements import GetElementsTool
 from langchain.tools.playwright.navigate import NavigateTool
 from langchain.tools.playwright.navigate_back import NavigateBackTool
-from langchain.tools.playwright.utils import create_playwright_browser
 
 if TYPE_CHECKING:
     from playwright.async_api import Browser as AsyncBrowser
+    from playwright.sync_api import Browser as SyncBrowser
+else:
+    try:
+        # We do this so pydantic can resolve the types when instantiating
+        from playwright.async_api import Browser as AsyncBrowser
+        from playwright.sync_api import Browser as SyncBrowser
+    except ImportError:
+        pass
 
 
 class PlayWrightBrowserToolkit(BaseToolkit):
     """Toolkit for web browser tools."""
 
-    browser: AsyncBrowser = Field(default_factory=create_playwright_browser)
+    sync_browser: Optional["SyncBrowser"] = None
+    async_browser: Optional["AsyncBrowser"] = None
 
     class Config:
         """Configuration for this pydantic object."""
@@ -33,15 +44,11 @@ class PlayWrightBrowserToolkit(BaseToolkit):
         arbitrary_types_allowed = True
 
     @root_validator
-    def check_args(cls, values: dict) -> dict:
+    def validate_imports_and_browser_provided(cls, values: dict) -> dict:
         """Check that the arguments are valid."""
-        try:
-            from playwright.async_api import Browser as AsyncBrowser  # noqa: F401
-        except ImportError:
-            raise ValueError(
-                "The 'playwright' package is required to use this tool."
-                " Please install it with 'pip install playwright'."
-            )
+        lazy_import_playwright_browsers()
+        if values.get("async_browser") is None and values.get("sync_browser") is None:
+            raise ValueError("Either async_browser or sync_browser must be specified.")
         return values
 
     def get_tools(self) -> List[BaseTool]:
@@ -56,11 +63,21 @@ class PlayWrightBrowserToolkit(BaseToolkit):
             CurrentWebPageTool,
         ]
 
-        return [tool_cls.from_browser(self.browser) for tool_cls in tool_classes]
+        tools = [
+            tool_cls.from_browser(
+                sync_browser=self.sync_browser, async_browser=self.async_browser
+            )
+            for tool_cls in tool_classes
+        ]
+        return cast(List[BaseTool], tools)
 
     @classmethod
-    def from_browser(cls, browser: AsyncBrowser) -> PlayWrightBrowserToolkit:
-        from playwright.async_api import Browser as AsyncBrowser
-
-        cls.update_forward_refs(AsyncBrowser=AsyncBrowser)
-        return cls(browser=browser)
+    def from_browser(
+        cls,
+        sync_browser: Optional[SyncBrowser] = None,
+        async_browser: Optional[AsyncBrowser] = None,
+    ) -> PlayWrightBrowserToolkit:
+        """Instantiate the toolkit."""
+        # This is to raise a better error than the forward ref ones Pydantic would have
+        lazy_import_playwright_browsers()
+        return cls(sync_browser=sync_browser, async_browser=async_browser)

langchain/tools/__init__.py

@@ -16,7 +16,6 @@ from langchain.tools.ifttt import IFTTTWebhook
 from langchain.tools.openapi.utils.api_models import APIOperation
 from langchain.tools.openapi.utils.openapi_utils import OpenAPISpec
 from langchain.tools.playwright import (
-    BaseBrowserTool,
     ClickTool,
     CurrentWebPageTool,
     ExtractHyperlinksTool,
@@ -32,7 +31,6 @@ from langchain.tools.shell.tool import ShellTool
 __all__ = [
     "AIPluginTool",
     "APIOperation",
-    "BaseBrowserTool",
     "BaseTool",
     "BaseTool",
     "BingSearchResults",

langchain/tools/playwright/__init__.py

@@ -1,6 +1,5 @@
 """Browser tools and toolkit."""
 
-from langchain.tools.playwright.base import BaseBrowserTool
 from langchain.tools.playwright.click import ClickTool
 from langchain.tools.playwright.current_page import CurrentWebPageTool
 from langchain.tools.playwright.extract_hyperlinks import ExtractHyperlinksTool
@@ -15,7 +14,6 @@ __all__ = [
     "ExtractTextTool",
     "ExtractHyperlinksTool",
     "GetElementsTool",
-    "BaseBrowserTool",
     "ClickTool",
     "CurrentWebPageTool",
 ]

langchain/tools/playwright/base.py

@@ -1,40 +1,55 @@
 from __future__ import annotations
 
-from typing import TYPE_CHECKING, Any
+from typing import TYPE_CHECKING, Optional, Tuple, Type
 
-from pydantic import Field, root_validator
+from pydantic import root_validator
 
 from langchain.tools.base import BaseTool
-from langchain.tools.playwright.utils import create_playwright_browser, run_async
 
 if TYPE_CHECKING:
     from playwright.async_api import Browser as AsyncBrowser
+    from playwright.sync_api import Browser as SyncBrowser
+else:
+    try:
+        # We do this so pydantic can resolve the types when instantiating
+        from playwright.async_api import Browser as AsyncBrowser
+        from playwright.sync_api import Browser as SyncBrowser
+    except ImportError:
+        pass
+
+
+def lazy_import_playwright_browsers() -> Tuple[Type[AsyncBrowser], Type[SyncBrowser]]:
+    try:
+        from playwright.async_api import Browser as AsyncBrowser  # noqa: F401
+        from playwright.sync_api import Browser as SyncBrowser  # noqa: F401
+    except ImportError:
+        raise ValueError(
+            "The 'playwright' package is required to use the playwright tools."
+            " Please install it with 'pip install playwright'."
+        )
+    return AsyncBrowser, SyncBrowser
 
 
 class BaseBrowserTool(BaseTool):
     """Base class for browser tools."""
 
-    browser: AsyncBrowser = Field(default_factory=create_playwright_browser)
+    sync_browser: Optional["SyncBrowser"] = None
+    async_browser: Optional["AsyncBrowser"] = None
 
     @root_validator
-    def check_args(cls, values: dict) -> dict:
+    def validate_browser_provided(cls, values: dict) -> dict:
         """Check that the arguments are valid."""
-        try:
-            from playwright.async_api import Browser as AsyncBrowser  # noqa: F401
-        except ImportError:
-            raise ValueError(
-                "The 'playwright' package is required to use this tool."
-                " Please install it with 'pip install playwright'."
-            )
+        lazy_import_playwright_browsers()
+        if values.get("async_browser") is None and values.get("sync_browser") is None:
+            raise ValueError("Either async_browser or sync_browser must be specified.")
         return values
 
-    def _run(self, *args: Any, **kwargs: Any) -> str:
-        """Use the tool."""
-        return run_async(self._arun(*args, **kwargs))
-
     @classmethod
-    def from_browser(cls, browser: AsyncBrowser) -> BaseBrowserTool:
-        from playwright.async_api import Browser as AsyncBrowser
-
-        cls.update_forward_refs(AsyncBrowser=AsyncBrowser)
-        return cls(browser=browser)
+    def from_browser(
+        cls,
+        sync_browser: Optional[SyncBrowser] = None,
+        async_browser: Optional[AsyncBrowser] = None,
+    ) -> BaseBrowserTool:
+        """Instantiate the tool."""
+        lazy_import_playwright_browsers()
+        return cls(sync_browser=sync_browser, async_browser=async_browser)

langchain/tools/playwright/click.py

@@ -6,6 +6,7 @@ from pydantic import BaseModel, Field
 
 from langchain.tools.playwright.base import BaseBrowserTool
 from langchain.tools.playwright.utils import (
+    aget_current_page,
     get_current_page,
 )
 
@@ -21,9 +22,20 @@ class ClickTool(BaseBrowserTool):
     description: str = "Click on an element with the given CSS selector"
     args_schema: Type[BaseModel] = ClickToolInput
 
+    def _run(self, selector: str) -> str:
+        """Use the tool."""
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+        page = get_current_page(self.sync_browser)
+        # Navigate to the desired webpage before using this tool
+        page.click(selector)
+        return f"Clicked element '{selector}'"
+
     async def _arun(self, selector: str) -> str:
         """Use the tool."""
-        page = await get_current_page(self.browser)
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
+        page = await aget_current_page(self.async_browser)
         # Navigate to the desired webpage before using this tool
         await page.click(selector)
         return f"Clicked element '{selector}'"

langchain/tools/playwright/current_page.py

@@ -6,6 +6,7 @@ from pydantic import BaseModel
 
 from langchain.tools.playwright.base import BaseBrowserTool
 from langchain.tools.playwright.utils import (
+    aget_current_page,
     get_current_page,
 )
 
@@ -15,7 +16,16 @@ class CurrentWebPageTool(BaseBrowserTool):
     description: str = "Returns the URL of the current page"
     args_schema: Type[BaseModel] = BaseModel
 
+    def _run(self) -> str:
+        """Use the tool."""
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+        page = get_current_page(self.sync_browser)
+        return str(page.url)
+
     async def _arun(self) -> str:
         """Use the tool."""
-        page = await get_current_page(self.browser)
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
+        page = await aget_current_page(self.async_browser)
         return str(page.url)

langchain/tools/playwright/extract_hyperlinks.py

@@ -1,12 +1,12 @@
 from __future__ import annotations
 
 import json
-from typing import TYPE_CHECKING, Type
+from typing import TYPE_CHECKING, Any, Type
 
 from pydantic import BaseModel, Field, root_validator
 
 from langchain.tools.playwright.base import BaseBrowserTool
-from langchain.tools.playwright.utils import get_current_page
+from langchain.tools.playwright.utils import aget_current_page, get_current_page
 
 if TYPE_CHECKING:
     pass
@@ -29,7 +29,7 @@ class ExtractHyperlinksTool(BaseBrowserTool):
     args_schema: Type[BaseModel] = ExtractHyperlinksToolInput
 
     @root_validator
-    def check_args(cls, values: dict) -> dict:
+    def check_bs_import(cls, values: dict) -> dict:
         """Check that the arguments are valid."""
         try:
             from bs4 import BeautifulSoup  # noqa: F401
@@ -40,15 +40,12 @@ class ExtractHyperlinksTool(BaseBrowserTool):
             )
         return values
 
-    async def _arun(self, absolute_urls: bool = False) -> str:
-        """Use the tool."""
+    @staticmethod
+    def scrape_page(page: Any, html_content: str, absolute_urls: bool) -> str:
         from urllib.parse import urljoin
 
         from bs4 import BeautifulSoup
 
-        page = await get_current_page(self.browser)
-        html_content = await page.content()
-
         # Parse the HTML content with BeautifulSoup
         soup = BeautifulSoup(html_content, "lxml")
 
@@ -59,6 +56,21 @@ class ExtractHyperlinksTool(BaseBrowserTool):
             links = [urljoin(base_url, anchor.get("href", "")) for anchor in anchors]
         else:
             links = [anchor.get("href", "") for anchor in anchors]
-
         # Return the list of links as a JSON string
         return json.dumps(links)
+
+    def _run(self, absolute_urls: bool = False) -> str:
+        """Use the tool."""
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+        page = get_current_page(self.sync_browser)
+        html_content = page.content()
+        return self.scrape_page(page, html_content, absolute_urls)
+
+    async def _arun(self, absolute_urls: bool = False) -> str:
+        """Use the tool asynchronously."""
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
+        page = await aget_current_page(self.async_browser)
+        html_content = await page.content()
+        return self.scrape_page(page, html_content, absolute_urls)

langchain/tools/playwright/extract_text.py

@@ -5,7 +5,7 @@ from typing import Type
 from pydantic import BaseModel, root_validator
 
 from langchain.tools.playwright.base import BaseBrowserTool
-from langchain.tools.playwright.utils import get_current_page
+from langchain.tools.playwright.utils import aget_current_page, get_current_page
 
 
 class ExtractTextTool(BaseBrowserTool):
@@ -14,7 +14,7 @@ class ExtractTextTool(BaseBrowserTool):
     args_schema: Type[BaseModel] = BaseModel
 
     @root_validator
-    def check_args(cls, values: dict) -> dict:
+    def check_acheck_bs_importrgs(cls, values: dict) -> dict:
         """Check that the arguments are valid."""
         try:
             from bs4 import BeautifulSoup  # noqa: F401
@@ -25,12 +25,30 @@ class ExtractTextTool(BaseBrowserTool):
             )
         return values
 
+    def _run(self) -> str:
+        """Use the tool."""
+        # Use Beautiful Soup since it's faster than looping through the elements
+        from bs4 import BeautifulSoup
+
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+
+        page = get_current_page(self.sync_browser)
+        html_content = page.content()
+
+        # Parse the HTML content with BeautifulSoup
+        soup = BeautifulSoup(html_content, "lxml")
+
+        return " ".join(text for text in soup.stripped_strings)
+
     async def _arun(self) -> str:
         """Use the tool."""
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
         # Use Beautiful Soup since it's faster than looping through the elements
         from bs4 import BeautifulSoup
 
-        page = await get_current_page(self.browser)
+        page = await aget_current_page(self.async_browser)
         html_content = await page.content()
 
         # Parse the HTML content with BeautifulSoup

langchain/tools/playwright/get_elements.py

@@ -6,10 +6,11 @@ from typing import TYPE_CHECKING, List, Optional, Sequence, Type
 from pydantic import BaseModel, Field
 
 from langchain.tools.playwright.base import BaseBrowserTool
-from langchain.tools.playwright.utils import get_current_page
+from langchain.tools.playwright.utils import aget_current_page, get_current_page
 
 if TYPE_CHECKING:
     from playwright.async_api import Page as AsyncPage
+    from playwright.sync_api import Page as SyncPage
 
 
 class GetElementsToolInput(BaseModel):
@@ -25,7 +26,7 @@ class GetElementsToolInput(BaseModel):
     )
 
 
-async def _get_elements(
+async def _aget_elements(
     page: AsyncPage, selector: str, attributes: Sequence[str]
 ) -> List[dict]:
     """Get elements matching the given CSS selector."""
@@ -45,6 +46,26 @@ async def _get_elements(
     return results
 
 
+def _get_elements(
+    page: SyncPage, selector: str, attributes: Sequence[str]
+) -> List[dict]:
+    """Get elements matching the given CSS selector."""
+    elements = page.query_selector_all(selector)
+    results = []
+    for element in elements:
+        result = {}
+        for attribute in attributes:
+            if attribute == "innerText":
+                val: Optional[str] = element.inner_text()
+            else:
+                val = element.get_attribute(attribute)
+            if val is not None and val.strip() != "":
+                result[attribute] = val
+        if result:
+            results.append(result)
+    return results
+
+
 class GetElementsTool(BaseBrowserTool):
     name: str = "get_elements"
     description: str = (
@@ -52,11 +73,22 @@ class GetElementsTool(BaseBrowserTool):
     )
     args_schema: Type[BaseModel] = GetElementsToolInput
 
+    def _run(self, selector: str, attributes: Sequence[str] = ["innerText"]) -> str:
+        """Use the tool."""
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+        page = get_current_page(self.sync_browser)
+        # Navigate to the desired webpage before using this tool
+        results = _get_elements(page, selector, attributes)
+        return json.dumps(results)
+
     async def _arun(
         self, selector: str, attributes: Sequence[str] = ["innerText"]
     ) -> str:
         """Use the tool."""
-        page = await get_current_page(self.browser)
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
+        page = await aget_current_page(self.async_browser)
         # Navigate to the desired webpage before using this tool
-        results = await _get_elements(page, selector, attributes)
+        results = await _aget_elements(page, selector, attributes)
         return json.dumps(results)

langchain/tools/playwright/navigate.py

@@ -6,6 +6,7 @@ from pydantic import BaseModel, Field
 
 from langchain.tools.playwright.base import BaseBrowserTool
 from langchain.tools.playwright.utils import (
+    aget_current_page,
     get_current_page,
 )
 
@@ -21,9 +22,20 @@ class NavigateTool(BaseBrowserTool):
     description: str = "Navigate a browser to the specified URL"
     args_schema: Type[BaseModel] = NavigateToolInput
 
+    def _run(self, url: str) -> str:
+        """Use the tool."""
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+        page = get_current_page(self.sync_browser)
+        response = page.goto(url)
+        status = response.status if response else "unknown"
+        return f"Navigating to {url} returned status code {status}"
+
     async def _arun(self, url: str) -> str:
         """Use the tool."""
-        page = await get_current_page(self.browser)
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
+        page = await aget_current_page(self.async_browser)
         response = await page.goto(url)
         status = response.status if response else "unknown"
         return f"Navigating to {url} returned status code {status}"

langchain/tools/playwright/navigate_back.py

@@ -6,6 +6,7 @@ from pydantic import BaseModel
 
 from langchain.tools.playwright.base import BaseBrowserTool
 from langchain.tools.playwright.utils import (
+    aget_current_page,
     get_current_page,
 )
 
@@ -17,15 +18,32 @@ class NavigateBackTool(BaseBrowserTool):
     description: str = "Navigate back to the previous page in the browser history"
     args_schema: Type[BaseModel] = BaseModel
 
+    def _run(self) -> str:
+        """Use the tool."""
+        if self.sync_browser is None:
+            raise ValueError(f"Synchronous browser not provided to {self.name}")
+        page = get_current_page(self.sync_browser)
+        response = page.go_back()
+
+        if response:
+            return (
+                f"Navigated back to the previous page with URL '{response.url}'."
+                f" Status code {response.status}"
+            )
+        else:
+            return "Unable to navigate back; no previous page in the history"
+
     async def _arun(self) -> str:
         """Use the tool."""
-        page = await get_current_page(self.browser)
+        if self.async_browser is None:
+            raise ValueError(f"Asynchronous browser not provided to {self.name}")
+        page = await aget_current_page(self.async_browser)
         response = await page.go_back()
 
         if response:
             return (
                 f"Navigated back to the previous page with URL '{response.url}'."
-                " Status code {response.status}"
+                f" Status code {response.status}"
             )
         else:
             return "Unable to navigate back; no previous page in the history"

langchain/tools/playwright/utils.py

@@ -7,9 +7,11 @@ from typing import TYPE_CHECKING, Any, Coroutine, TypeVar
 if TYPE_CHECKING:
     from playwright.async_api import Browser as AsyncBrowser
     from playwright.async_api import Page as AsyncPage
+    from playwright.sync_api import Browser as SyncBrowser
+    from playwright.sync_api import Page as SyncPage
 
 
-async def get_current_page(browser: AsyncBrowser) -> AsyncPage:
+async def aget_current_page(browser: AsyncBrowser) -> AsyncPage:
     if not browser.contexts:
         context = await browser.new_context()
         return await context.new_page()
@@ -20,13 +22,31 @@ async def get_current_page(browser: AsyncBrowser) -> AsyncPage:
     return context.pages[-1]
 
 
-def create_playwright_browser() -> AsyncBrowser:
+def get_current_page(browser: SyncBrowser) -> SyncPage:
+    if not browser.contexts:
+        context = browser.new_context()
+        return context.new_page()
+    context = browser.contexts[0]  # Assuming you're using the default browser context
+    if not context.pages:
+        return context.new_page()
+    # Assuming the last page in the list is the active one
+    return context.pages[-1]
+
+
+def create_async_playwright_browser() -> AsyncBrowser:
     from playwright.async_api import async_playwright
 
     browser = run_async(async_playwright().start())
     return run_async(browser.chromium.launch(headless=True))
 
 
+def create_sync_playwright_browser() -> SyncBrowser:
+    from playwright.sync_api import sync_playwright
+
+    browser = sync_playwright().start()
+    return browser.chromium.launch(headless=True)
+
+
 T = TypeVar("T")