Skip to content

Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.

better descripiton for extract

Magnus Müller13836049618d
diff --git a/browser_use/tools/service.py b/browser_use/tools/service.py
index 917b31107..06145e815 100644
--- a/browser_use/tools/service.py
+++ b/browser_use/tools/service.py
@@ -38,6 +38,7 @@ from browser_use.tools.views import (
ClickElementAction,
CloseTabAction,
DoneAction,
+ ExtractAction,
GetDropdownOptionsAction,
InputTextAction,
NavigateAction,
@@ -577,15 +578,16 @@ class Tools(Generic[Context]):
"""LLM extracts structured data from page markdown. Use when: on right page, know what to extract, haven't called before on same page+query. Can't get interactive elements. Set extract_links=True for URLs. Use start_from_char if truncated. If fails, use find_text instead.""",
)
async def extract(
- query: str,
+ params: ExtractAction,
browser_session: BrowserSession,
page_extraction_llm: BaseChatModel,
file_system: FileSystem,
- extract_links: bool = False,
- start_from_char: int = 0,
):
# Constants
MAX_CHAR_LIMIT = 30000
+ query = params.query
+ extract_links = params.extract_links
+ start_from_char = params.start_from_char
# Extract clean markdown using the unified method
try:
@@ -603,7 +605,7 @@ class Tools(Generic[Context]):
if start_from_char > 0:
if start_from_char >= len(content):
return ActionResult(
- error=f'start_from_char ({start_from_char}) exceeds content length ({len(content)}). Content has {final_filtered_length} characters after filtering.'
+ error=f'start_from_char ({start_from_char}) exceeds content length {final_filtered_length} characters.'
)
content = content[start_from_char:]
content_stats['started_from_char'] = start_from_char
diff --git a/browser_use/tools/views.py b/browser_use/tools/views.py
index 5775175a0..008c8f5f0 100644
--- a/browser_use/tools/views.py
+++ b/browser_use/tools/views.py
@@ -4,6 +4,16 @@ from pydantic import BaseModel, ConfigDict, Field
# Action Input Models
+class ExtractAction(BaseModel):
+ query: str
+ extract_links: bool = Field(
+ default=False, description='Set True to true if the query requires links, else false to safe tokens'
+ )
+ start_from_char: int = Field(
+ default=0, description='Use this for long markdowns to start from a specific character (not index in browser_state)'
+ )
+
+
class SearchAction(BaseModel):
query: str
engine: str = Field(