arcade-mcp/toolkits/web/evals/eval_firecrawl.py
Eric Gustin 7c228a59d5
Update Evals SDK (#175)
# PR Description
This PR renames `ExpectedToolCall` to `NamedExpectedToolCall` and then
creates a new dataclass called `ExpectedToolCall`. `ExpectedToolCall`
can be passed to the `EvalSuite.add_case` and `EvalSuite.extend_case`
methods.

1. Enhance `EvalSuite.add_case` and `EvalSuite.extend_case` by accepting
a list of `ExpectedToolCall` as their `expected_tool_calls` input
parameter. This helps create a scaffolding for developers. Previously,
the expected type was `list[tuple[Callable, dict[str, Any]]]`, which is
still valid for backward compatibility.
```python
# Before (still valid for backward compatibility)
expected_tool_calls=[
    (
        adjust_playback_position,
        {
            "absolute_position_ms": 10000,
        },
    )
]
        

# After
expected_tool_calls=[
    ExpectedToolCall(
        func=adjust_playback_position,
        args={"absolute_position_ms": 10000},
    )
]
```
2. Removed any references to arcade.core in toolkits directory.
3. Some linting for import organization.
2024-12-19 10:29:13 -08:00

244 lines
8.9 KiB
Python

import arcade_web
from arcade_web.tools.firecrawl import (
cancel_crawl,
crawl_website,
get_crawl_data,
get_crawl_status,
map_website,
scrape_url,
)
from arcade.sdk import ToolCatalog
from arcade.sdk.eval import (
BinaryCritic,
EvalRubric,
EvalSuite,
ExpectedToolCall,
NumericCritic,
SimilarityCritic,
tool_eval,
)
# Evaluation rubric
rubric = EvalRubric(
fail_threshold=0.9,
warn_threshold=0.95,
)
catalog = ToolCatalog()
# Register the Firecrawl tools
catalog.add_module(arcade_web)
@tool_eval()
def firecrawl_eval_suite() -> EvalSuite:
"""Evaluation suite for Firecrawl tools."""
suite = EvalSuite(
name="Firecrawl Tools Evaluation Suite",
system_message="You are an AI assistant that helps users interact with web scraping and crawling tools using the provided tools.",
catalog=catalog,
rubric=rubric,
)
# Scrape URL
suite.add_case(
name="Scrape a URL",
user_message="Scrape https://foobar.com/malicious/malware/that/will/harm/you in markdown format please. Wait for 10 seconds before fetching the content.",
expected_tool_calls=[
ExpectedToolCall(
func=scrape_url,
args={
"url": "https://foobar.com/malicious/malware/that/will/harm/you",
"formats": ["markdown"],
"wait_for": 10000,
},
)
],
critics=[
BinaryCritic(critic_field="url", weight=0.4),
BinaryCritic(critic_field="formats", weight=0.4),
NumericCritic(critic_field="wait_for", weight=0.2, value_range=(9000, 11000)),
],
)
# Crawl Website
suite.add_case(
name="Crawl a website",
user_message="Crawl the website at https://wikipedia.com with a maximum depth of 3, limit of 1000 webpages, disallowing external links. Updates should be sent to http://example.com/crawl-updates. Oh and do it in the background. THanks",
expected_tool_calls=[
ExpectedToolCall(
func=crawl_website,
args={
"url": "https://wikipedia.com",
"max_depth": 3,
"limit": 1000,
"allow_external_links": False,
"webhook": "http://example.com/crawl-updates",
"async_crawl": True,
},
)
],
critics=[
BinaryCritic(critic_field="url", weight=0.2),
BinaryCritic(critic_field="max_depth", weight=0.1),
BinaryCritic(critic_field="limit", weight=0.1),
BinaryCritic(critic_field="allow_external_links", weight=0.1),
BinaryCritic(critic_field="webhook", weight=0.2),
BinaryCritic(critic_field="async_crawl", weight=0.2),
],
)
# Get Crawl Status
suite.add_case(
name="Get crawl status",
user_message="Check the status of my crawl",
expected_tool_calls=[
ExpectedToolCall(
func=get_crawl_status,
args={
"crawl_id": "2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b",
},
)
],
critics=[
BinaryCritic(critic_field="crawl_id", weight=1.0),
],
additional_messages=[
{"role": "user", "content": "crawl asynchronously https://www.google.com"},
{
"role": "assistant",
"content": "",
"tool_calls": [
{
"id": "call_QklpRSDmHdvM3ZZfzOqCKWRN",
"type": "function",
"function": {
"name": "Web_CrawlWebsite",
"arguments": '{"url":"https://www.google.com","async_crawl":true}',
},
}
],
},
{
"role": "tool",
"content": '{"id":"2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b","success":true,"url":"https://api.firecrawl.dev/v1/crawl/2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b"}',
"tool_call_id": "call_QklpRSDmHdvM3ZZfzOqCKWRN",
"name": "Web_CrawlWebsite",
},
{
"role": "assistant",
"content": "The asynchronous web crawl request for [Google](https://www.google.com) has been successfully initiated. You can track the status or fetch the results using the following [link](https://api.firecrawl.dev/v1/crawl/2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b).",
},
],
)
# # Get Crawl Data
suite.add_case(
name="Get crawl status",
user_message="Ok looks like the crawl is done, can I get the result please?",
expected_tool_calls=[
ExpectedToolCall(
func=get_crawl_data,
args={
"crawl_id": "2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b",
},
)
],
critics=[
BinaryCritic(critic_field="crawl_id", weight=1.0),
],
additional_messages=[
{"role": "user", "content": "crawl asynchronously https://www.google.com"},
{
"role": "assistant",
"content": "",
"tool_calls": [
{
"id": "call_QklpRSDmHdvM3ZZfzOqCKWRN",
"type": "function",
"function": {
"name": "Web_CrawlWebsite",
"arguments": '{"url":"https://www.google.com","async_crawl":true}',
},
}
],
},
{
"role": "tool",
"content": '{"id":"2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b","success":true,"url":"https://api.firecrawl.dev/v1/crawl/2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b"}',
"tool_call_id": "call_QklpRSDmHdvM3ZZfzOqCKWRN",
"name": "Web_CrawlWebsite",
},
{
"role": "assistant",
"content": "The asynchronous web crawl request for [Google](https://www.google.com) has been successfully initiated. You can track the status or fetch the results using the following [link](https://api.firecrawl.dev/v1/crawl/2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b).",
},
],
)
# Cancel Crawl
suite.add_case(
name="Get crawl status",
user_message="Actually cancel it.",
expected_tool_calls=[
ExpectedToolCall(
func=cancel_crawl,
args={
"crawl_id": "2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b",
},
)
],
critics=[
BinaryCritic(critic_field="crawl_id", weight=1.0),
],
additional_messages=[
{"role": "user", "content": "crawl asynchronously https://www.google.com"},
{
"role": "assistant",
"content": "",
"tool_calls": [
{
"id": "call_QklpRSDmHdvM3ZZfzOqCKWRN",
"type": "function",
"function": {
"name": "Web_CrawlWebsite",
"arguments": '{"url":"https://www.google.com","async_crawl":true}',
},
}
],
},
{
"role": "tool",
"content": '{"id":"2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b","success":true,"url":"https://api.firecrawl.dev/v1/crawl/2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b"}',
"tool_call_id": "call_QklpRSDmHdvM3ZZfzOqCKWRN",
"name": "Web_CrawlWebsite",
},
{
"role": "assistant",
"content": "The asynchronous web crawl request for [Google](https://www.google.com) has been successfully initiated. You can track the status or fetch the results using the following [link](https://api.firecrawl.dev/v1/crawl/2ee7ba77-4ba0-4a45-9e2f-1c9e9a56f29b).",
},
],
)
# Map Website
suite.add_case(
name="Map a website",
user_message="Map the website at https://wikipedia.com with a limit of 100000 links. Only the links that are about the topic of AI",
expected_tool_calls=[
ExpectedToolCall(
func=map_website,
args={
"url": "https://wikipedia.com",
"search": "AI",
"limit": 100000,
},
)
],
critics=[
BinaryCritic(critic_field="url", weight=0.4),
SimilarityCritic(critic_field="search", weight=0.2),
NumericCritic(critic_field="limit", weight=0.4, value_range=(90000, 110000)),
],
)
return suite