Add clinical trials search tool #777
Conversation
dosubot
bot
added
size:XXL
| @@ -33,8 +39,20 @@ | |||
|
|
|||
| POPULATE_FROM_SETTINGS = None | |||
|
|
|||
| DEFAULT_TOOL_NAMES: list[str] = [ | |||
There was a problem hiding this comment.
I like this, nice.
What do you think of moving this to be co-located with AVAILABLE_TOOL_NAME_TO_CLASS in tools?
| class ClinicalTrialsSearch(NamedTool): | ||
| TOOL_FN_NAME = "clinical_trials_search" | ||
|
|
||
| model_config = ConfigDict(extra="forbid", arbitrary_types_allowed=True) |
There was a problem hiding this comment.
Why arbitrary_types_allowed=True? I don't think we need it (may be wrong here tho)
| """ | ||
| ) | ||
|
|
||
| async def clinical_trials_search(self, query: str, state: EnvironmentState): |
There was a problem hiding this comment.
| async def clinical_trials_search(self, query: str, state: EnvironmentState): | |
| async def clinical_trials_search(self, query: str, state: EnvironmentState) -> str: |
Even though we don't use return type now, it's still nice to have for IDE usage
| settings: Settings = Field(default_factory=Settings) | ||
|
|
||
| # Gather evidence tool must be modified to understand the new evidence | ||
| GATHER_EVIDENCE_TOOL_PROMPT_OVERRIDE: ClassVar[str] = ( |
There was a problem hiding this comment.
Did you try printing this? I think it will have extra spaces unless you do """Gather evidence...
| }, | ||
| ) as response: | ||
| if response.status == MALFORMATTED_QUERY_STATUS: | ||
| # the 400s from clinicaltrials.gov are not JSON |
There was a problem hiding this comment.
Do you mind moving this comment to live with MALFORMATTED_QUERY_STATUS, or renaming MALFORMATTED_QUERY_STATUS to be CLINICAL_TRIALS_DIDNT_GIVE_JSON = 400
| Returns: | ||
| tuple[int, int, str | None]: | ||
| Total number of trials found, number of trials added, and error message if any. | ||
|
|
|
|
||
|
|
||
| # SEE: https://regex101.com/r/L0L5MH/1 | ||
| CLINICAL_STATUS_SEARCH_REGEX_PATTERN: str = ( |
There was a problem hiding this comment.
Can you co-locate this with the status generation code? Makes sense to have next to each other in the code
This upstreams FutureHouse's clinical trials search tool to make it open source. The tool is not turned on by default, but I'm going to make more docs on how to use the tool as well.
I also found many typing regressions -- I've added those back in where necessary.