resolve-parameter-values

Resolving parameter values with db_lookup

A parameter the user types from memory will not match stored data exactly. db_lookup puts a resolution step between extraction and execution: an exact match passes, a close match is corrected silently, an ambiguous one comes back as suggestions, and a genuine miss falls through.

Declaring it

class Signature:
    class Input(BaseModel):
        user_name: str = Field(
            description="Name of a person",
            examples=['John', 'Jane Doe'],
            json_schema_extra={'db_lookup': True},
        )

The hook

A static method on the same Signature, dispatching on field name:

from fastworkflow.utils.signatures import DatabaseValidator

    @staticmethod
    def db_lookup(workflow: fastworkflow.Workflow,
                  field_name: str,
                  field_value: str) -> tuple[bool, str | None, list[str]]:
        if field_name == 'user_name':
            chatroom = workflow.command_context_for_response_generation
            key_values = chatroom.list_users()
            return DatabaseValidator.fuzzy_match(field_value, key_values)
        return (False, '', [])

One hook serves every db_lookup field on the command, so always dispatch on field_name and always return the neutral (False, '', []) for anything unrecognised.

Fetch the key set at call time from the live context — workflow.command_context_for_response_generation — never from a module-level constant. The valid set depends on where the user is.

The return contract

(matched, corrected_value, suggestions), and the three outcomes are genuinely different:

Return	Effect
(True, value, [])	the parameter is overwritten with value and validation continues
(False, None, [s1, s2])	the parameter is marked invalid; the caller is asked "Did you mean one of these …?"
(False, '', [])	validation passes — absence from the key set is not by itself an error