compass.validation.location.JurisdictionWebsiteValidator#

class JurisdictionWebsiteValidator(browser_semaphore=None, file_loader_kwargs=None, **kwargs)[source]#

Bases: object

Validate whether a website is the primary jurisdiction portal

Notes

The validator stores the initialization arguments so they can be reused across many documents without reconfiguration.

Parameters:

  • browser_semaphore (asyncio.Semaphore, optional) – Semaphore constraining concurrent Playwright usage. None applies no concurrency limit. Default is None.

  • file_loader_kwargs (dict, optional) – Keyword arguments passed to elm.web.file_loader.AsyncWebFileLoader. Default is None.

  • **kwargs – Additional keyword arguments cached for downstream LLM calls triggered during validation.

Methods

check(url, jurisdiction)

Determine whether a website serves as a jurisdiction's portal

Attributes

WEB_PAGE_CHECK_SYSTEM_MESSAGE

System message for main jurisdiction website validation calls
WEB_PAGE_CHECK_SYSTEM_MESSAGE = 'You are an expert data analyst that examines website text to determine if the website is the main website for a given jurisdiction. Only ever answer based on the information from the website itself. Do not treat affiliated but separate entities as the main jurisdiction website (i.e. such as school district websites, courthouse or court system websites, library websites, or other department and agency websites are **not** the main website).'#

System message for main jurisdiction website validation calls

async check(url, jurisdiction)[source]#

Determine whether a website serves as a jurisdiction’s portal

The validator first performs an inexpensive URL classification before downloading page content. Only when the URL fails the initial check does it fetch and inspect the page text using a generic LLM caller.

Parameters:

  • url (str) – URL to inspect. Empty values return False immediately.

  • jurisdiction (Jurisdiction) – Target jurisdiction descriptor used to frame the validation prompts.

Returns:

bool – True when either the URL quick check or the full page evaluation indicates the site is the official main website for the jurisdiction.

Raises:

compass.exceptions.COMPASSError – Propagated from BaseLLMCaller if configured to raise on LLM failures.

Examples

>>> validator = JurisdictionWebsiteValidator()
>>> await validator.check("https://county.gov", jurisdiction)
True