Module integrators.docuvision_utils

Utility and helper functions for docuvision_integrator.py

Functions

def blank_response(*, task_dict: dict[str, typing.Any], error_message: str, **kwargs) ‑> GetTaskResponse

get an empty template for a response with the minimum fields needed to avoid lookup errors when compiling results. also include diagnostic info in the final output if required/available.

def check_cascaded_pids(txt_pages: list[str], response: Any)

** MODIFIES RESPONSE IN PLACE **** updates pid value and found status for pages with cascaded PIDs.

Builds a list of names discovered in results for the found PIDs immediately before and after each set of contiguous pages without a found PID. Tests both lists of names against each line of OCR text in each page. The PID whose names produce the best sequence match ratio is selected for the page if the margin between the non-selected best match and selected best match is greater than 0.2. Otherwise, the PID is set to unknown to allow a user to sort out the issue.

def format_dotmap(dotmap: str) ‑> str

format dotmap

def format_value(value: str, dotmap: str)

Apply standard formatting for specific value types or return value if supplied value or label does not constitute special handling.

def valid_dos(value: str | None) ‑> bool

False if supplied date str is future or past by more than 30 days

def validate_confidence(confidence: float, value: str, dotmap: str) ‑> float

Check validity of extracted value and set confidence to 0.0 if value is invalid (i.e. a "date" field whose value is not a valid date should have confidence = 0.0)

Classes

class GetTaskResponse (id: int = 0, name: str = 'Error', error: str | None = None, state: str = 'Error', result: GetTaskResult = <factory>, status: str | None = None, service: str = 'docuvision-1', metadata: dict[str, typing.Any] = <factory>, statistics: dict[str, typing.Any] = <factory>, processedDocument: dict[str, typing.Any] = <factory>, default_dos: str = '2024-08-27', force_default_dos: dataclasses.InitVar[bool] = False)

dataclass representation of a docuvision response

Class variables

var default_dos : str

var error : str | None

var force_default_dos : dataclasses.InitVar[bool]

var id : int

var metadata : dict[str, typing.Any]

var name : str

var processedDocument : dict[str, typing.Any]

var result : GetTaskResult

var service : str

var state : str

var statistics : dict[str, typing.Any]

var status : str | None

class GetTaskResult (PIDS: list[str] = <factory>, PAGES: list[dict[str, typing.Any]] = <factory>, RESULT: list[dict[str, typing.Any]] = <factory>, METADATA: dict[str, typing.Any] = <factory>)

dataclass representation of a docuvision response

Class variables

var METADATA : dict[str, typing.Any]

var PAGES : list[dict[str, typing.Any]]

var PIDS : list[str]

var RESULT : list[dict[str, typing.Any]]

class PidTestCase (before_pid: str, unknown: list[list[str]], after_pid: str, start_offset: int)

tuples of (previous, concatenated unknown page text, next)

Ancestors

• builtins.tuple

Instance variables

var after_pid : str

Alias for field number 2

var before_pid : str

Alias for field number 0

var start_offset : int

Alias for field number 3

var unknown : list[list[str]]

Alias for field number 1