Remittance Slip Digit Recognition

Remittance Slip Digit Recognition

Your customer's bank returns remittance slips as flat text: the 9-digit payment reference is printed as seven-segment ASCII art across three rows, and the scanner occasionally drops or adds a single stroke. Your job is the recognition pipeline distilled to its core algorithm: read the digits, validate the weighted check digit, and — when a scan is broken — work out whether exactly one single-segment repair can rescue it or whether the slip must go to a human.

Requirements

Create a codeunit named "Payment Slip OCR" with three public procedures:

procedure RecognizeDigits(Top: Text; Middle: Text; Bottom: Text): Text
procedure HasValidCheckDigit(Reference: Text): Boolean
procedure Scan(Top: Text; Middle: Text; Bottom: Text): Text

The slip format

Each scan arrives as three rows of exactly 27 characters containing only spaces, underscores (_), and pipes (|).

Cell i (1..9, left to right) is columns 3i-2 to 3i of all three rows — a 3×3 block of 9 characters.

The canonical renderings of the ten digits, 0 through 9, three columns per digit:

 _     _  _     _  _  _  _  _ 
| |  | _| _||_||_ |_   ||_||_|
|_|  ||_  _|  | _||_|  ||_| _|

A full scan of the reference 123456789 looks like this:

    _  _     _  _  _  _  _ 
  | _| _||_||_ |_   ||_||_|
  ||_  _|  | _||_|  ||_| _|

RecognizeDigits

Return exactly 9 characters, one per cell, left to right.

A cell that matches a canonical digit rendering exactly (all 9 characters) becomes that digit; any other cell becomes ?.