primarilysnark / katabankocr Goto Github PK

In the event that an account number is not recognized, the program currently simply returns the account number and "ERR". Proper handling involves checking for bad OCR data.

To properly check for bad OCR data, we need to map across all possible account numbers that can be created by changing the OCR'd input by one character. Since this change will only change one character, we can build a mapping for each character of each character it can match with only one OCR'd character change.

Example:

 _         _
| |   ->  |_|
|_|       |_|

We can then permute over the list of characters, pushing each possible variation onto a list and checking for each validity. The entire thing should be able to be done with one .SelectMany() and one .Where() to check for valid alternatives.

Currently the output prints to Console. We need to write the output to a file. This is low hanging fruit if anyone wants it. We can start with a hard-coded file and change it to be taken as an argument later on.