Convert Numbers In Text To Words By Language

Assumptions

The definition of a string of a valid number is given in base.py
File size should not exceed available memory allowed by OS
The supported number does not exceed 999,999,999,999
At the moment only English is supported

Design and Implementation

Scope(Boundary) of the problem
- The input is a string of text. The text can be in any natural language. The source of intput can be read from memory, file, network, etc.
- The output is a string of numbers in words. The words can be in any natural languages.
- The string of a valid number can be of type int, float, scientific notion, etc. In this version, int is supported.
- Given target language English, the output is made up of a finite set of number words. Therefore a full enumeration of building words should be provided
Test Setup(TDD cycle)

the test should start from passing simple common testcases. for instance 0, 1, 10, in order to make sure the project can always pass the test while deleveoping further on. In summary, the testsuite should include
- string of invalid numbers
- string of unique-spelling numbers
- string of non-unique-spelling numbers
- string of mixed of words and numbers
- string of text of any forms
Top-down approach

At the beginning of prototyping, I started from identifying objects and defining high level interfaces. In this problem, A generic converter, which handles text parsing and number formatting, and a language specific converter, which handles parsing of unique number string to this language have been identified. By doing so, the project is scalable horizontally at adding more language specific converters while vertically scallable at handling IO, formatting, more number types.

Core algorithm

Given a valid number
    if type is int
        parse int by 3 digits groups
    if type is float
        parse int by 3 digits groups
        parse decimals
        concat
    if scientific
        pasrse scientific

Given the first iteration, the implementation of integer parsing is delivered. More number types can be added in the following versions.

Future work
- add network based reader to io utils, add writer to files
- add more language specific converters
- add converting function for floats, and scientific notations
- extend the max number to be supported

Instructions to Run

Project

Install this package by running bin/install.sh
In terminal, you can run n2w by using below command examples.

$ n2w -t 'abc 123'
one hundred and twenty-three
$ n2w -f 'test/test_input.txt'
eleven
twenty-one

Test Suites

Navigate to project root and run pytest

$pytest
=================================== test session starts ====================================
platform darwin -- Python 3.5.0, pytest-5.4.3, py-1.9.0, pluggy-0.13.1
rootdir: /Users/justin/PycharmProjects/n2w
collected 10 items                                                                         

test/test_num2word.py ..........                                                     [100%]

==================================== 10 passed in 0.06s ====================================

derxir/n2w_public