datafilter 0.3.0

Quickly find tokens (words, phrases, etc) within your data.

Data Filter

Quickly find tokens (words, phrases, etc) within your data.

Data Filter is a lightweight data cleansing framework that can be easily extended to support different data types, structures or processing requirements. It natively supports the following data types:

• CSV files
• Text files
• Text strings

Table of Contents

• Requirements
• Installation
• Basic Usage
• Features
  • Base
  • Filters
    • CSV
    • Text
    • TextFile

Requirements

• Python 3.6+

Installation

To install, simply use pipenv (or pip):

>>> pipenv install datafilter

Basic Usage

Each example below returns a generator that yields parsed data.

CSV

from datafilter import CSV

tokens = ["Lorem", "ipsum", "Volutpat est", "mi sit amet"]
data = CSV("test.csv", tokens=tokens)
print(next(data.results()))

Text

from datafilter import Text

text = "Lorem ipsum dolor sit amet, consectetur adipiscing elit"
data = Text(text, tokens=["Lorem"])
print(next(data.results()))

Text File

from datafilter import TextFile

data = TextFile("test.txt", tokens=["Lorem", "ipsum"], re_split=r"(?<=\.)")
print(next(data.results()))

Features

Data Filter was designed to be highly extensible. Common or useful filters can be easily reused and shared. A few example use cases include:

• Filters that can handle different data types such as Microsoft Word, Google Docs, etc.
• Filters that can handle incoming data from external APIs.

Base

Abstract base class that's subclassed by every filter.

Base includes several methods to ensure data is properly normalized, formatted and returned. The results method is an @abstractmethod to enforce its use in subclasses.

Parameters

tokens

type <list>

A list of strings that will be searched for within a set of data.

translations

type <list>

A list of strings that will be removed during normalization.

Default

['0123456789', '(){}[]<>!?.:;,`\'"@#$%^&*+-|=~–—/\\_', '\t\n\r\x0c\x0b']

bidirectional

type <bool>

When True, token matching will be bidirectional.

Default

True

Note:

A common obfuscation method is to reverse the offending string or phrase. This helps detect those instances.

caseinsensitive

type <bool>

When True, tokens and data are converted to lowercase during normalization.

Default

True

Methods

results

Abstract method used to return results within a filter. This is defined by a Base subclass

maketrans

Returns a translation table used during normalization.

Returns

type <dict>

normalize

Returns normalized data. Normalization includes converting data to lowercase and removing strings.

Returns

type <tuple>

Note:

Normalized data is returned as a tuple. The first element is the original data. The second element is the normalized data.

parse

Returns parsed and properly formatted data.

Returns

type <dict>

Example:

Assume we're searching for the token "Lorem" in a very short string.

data = Text("Lorem ipsum dolor sit amet", tokens=["Lorem"])
print(next(data.results()))

The returned result would be formatted as:

{
    "data": "Lorem ipsum dolor sit amet",
    "flagged": True,
    "describe": {
        "tokens": {
            "detected": ["Lorem"],
            "count": 1,
            "frequency": {
                "Lorem": 1,
            },
        }
    },
}

Filters

Filters subclass and extend the Base class to support various data types and structure. This extensibility allows for the creation of powerful custom filters specifically tailored to a given task, data type or structure.

CSV

Parameters

CSV is a subclass of Base and inherits all parameters.

path

type <str>

Path to a CSV file.

Methods

CSV is a subclass of Base and inherits all methods.

Text

Parameters

Text is a subclass of Base and inherits all parameters.

text

type <str>

A text string.

re_split

type <str>

A regular expression pattern or string that will be applied to text with re.split before normalization.

Methods

Text is a subclass of Base and inherits all methods.

TextFile

Parameters

TextFile is a subclass of Base and inherits all parameters.

path

type <str>

Path to a text file.

re_split

type <str>

A regular expression pattern or string that will be applied to text with re.split before normalization.

Methods

TextFile is a subclass of Base and inherits all methods.