Overview

Introduction and Examples

The fuzzyparsers library provides a small collection of functions to sanitize free form user input. For the moment its chief value is the flexible date parser. The library strives to be locale smart in parsing dates.

The library has two main parsers. The first is a prefix parser which compares a string to a list of strings and returns the unique element of the list which matches the prefix. An exception is thrown if the match is not unique.

>>> from fuzzyparsers import fuzzy_match
>>> fuzzy_match(['aab','bba','abc'],'aa')
'aab'
>>> fuzzy_match(['aab','bba','abc'],'a')  # two strings starting with 'a'.
Traceback (most recent call last):
...
ValueError: ambiguous match for 'a'

The second parser parses dates in various formats and returns a datetime.date object. Accepted formats include:

jan 12, 2003
jan 5
2004-3-5
+34 -- 34 days in the future (relative to todays date)
-4 -- 4 days in the past (relative to todays date)

For instance:

>>> from fuzzyparsers import parse_date
>>> parse_date('nov 30 2012') # my youngest son's birthday
datetime.date(2012, 11, 30)

The library allows setting a default date to fill in specified components of a date (e.g. the year). By default, a date with-out a year to will give the current year.

>>> from fuzzyparsers import DateParser
>>> import datetime
>>> DateParser(today=datetime.date(2013, 3, 1)).parse_date('feb 3')
datetime.date(2013, 2, 3)

TODO

We'd like to support the following features:

  • Parsing time strings like "10 am" and "3 pm"
  • A "[0-9]*.[0-9]*" with the first hunk a month and the second hunk a day should return the month/day combination which is nearest. For example, "12-3" would return december 3 of this year or last year.

Changelog

  • 0.9.0 - support Python 3
  • 0.8.0 - switch to MIT license because I couldn't figure out why I cared about GPL for this package.
  • 0.7.3 - locale month-day order issues fixed (thanks to Treeve for getting this started.
  • 0.7.2 - added doc-tests and "march 2012" date format; doc-test scripts
  • 0.7.1 - install fixes
  • 0.7 - overhaul of date parsing api to support relative dates (not necessarily relative to the current date)
  • 0.6.x - initial public release and series of doc/install corrections

Installation

Fuzzyparsers is written by Joel B. Mohler and distributed under the terms of the MIT license.

Use the following commands to run the extensive doc-tests:

py.test fuzzyparsers --doctest-module
python -m doctest README.txt

To install fuzzyparsers, do the normal python thing (probably as root):

python setup.py install

or:

pip install fuzzyparsers