yushiyangk/parsetypes

Parse serialised data to recover their original underlying types

parsetypes

This Python package provides tools for parsing serialised data to recover their original underlying types.

Overview

The TypeParser class provides configurable type inference and parsing. This can be initialised with different settings to, for example:

• allow None (null values) or not
• treat inf as either a float or a normal string
• give exact Decimal values instead of floats
• detect inline lists

Install

pip install parsetypes

Basic examples

Import parser:

from parsetypes import TypeParser

Parse a single value:

parser = TypeParser()
parser.parse("1.2")   # 1.2
parser.parse("true")  # True
parser.parse("")      # None

Parse a series so that it has a consistent type:

parser = TypeParser()
parser.parse_series(["0", "1", "2"])         # [0, 1, 2]
parser.parse_series(["0", "1.2", ""])        # [0.0, 1.2, None]
parser.parse_series(["false", "true", ""])   # [False, True, None]
parser.parse_series(["false", "true", "2"])  # [0, 1, 2]
parser.parse_series(["1", "2.3", "abc"])     # ["1", "2.3", "abc"]

Parse a table so that each column is of a consistent type:

parser = TypeParser()
table = parser.parse_table([
	["0", "3",   "false", "false", "7"],
	["1", "4.5", "true",  "true",  "8.9"],
	["2", "",    "",      "6",     "abc"],
]):
assert table == [
	[0, 3.0,  False, 0, "7"],
	[1, 4.5,  True,  1, "8.9"],
	[2, None, None,  6, "abc"],
]

The main contribution of this module lies in the infer_series() and infer_table() functions, which are also called by parse_series() and parse_table().

Issues

Found a bug? Please report an issue, or, better yet, contribute a bugfix.