Python library to convert between SI unit prefixes

Asked 10/6, 2012 at 15:3 Answered 26/4, 2023 at 10:25

I'm looking for a python library which comes with support to convert numbers between various SI prefixes, for example, kilo to pico, nano to giga and so on.What would you recommend?

Sentinel answered 10/6, 2012 at 15:3 Comment(4)

possible duplicate of Unit Conversion in Python – Motivation 10/6, 2012 at 15:7

@GWW: Not really, that question wants to convert units, this is about prefixes. – Kurtkurth 10/6, 2012 at 15:28

@Zed: It isn't really clear what you want to do. What is the format of the input, for example? Strings? Numbers and strings? Numbers and a prefix index? It might help if you gave an example of what you want to do exactly. – Kurtkurth 10/6, 2012 at 15:30

But since the number of prefixes is fairly limited, you might be better of with a simple dictionary (e.g. {'giga':1e9, 'kilo':1e3, 'milli':1e-3, ...}) – Kurtkurth 10/6, 2012 at 15:32

I ported a simple function (original C version written by Jukka “Yucca” Korpela) to Python for formatting numbers according to SI standards. I use it often, for example, to set tick labels on plots, etc.

You can install it with:

pip install si-prefix

The source is available on GitHub.

Example usage:

from si_prefix import si_format

print si_format(.5)
# 500.0m  (default precision is 1)

print si_format(.01331, precision=2)
# 13.31m

print si_format(1331, precision=2)
# 1.33k

print si_format(1331, precision=0)
# 1k

Knighthead answered 19/8, 2015 at 10:33 Comment(2)

This doesn't seem to have a way to go REVERSE (from SI string to number) – Redpoll 5/10, 2022 at 23:21

@K.S., please check out the si_prefix.si_parse() function (see here for the definition). This function was added in version 0.4. – Knighthead 7/10, 2022 at 13:14

QuantiPhy is a new package that converts to and from numbers with SI scale factors. It is often a better choice that the unit packages such as Unum and Magnitude that are heavier and focused on the units rather than the scale factors.

QuantiPhy provides Quantity, which is an object that combines a number with its unit of measure (the units are optional). When creating a quantity you can use SI unit prefixes. Once you have a Quantity you can use it in expressions, where it acts as a float. Or you can convert it to a string, in which case it uses the SI unit prefixes by default.

>>> from quantiphy import Quantity

# convert strings to quantities
>>> duration = Quantity('0.12 ks')
>>> print(duration)
120 s

# convert to other units when rendering to a string
>>> print(duration.render(scale='min'))
2 min

# quantities act like floats in expressions
>>> rate = 1/duration
>>> print(rate)
0.008333333333333333

# convert floats to quantities
>>> rate = Quantity(rate, 'Hz')
>>> print(rate)
8.3333 mHz

# can be used in format strings
>>> print(f'Duration = {duration:<12.3} Rate = {rate}')
Duration = 120 s        Rate = 8.3333 mHz

By default QuantiPhy uses the natural prefix when rendering to a string, which is probably what you want. But you can force it to render to a specific prefix using scaling:

>>> mass = Quantity('1000 g')
>>> print(mass)
1 kg

>>> print(mass.render(show_si=False))
1e3 g

>>> print(mass.render(show_si=False, scale=(1e-12, 'pg')))
1e9 pg

In this case you must turn off SI unit prefixes to avoid getting multiple prefixes: '1 npg'.

A more natural example might be where you are converting units:

>>> l = Quantity('2um')                                                      
>>> print(l.render(scale='Å'))                                               
20 kÅ                                                                        

>>> print(f'{l:sÅ}')                                                         
20 kÅ

The last example shows that you can place your desired units in the format string after the type and the conversion will be done for you automatically.

Credenza answered 15/7, 2017 at 21:23 Comment(0)

Dictionaries

If you don't want to use any 3rd-party library like the ones listed below, you can actually implement your own parsing function.

Use a dictionary to match up the prefixes to their values. I've done it for you already:

_prefix = {'y': 1e-24,  # yocto
           'z': 1e-21,  # zepto
           'a': 1e-18,  # atto
           'f': 1e-15,  # femto
           'p': 1e-12,  # pico
           'n': 1e-9,   # nano
           'u': 1e-6,   # micro
           'm': 1e-3,   # mili
           'c': 1e-2,   # centi
           'd': 1e-1,   # deci
           'k': 1e3,    # kilo
           'M': 1e6,    # mega
           'G': 1e9,    # giga
           'T': 1e12,   # tera
           'P': 1e15,   # peta
           'E': 1e18,   # exa
           'Z': 1e21,   # zetta
           'Y': 1e24,   # yotta
    }

Then you can use regex (as described by my answer here) to search or parse the input and use the dictionary for getting the appropriate value.

Unum

Unum is well finished and thoroughly documented library.

Pros:

allows you to define arbitrary units (magnitude only supports user-defined units as long as they are a combination of the base units).

Cons:

doesn't handle prefixes well
clutters your namespace with all its unit definitions (you end up with variables named M, S etc. in your namespace)

Magnitude

You can also use Magnitude, another library. It supports all the kinds of SI unit prefixes you're talking about, plus it'll handle the parsing as well. From the site:

A physical quantity is a number with a unit, like 10 km/h. Units are specified as strings. They can be any of the SI units, plus a bunch of non-SI, bits, dollars, and any combination of them. They can include the standard SI prefixes.
...
All standard prefixes are understood, from yocto to yotta and from kibi to exbi.

Prelusive answered 10/6, 2012 at 17:45 Comment(2)

Regex doesn't sound like the best way to detect which prefix to use. Doesn't log(abs(value)) give you the size of your value in a form you can use to select a prefix. – Gardell 12/11, 2013 at 11:22

Updated Unum link: pypi.org/project/Unum. – Conformation 31/7 at 10:18

I don't know if this is the best answer but it is working in my case. Feel free to verify my solution. I am working for first time with Python and constructive criticism is welcome... along with positive feedback :D
This is my code:

class Units:
def __init__(self):
    global si;
    si = {
          -18 : {'multiplier' : 10 ** 18, 'prefix' : 'a'},
          -17 : {'multiplier' : 10 ** 18, 'prefix' : 'a'},
          -16 : {'multiplier' : 10 ** 18, 'prefix' : 'a'},
          -15 : {'multiplier' : 10 ** 15, 'prefix' : 'f'},
          -14 : {'multiplier' : 10 ** 15, 'prefix' : 'f'},
          -13 : {'multiplier' : 10 ** 15, 'prefix' : 'f'},
          -12 : {'multiplier' : 10 ** 12, 'prefix' : 'p'},
          -11 : {'multiplier' : 10 ** 12, 'prefix' : 'p'},
          -10 : {'multiplier' : 10 ** 12, 'prefix' : 'p'},
          -9 : {'multiplier' : 10 ** 9, 'prefix' : 'n'},
          -8 : {'multiplier' : 10 ** 9, 'prefix' : 'n'},
          -7 : {'multiplier' : 10 ** 9, 'prefix' : 'n'},
          -6 : {'multiplier' : 10 ** 6, 'prefix' : 'u'},
          -5 : {'multiplier' : 10 ** 6, 'prefix' : 'u'},
          -4 : {'multiplier' : 10 ** 6, 'prefix' : 'u'},
          -3 : {'multiplier' : 10 ** 3, 'prefix' : 'm'},
          -2 : {'multiplier' : 10 ** 2, 'prefix' : 'c'},
          -1 : {'multiplier' : 10 ** 1, 'prefix' : 'd'},
           0 : {'multiplier' : 1, 'prefix' : ''},
           1 : {'multiplier' : 10 ** 1, 'prefix' : 'da'},
           2 : {'multiplier' : 10 ** 3, 'prefix' : 'k'},
           3 : {'multiplier' : 10 ** 3, 'prefix' : 'k'},
           4 : {'multiplier' : 10 ** 3, 'prefix' : 'k'},
           5 : {'multiplier' : 10 ** 3, 'prefix' : 'k'},
           6 : {'multiplier' : 10 ** 6, 'prefix' : 'M'},
           7 : {'multiplier' : 10 ** 6, 'prefix' : 'M'},
           8 : {'multiplier' : 10 ** 6, 'prefix' : 'M'},
           9 : {'multiplier' : 10 ** 9, 'prefix' : 'G'},
          10 : {'multiplier' : 10 ** 9, 'prefix' : 'G'},
          11 : {'multiplier' : 10 ** 9, 'prefix' : 'G'},
          12 : {'multiplier' : 10 ** 12, 'prefix' : 'T'},
          13 : {'multiplier' : 10 ** 12, 'prefix' : 'T'},
          14 : {'multiplier' : 10 ** 12, 'prefix' : 'T'},
          15 : {'multiplier' : 10 ** 15, 'prefix' : 'P'},
          16 : {'multiplier' : 10 ** 15, 'prefix' : 'P'},
          17 : {'multiplier' : 10 ** 15, 'prefix' : 'P'},
          18 : {'multiplier' : 10 ** 18, 'prefix' : 'E'},
          }

def convert(self, number):
    # Checking if its negative or positive
    if number < 0:
        negative = True;
    else:
        negative = False;

    # if its negative converting to positive (math.log()....)
    if negative:
        number = number - (number*2);

    # Taking the exponent
    exponent = int(math.log10(number));

    # Checking if it was negative converting it back to negative
    if negative:
        number = number - (number*2);

    # If the exponent is smaler than 0 dividing the exponent with -1
    if exponent < 0:
        exponent = exponent-1;
        return [number * si[exponent]['multiplier'], si[exponent]['prefix']]; 
    # If the exponent bigger than 0 just return it
    elif exponent > 0:
        return [number / si[exponent]['multiplier'], si[exponent]['prefix']]; 
    # If the exponent is 0 than return only the value
    elif exponent == 0:
        return [number, ''];

And this is how it works:

c1 = +1.189404E-010
fres = -4.07237500000000E+007;
ls = +1.943596E-005;

units = sci.Units();
rValue, rPrefix = units.convert(c1);
print rValue;
print rPrefix;

print units.convert(fres);
print units.convert(ls);

And the response is:

118.9404
p
[-40.72375, 'M']
[19.435959999999998, 'u']

I don't know if anyone will find this helpful or not. I hope you do. I've posted here so the people who want help to see it also to give them an idea maybe they can optimize it :)

Narthex answered 6/12, 2013 at 15:24 Comment(1)

Improvement suggestion: You use 'd' as the prefix for both deci and deca, which is both confusing and incorrect. Deca should have the prefix "da". – Fenestration 3/5, 2019 at 8:43

I know this is an old thread, but I'd just like to throw out a reference to a python library I wrote which handles all manner of prefix unit conversion handling

Here's the major feature list:

Converting between SI and NIST prefix units (kB to GiB)
Converting between units of the same type (SI to SI, or NIST to NIST)
Automatic human-readable prefix selection (like in hurry.filesize https://pypi.python.org/pypi/hurry.filesize)
Basic arithmetic operations (subtracting 42KiB from 50GiB)
Rich comparison operations (1024 Bytes == 1KiB)
bitwise operations (<<, >>, &, |, ^)
Reading a device's storage capacity (Linux/OS X support only)
argparse https://docs.python.org/2/library/argparse.html integration as a custom type
progressbar https://code.google.com/p/python-progressbar/ integration as a better file transfer speed widget
String parsing
Sorting

Demimondaine answered 7/1, 2016 at 16:32 Comment(1)

Super! Thank you! Testing this with a sum of ZFS filesystem volume sizes, in bytes -- certainly, any bytes source would do -- I like the exacting syntax for the units conversion here: GiB.from_other(Byte(used))) and it retains the decimal part! (This was my favorite part of HS chem tbt) – Highflown 13/12, 2023 at 3:8

You can use Prefixed, which has a float type with additional formatting options.

You can create float-like numbers by including the prefix

>>> from prefixed import Float

>>> Float('2k')
Float(2000.0)

prefixed.Float is a subclass of float, so you can use it just like a float, but when you want to output, it supports additional format specifiers.

num = Float('2k')
>>> f'{num}'
'2000.0'
>>> f'{num:.2h}'
'2.00k'

Binary prefixes are also supported and some additional formatting options. See the docs for more info.

Suppurative answered 18/11, 2022 at 18:40 Comment(0)

Since the question is not yet flagged as answered, I'll give you here my solution, which is largely inspired from si-prefix, but I wanted to limit the number of dependencies of my package.

You can use the following function :

def format_prefix(value, precision=2, unit="m"):
    """
   Formats a numerical value with an appropriate SI prefix.

   Args:
   - value: A float representing the numerical value to be formatted.
   - precision: An integer representing the number of decimal places to include in the formatted value (default: 2).
   - unit: A string representing the unit of measurement to be appended to the formatted value (default: "m").

   Returns:
   - A string representing the formatted value with the appropriate SI prefix and unit.

   Raises:
   - ValueError: If the exponent is out of range of the available prefixes.
    """
    SI_PREFIX_UNITS = u"yzafpnµm kMGTPEZY"
    negative = False
    digits = precision + 2

    if value == 0.:
        expof10 = 0
    else:
        if value < 0.:
            value = -value
            negative = True
        expof10 = int(np.log10(value))
        if expof10 > 0:
            expof10 = (expof10 // 3) * 3
        else:
            expof10 = (-expof10 + 3) // 3 * (-3)

        value *= 10 ** (-expof10)

        if value >= 1000.:
            value /= 1000.0
            expof10 += 3
        elif value >= 100.0:
            digits -= 2
        elif value >= 10.0:
            digits -= 1

        if negative:
            value *= -1
    expof10 = int(expof10)
    prefix_levels = (len(SI_PREFIX_UNITS) - 1) // 2
    si_level = expof10 // 3

    if abs(si_level) > prefix_levels:
        raise ValueError("Exponent out range of available prefixes.")
    return f"{round(value*10**digits)/10**digits} "+SI_PREFIX_UNITS[si_level + prefix_levels].strip()+unit

The function outputs a string like this :

>>> format_prefix(123687.2e-9, unit="m")
'123.69 µm'
>>> format_prefix(123687.2e-9, precision=12, unit="rad")
'123.6872 µrad'

Steamship answered 26/4, 2023 at 10:25 Comment(0)

@naitsirhc, thanks for your package. i have added a little function idea to use your package

import pandas as pd
import collections
Measure = collections.namedtuple('Measure', 'SLOT TEXT AVG HIGH LAST LOW SIGMA SWEEPS')
d=[Measure(SLOT='1', TEXT='CH1,AMPLITUDE', AVG='584.4782173493248E-3', HIGH='603.9744119119119E-3', LAST='594.125218968969E-3', LOW='561.1797735235235E-3', SIGMA='5.0385410346638E-3', SWEEPS='237996'), Measure(SLOT='2', TEXT='CH1,FREQUENCY', AVG='873.9706607717992E+6', HIGH='886.1564731675113E+6', LAST='873.9263571643770E+6', LOW='854.8833348698727E+6', SIGMA='4.382200567330E+6', SWEEPS='20705739'), Measure(SLOT='3', TEXT='CH4,PERIOD', AVG='1.1428492411436E-9', HIGH='1.1718844685593E-9', LAST='1.1432428766843E-9', LOW='1.1261916413092E-9', SIGMA='6.6735923746950E-12', SWEEPS='20680921'), Measure(SLOT='4', TEXT='CH4,FREQUENCY', AVG='875.0358282079155E+6', HIGH='887.9483414008331E+6', LAST='874.780693212961E+6', LOW='853.3264385945507E+6', SIGMA='5.0993358972092E+6', SWEEPS='20681008')]

from si_prefix import si_format

import si_prefix
si_prefix.SI_PREFIX_UNITS="yzafpnum kMGTPEZY"

def siSuffixNotation(element):
    try:
        ret=float(element)
        return str(si_format(ret)).replace(' ','')
    except ValueError:
        return element

df=pd.DataFrame(d)

df.T.applymap(siSuffixNotation) #<= nice pretty print output table
                    0              1           2              3
SLOT              1.0            2.0         3.0            4.0
TEXT    CH1,AMPLITUDE  CH1,FREQUENCY  CH4,PERIOD  CH4,FREQUENCY
AVG            584.5m         874.0M        1.1n         875.1M
HIGH           604.0m         885.6M        1.2n         887.9M
LAST           586.5m         874.2M        1.1n         874.9M
LOW            561.2m         854.9M        1.1n         854.1M
SIGMA            5.0m           4.4M        6.7p           5.1M
SWEEPS         191.5k          16.7M       16.6M          16.6M

Thanks to you, i can know have a pretty print table as i like it. (i don't like space between the number and the suffix, and i do not like the unicode type, i prefer u for micro) ++

Nitaniter answered 6/1, 2022 at 14:36 Comment(0)

Hot tags

Godot Unity Godot Help Programming Godot 4.X GUI GDScript 3D 2D Physics CSharp Godot 3.X VR XR Projects C++

Dictionaries

Unum

Magnitude

Recommended topics

Hot tags