Faux-lars

fauxlars is a speedy fake data generation library.

Example

Install the library

python -m venv .venv && source .venv/bin/activate && python -m pip install faux_lars

Let's generate some data

import polars as pl
from faux_lars import generate_lazyframe

rows =5000
df = (
    generate_lazyframe(
        {
            "clicks": pl.UInt8,
            "Name": "name",
            "City Name": "city_name",
            "Phone": "mobile_number",
        },
        rows,
        "en",
    ).collect()
)
print(df)

┌────────┬──────────────────┬──────────────────────────┬────────────────┐
│ clicks ┆ Name             ┆ City Name                ┆ Phone          │
│ ---    ┆ ---              ┆ ---                      ┆ ---            │
│ u8     ┆ str              ┆ str                      ┆ str            │
╞════════╪══════════════════╪══════════════════════════╪════════════════╡
│ 207    ┆ Erik Conn        ┆ Beier stad               ┆ 512-797-9451   │
│ 66     ┆ Hildegard Murphy ┆ Wintheiser bury          ┆ (907) 011-3125 │
│ 38     ┆ Tremayne Casper  ┆ Okuneva land             ┆ 933.955.5987   │
│ 157    ┆ Kamille Haley    ┆ McGlynn furt             ┆ 329.524.7080   │
│ 123    ┆ Erika Kozey      ┆ East Favian Fisher burgh ┆ 1-846-364-8772 │
│ …      ┆ …                ┆ …                        ┆ …              │
│ 25     ┆ Chadd Rosenbaum  ┆ Nader furt               ┆ 201-837-7966   │
│ 50     ┆ Tevin Jerde      ┆ Davis shire              ┆ 1-125-407-6570 │
│ 22     ┆ Antwan Jones     ┆ West Sammie Hirthe shire ┆ 365-256-8860   │
│ 120    ┆ Eden McCullough  ┆ Goldner shire            ┆ 1-950-582-2326 │
│ 76     ┆ Tevin Batz       ┆ Bauch mouth              ┆ 560-202-3844   │
└────────┴──────────────────┴──────────────────────────┴────────────────┘

Docs

The faux_lars python API is deliberately small and lightweight.

There are two functions:

• generate_lazyframe
• generate_dataframe

Each receive the following arguments:

• schema: A dictionary of column_name, data_type key, value pairs. data_type can be a string data_type in the table below, or a non-complex polars DataType object.
• rows: The number of rows to generate.
• language: Any of the supported locales; see the table below.

Supported Locales

Data Type	en	fr	jp	ar	pt_br	zh_tw	zh_cn
polars_datatypes	✓*	✓*	✓*	✓*	✓*	✓*	✓*
name	✓	✓	✓	✓	✓	✓	✓
first_name	✓	✓	✓	✓	✓	✓	✓
last_name	✓	✓	✓	✓	✓	✓	✓
building_number	✓*	✓*	✓*	✓*	✓*	✓*	✓*
city_name	✓
country_code	✓*	✓*	✓*	✓*	✓*	✓*	✓*
country_name	✓
latitude	✓
longitude	✓
postcode	✓
secondary_address	✓
secondary_address_type	✓
state_abbr	✓
state_name	✓				✓
street_name	✓				✓
time_zone	✓
zip_code	✓
isbn	✓*	✓*	✓*	✓*	✓*	✓*	✓*
isbn_10	✓*	✓*	✓*	✓*	✓*	✓*	✓*
isbn_13	✓*	✓*	✓*	✓*	✓*	✓*	✓*
bs	✓
company_name	✓	✓	✓	✓	✓	✓	✓
industry	✓
profession	✓
credit_card_number	✓*	✓*	✓*	✓*	✓*	✓*	✓*
currency_code	✓*	✓*	✓*	✓*	✓*	✓*	✓*
currency_name	✓
currency_symbol	✓*	✓*	✓*	✓*	✓*	✓*	✓*
dir_path	✓
file_ext	✓
file_name	✓
file_path	✓
mime_type	✓
semver	✓
semver_stable	✓
semver_unstable	✓
licence_plate		✓
health_insurance_code		✓
free_email	✓	✓			✓
bic	✓*	✓*	✓*	✓*	✓*	✓*	✓*
isin	✓*	✓*	✓*	✓*	✓*	✓*	✓*
ip	✓*	✓*	✓*	✓*	✓*	✓*	✓*
ipv4	✓*	✓*	✓*	✓*	✓*	✓*	✓*
ipv6	✓*	✓*	✓*	✓*	✓*	✓*	✓*
mac_address	✓*	✓*	✓*	✓*	✓*	✓*	✓*
password	✓	✓			✓
safe_email	✓	✓	✓	✓	✓	✓	✓
user_agent	✓*	✓*	✓*	✓*	✓*	✓*	✓*
user_name	✓*	✓*	✓*	✓*	✓*	✓*	✓*
field	✓		✓				✓
position	✓		✓				✓
seniority	✓		✓				✓
title	✓		✓				✓
lorem	✓*	✓*	✓*	✓*	✓*	✓*	✓*
cell_number	✓*	✓*	✓*	✓*	✓*	✓*	✓*
phone_number	✓*	✓*	✓*	✓*	✓*	✓*	✓*

Key:

• ✓
• ✓ : supported for this locale.
• ✓* : supported, but this value does not vary by locale.

All non-complex polars DataType objects are supported. These can be passed by name as a string, or as a polars DataType object (see the example above).

For non-sring types, locale is irrelevant. A complex polars Datatype is a struct, array, list, or enum.

Locales:

• en: English
• fr: French
• jp: Japanese
• ar: Saudi Arabian Arabic
• pt_br: Brazilian Portugese
• zh_tw: Traditional Chinese
• zh_cn: Simplified Chinese

Benchmarks

• On a laptop with 16GB of RAM and 8 cores, 5,000 rows with 1 utf8 column and three string columns generates in under 0.25 seconds.
• 1,000,000 rows generates in under 7.
• Take a look at benchmarks for comparison with two popular python fake data generation libraries: mimesis and faker.