Polars Expression Transformer

Transform string-based expressions into Polars DataFrame operations. Write simple, SQL-like expressions and let the library convert them to optimized Polars code.

Playground & docs

There is an interactive playground and function reference that runs the library in the browser through Pyodide. You can try expressions on sample data and see the generated Polars and FlowFrame code without installing anything.

The site lives in docs/; the function reference is generated from the docstrings with python generate_docs.py.

Quick Start

import polars as pl
from polars_expr_transformer import simple_function_to_expr

df = pl.DataFrame({
    'first_name': ['John', 'Jane', 'Bob'],
    'last_name': ['Doe', 'Smith', 'Johnson'],
    'age': [30, 25, 45],
    'salary': [50000, 60000, 75000]
})

# Concatenate columns
df.select(simple_function_to_expr('concat([first_name], " ", [last_name])').alias('full_name'))

# Conditional logic
df.select(simple_function_to_expr('if [age] > 30 then "Senior" else "Junior" endif').alias('level'))

# Math operations
df.select(simple_function_to_expr('[salary] * 1.1').alias('new_salary'))

# Combine multiple operations
df.select(simple_function_to_expr('uppercase(left([last_name], 3))').alias('code'))

Installation

pip install polars-expr-transformer

Why Use This Library?

Use Case	Recommendation
Building applications with user-defined transformations	✅ Yes - Users can write expressions without Python knowledge
SQL/Tableau users transitioning to Polars	✅ Yes - Familiar syntax
Need a simple expression language for configs	✅ Yes - Easy to serialize and store
Writing performance-critical Polars code	❌ No - Use Polars directly
Need all Polars features	❌ No - This covers common operations only

Expression Syntax

Column References

Reference DataFrame columns using square brackets:

'[column_name]'           # Reference a column
'[Column With Spaces]'    # Columns with spaces work too

Values & Data Types

Besides column references, you can write literal values directly. Five literal types are supported:

Type	How to write it	Examples
String	Single or double quotes	`"hello"`, `'world'`
Integer	Bare whole numbers (negatives allowed)	`42`, `-7`
Float	Bare decimal numbers	`3.14`, `-0.5`
Boolean	`true` or `false` (case-insensitive)	`true`, `False`
Null	`null` (case-insensitive) — the missing value	`null`

'if [active] = true then "yes" else "no" endif'   # boolean literal
'[price] * 1.1'                                     # float literal
'coalesce([nickname], null)'                        # null literal

Operators

Operator	Description	Example
`+`	Addition	`[a] + [b]`
`-`	Subtraction	`[a] - 10`
`*`	Multiplication	`[price] * [quantity]`
`/`	Division	`[total] / [count]`
`%`	Modulo	`[value] % 2`
`=` or `==`	Equals	`[status] = "active"`
`!=`	Not equals	`[type] != "deleted"`
`>`, `>=`, `<`, `<=`	Comparisons	`[age] >= 18`
`and`	Logical AND	`[a] > 0 and [b] > 0`
`or`	Logical OR	`[x] = 1 or [y] = 1`
`in ( ... )`	Membership	`[name] in ("alice", "carol")`
`not in ( ... )`	Non-membership	`[status] not in ("failed", "cancelled")`
`in`	Substring, against a plain value	`"li" in [name]`

Membership needs the parentheses: [x] in ("a", "b") tests whether [x] is one of the listed values, while "a" in [x] tests whether [x] contains the text "a". Members must all be of one type, and a member may reference a column. As in Polars, a null input yields null rather than false, and an empty list () matches nothing.

Conditional Expressions

# Simple if-then-else
'if [age] >= 18 then "Adult" else "Minor" endif'

# Multiple conditions with elseif
'if [score] >= 90 then "A" elseif [score] >= 80 then "B" elseif [score] >= 70 then "C" else "F" endif'

# Nested conditions
'if [type] = "A" then (if [value] > 100 then "High A" else "Low A" endif) else "Other" endif'

Comments

# Single-line comments with //
'[column] + 1 // This adds one to the column'

# Multi-line expressions with comments
'''
[price] * [quantity]  // Calculate subtotal
- [discount]          // Apply discount
'''

Available Functions

String Functions

Function	Description	Example
`concat(a, b, ...)`	Concatenate strings	`concat([first], " ", [last])`
`length(text)`	String length	`length([name])`
`uppercase(text)`	Convert to uppercase	`uppercase([code])`
`lowercase(text)`	Convert to lowercase	`lowercase([email])`
`titlecase(text)`	Convert to title case	`titlecase([name])`
`left(text, n)`	First n characters	`left([phone], 3)`
`right(text, n)`	Last n characters	`right([id], 4)`
`mid(text, start, len)`	Substring from position	`mid([code], 2, 3)`
`substring(text, start, len)`	Alias for mid	`substring([text], 0, 10)`
`trim(text)`	Remove leading/trailing spaces	`trim([input])`
`left_trim(text)`	Remove leading spaces	`left_trim([text])`
`right_trim(text)`	Remove trailing spaces	`right_trim([text])`
`replace(text, find, replace)`	Replace text	`replace([name], ".", "")`
`find_position(text, search)`	Find substring position	`find_position([text], "@")`
`pad_left(text, len, char)`	Pad string on left	`pad_left([id], 5, "0")`
`pad_right(text, len, char)`	Pad string on right	`pad_right([code], 10, " ")`
`starts_with(text, prefix)`	Check prefix	`starts_with([url], "https")`
`ends_with(text, suffix)`	Check suffix	`ends_with([file], ".csv")`
`reverse(text)`	Reverse string	`reverse([text])`
`repeat(text, n)`	Repeat string n times	`repeat("*", 5)`
`split(text, delimiter)`	Split into list	`split([tags], ",")`
`count_match(text, pattern)`	Count occurrences	`count_match([text], "a")`
`string_similarity(a, b, method)`	Similarity score (0-1)	`string_similarity([a], [b], "levenshtein")`

String functions read a non-text column as text rather than failing on it, so contains([order_date], "2026-09") works on a Date column. A Date reads as 2026-09-24, a Datetime as 2026-09-24 10:00:00 (no fractional seconds, so ends_with([ordered_at], "10:00:00") matches), and everything else the way cast(pl.String) renders it. Text columns are untouched. Use format_date for any other date layout.

Math Functions

Function	Description	Example
`abs(n)`	Absolute value	`abs([difference])`
`round(n, decimals)`	Round to decimals	`round([price], 2)`
`ceil(n)`	Round up	`ceil([value])`
`floor(n)`	Round down	`floor([value])`
`power(base, exp)`	Exponentiation	`power([x], 2)`
`pow(base, exp)`	Alias for power	`pow(2, [n])`
`sqrt(n)`	Square root	`sqrt([area])`
`log(n)`	Natural logarithm	`log([value])`
`log10(n)`	Base-10 logarithm	`log10([value])`
`log2(n)`	Base-2 logarithm	`log2([value])`
`exp(n)`	e^n	`exp([rate])`
`mod(a, b)`	Modulo	`mod([value], 10)`
`sign(n)`	Sign (-1, 0, 1)	`sign([change])`
`negation(n)`	Negate value	`negation([amount])`
`sin(n)`, `cos(n)`, `tan(n)`	Trigonometric	`sin([angle])`
`asin(n)`, `acos(n)`, `atan(n)`	Inverse trig	`asin([ratio])`
`tanh(n)`	Hyperbolic tangent	`tanh([x])`
`random_int(min, max)`	Random integer	`random_int(1, 100)`

Date Functions

Function	Description	Example
`now()`	Current datetime	`now()`
`today()`	Current date	`today()`
`year(date)`	Extract year	`year([created_at])`
`month(date)`	Extract month (1-12)	`month([date])`
`day(date)`	Extract day (1-31)	`day([date])`
`hour(datetime)`	Extract hour (0-23)	`hour([timestamp])`
`minute(datetime)`	Extract minute	`minute([time])`
`second(datetime)`	Extract second	`second([time])`
`week(date)`	ISO week number (1-53)	`week([date])`
`weekday(date)`	Day of week (1=Mon, 7=Sun)	`weekday([date])`
`dayofweek(date)`	Alias for weekday	`dayofweek([date])`
`quarter(date)`	Quarter (1-4)	`quarter([date])`
`dayofyear(date)`	Day of year (1-366)	`dayofyear([date])`
`add_days(date, n)`	Add days	`add_days([start], 30)`
`add_weeks(date, n)`	Add weeks	`add_weeks([date], 2)`
`add_months(date, n)`	Add months	`add_months([date], 6)`
`add_years(date, n)`	Add years	`add_years([birth], 18)`
`add_hours(dt, n)`	Add hours	`add_hours([time], 3)`
`add_minutes(dt, n)`	Add minutes	`add_minutes([time], 30)`
`add_seconds(dt, n)`	Add seconds	`add_seconds([time], 60)`
`date_diff_days(a, b)`	Days between dates	`date_diff_days([end], [start])`
`datetime_diff_seconds(a, b)`	Seconds between	`datetime_diff_seconds([a], [b])`
`format_date(date, fmt)`	Format as string	`format_date([date], "%Y-%m-%d")`
`start_of_month(date)`	First of month	`start_of_month([date])`
`end_of_month(date)`	Last of month	`end_of_month([date])`
`date_truncate(date, unit)`	Truncate to unit	`date_truncate([dt], "1day")`

Logic & Null Handling

Function	Description	Example
`equals(a, b)`	Check equality	`equals([status], "active")`
`does_not_equal(a, b)`	Check inequality	`does_not_equal([type], "deleted")`
`is_empty(value)`	Check if null	`is_empty([email])`
`is_not_empty(value)`	Check if not null	`is_not_empty([phone])`
`coalesce(a, b, ...)`	First non-null	`coalesce([nickname], [name], "Unknown")`
`ifnull(value, default)`	Replace null	`ifnull([count], 0)`
`nvl(value, default)`	Alias for ifnull	`nvl([value], 0)`
`nullif(a, b)`	Null if equal	`nullif([value], 0)`
`between(val, min, max)`	Range check (inclusive)	`between([age], 18, 65)`
`greatest(a, b, ...)`	Maximum value	`greatest([a], [b], [c])`
`least(a, b, ...)`	Minimum value	`least([price1], [price2])`
`contains(text, search)`	Contains substring	`contains([desc], "sale")`
`_in(value, text)`	Value in text	`_in("admin", [roles])`
`_not(value)`	Logical NOT	`_not([is_deleted])`
`is_string(value)`	Type check	`is_string([field])`

Type Conversions

Function	Description	Example
`to_string(value)`	Convert to string	`to_string([id])`
`to_integer(value)`	Convert to integer	`to_integer([count])`
`to_float(value)`	Convert to float	`to_float([price])`
`to_number(value)`	Alias for to_float	`to_number([value])`
`to_boolean(value)`	Convert to boolean	`to_boolean([flag])`
`to_date(text, format)`	Parse date	`to_date([date_str], "%Y-%m-%d")`
`to_datetime(text, format)`	Parse datetime	`to_datetime([ts], "%Y-%m-%d %H:%M:%S")`
`to_decimal(value, precision)`	Convert with precision	`to_decimal([amount], 2)`

Hashing

One-way fingerprints: the original value cannot be recovered from the result.

Function	Description	Example
`hash(value)`	Fast numeric fingerprint	`hash([email])`
`md5(text)`	MD5 hash (32 hex characters)	`md5([email])`
`sha1(text)`	SHA-1 hash (40 hex characters)	`sha1([email])`
`sha256(text)`	SHA-256 hash (64 hex characters)	`sha256([email])`
`sha512(text)`	SHA-512 hash (128 hex characters)	`sha512([email])`

hash() is a fast, non-cryptographic hash whose numbers may change between Polars versions; use sha256() when the fingerprint has to stay the same over time. The cryptographic digests run through Python's hashlib, so generated code for them needs import hashlib alongside import polars as pl.

Encoding

Reversible text encodings: decode turns whatever encode produced back into the original text.

Function	Description	Example
`encode(text, encoding)`	Encode as base64 or hex	`encode([name], "hex")`
`decode(text, encoding)`	Decode from base64 or hex	`decode([token], "base64")`
`base64_encode(text)`	Encode as base64	`base64_encode([name])`
`base64_decode(text)`	Decode from base64	`base64_decode([token])`
`hex_encode(text)`	Encode as hexadecimal	`hex_encode([name])`
`hex_decode(text)`	Decode from hexadecimal	`hex_decode([token])`

API Reference

`simple_function_to_expr(expression: str) -> pl.Expr`

Converts a string expression to a Polars expression.

from polars_expr_transformer import simple_function_to_expr

expr = simple_function_to_expr('[price] * [quantity]')
df.select(expr.alias('total'))

`build_func(expression: str) -> Func`

Returns the intermediate function object for inspection/debugging.

from polars_expr_transformer import build_func

func = build_func('concat([a], [b])')
print(func.get_readable_pl_function())  # See the Polars translation

`get_all_expressions() -> List[str]`

Returns a list of all available function names.

from polars_expr_transformer import get_all_expressions

functions = get_all_expressions()
print(functions)  # ['concat', 'length', 'uppercase', ...]

`get_expression_overview() -> List[ExpressionsOverview]`

Returns functions grouped by category with descriptions.

from polars_expr_transformer import get_expression_overview

for category in get_expression_overview():
    print(f"\n{category.category}:")
    for expr in category.expressions:
        print(f"  {expr.name}: {expr.description}")

Error Handling

The library validates expressions before parsing and raises ExpressionSyntaxError (a subclass of ValueError) with the exact position of the problem and a hint:

# Misspelled keyword
simple_function_to_expr('f [age] > 30 then "Senior" else "Junior" endif')
# ExpressionSyntaxError:
# Found 'then' at position 14, but there is no 'if' before it.
# f [age] > 30 then "Senior" else "Junior" endif
#              ^
# Hint: Every condition starts with 'if': if <condition> then <value> else <value> endif.
# Check that 'if' is present and spelled correctly.

# Unbalanced parentheses
simple_function_to_expr('((1)')
# ExpressionSyntaxError:
# Unbalanced parentheses: '(' at position 1 is never closed.
# ((1)
# ^
# Hint: Add a matching ')'.

# Unknown function
simple_function_to_expr('unknown_func([col])')
# ExpressionSyntaxError: Expected a single value, but found 2. This usually means
# a function name is misspelled or unknown, or an operator is missing between two values.

Catch errors with except ExpressionSyntaxError (importable from the package root) or simply except ValueError.

Built on Polars

This library is built on top of Polars, a blazingly fast DataFrame library written in Rust. All expressions are converted to native Polars operations, ensuring optimal performance.

Contributing

Contributions are welcome! Please feel free to submit issues and pull requests on GitHub.

License

MIT License - see LICENSE file for details.

Acknowledgements

Thanks to the Polars team for creating such an amazing library.

Metadata

Release files for polars-expr-transformer 0.6.4

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for polars-expr-transformer 0.6.4
File	Size	Uploaded
polars_expr_transformer-0.6.4.tar.gz	62.8 kB	Oct 1, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for polars-expr-transformer 0.6.4
File	Interpreter	ABI	Platform	Reset
polars_expr_transformer-0.6.4-py3-none-any.whl 70.6 kB Oct 1, 2026	Python 3	none	any	Details

Total release size: 133.4 kB

Release files / polars_expr_transformer-0.6.4.tar.gz

Download URL	polars_expr_transformer-0.6.4.tar.gz
Size	62.8 kB
Tags	Source
SHA-256 checksum How to use checksums	`c71809876a14e535ef0c1d54ed9dc173b5e5cab6ad00cdbc4a8e55fae00feeb5`
BLAKE2b-256 checksum How to use checksums	`6dc62bb9c7d4fb058e1d3bba2656acc76d8a82290a6458feb2474045cbb61302`
Upload date	Oct 1, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/Edwardvaneechoud/polars_expr_transformer
Publishing commit: github.com/Edwardvaneechoud/polars_expr_transformer/tree/18bc46bf8e89995ae271f06c700d1790bff7774a

Workflow

Publishing configuration: github.com/Edwardvaneechoud/polars_expr_transformer/blob/18bc46bf8e89995ae271f06c700d1790bff7774a/.github/workflows/ci-cd.yml
Publishing logs: github.com/Edwardvaneechoud/polars_expr_transformer/actions/runs/36838563881/attempts/2

Artifact

Subject: polars_expr_transformer-0.6.4.tar.gz
SHA-256 checksum: c71809876a14e535ef0c1d54ed9dc173b5e5cab6ad00cdbc4a8e55fae00feeb5 Verifying attestations

Release files / polars_expr_transformer-0.6.4-py3-none-any.whl

Download URL	polars_expr_transformer-0.6.4-py3-none-any.whl
Size	70.6 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`bad573a400733a6a6b8205f4a882df9a538a9730f7db8327c73f5b4e3701bfb9`
BLAKE2b-256 checksum How to use checksums	`c5c11b17f9a6ec0236c745f40c5faf3825474324988fee4aa385d66793186ac0`
Upload date	Oct 1, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/Edwardvaneechoud/polars_expr_transformer
Publishing commit: github.com/Edwardvaneechoud/polars_expr_transformer/tree/18bc46bf8e89995ae271f06c700d1790bff7774a

Workflow

Publishing configuration: github.com/Edwardvaneechoud/polars_expr_transformer/blob/18bc46bf8e89995ae271f06c700d1790bff7774a/.github/workflows/ci-cd.yml
Publishing logs: github.com/Edwardvaneechoud/polars_expr_transformer/actions/runs/36838563881/attempts/2

Artifact

Subject: polars_expr_transformer-0.6.4-py3-none-any.whl
SHA-256 checksum: bad573a400733a6a6b8205f4a882df9a538a9730f7db8327c73f5b4e3701bfb9 Verifying attestations