Skip to content

API Reference

This document details the classes and functions exposed by the PardoX Python SDK.

Top-Level Functions

read_csv

Reads a Comma Separated Value (CSV) file into a DataFrame using the multi-threaded Rust engine.

def read_csv(path: str, has_headers: bool = True) -> DataFrame

Parameters:

Parameter Type Description
path str Path to the .csv file.
has_headers bool Whether the first row contains column names. Default: True.

Returns: DataFrame

read_sql

Executes a SQL query against a Postgres database and loads the result directly.

def read_sql(connection_string: str, query: str) -> DataFrame

Parameters:

Parameter Type Description
connection_string str URI format: postgres://user:pass@host:port/db
query str The SQL SELECT statement to execute.

Returns: DataFrame

read_prdx

Loads a native PardoX binary file (.prdx).

def read_prdx(path: str) -> DataFrame

Parameters:

Parameter Type Description
path str Path to the .prdx file.

Returns: DataFrame

from_arrow

Zero-Copy conversion from a PyArrow Table.

def from_arrow(table: pyarrow.Table) -> DataFrame

Parameters:

Parameter Type Description
table pyarrow.Table PyArrow Table to convert.

Returns: DataFrame

Class: DataFrame

The main data structure holding the HyperBlock memory manager.

Properties

  • shape: Returns a tuple (rows, cols) representing the dimensions.
  • columns: Returns a list of column names.

Methods

head(n=5)

Returns the first n rows as a list of dictionaries. Useful for inspection.

df.head(10)  # Show first 10 rows

tail(n=5)

Returns the last n rows as a list of dictionaries.

df.tail(10)  # Show last 10 rows

to_prdx(path)

Saves the current DataFrame state to a binary file.

df.to_prdx("output.prdx")

Parameters:

Parameter Type Description
path str Path to save the .prdx file.

fillna(value)

Fills NaN / null values in all compatible columns with the given scalar.

df.fillna(0.0)  # Replace all nulls with 0

Current Limitation

Currently supports filling numeric columns with float values.

Parameters:

Parameter Type Description
value float The value to replace nulls with.

Class: Series

Represents a single column within a DataFrame. Returned when selecting a column (e.g., df['price']).

Arithmetic

Supported operators: +, -, *, /

Operations are vectorized (SIMD) and return a new Series or modify in-place if assigned back.

# Vector arithmetic
df['total'] = df['price'] * df['quantity']
df['tax'] = df['total'] * 0.16

Aggregations

sum()

Returns the sum of all values.

total = df['amount'].sum()

Returns: float

mean()

Returns the arithmetic average.

average = df['amount'].mean()

Returns: float

min()

Returns the minimum value.

lowest = df['amount'].min()

Returns: float

max()

Returns the maximum value.

highest = df['amount'].max()

Returns: float

std()

Returns the standard deviation (population).

volatility = df['amount'].std()

Returns: float

count()

Returns the count of non-null values.

valid_count = df['amount'].count()

Returns: int

Transformations

round(decimals)

Rounds values to the specified number of decimal places in-place.

df['price'].round(2)  # Round to 2 decimals

Parameters:

Parameter Type Description
decimals int Number of decimal places.