jmpio

A Python package for reading and writing SAS JMP files.

Description

jmpio is a mostly-faithful Python port of the Julia package JMPReader.jl by @jaakkor2. It aims to be able to read and write binary JMP files from SAS JMP statistical software. There's no way this package would exist without @jaakor2's efforts.

Warning

jmpio write support is still experimental. jmpio is not particularly efficient and uses pandas DataFrames internally.

Features

• Read JMP files into pandas DataFrames
• Write pandas DataFrames to JMP files (experimental)
  • Optional JSL workflow using a local JMP installation
  • Embed Scripts in those files for plotting data.
• Support for all JMP data types:
  • Numeric (Float64, Int8, Int16, Int32)
  • Character (fixed width and variable width strings)
  • Date/Time (Date, DateTime, Time, Duration)
  • Geographic (Latitude/Longitude)
  • Row states (markers and colors)
  • Currency
• Support for compressed and uncompressed JMP files
• Column selection/filtering when reading
• Strong typing throughout the codebase

Installation

pip install jmpio

Basic Usage

Reading a JMP file

import jmpio
import pandas as pd

# Read a JMP file into a pandas DataFrame
df = jmpio.read_jmp("path/to/file.jmp")

# Select specific columns
df = jmpio.read_jmp("path/to/file.jmp", select=["Column1", "Column2"])

# Drop specific columns
df = jmpio.read_jmp("path/to/file.jmp", drop=["Column1", "Column2"])

# Find all JMP files in a directory
file_info = jmpio.scan_directory("path/to/directory")

Writing a JMP file

import jmpio
import pandas as pd
from datetime import datetime, date

# Create a DataFrame
df = pd.DataFrame({
    'integers': [1, 2, 3, 4],
    'floats': [1.1, 2.2, 3.3, 4.4],
    'strings': ['a', 'bb', 'ccc', 'dddd'],
    'dates': [date(2023, 1, 1), date(2023, 2, 1), date(2023, 3, 1), date(2023, 4, 1)]
})

# Write to a JMP file with compression using the native Python writer (default)
jmpio.write_jmp(df, "output.jmp")

# Write to a JMP file without compression
jmpio.write_jmp(df, "output_uncompressed.jmp", compress=False)

# Use a local JMP installation through generated JSL scripts
jmpio.write_jmp(df, "output_native.jmp", engine="jsl")

# Write with the Python writer, validate with JMP when available, and fall back
# to JSL if JMP cannot open the Python-written file
jmpio.write_jmp(df, "output_auto.jmp", engine="auto")

The JSL workflow is optional and requires SAS JMP to be installed locally. By default, jmpio searches common Windows install paths and JMPIO_JMP_EXE; you can also pass jmp_executable="C:/Program Files/SAS/JMP/17/jmp.exe".

Applying JMP column properties

When JMP is installed locally, jmpio can use JSL automation to apply or dump column properties that are not written by the native Python binary writer:

import jmpio

jmpio.apply_column_properties_with_jsl(
    "output.jmp",
    [
        {
            "column": "Protein Concentration (mg/mL)",
            "lsl": 90,
            "usl": 110,
            "show_limits": 0,
        },
        {
            "column": "Aggregate (%)",
            "control_limits": "{XBar(Avg(0.4), LCL(0.35), UCL(0.55))}",
        },
        {"column": "Total Purity (%)", "sigma": "0.007"},
        {
            "column": "PC^2",
            "formula": ':"Protein Concentration (mg/mL)"n * :"Protein Concentration (mg/mL)"n',
        },
    ],
)

properties = jmpio.dump_column_properties_with_jsl("output.jmp")

Supported property fields are lsl, usl, target, show_limits, control_limits, sigma, and formula.

Supported Data Types

Reading

The following data types are supported when reading JMP files:

JMP Data Type	Python Data Type
Numeric	float64, int8, int16, int32
Character	str (fixed or variable width)
Date	datetime64[D] / datetime.date
Time	datetime64[s] / datetime.time
DateTime	datetime64[s] / datetime.datetime
Duration	timedelta64[ms] / datetime.timedelta
Row State	jmpio.RowState
Geographic	float64
Currency	float64

Writing

The following data types are supported when writing to JMP files:

Python Data Type	JMP Data Type
int, Int8, Int16, Int32	Integer
float, Float64	Numeric
str	Character
datetime64, datetime.datetime	DateTime
datetime64[D], datetime.date	Date
datetime.time	Time
timedelta64, datetime.timedelta	Duration
jmpio.RowState	Row State

Requirements

• Python 3.10+
• NumPy
• pandas

License

MIT

Acknowledgments

This package is a Python port of the Julia package JMPReader.jl. JMP is a registered trademark of SAS Institute Inc. This project is not affiliated with, sponsored by, or endorsed by SAS Institute Inc.