Package 'jmpread'

Title: Read 'JMP' Binary Data Tables in Native R
Description: A reader for 'JMP' binary data tables, implemented entirely in R. Supports common numeric, character, temporal and row-state column encodings, including gzip compression. Returns data frames with source metadata and reports unsupported or malformed encodings. Based on binary format work in 'jumpio' <https://github.com/abardagjy/jumpio> and 'JMPReader.jl' <https://github.com/jaakkor2/JMPReader.jl>. No 'JMP' installation or 'Python' runtime is required.
Authors: Joseph Hill [aut, cre], Andy Bardagjy [ctb, cph] (jumpio algorithms adapted under the MIT license), Jaakko Ruohio [ctb, cph] (JMPReader.jl format work and MIT-licensed fixtures)
Maintainer: Joseph Hill <[email protected]>
License: MIT + file LICENSE
Version: 0.1.0
Built: 2026-11-19 03:51:11 UTC
Source: https://github.com/cran/jmpread

Help Index


Read a JMP Binary Data Table

Description

Read JMP binary files using native R. Handles common numeric, fixed and variable character, pooled character, compressed, temporal and row-state storage. Unsupported encodings raise errors.

Usage

read_jmp(path, select = NULL, exclude = NULL, encoding = NULL,
  temporal = c("convert", "raw"), metadata = TRUE,
  max_bytes = 256 * 1024^2)

Arguments

path

One path to a local JMP file.

select, exclude

Character vectors of column names to retain or omit, respectively. Mutually exclusive. Source column order is preserved.

encoding

A character encoding override, or NULL to use the header.

temporal

Convert recognized date, datetime and time display formats to UTC POSIXct, and durations to difftime, or return raw JMP seconds.

metadata

Attach file and column metadata attributes.

max_bytes

Maximum input file size in bytes. Defaults to 256 MiB.

Details

Numeric missing sentinels become NA. Empty character strings remain empty strings, explicit pooled missing sentinels become NA. Date-display columns retain their time of day, using POSIXct rather than Date. JMP's epoch is 1904-01-01. Unknown numeric display formats remain numeric, their type bytes are preserved in metadata.

Row-state columns are list columns of raw byte vectors. Palette meanings, hidden/excluded row flags and JSL properties are not interpreted. Column metadata retains the raw post-header block, including its data payload, this may substantially increase memory use. Formulas are never executed. Use metadata = FALSE to avoid retaining raw blocks.

The full input is read into memory. The option jmpread.max_inflated_bytes limits each declared inflated column size (default 256 MiB). It is not a streaming decompression memory limit; this reader is not hardened against adversarial compressed files. Embedded NULs in text are rejected because R strings cannot represent them.

Value

A data.frame preserving original column names. With metadata = TRUE, attr(result, "jmp_metadata") describes the file and attr(result[[i]], "jmp_metadata") describes each column.

Examples

path <- system.file("extdata", "jmpreader_jl", "example1", "example1.jmp",
                    package = "jmpread")
table <- read_jmp(path)
table[, c("ints", "floats")]
attr(table[["date"]], "jmp_metadata")$storage
read_jmp(path, select = c("ints", "char utf8"), metadata = FALSE)