Julia Csv

Julia Csv

Use CSV.jl for high-performance reading/writing of delimited data with explicit control over parsing behavior.

Choose the Read Mode

• CSV.File(input; kw...): full-column parse, table-like object, best default.
• CSV.read(input, sink; kw...): parse and hand columns directly to sink (e.g. DataFrame) without intermediate copies.
• CSV.Rows(input; kw...): row-streaming with lower memory footprint; supports reusebuffer=true.
• CSV.Chunks(input; ntasks=...): chunked parsing for very large files.

using CSV, DataFrames

f = CSV.File("data.csv")
df = CSV.read("data.csv", DataFrame)

Parse Schema and Layout Explicitly

Common schema/layout controls:

• header, normalizenames, skipto, footerskip
• select / drop, limit
• types, typemap, stringtype, pool, downcast
• strict, silencewarnings, maxwarnings

Delimiter and cell interpretation:

• delim, ignorerepeated (fixed-width style data)
• missingstring
• quoted, quotechar / openquotechar / closequotechar, escapechar
• dateformat, decimal, groupmark, truestrings, falsestrings

Threading and chunking:

• ntasks, rows_to_check

Support Multiple Input Forms

CSV.jl supports:

• filename/filepath
• IO / Cmd
• Vector{UInt8}
• vector of inputs for vertical concatenation (matching schema)

Gzip (.gz) input is handled automatically; use buffer_in_memory=true if needed.

Write CSV Output

CSV.write("out.csv", table)
rows = CSV.RowWriter(table)

Useful write controls:

• delim, quotechar, openquotechar, closequotechar, escapechar
• missingstring, dateformat, decimal
• append, header, newline, quotestrings
• transform, bom, compress, partition, bufsize

Reference

• references/csv-patterns.md - read/write mode selection and option patterns

Related Skills

• julia-prettytables - rendering parsed tabular data

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Krastanov
• Source: Krastanov/JuliaLLMAgentSkills
• License: Unlicense

Install and usage instructions live in the source repository linked above.