awk-csv

This is an Awk library for reading and writing CSV (Comma Separated Value) data. Quoted fields in CSV may contain commas and newlines, so naively using split() to parse CSV is not always going to work.

To use this library, you can add "-f /usr/share/awk/csv.awk" to your awk command line before your own script, for example: 


	awk -f /usr/share/awk/csv.awk -f script-using-csv.awk

Input Functions

csv_read(record)
Read CSV from current input file (sets FNR and NR)
csv_read_command(record, command)
Read CSV from output of shell command
csv_read_file(record, filename)
Read CSV from named file

All input functions return 0 on end of file, -1 on error, or the number of fields. Fields are placed in numbered fields of record, starting from 1: record[1], record[2], ... record[n]. There is always at least one field per input record - an empty input record yields a single empty field. For csv_read_command or csv_read_file, on end of file, it is necessary to call close(). CRLF sequences within fields are converted to plain LF.

Convenience Function

csv_header(record)
Reverse the indices and values of an array

Traditionally, CSV files will have a first record indicating the names of the fields for each record. This is simply a convenience function to use on that first record, so that you can look up the number of a field by its name. It will convert an array such as: 


	header[1] = "foo"
	header[2] = "bar"
	header[3] = "baz"
	header[4] = "quux"

into:


	header["foo"] = 1
	header["bar"] = 2
	header["baz"] = 3
	header["quux"] = 4

Output Functions

csv_write(record)
Write CSV to standard output
csv_write_command(record, command)
Write CSV to input of shell command
csv_write_file(record, filename)
Write CSV to named file
csv_append_file(record, filename)
Append CSV to named file

Output functions take a record array with numbered fields, starting with 1: record[1], record[2], ... If the record is missing a field n, all fields numbered greater than n will be ignored. If the record has no field 1, a record with a single empty field will be output. It is necessary to call close() when output is finished.

The CSV format understood by this library

Each record is separated by either CRLF or LF alone (the CR is dropped). Within each record, fields are separated by commas. Fields (or even parts of fields) may be quoted with double quotes, which is necessary when fields contain commas, newlines, or double quotes. A double quote within a quoted field may be escaped by preceding it with another double quote. This is the same as RFC 4180, except that the RFC requires CRLF record separators and does not allow LF alone.

This code relies on no more than the POSIX awk specification. It has been successfully tested with: