#!/bin/bash

# Some basic CSV processing functions for shell scripts
#
# Stephen Thomas <flabdablet@gmail.com> 12-Jun-2009
#
# This is free software. You can do whatever you want with it
# except hold me accountable for any grief it causes you.



# Regex components

q='"'           # quote
w=$' \t'        # whitespace
nq="^$q"	# not quote
nwc="^$w,"      # not whitespace or comma
nwqc="^$w$q,"   # not whitespace, quote or comma

uqf="[$nwqc]([$w]*[$nwc])*"     # unquoted field
qf="($q[$nq]*$q)*"              # quoted field
iqf="($q[$nq]*$q)*$q[$nq]*"     # incomplete quoted field

csv="[$w]*($qf|$uqf)[$w]*,[$w]*(.*)"    # comma separated values



# Read one CSV record from standard input and make its values available
# in the fields[] array. Set corresponding entries in the quoted[] array
# to 1 if the value was quoted in the CSV record, 0 if it was bare.

function csvread () {
	unset fields
	unset quoted
	nf=0
	hold=
	while
		IFS= read -r line || return
		while [[ "$hold$line" =~ ^$csv$ ]]
		do
			fields[nf++]="${BASH_REMATCH[1]}"
			line="${BASH_REMATCH[4]}"
			hold=
		done
		[[ "$hold$line" =~ ^$iqf$ ]]
	do
		hold="$hold$line"$'\n'
	done
	fields[nf++]="$hold$line"

	for (( i=0; i<nf; i++ ))
	do
		if [[ "${fields[i]:0:1}" == "$q" ]]
		then
			quoted[i]=1
			fields[i]="${fields[i]:1:${#fields[i]}-2}"
			fields[i]="${fields[i]//$q$q/$q}"
		else
			quoted[i]=0
		fi
	done
}



# Dump the quoted[] and fields[] arrays along with their indices.
# Might be useful for debugging.

function csvdump () {
	for i in ${!fields[*]}
	do
		printf '%d: %d %q\n' $i ${quoted[i]} "${fields[i]}"
	done
}



# Write the fields[] array to standard output as a CSV record.
# Any field whose corresponding element in quoted[] is set to 1
# will be wrapped in quotes and have internal quotes doubled.

function csvwrite () {
	unset output
	for i in ${!fields[*]}
	do
		if let quoted[i]
		then
			output[i]="$q${fields[i]//$q/$q$q}$q"
		else
			output[i]="${fields[i]}"
		fi
	done
	local IFS=,
	echo "${output[*]}"
}