parse csv file use state machine

csv.lua

-- vi: syntax=lua ts=4 sw=4 et:
--
--         FILE:  csv.lua
--       AUTHOR:  LeeoNix
--  DESCRIPTION:  parse csv file use state machine
--        NOTES:  ---
--

local C_INV   = 1 -- invalid characters
local C_COMMA = 2 -- ,
local C_DQUOT = 3 -- "
local C_LF    = 4 -- \n
local C_CR    = 5 -- \r
local C_ETC   = 6 -- the rest
local C_MAX   = 7

--
-- This array maps the first 96 ASCII characters into character classes
-- The remaining characters should be mapped to C_ETC
--
local ascii_class = {
    C_INV,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,
    C_ETC,   C_ETC,   C_LF,    C_ETC,   C_ETC,   C_CR,    C_ETC,   C_ETC,
    C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,
    C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,

    C_ETC,   C_ETC,   C_DQUOT, C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,
    C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_COMMA, C_ETC,   C_ETC,   C_ETC,
    C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,
    C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,   C_ETC,
};

local S_NLF = 1
local S_NCR = 2
local S_NCL = 3
local S_NST = 4
local S_STR = 5
local S_QST = 6
local S_QT1 = 7
local S_QT2 = 8
local S_MAX = 9
local S_ERR = S_MAX

local state_table = {
--[[             inv     ,      "      \n     \r    etc ]]
--[[ S_NLF ]] { S_ERR, S_NST, S_QT1, S_NLF, S_NCR, S_STR },
--[[ S_NCR ]] { S_ERR, S_NST, S_QT1, S_NCL, S_NCR, S_STR },
--[[ S_NCL ]] { S_ERR, S_NST, S_QT1, S_NLF, S_NCR, S_STR },
--[[ S_NST ]] { S_ERR, S_NST, S_QT1, S_NLF, S_NCR, S_STR },
--[[ S_STR ]] { S_ERR, S_NST, S_ERR, S_NLF, S_NCR, S_STR },
--[[ S_QST ]] { S_ERR, S_QST, S_QT2, S_QST, S_QST, S_QST },
--[[ S_QT1 ]] { S_ERR, S_QST, S_QT2, S_QST, S_QST, S_QST },
--[[ S_QT2 ]] { S_ERR, S_NST, S_QST, S_NLF, S_NCR, S_ERR }
};

local csv = {
    state     = S_NLF,
    cache     = '',
    cache_row = nil,
    result    = {},
}

function csv:clear()
    self.state     = S_NLF
    self.cache     = ''
    self.cache_row = nil
    self.result    = {}
end -- end function

function csv:add(str)
    local row

    local state = self.state
    if state == S_STR or state == S_QST then
        str = self.cache .. str
        row = self.cache_row
    else
        row = {}
    end -- end if

    local first = 1
    local last = 1
    for i = 1, #str do
        local c = str:byte(i)
        local asc
        if c < 64 then
            asc = ascii_class[c + 1]
        else
            asc = C_ETC
        end -- end if else

        state = state_table[state][asc]
        if state == S_STR or state == S_QST then
            last = i
        elseif state == S_NST then
            table.insert(row, str:sub(first, last))
            first = i + 1
        elseif state == S_NLF or state == S_NCR then
            table.insert(row, str:sub(first, last))
            first = i + 1
            table.insert(self.result, row)
            row = {}
        elseif state == S_QT1 then
            first = first + 1
        elseif state == S_ERR then
            return false, "Parse Error"
        end -- end if else
    end -- end for

    if state == S_STR or state == S_QST then
        self.cache = str:sub(first)
        self.cache_row = row
    elseif #row ~= 0 then
        table.insert(self.result, row)
    end -- end if
    self.state = state
    return true
end -- end function

function csv:finish()
    local state = self.state
    if state < S_NLF and state > S_NST then
        return nil, "Finish Error"
    end -- end if
    self.state = S_ERR;
    return self.result
end -- end function

function csv:parse_string(s)
    local last = s:byte(#s)
    if last ~= 10 and last ~= 13 then
        s = s .. '\n'
    end -- end

    local r, err = self:add(s)
    if not r then
        return r, err
    end -- end if

    return self:finish()
end -- end function

function csv:parse_file(name)
    local f = assert(io.open(name, "r"))
    local s = f:read('*a')
    assert(io.close(f))
    return self:parse_string(s)
end -- end function

function csv:dump(t, write, nl)
    t = t or self.result
    write = write or io.write
    nl = nl or '\n'
    for _, row in ipairs(t) do
        write(table.concat(row, ','))
        write(nl)
    end -- end for
end -- end function

local r, err = csv:parse_file(arg[1])
csv:dump()

return csv