···<date: 2012-11-22, Thursday>···<from: Hans Hagen>···

> On 11/22/2012 3:36 PM, Marcin Borkowski wrote:
> >Dnia 2012-11-22, o godz. 13:26:52
> >Mojca Miklavec <mojca.miklavec.lists@gmail.com> napisał(a):
> >
> >>On Thu, Nov 22, 2012 at 12:57 PM, Marcin Borkowski wrote:
> >>>le) in order to
> >>>typeset a (nice) TABLE. However, I have a few columns I'd like to
> >>>omit.  I can (of course) hand-edit the csv file; but is there a way
> >>>to do it automatically?  Something like
> >>>
> >>>\setupTABLE[column][3,4,5][kill]
> >>>
> >>>I did [empty=yes,width=0pt] in place of [kill], and it worked, but
> >>>it seems like a hack (and I don't know whether it does actually
> >>>process the cells I'm omitting - which in my case wouldn't bother
> >>>me, but might be an additional layer of inelegance;)).  And better
> >>>ideas?
> >>
> >>If you have up to 9 columns, you could use
> >>
> >>     \def\ProcessingLine#1#2#3#4#5#6#7{%
> >>         \bTR\bTD#1\eTD\bTD#2\eTD\bTD#6\eTD\bTD#7\eTD}
> >>
> >>and then [command=\ProcessingLine]
> >
> >Well, something like 20 columns (on A4 landscape). ;)
> >
> >It turns out that my method somehow doesn't work well without setting
> >also height=0pt; then it's fine, but I'm still wondering about a
> >cleaner way.
> 
> I've added a splutter to the core:
> 
> \startluacode
> 
> local mycsvsplitter = utilities.parsers.csvsplitter {
>     separator = ",",
>     quote     = '"',
> }
> 
> local crap = [[
> "1","2","3","4"
> "a","b","c","d"
> ]]
> 
> local mycrap = mycsvsplitter(crap)
> 
> context.bTABLE()
>     for i=1,#mycrap do
>         context.bTR()
>         local c = mycrap[i]
>         for i=1,#c do
>             context.bTD()
>             context(c[i])
>             context.eTD()
>         end
>         context.eTR()
>     end
> context.eTABLE()
> 
> \stopluacode

Hi Hans,

although csv is not a standard per se there is nevertheless an
rfc: <http://tools.ietf.org/html/rfc4180>. Can we have an
optional rfc-compliant parser as well? That entails interpreting
the first line as field header if it consists entirely of
unquoted fields -- neat feature as one can treat these as column
identifiers to query fields in a more natural fashion.

·································································

\starttext
\startluacode

local P, R, V, S, C, Ct, Cs, Carg, Cc, Cg, Cf = lpeg.P, lpeg.R, lpeg.V, lpeg.S, lpeg.C, lpeg.Ct, lpeg.Cs, lpeg.Carg, lpeg.Cc, lpeg.Cg, lpeg.Cf

local lpegmatch = lpeg.match
local patterns = lpeg.patterns
local newline  = patterns.newline

-----------------------------------------------------------------------
--                        RFC 4180 csv parser
-----------------------------------------------------------------------

-- based on http://tools.ietf.org/html/rfc4180
-- notable deviations from the RFC grammar:
--   · allow overriding “comma” and “quote” chars in spec
--   · the “textdata” rule is extrapolated to mean, basically,
--     “everything but the quote character”, not just the character
--     range as in the RFC
--   · newline from l-lpeg.lua instead of “crlf”

local rfc4180_spec = { separator = ",", quote     = [["]] }

utilities.parsers.rfc4180_splitter = function (specification)
    specification     = specification or rfc4180_spec
    local separator   = specification.separator --> rfc: COMMA
    local quotechar   = P(specification.quote)  -->      DQUOTE
    local dquotechar  = quotechar * quotechar   -->      2DQUOTE
                      / specification.quote

    local separator   = S(separator ~= "" and separator or ",")
    local escaped     = quotechar
                      * Cs((dquotechar + (1 - quotechar))^0)
                      * quotechar
    local non_escaped = C((1 - quotechar - newline - separator)^1)
    local field       = escaped + non_escaped
    local record      = Ct(field * (separator * field)^0)
    local name        = field -- sic!
    local header      = Cg(Ct(name * (separator * name)^0), "header")
    local file        = Ct((header * newline)^-1 -- optional header
                         * record
                         * (newline * record)^0
                         * newline^0)
    return function (data) return lpegmatch (file, data) end
end

-----------------------------------------------------------------------
--                  example writer (natural table)
-----------------------------------------------------------------------

local print_csv_table = function (tab)
  local header = tab.header
  context.setupTABLE({ frame = "off" })
  context.setupTABLE({"header"}, { background  = "screen" })
  context.bTABLE()
    if header then
      context.bTABLEhead()
      context.bTR()
      for i=1,#header do
        context.bTH()
        context(header[i])
        context.eTH()
      end
      context.eTR()
      context.eTABLEhead()
    end
    context.bTABLEbody()
    for i=1,#tab do
      context.bTR()
      local c = tab[i]
      for i=1,#c do
        context.bTD()
        context(c[i])
        context.eTD()
      end
      context.eTR()
    end
    context.eTABLEbody()
  context.eTABLE()
end

-----------------------------------------------------------------------
--                               usage
-----------------------------------------------------------------------

local mycsvsplitter = utilities.parsers.rfc4180_splitter ()

local crap = [[
first,second,third,fourth
"1","2","3","4"
"a","b","c","d"
"foo","bar""baz","boogie","xyzzy"
]]

local mycrap = mycsvsplitter(crap)

print_csv_table (mycrap)

-----------------------------------------------------------------------

\stopluacode
\stoptext