<?php

# Written by Jeremy Brand.
# Purpose: to parse fixed column width (Gov't) files and produce 
#          output in delimited format suitable for excel or access, etc.
# Input: stdin
# Output: stdout
# 
# example command line usage:
# > php -q fixed_column_split.php < inputfile.txt > outputfile.txt
# 
#
# You must first change the sp_data_definition() function
#                           --------------------
# to match your input data.  You will need the key to translate 
# the data first.
#


define('DELIMITER',",");     # What to split fields by.  Usually a tab ("\t")
                             # or a comma (",").
                             

define('ENCAPSULATOR','"');  # Do you want to wrap each field in something? 
                             # A quote maybe...  

define('EOL',"\n");          # Unix end-of-line is the \n (newline) char.
                             # Other options are \r\n (DOS) \r (MAC)

define('OFFSET',1);          # if the "start byte" is 1, then set this...
                             # else, set it to 0.


# Define the fixed length data here.
#
# There are three fields.  
# -  First field is a NAME (can be anything except a pipe (|).
# -  Second field is the start column.
# -  Third field is the column length.
# Separete by a pipe (|).
function sp_data_definition()
{
  return array(

    'county code|1|3',
    'reg #|4|8',
    'votes status|12|1',
    'name l|13|20',
    'name f|33|20',
    'name m|53|20',
    'suffix|73|3',
    'title|76|3',
    'rs address #|79|6',
    'street name|85|30',
    'rcs. street suffix|115|2',
    'apt #|117|8',
    'city|125|17',
    'zip|142|9',
    'flag mail|151|2',
    'transsomething...|153|1',
    'transea?...|154|6',
    'filler..|160|2',
    'DOB|162|8',
    'reg date|170|8',
    'race|178|1',
    'gender|179|1',
    'filler|180|1',
    'land district|181|3',
    'land lot|184|4',
    'old reg date|188|8',
    'reason status|196|9',
    'filler|205|5',
    'county presinct ID|210|5',
    'city precinct ID|215|5',
    'voter district|220|166',
    'date last voted|386|8',
    'type of election|394|3',
    'party last elected|397|3',

  );
}

##################################################################
##  You should not need to edit anything below here
##################################################################

function sp_read_stdin_line($line_length=4096)
{
  $buf = '';
  $fd = fopen('php://stdin', 'r');
  if ($fd)
  {
    $buf = fgets($fd, $line_length);
    fclose($fd); 
  }
  if (strlen($buf))
    return $buf;
}

function sp_stderr($buf)
{
  $fd = fopen('php://stderr','w');
  if ($fd)
  {
    fwrite($fd, $buf, strlen($buf));
    fclose($fd);
  }
}

set_time_limit(0);

function sp_do_procedure($buf)
{
  static $done_label;

  $object = sp_data_definition();
  reset($object);
  $newline = '';
  $label_line = '';
  $out = '';
  while(list(,$obj_row) = each($object))
  {
    list($label,$start_char,$char_len) = split('\|',$obj_row,3);
    $start_char -= OFFSET;
    $newline .= ENCAPSULATOR.trim(substr($buf,$start_char,$char_len)).ENCAPSULATOR.DELIMITER;
    if (!$done_label)
    {
      $label_line .= ENCAPSULATOR.trim($label).ENCAPSULATOR.DELIMITER;
    }
  }
  if (!$done_label)
  {
    $done_label = TRUE;
    $label_line = substr($label_line, 0, strlen($label_line)-1);
    print $label_line . EOL;
  }
  $newline = substr($newline, 0, strlen($newline)-1);
  print $newline . EOL;

}

$fd = fopen('php://stdin','r');
if ($fd)
{
  while($line = fgets($fd, 10000))
  {
    sp_do_procedure($line);
  }
  fclose($fd);
}
else
{
  sp_stderr("No stdin. Exiting.\n");
}


?>