#!/usr/local/bin/php -q
<?php
/*
 * this is just a quick script made in a few minutes to help a friend to 
 * strip quotes ('>' characters) off mails - or any text files -, or to 
 * unwrap text inside paragraphs, which is useful if the text has been 
 * wrapped for lines of maximum length 80. there's a switch to output crude 
 * html, too. requires php binary (php-cgi), change the path on the first line.
 *
 * Usage: ./striptext.phpsh <filename> [--dequote|--dewrap|--html] 
 *          (writes output to <filename>.new)
 *    OR: ./striptext.phpsh [--dequote|--dewrap|--html] < filename
 *          (reads input from standard input and writes to standard output)
 *
 * v1.0 - Mon Sep 22 19:21:52 EEST 2003 blitzer@meizo.com
 *
 */

  $argv = $_SERVER ["argv"];

  if (in_array ("--html", $argv))
    $html = 1;

  if (in_array ("--dequote", $argv))
    $dequote = 1;

  if (in_array ("--dewrap", $argv))
    $dewrap = 1;

  if (sizeof ($argv) < 2 || (!$html && !$dequote && !$dewrap))
    {
      echo ("Usage: " . $argv [0] . 
            " <filename> [--dequote|--dewrap|--html]\n" .
            "         (writes output to <filename>.new)\n" .
            "   OR: " . $argv [0] . 
            " [--dequote|--dewrap|--html]\n" .
            "         (reads input from standard input and writes to standard output)\n");
      exit;
    }
  
  if (!ereg ("^--", $argv [1]))
    $f = @fopen ($argv [1], "r");
  else
    $f = fopen ("php://stdin", "r");

  if (!$f)
    exit;

  $data = fread ($f, 1000000);
  fclose ($f);

  $data = str_replace ("\r\n", "\n", $data);

  function is_unwrapped ($a)
    { 
      foreach ($a as $line)
        {
          if (strlen ($line) >= 55)
            return 0;
        }
 
      return 1;
    }

  if ($dequote)
    {
      $arr = explode ("\n", $data);

      for ($i = 0; $i < sizeof ($arr); $i++)
        {
          while (ereg ("^[ \t]*>[ \t]*(.*)$", $arr [$i], $reg_arr))
            $arr [$i] = $reg_arr [1];
        }

      $data = implode ("\n", $arr);
    }

  $paragraphs = preg_split ("#\n[ \t]*\n#", $data);

  $ret = "";

  foreach ($paragraphs as $s) 
    { 
      if (strlen ($ret))
        $ret .= "\n\n";

      $a = explode ("\n", $s);

      $unwrapped = is_unwrapped ($a);

      for ($i = 0; $i < sizeof ($a); $i++)
        {
          $line = $a [$i];

          if ($html)
            $a [$i] = htmlspecialchars ($a [$i]);

          if ($dewrap)
            {
              if ($unwrapped ||
                  ereg ("^[^ :]+: ", $line) || 
                  ereg ("^------", $line) ||
                  ereg ("^From .*@.* [0-9][0-9]:[0-9][0-9]:[0-9][0-9]",
                        $line))
                {
                  if ($html) 
                    $a [$i] .= "<br>";

                  $a [$i] .= "\n";
                }

              else
                $a [$i] .= " ";
           }
        }

      if ($dewrap)
        {
          $s = implode ("", $a);
          $s = str_replace ("\t", " ", $s);

          while (strstr ($s, "  "))
            $s = str_replace ("  ", " ", $s);
        }
      else
        $s = implode ("\n", $a);

      if ($html)
        $ret .= "<p>";

      $ret .= $s;

      if ($html)
        $ret .= "\n</p>\n";
    }

  if (!ereg ("^--", $argv [1]))
    {
      $f = fopen ($argv [1] . ".new", "w");
      fwrite ($f, $ret);
      fclose ($f);
      echo ("Stripped file '" . $argv [1] . "' into '" . $argv [1] . ".new'\n");
    }
  else
    echo ($ret);

  exit;
?>