private function chunker(&$text, $style = "") { $errors = null; $text = collapse_spaces($text); $textarr = split_string($text); $chunksarr = split_on_spaces($textarr, $this->spaces); if ($style == "clean") { $chunksarr = remove_junk($chunksarr); } $chunkhashes = null; foreach ($chunksarr as $end => &$chunkarr) { $chunkhashes[$end] = count_words($chunkarr); } if (!$chunkhashes) { $errors[] = "Could not hash chunks. Huh."; trigger_error("Could not hash chunks. Huh."); return $errors; } // if cleaned style, remove all bad words and lc $max = array_pop(array_keys($textarr)) + 1; $maxlen = strlen("{$max}"); $pad = "%0{$maxlen}s"; foreach ($chunksarr as $end => $chunkarr) { $endpad = $end + 1; $endpad = sprintf($pad, $endpad); $out = $this->write_txt($chunkarr, $endpad, $style); $out2 = $this->write_csv($chunkhashes[$end], $endpad, $style); if ($out || $out2) { $errors = array_merge($out, $out2, $errors); } } if ($errors) { rrmdir($this->folder); trigger_error("Something in the chunking process went wrong."); } return $errors; }
function character_cleanup($str) { # data validation data_validation($str, "|required|"); return trim(collapse_spaces(eregi_replace("[^[:alnum:]!@#\$%&*()-_+=:;'\",.<>?/\\]", " ", $str))); }