[Sommaire] [Imprimer]

   1  <?php
   2  
   3  require_once 'PEAR.php';
   4  require_once 'File.php';
   5  
   6  /**
   7   * The File_CSV package allows reading and creating of CSV data and files.
   8   *
   9   * $Horde: framework/File_CSV/CSV.php,v 1.4.2.7 2006/03/30 10:15:29 selsky Exp $
  10   *
  11   * Copyright 2002-2003 Tomas Von Veschler Cox <cox@idecnet.com>
  12   * Copyright 2005-2006 Jan Schneider <jan@horde.org>
  13   *
  14   * This source file is subject to version 2.0 of the PHP license, that is
  15   * bundled with this package in the file LICENSE, and is available at through
  16   * the world-wide-web at http://www.php.net/license/2_02.txt.  If you did not
  17   * receive a copy of the PHP license and are unable to obtain it through the
  18   * world-wide-web, please send a note to license@php.net so we can mail you a
  19   * copy immediately.
  20   *
  21   * @author  Tomas Von Veschler Cox <cox@idecnet.com>
  22   * @author  Jan Schneider <jan@horde.org>
  23   * @since   Horde 3.1
  24   * @package File_CSV
  25   */
  26  class File_CSV {
  27  
  28      /**
  29       * Discovers the format of a CSV file (the number of fields, the separator,
  30       * the quote string, and the line break).
  31       *
  32       * We can't use the auto_detect_line_endings PHP setting, because it's not
  33       * supported by fgets() contrary to what the manual says.
  34       *
  35       * @static
  36       *
  37       * @param string  The CSV file name
  38       * @param array   Extra separators that should be checked for.
  39       *
  40       * @return array  The format hash.
  41       */
  42      function discoverFormat($file, $extraSeps = array())
  43      {
  44          if (!$fp = @fopen($file, 'r')) {
  45              return PEAR::raiseError('Could not open file: ' . $file);
  46          }
  47  
  48          $seps = array("\t", ';', ':', ',');
  49          $seps = array_merge($seps, $extraSeps);
  50          $matches = array();
  51          $crlf = null;
  52          $conf = array();
  53  
  54          /* Take the first 10 lines and store the number of ocurrences for each
  55           * separator in each line. */
  56          for ($i = 0; ($i < 10) && ($line = fgets($fp));) {
  57              /* Do we have Mac line endings? */
  58              $lines = preg_split('/\r(?!\n)/', $line, 10);
  59              $j = 0;
  60              $c = count($lines);
  61              if ($c > 1) {
  62                  $crlf = "\r";
  63              }
  64              while ($i < 10 && $j < $c) {
  65                  $line = $lines[$j];
  66                  if (!isset($crlf)) {
  67                      foreach (array("\r\n", "\n") as $c) {
  68                          if (substr($line, -strlen($c)) == $c) {
  69                              $crlf = $c;
  70                              break;
  71                          }
  72                      }
  73                  }
  74                  $i++;
  75                  $j++;
  76                  foreach ($seps as $sep) {
  77                      $matches[$sep][$i] = substr_count($line, $sep);
  78                  }
  79              }
  80          }
  81          if (isset($crlf)) {
  82              $conf['crlf'] = $crlf;
  83          }
  84  
  85          /* Group the results by amount of equal ocurrences. */
  86          $fields = array();
  87          $amount = array();
  88          foreach ($matches as $sep => $lines) {
  89              $times = array();
  90              $times[0] = 0;
  91              foreach ($lines as $num) {
  92                  if ($num > 0) {
  93                      $times[$num] = (isset($times[$num])) ? $times[$num] + 1 : 1;
  94                  }
  95              }
  96              arsort($times);
  97              $fields[$sep] = key($times);
  98              $amount[$sep] = $times[key($times)];
  99          }
 100          arsort($amount);
 101          $sep = key($amount);
 102  
 103          $conf['fields'] = $fields[$sep] + 1;
 104          $conf['sep']    = $sep;
 105  
 106          /* Test if there are fields with quotes arround in the first 5
 107           * lines. */
 108          $quotes = '"\'';
 109          $quote  = '';
 110          rewind($fp);
 111          for ($i = 0; ($i < 5) && ($line = fgets($fp)); $i++) {
 112              if (preg_match("|$sep([$quotes]).*([$quotes])$sep|U", $line, $match)) {
 113                  if ($match[1] == $match[2]) {
 114                      $quote = $match[1];
 115                      break;
 116                  }
 117              }
 118              if (preg_match("|^([$quotes]).*([$quotes])$sep|", $line, $match) ||
 119                  preg_match("|([$quotes]).*([$quotes])$sep\s$|Us", $line, $match))
 120              {
 121                  if ($match[1] == $match[2]) {
 122                      $quote = $match[1];
 123                      break;
 124                  }
 125              }
 126          }
 127          $conf['quote'] = $quote;
 128  
 129          fclose($fp);
 130  
 131          // XXX What about trying to discover the "header"?
 132          return $conf;
 133      }
 134  
 135      /**
 136       * Reads a row from a CSV file and returns it as an array.
 137       *
 138       * @param string $file  The name of the CSV file.
 139       * @param array $conf   The configuration for the CSV file.
 140       *
 141       * @return array|boolean  The CSV data or false if no more data available.
 142       */
 143      function read($file, &$conf)
 144      {
 145          $fp = File_CSV::getPointer($file, $conf, FILE_MODE_READ);
 146          if (is_a($fp, 'PEAR_Error')) {
 147              return $fp;
 148          }
 149  
 150          $line = fgets($fp);
 151          if (!$line) {
 152              return false;
 153          }
 154  
 155          /* Use readQuoted() if we have Mac line endings. */
 156          if (preg_match('/\r(?!\n)/', $line)) {
 157              fseek($fp, -strlen($line), SEEK_CUR);
 158              return File_CSV::readQuoted($file, $conf);
 159          }
 160  
 161          File_CSV::_line(File_CSV::_line() + 1);
 162  
 163          if ($conf['fields'] == 1) {
 164              return array($line);
 165          }
 166  
 167          $fields = explode($conf['sep'], $line);
 168          if ($conf['quote']) {
 169              $last = &$fields[count($fields) - 1];
 170              /* Fallback to read the line with readQuoted() when guess that the
 171               * simple explode won't work right. */
 172              if ((substr($last, -strlen($conf['crlf'])) == $conf['crlf'] &&
 173                   $last{0} == $conf['quote'] &&
 174                   $last{strlen(rtrim($last)) - 1} != $conf['quote']) ||
 175                  (count($fields) != $conf['fields'])
 176                      // XXX perhaps there is a separator inside a quoted field
 177                      //preg_match("|{$conf['quote']}.*{$conf['sep']}.*{$conf['quote']}|U", $line)
 178                  )
 179              {
 180                  fseek($fp, -strlen($line), SEEK_CUR);
 181                  return File_CSV::readQuoted($file, $conf);
 182              } else {
 183                  $last = rtrim($last);
 184                  foreach ($fields as $k => $v) {
 185                      $fields[$k] = File_CSV::unquote($v, $conf['quote'], $conf['crlf']);
 186                  }
 187              }
 188          } else {
 189              $fields[count($fields) -1] = rtrim($fields[count($fields) -1]);
 190          }
 191  
 192          if (count($fields) < $conf['fields']) {
 193              File_CSV::warning(sprintf(_("Wrong number of fields in line %d. Expected %d, found %d."), File_CSV::_line(), $conf['fields'], count($fields)));
 194              $fields = array_merge($fields, array_fill(0, $conf['fields'] - count($fields), ''));
 195          } elseif (count($fields) > $conf['fields']) {
 196              File_CSV::warning(sprintf(_("More fields found in line %d than the expected %d."), File_CSV::_line(), $conf['fields']));
 197              array_splice($fields, $conf['fields']);
 198          }
 199  
 200          return $fields;
 201      }
 202  
 203      /**
 204       * Reads a row from a CSV file and returns it as an array.
 205       * This method is able to read fields with multiline data and normalizes
 206       * linebreaks to single newline characters (0x0a).
 207       *
 208       * @param string $file  The name of the CSV file.
 209       * @param array $conf   The configuration for the CSV file.
 210       *
 211       * @return array|boolean  The CSV data or false if no more data available.
 212       */
 213      function readQuoted($file, &$conf)
 214      {
 215          $fp = File_CSV::getPointer($file, $conf, FILE_MODE_READ);
 216          if (is_a($fp, 'PEAR_Error')) {
 217              return $fp;
 218          }
 219  
 220          /* A buffer with all characters of the current field read so far. */
 221          $buff = '';
 222          /* The current character. */
 223          $c = null;
 224          /* The read fields. */
 225          $ret = array();
 226          /* The number of the current field. */
 227          $i = 1;
 228          /* Are we inside a quoted field? */
 229          $in_quote = false;
 230  
 231          while (($ch = fgetc($fp)) !== false) {
 232              /* Previous character. */
 233              $prev = $c;
 234              /* Current character. */
 235              $c = $ch;
 236  
 237              /* Simple character. */
 238              if ($c != $conf['quote'] &&
 239                  $c != $conf['sep'] &&
 240                  strpos($conf['crlf'], $c) === false) {
 241                  $buff .= $c;
 242                  continue;
 243              }
 244  
 245              if ($conf['quote'] && $c == $conf['quote'] &&
 246                  ($prev == $conf['sep'] ||
 247                   $prev == substr($conf['crlf'], -1) ||
 248                   $prev === null))
 249              {
 250                  /* Quote begin. */
 251                  $in_quote = true;
 252              } elseif ($in_quote) {
 253                  /* Quote end? */
 254                  if ($c == $conf['sep'] && $prev == $conf['quote']) {
 255                      $in_quote = false;
 256                  } elseif ($c == substr($conf['crlf'], -1)) {
 257                      $sub = (strlen($conf['crlf']) > 1 &&
 258                              $prev == substr($conf['crlf'], -2, 1))
 259                          ? 2
 260                          : 1;
 261                      if ((strlen($buff) >= $sub) &&
 262                          ($buff{strlen($buff) - $sub} == $conf['quote']))
 263                      {
 264                          $in_quote = false;
 265                      } else {
 266                          File_CSV::_line(File_CSV::_line() + 1);
 267                      }
 268                  }
 269              }
 270  
 271              /* End of line or end of field. */
 272              if (!$in_quote &&
 273                  ($c == $conf['sep'] || $c == substr($conf['crlf'], -1))) {
 274                  /* More fields than expected. */
 275                  if ($c == $conf['sep'] &&
 276                      (count($ret) + 1) == $conf['fields']) {
 277                      while ($c !== false && $c != substr($conf['crlf'], -1)) {
 278                          $c = fgetc($fp);
 279                      }
 280                      File_CSV::warning(sprintf(_("More fields found in line %d than the expected %d."), File_CSV::_line(), $conf['fields']));
 281                  }
 282  
 283                  if ($prev == substr($conf['crlf'], -2, 1)) {
 284                      $buff = substr($buff, 0, -1);
 285                  }
 286  
 287                  /* Less fields than expected. */
 288                  if (($c == substr($conf['crlf'], -1)) &&
 289                      ($i != $conf['fields'])) {
 290                      if ($i == 1) {
 291                          /* Skip empty lines. */
 292                          return $ret;
 293                      }
 294                      File_CSV::warning(sprintf(_("Wrong number of fields in line %d. Expected %d, found %d."), File_CSV::_line(), $conf['fields'], $i));
 295  
 296                      $ret[] = File_CSV::unquote($buff, $conf['quote'], $conf['crlf']);
 297                      $ret = array_merge($ret, array_fill(0, $conf['fields'] - $i, ''));
 298                      return $ret;
 299                  }
 300  
 301                  $ret[] = File_CSV::unquote($buff, $conf['quote'], $conf['crlf']);
 302                  if (count($ret) == $conf['fields']) {
 303                      return $ret;
 304                  }
 305  
 306                  $buff = '';
 307                  $i++;
 308                  continue;
 309              }
 310              $buff .= $c;
 311          }
 312  
 313          return feof($fp) ? false : $ret;
 314      }
 315  
 316      /**
 317       * Writes a hash into a CSV file.
 318       *
 319       * @param string $file   The name of the CSV file.
 320       * @param array $fields  The CSV data.
 321       * @param array $conf    The configuration for the CSV file.
 322       *
 323       * @return boolean  True on success, PEAR_Error on failure.
 324       */
 325      function write($file, $fields, &$conf)
 326      {
 327          if (is_a($fp = File_CSV::getPointer($file, $conf, FILE_MODE_WRITE), 'PEAR_Error')) {
 328              return $fp;
 329          }
 330  
 331          if (count($fields) != $conf['fields']) {
 332              return PEAR::raiseError(sprintf(_("Wrong number of fields. Expected %d, found %d."), $conf['fields'], count($fields)));
 333          }
 334  
 335          $write = '';
 336          for ($i = 0; $i < count($fields); $i++) {
 337              if (!is_numeric($fields[$i]) && $conf['quote']) {
 338                  $write .= $conf['quote'] . $fields[$i] . $conf['quote'];
 339              } else {
 340                  $write .= $fields[$i];
 341              }
 342              if ($i < (count($fields) - 1)) {
 343                  $write .= $conf['sep'];
 344              } else {
 345                  $write .= $conf['crlf'];
 346              }
 347          }
 348  
 349          if (!fwrite($fp, $write)) {
 350              return PEAR::raiseError(sprintf(_("Cannot write to file \"%s\""), $file));
 351          }
 352  
 353          return true;
 354      }
 355  
 356      /**
 357       * Removes surrounding quotes from a string and normalizes linebreaks.
 358       *
 359       * @param string $field  The string to unquote.
 360       * @param string $quote  The quote character.
 361       * @param string $crlf   The linebreak character.
 362       *
 363       * @return string  The unquoted data.
 364       */
 365      function unquote($field, $quote, $crlf)
 366      {
 367          /* Skip empty fields (form: ;;) */
 368          if (!strlen($field)) {
 369              return $field;
 370          }
 371          if ($quote && $field{0} == $quote && $field{strlen($field)-1} == $quote) {
 372              return str_replace($crlf, "\n", substr($field, 1, -1));
 373          }
 374          return $field;
 375      }
 376  
 377      /**
 378       * Sets or gets the current line being parsed.
 379       *
 380       * @param integer $line  If specified, the current line.
 381       *
 382       * @return integer  The current line.
 383       */
 384      function _line($line = null)
 385      {
 386          static $current_line = 0;
 387  
 388          if (!is_null($line)) {
 389              $current_line = $line;
 390          }
 391  
 392          return $current_line;
 393      }
 394  
 395      /**
 396       * Adds a warning to or retrieves and resets the warning stack.
 397       *
 398       * @param string  A warning string.  If not specified, the existing
 399       *                warnings will be returned instead and the warning stack
 400       *                gets emptied.
 401       *
 402       * @return array  If no parameter has been specified, the list of existing
 403       *                warnings.
 404       */
 405      function warning($warning = null)
 406      {
 407          static $warnings = array();
 408  
 409          if (is_null($warning)) {
 410              $return = $warnings;
 411              $warnings = array();
 412              return $return;
 413          }
 414  
 415          $warnings[] = $warning;
 416      }
 417  
 418      /**
 419       * Returns or creates the file descriptor associated with a file.
 420       *
 421       * @static
 422       *
 423       * @param string $file  The name of the file
 424       * @param array $conf   The configuration
 425       * @param string $mode  The open mode. FILE_MODE_READ or FILE_MODE_WRITE.
 426       *
 427       * @return resource  The file resource or PEAR_Error on error.
 428       */
 429      function getPointer($file, &$conf, $mode = FILE_MODE_READ)
 430      {
 431          static $resources = array();
 432          static $config = array();
 433  
 434          $signature = $file . serialize($conf);
 435          if (isset($resources[$signature])) {
 436              $conf = $config[$signature];
 437              return $resources[$signature];
 438          }
 439          if (is_a($error = File_CSV::_checkConfig($conf), 'PEAR_Error')) {
 440              return $error;
 441          }
 442          $config[$signature] = $conf;
 443  
 444          $fp = @fopen($file, $mode);
 445          if (!is_resource($fp)) {
 446              return PEAR::raiseError(sprintf(_("Can't open file \"%s\"."), $file));
 447          }
 448          $resources[$signature] = $fp;
 449          File_CSV::_line(0);
 450  
 451          if ($mode == FILE_MODE_READ && !empty($conf['header'])) {
 452              if (is_a($header = File_CSV::read($file, $conf), 'PEAR_Error')) {
 453                  return $header;
 454              }
 455          }
 456  
 457          return $fp;
 458      }
 459  
 460      /**
 461       * Checks the configuration given by the user.
 462       *
 463       * @param array $conf    The configuration assoc array
 464       * @param string $error  The error will be written here if any
 465       */
 466      function _checkConfig(&$conf)
 467      {
 468          // check conf
 469          if (!is_array($conf)) {
 470              return PEAR::raiseError('Invalid configuration.');
 471          }
 472  
 473          if (!isset($conf['fields']) || !is_numeric($conf['fields'])) {
 474              return PEAR::raiseError(_("The number of fields must be numeric."));
 475          }
 476  
 477          if (isset($conf['sep'])) {
 478              if (strlen($conf['sep']) != 1) {
 479                  return PEAR::raiseError(_("The separator must be one single character."));
 480              }
 481          } elseif ($conf['fields'] > 1) {
 482              return PEAR::raiseError(_("No separator specified."));
 483          }
 484  
 485          if (!empty($conf['quote'])) {
 486              if (strlen($conf['quote']) != 1) {
 487                  return PEAR::raiseError(_("The quote character must be one single character."));
 488              }
 489          } else {
 490              $conf['quote'] = '';
 491          }
 492  
 493          if (!isset($conf['crlf'])) {
 494              $conf['crlf'] = "\n";
 495          }
 496      }
 497  
 498  }