CRM/Import/DataSource/CSV.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | Copyright CiviCRM LLC. All rights reserved.                        |
   5  |                                                                    |
   6  | This work is published under the GNU AGPLv3 license with some      |
   7  | permitted exceptions and without any warranty. For full license    |
   8  | and copyright information, see https://civicrm.org/licensing       |
   9  +--------------------------------------------------------------------+
  10  */
  11
  12 /**
  13  *
  14  * @package CRM
  15  * @copyright CiviCRM LLC https://civicrm.org/licensing
  16  */
  17 class CRM_Import_DataSource_CSV extends CRM_Import_DataSource {
  18   const
  19     NUM_ROWS_TO_INSERT = 100;
  20
  21   /**
  22    * Form fields declared for this datasource.
  23    *
  24    * @var string[]
  25    */
  26   protected $submittableFields = ['skipColumnHeader', 'uploadField'];
  27
  28   /**
  29    * Provides information about the data source.
  30    *
  31    * @return array
  32    *   collection of info about this data source
  33    */
  34   public function getInfo(): array {
  35     return ['title' => ts('Comma-Separated Values (CSV)')];
  36   }
  37
  38   /**
  39    * This is function is called by the form object to get the DataSource's form snippet.
  40    *
  41    * It should add all fields necessary to get the data
  42    * uploaded to the temporary table in the DB.
  43    *
  44    * @param CRM_Core_Form $form
  45    *
  46    * @throws \CRM_Core_Exception
  47    */
  48   public function buildQuickForm(&$form) {
  49     $form->add('hidden', 'hidden_dataSource', 'CRM_Import_DataSource_CSV');
  50
  51     $uploadFileSize = CRM_Utils_Number::formatUnitSize(Civi::settings()->get('maxFileSize') . 'm', TRUE);
  52     //Fetch uploadFileSize from php_ini when $config->maxFileSize is set to "no limit".
  53     if (empty($uploadFileSize)) {
  54       $uploadFileSize = CRM_Utils_Number::formatUnitSize(ini_get('upload_max_filesize'), TRUE);
  55     }
  56     $uploadSize = round(($uploadFileSize / (1024 * 1024)), 2);
  57     $form->assign('uploadSize', $uploadSize);
  58     $form->add('File', 'uploadFile', ts('Import Data File'), NULL, TRUE);
  59     $form->setMaxFileSize($uploadFileSize);
  60     $form->addRule('uploadFile', ts('File size should be less than %1 MBytes (%2 bytes)', [
  61       1 => $uploadSize,
  62       2 => $uploadFileSize,
  63     ]), 'maxfilesize', $uploadFileSize);
  64     $form->addRule('uploadFile', ts('Input file must be in CSV format'), 'utf8File');
  65     $form->addRule('uploadFile', ts('A valid file must be uploaded.'), 'uploadedfile');
  66
  67     $form->addElement('checkbox', 'skipColumnHeader', ts('First row contains column headers'));
  68   }
  69
  70   /**
  71    * Process the form submission.
  72    *
  73    * @param array $params
  74    * @param string $db
  75    * @param \CRM_Core_Form $form
  76    *
  77    * @throws \API_Exception
  78    * @throws \CRM_Core_Exception
  79    */
  80   public function postProcess(&$params, &$db, &$form) {
  81     $file = $params['uploadFile']['name'];
  82     $firstRowIsColumnHeader = $params['skipColumnHeader'] ?? FALSE;
  83     $result = self::_CsvToTable(
  84       $file,
  85       $firstRowIsColumnHeader,
  86       CRM_Utils_Array::value('fieldSeparator', $params, ',')
  87     );
  88
  89     $form->set('originalColHeader', CRM_Utils_Array::value('column_headers', $result));
  90     $form->set('importTableName', $result['import_table_name']);
  91     $this->updateUserJobMetadata('DataSource', [
  92       'table_name' => $result['import_table_name'],
  93       'column_headers' => $firstRowIsColumnHeader ? $result['column_headers'] : [],
  94       'number_of_columns' => $result['number_of_columns'],
  95     ]);
  96   }
  97
  98   /**
  99    * Create a table that matches the CSV file and populate it with the file's contents
 100    *
 101    * @param string $file
 102    *   File name to load.
 103    * @param bool $headers
 104    *   Whether the first row contains headers.
 105    * @param string $fieldSeparator
 106    *   Character that separates the various columns in the file.
 107    *
 108    * @return array
 109    *   name of the created table
 110    * @throws \CRM_Core_Exception
 111    */
 112   private static function _CsvToTable(
 113     $file,
 114     $headers = FALSE,
 115     $fieldSeparator = ','
 116   ) {
 117     $result = [];
 118     $fd = fopen($file, 'r');
 119     if (!$fd) {
 120       throw new CRM_Core_Exception("Could not read $file");
 121     }
 122     if (filesize($file) == 0) {
 123       throw new CRM_Core_Exception("$file is empty. Please upload a valid file.");
 124     }
 125
 126     // support tab separated
 127     if (strtolower($fieldSeparator) === 'tab' ||
 128       strtolower($fieldSeparator) === '\t'
 129     ) {
 130       $fieldSeparator = "\t";
 131     }
 132
 133     $firstrow = fgetcsv($fd, 0, $fieldSeparator);
 134
 135     // create the column names from the CSV header or as col_0, col_1, etc.
 136     if ($headers) {
 137       //need to get original headers.
 138       $result['column_headers'] = $firstrow;
 139
 140       $strtolower = function_exists('mb_strtolower') ? 'mb_strtolower' : 'strtolower';
 141       $columns = array_map($strtolower, $firstrow);
 142       $columns = str_replace(' ', '_', $columns);
 143       $columns = preg_replace('/[^a-z_]/', '', $columns);
 144
 145       // need to take care of null as well as duplicate col names.
 146       $duplicateColName = FALSE;
 147       if (count($columns) != count(array_unique($columns))) {
 148         $duplicateColName = TRUE;
 149       }
 150
 151       // need to truncate values per mysql field name length limits
 152       // mysql allows 64, but we need to account for appending colKey
 153       // CRM-9079
 154       foreach ($columns as $colKey => & $colName) {
 155         if (strlen($colName) > 58) {
 156           $colName = substr($colName, 0, 58);
 157         }
 158       }
 159
 160       if (in_array('', $columns) || $duplicateColName) {
 161         foreach ($columns as $colKey => & $colName) {
 162           if (!$colName) {
 163             $colName = "col_$colKey";
 164           }
 165           elseif ($duplicateColName) {
 166             $colName .= "_$colKey";
 167           }
 168         }
 169       }
 170
 171       // CRM-4881: we need to quote column names, as they may be MySQL reserved words
 172       foreach ($columns as & $column) {
 173         $column = "`$column`";
 174       }
 175     }
 176     else {
 177       $columns = [];
 178       foreach ($firstrow as $i => $_) {
 179         $columns[] = "col_$i";
 180       }
 181     }
 182
 183     $table = CRM_Utils_SQL_TempTable::build()->setDurable();
 184     $tableName = $table->getName();
 185     CRM_Core_DAO::executeQuery("DROP TABLE IF EXISTS $tableName");
 186     $table->createWithColumns(implode(' text, ', $columns) . ' text');
 187
 188     $numColumns = count($columns);
 189
 190     // the proper approach, but some MySQL installs do not have this enabled
 191     // $load = "LOAD DATA LOCAL INFILE '$file' INTO TABLE $table FIELDS TERMINATED BY '$fieldSeparator' OPTIONALLY ENCLOSED BY '\"'";
 192     // if ($headers) {   $load .= ' IGNORE 1 LINES'; }
 193     // $db->query($load);
 194
 195     // parse the CSV line by line and build one big INSERT (while MySQL-escaping the CSV contents)
 196     if (!$headers) {
 197       rewind($fd);
 198     }
 199
 200     $sql = NULL;
 201     $first = TRUE;
 202     $count = 0;
 203     while ($row = fgetcsv($fd, 0, $fieldSeparator)) {
 204       // skip rows that dont match column count, else we get a sql error
 205       if (count($row) != $numColumns) {
 206         continue;
 207       }
 208       // A blank line will be array(0 => NULL)
 209       if ($row === [NULL]) {
 210         continue;
 211       }
 212
 213       if (!$first) {
 214         $sql .= ', ';
 215       }
 216
 217       $first = FALSE;
 218
 219       // CRM-17859 Trim non-breaking spaces from columns.
 220       $row = array_map(['CRM_Import_DataSource_CSV', 'trimNonBreakingSpaces'], $row);
 221       $row = array_map(['CRM_Core_DAO', 'escapeString'], $row);
 222       $sql .= "('" . implode("', '", $row) . "')";
 223       $count++;
 224
 225       if ($count >= self::NUM_ROWS_TO_INSERT && !empty($sql)) {
 226         CRM_Core_DAO::executeQuery("INSERT IGNORE INTO $tableName VALUES $sql");
 227
 228         $sql = NULL;
 229         $first = TRUE;
 230         $count = 0;
 231       }
 232     }
 233
 234     if (!empty($sql)) {
 235       CRM_Core_DAO::executeQuery("INSERT IGNORE INTO $tableName VALUES $sql");
 236     }
 237
 238     fclose($fd);
 239
 240     //get the import tmp table name.
 241     $result['import_table_name'] = $tableName;
 242     $result['number_of_columns'] = $numColumns;
 243     return $result;
 244   }
 245
 246   /**
 247    * Trim non-breaking spaces in a multibyte-safe way.
 248    * See also dev/core#2127 - avoid breaking strings ending in à or any other
 249    * unicode character sharing the same 0xA0 byte as a non-breaking space.
 250    *
 251    * @param string $string
 252    * @return string The trimmed string
 253    */
 254   public static function trimNonBreakingSpaces(string $string): string {
 255     $encoding = mb_detect_encoding($string, NULL, TRUE);
 256     if ($encoding === FALSE) {
 257       // This could mean a couple things. One is that the string is
 258       // ASCII-encoded but contains a non-breaking space, which causes
 259       // php to fail to detect the encoding. So let's just do what we
 260       // did before which works in that situation and is at least no
 261       // worse in other situations.
 262       return trim($string, chr(0xC2) . chr(0xA0));
 263     }
 264     elseif ($encoding !== 'UTF-8') {
 265       $string = mb_convert_encoding($string, 'UTF-8', [$encoding]);
 266     }
 267     return preg_replace("/^(\u{a0})+|(\u{a0})+$/", '', $string);
 268   }
 269
 270 }