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 $form
  76    *
  77    * @throws \API_Exception
  78    * @throws \CRM_Core_Exception
  79    * @throws \API_Exception
  80    */
  81   public function postProcess(&$params, &$db, &$form) {
  82     $firstRowIsColumnHeader = $params['skipColumnHeader'] ?? FALSE;
  83     $result = self::_CsvToTable(
  84       $this->getSubmittedValue('uploadFile')['name'],
  85       $this->getSubmittedValue('skipColumnHeader'),
  86       $this->getSubmittedValue('fieldSeparator') ?? ','
  87     );
  88     $this->addTrackingFieldsToTable($result['import_table_name']);
  89
  90     $form->set('originalColHeader', CRM_Utils_Array::value('column_headers', $result));
  91     $form->set('importTableName', $result['import_table_name']);
  92     $this->updateUserJobMetadata('DataSource', [
  93       'table_name' => $result['import_table_name'],
  94       'column_headers' => $firstRowIsColumnHeader ? $result['column_headers'] : [],
  95       'number_of_columns' => $result['number_of_columns'],
  96     ]);
  97   }
  98
  99   /**
 100    * Create a table that matches the CSV file and populate it with the file's contents
 101    *
 102    * @param string $file
 103    *   File name to load.
 104    * @param bool $headers
 105    *   Whether the first row contains headers.
 106    * @param string $fieldSeparator
 107    *   Character that separates the various columns in the file.
 108    *
 109    * @return array
 110    *   name of the created table
 111    * @throws \CRM_Core_Exception
 112    */
 113   private static function _CsvToTable(
 114     $file,
 115     $headers = FALSE,
 116     $fieldSeparator = ','
 117   ) {
 118     $result = [];
 119     $fd = fopen($file, 'r');
 120     if (!$fd) {
 121       throw new CRM_Core_Exception("Could not read $file");
 122     }
 123     if (filesize($file) == 0) {
 124       throw new CRM_Core_Exception("$file is empty. Please upload a valid file.");
 125     }
 126
 127     // support tab separated
 128     if (strtolower($fieldSeparator) === 'tab' ||
 129       strtolower($fieldSeparator) === '\t'
 130     ) {
 131       $fieldSeparator = "\t";
 132     }
 133
 134     $firstrow = fgetcsv($fd, 0, $fieldSeparator);
 135
 136     // create the column names from the CSV header or as col_0, col_1, etc.
 137     if ($headers) {
 138       //need to get original headers.
 139       $result['column_headers'] = $firstrow;
 140
 141       $strtolower = function_exists('mb_strtolower') ? 'mb_strtolower' : 'strtolower';
 142       $columns = array_map($strtolower, $firstrow);
 143       $columns = str_replace(' ', '_', $columns);
 144       $columns = preg_replace('/[^a-z_]/', '', $columns);
 145
 146       // need to take care of null as well as duplicate col names.
 147       $duplicateColName = FALSE;
 148       if (count($columns) != count(array_unique($columns))) {
 149         $duplicateColName = TRUE;
 150       }
 151
 152       // need to truncate values per mysql field name length limits
 153       // mysql allows 64, but we need to account for appending colKey
 154       // CRM-9079
 155       foreach ($columns as $colKey => & $colName) {
 156         if (strlen($colName) > 58) {
 157           $colName = substr($colName, 0, 58);
 158         }
 159       }
 160
 161       if (in_array('', $columns) || $duplicateColName) {
 162         foreach ($columns as $colKey => & $colName) {
 163           if (!$colName) {
 164             $colName = "col_$colKey";
 165           }
 166           elseif ($duplicateColName) {
 167             $colName .= "_$colKey";
 168           }
 169         }
 170       }
 171
 172       // CRM-4881: we need to quote column names, as they may be MySQL reserved words
 173       foreach ($columns as & $column) {
 174         $column = "`$column`";
 175       }
 176     }
 177     else {
 178       $columns = [];
 179       foreach ($firstrow as $i => $_) {
 180         $columns[] = "col_$i";
 181       }
 182     }
 183
 184     $table = CRM_Utils_SQL_TempTable::build()->setDurable();
 185     $tableName = $table->getName();
 186     CRM_Core_DAO::executeQuery("DROP TABLE IF EXISTS $tableName");
 187     $table->createWithColumns(implode(' text, ', $columns) . ' text');
 188
 189     $numColumns = count($columns);
 190
 191     // the proper approach, but some MySQL installs do not have this enabled
 192     // $load = "LOAD DATA LOCAL INFILE '$file' INTO TABLE $table FIELDS TERMINATED BY '$fieldSeparator' OPTIONALLY ENCLOSED BY '\"'";
 193     // if ($headers) {   $load .= ' IGNORE 1 LINES'; }
 194     // $db->query($load);
 195
 196     // parse the CSV line by line and build one big INSERT (while MySQL-escaping the CSV contents)
 197     if (!$headers) {
 198       rewind($fd);
 199     }
 200
 201     $sql = NULL;
 202     $first = TRUE;
 203     $count = 0;
 204     while ($row = fgetcsv($fd, 0, $fieldSeparator)) {
 205       // skip rows that dont match column count, else we get a sql error
 206       if (count($row) != $numColumns) {
 207         continue;
 208       }
 209       // A blank line will be array(0 => NULL)
 210       if ($row === [NULL]) {
 211         continue;
 212       }
 213
 214       if (!$first) {
 215         $sql .= ', ';
 216       }
 217
 218       $first = FALSE;
 219
 220       // CRM-17859 Trim non-breaking spaces from columns.
 221       $row = array_map(['CRM_Import_DataSource_CSV', 'trimNonBreakingSpaces'], $row);
 222       $row = array_map(['CRM_Core_DAO', 'escapeString'], $row);
 223       $sql .= "('" . implode("', '", $row) . "')";
 224       $count++;
 225
 226       if ($count >= self::NUM_ROWS_TO_INSERT && !empty($sql)) {
 227         CRM_Core_DAO::executeQuery("INSERT IGNORE INTO $tableName VALUES $sql");
 228
 229         $sql = NULL;
 230         $first = TRUE;
 231         $count = 0;
 232       }
 233     }
 234
 235     if (!empty($sql)) {
 236       CRM_Core_DAO::executeQuery("INSERT IGNORE INTO $tableName VALUES $sql");
 237     }
 238
 239     fclose($fd);
 240
 241     //get the import tmp table name.
 242     $result['import_table_name'] = $tableName;
 243     $result['number_of_columns'] = $numColumns;
 244     return $result;
 245   }
 246
 247   /**
 248    * Trim non-breaking spaces in a multibyte-safe way.
 249    * See also dev/core#2127 - avoid breaking strings ending in à or any other
 250    * unicode character sharing the same 0xA0 byte as a non-breaking space.
 251    *
 252    * @param string $string
 253    * @return string The trimmed string
 254    */
 255   public static function trimNonBreakingSpaces(string $string): string {
 256     $encoding = mb_detect_encoding($string, NULL, TRUE);
 257     if ($encoding === FALSE) {
 258       // This could mean a couple things. One is that the string is
 259       // ASCII-encoded but contains a non-breaking space, which causes
 260       // php to fail to detect the encoding. So let's just do what we
 261       // did before which works in that situation and is at least no
 262       // worse in other situations.
 263       return trim($string, chr(0xC2) . chr(0xA0));
 264     }
 265     elseif ($encoding !== 'UTF-8') {
 266       $string = mb_convert_encoding($string, 'UTF-8', [$encoding]);
 267     }
 268     return preg_replace("/^(\u{a0})+|(\u{a0})+$/", '', $string);
 269   }
 270
 271 }