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    * Provides information about the data source.
  23    *
  24    * @return array
  25    *   collection of info about this data source
  26    */
  27   public function getInfo(): array {
  28     return ['title' => ts('Comma-Separated Values (CSV)')];
  29   }
  30
  31   /**
  32    * Set variables up before form is built.
  33    *
  34    * @param CRM_Core_Form $form
  35    */
  36   public function preProcess(&$form) {
  37   }
  38
  39   /**
  40    * This is function is called by the form object to get the DataSource's form snippet.
  41    *
  42    * It should add all fields necessary to get the data
  43    * uploaded to the temporary table in the DB.
  44    *
  45    * @param CRM_Core_Form $form
  46    *
  47    * @throws \CRM_Core_Exception
  48    */
  49   public function buildQuickForm(&$form) {
  50     $form->add('hidden', 'hidden_dataSource', 'CRM_Import_DataSource_CSV');
  51
  52     $config = CRM_Core_Config::singleton();
  53
  54     $uploadFileSize = CRM_Utils_Number::formatUnitSize($config->maxFileSize . 'm', TRUE);
  55     //Fetch uploadFileSize from php_ini when $config->maxFileSize is set to "no limit".
  56     if (empty($uploadFileSize)) {
  57       $uploadFileSize = CRM_Utils_Number::formatUnitSize(ini_get('upload_max_filesize'), TRUE);
  58     }
  59     $uploadSize = round(($uploadFileSize / (1024 * 1024)), 2);
  60     $form->assign('uploadSize', $uploadSize);
  61     $form->add('File', 'uploadFile', ts('Import Data File'), NULL, TRUE);
  62     $form->setMaxFileSize($uploadFileSize);
  63     $form->addRule('uploadFile', ts('File size should be less than %1 MBytes (%2 bytes)', [
  64       1 => $uploadSize,
  65       2 => $uploadFileSize,
  66     ]), 'maxfilesize', $uploadFileSize);
  67     $form->addRule('uploadFile', ts('Input file must be in CSV format'), 'utf8File');
  68     $form->addRule('uploadFile', ts('A valid file must be uploaded.'), 'uploadedfile');
  69
  70     $form->addElement('checkbox', 'skipColumnHeader', ts('First row contains column headers'));
  71   }
  72
  73   /**
  74    * Process the form submission.
  75    *
  76    * @param array $params
  77    * @param string $db
  78    * @param \CRM_Core_Form $form
  79    *
  80    * @throws \CRM_Core_Exception
  81    */
  82   public function postProcess(&$params, &$db, &$form) {
  83     $file = $params['uploadFile']['name'];
  84     $result = self::_CsvToTable(
  85       $file,
  86       CRM_Utils_Array::value('skipColumnHeader', $params, FALSE),
  87       CRM_Utils_Array::value('import_table_name', $params),
  88       CRM_Utils_Array::value('fieldSeparator', $params, ',')
  89     );
  90
  91     $form->set('originalColHeader', CRM_Utils_Array::value('original_col_header', $result));
  92     $form->set('importTableName', $result['import_table_name']);
  93   }
  94
  95   /**
  96    * Create a table that matches the CSV file and populate it with the file's contents
  97    *
  98    * @param string $file
  99    *   File name to load.
 100    * @param bool $headers
 101    *   Whether the first row contains headers.
 102    * @param string $tableName
 103    *   Name of table from which data imported.
 104    * @param string $fieldSeparator
 105    *   Character that separates the various columns in the file.
 106    *
 107    * @return array
 108    *   name of the created table
 109    * @throws \CRM_Core_Exception
 110    */
 111   private static function _CsvToTable(
 112     $file,
 113     $headers = FALSE,
 114     $tableName = NULL,
 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['original_col_header'] = $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     if ($tableName) {
 184       CRM_Core_DAO::executeQuery("DROP TABLE IF EXISTS $tableName");
 185     }
 186     $table = CRM_Utils_SQL_TempTable::build()->setDurable();
 187     $tableName = $table->getName();
 188     CRM_Core_DAO::executeQuery("DROP TABLE IF EXISTS $tableName");
 189     $table->createWithColumns(implode(' text, ', $columns) . ' text');
 190
 191     $numColumns = count($columns);
 192
 193     // the proper approach, but some MySQL installs do not have this enabled
 194     // $load = "LOAD DATA LOCAL INFILE '$file' INTO TABLE $table FIELDS TERMINATED BY '$fieldSeparator' OPTIONALLY ENCLOSED BY '\"'";
 195     // if ($headers) {   $load .= ' IGNORE 1 LINES'; }
 196     // $db->query($load);
 197
 198     // parse the CSV line by line and build one big INSERT (while MySQL-escaping the CSV contents)
 199     if (!$headers) {
 200       rewind($fd);
 201     }
 202
 203     $sql = NULL;
 204     $first = TRUE;
 205     $count = 0;
 206     while ($row = fgetcsv($fd, 0, $fieldSeparator)) {
 207       // skip rows that dont match column count, else we get a sql error
 208       if (count($row) != $numColumns) {
 209         continue;
 210       }
 211       // A blank line will be array(0 => NULL)
 212       if ($row === [NULL]) {
 213         continue;
 214       }
 215
 216       if (!$first) {
 217         $sql .= ', ';
 218       }
 219
 220       $first = FALSE;
 221
 222       // CRM-17859 Trim non-breaking spaces from columns.
 223       $row = array_map(['CRM_Import_DataSource_CSV', 'trimNonBreakingSpaces'], $row);
 224       $row = array_map(['CRM_Core_DAO', 'escapeString'], $row);
 225       $sql .= "('" . implode("', '", $row) . "')";
 226       $count++;
 227
 228       if ($count >= self::NUM_ROWS_TO_INSERT && !empty($sql)) {
 229         CRM_Core_DAO::executeQuery("INSERT IGNORE INTO $tableName VALUES $sql");
 230
 231         $sql = NULL;
 232         $first = TRUE;
 233         $count = 0;
 234       }
 235     }
 236
 237     if (!empty($sql)) {
 238       CRM_Core_DAO::executeQuery("INSERT IGNORE INTO $tableName VALUES $sql");
 239     }
 240
 241     fclose($fd);
 242
 243     //get the import tmp table name.
 244     $result['import_table_name'] = $tableName;
 245
 246     return $result;
 247   }
 248
 249   /**
 250    * Trim non-breaking spaces in a multibyte-safe way.
 251    * See also dev/core#2127 - avoid breaking strings ending in à or any other
 252    * unicode character sharing the same 0xA0 byte as a non-breaking space.
 253    *
 254    * @param string $string
 255    * @return string The trimmed string
 256    */
 257   public static function trimNonBreakingSpaces(string $string): string {
 258     $encoding = mb_detect_encoding($string, NULL, TRUE);
 259     if ($encoding === FALSE) {
 260       // This could mean a couple things. One is that the string is
 261       // ASCII-encoded but contains a non-breaking space, which causes
 262       // php to fail to detect the encoding. So let's just do what we
 263       // did before which works in that situation and is at least no
 264       // worse in other situations.
 265       return trim($string, chr(0xC2) . chr(0xA0));
 266     }
 267     elseif ($encoding !== 'UTF-8') {
 268       $string = mb_convert_encoding($string, 'UTF-8', [$encoding]);
 269     }
 270     return preg_replace("/^(\u{a0})+|(\u{a0})+$/", '', $string);
 271   }
 272
 273 }