/**
* Get a new CSVParser using defined settings.
* @return Iterator
*/
public function getIterator()
{
if (!file_exists($this->filepath)) {
//TODO: throw exception instead?
return null;
}
$header = $this->hasheader ? $this->getFirstRow() : null;
$output = array();
$config = new LexerConfig();
$config->setDelimiter($this->delimiter);
$config->setEnclosure($this->enclosure);
$config->setIgnoreHeaderLine($this->hasheader);
$interpreter = new Interpreter();
// Ignore row column count consistency
$interpreter->unstrict();
$interpreter->addObserver(function (array $row) use(&$output, $header) {
if ($header) {
//create new row using headings as keys
$newrow = array();
foreach ($header as $k => $heading) {
if (isset($row[$k])) {
$newrow[$heading] = $row[$k];
}
}
$row = $newrow;
}
$output[] = $row;
});
$lexer = new Lexer($config);
$lexer->parse($this->filepath, $interpreter);
return new ArrayIterator($output);
}
/**
* use un-strict won't throw exception with inconsistent columns
*
*/
public function testInconsistentColumnsWithUnStrict()
{
$lines[] = array('test', 'test', 'test');
$lines[] = array('test', 'test');
$interpreter = new Interpreter();
$interpreter->unstrict();
foreach ($lines as $line) {
$interpreter->interpret($line);
}
}
/**
* @param $file
* @param Closure|null $callback
* @return array|null
*/
public function import($file, Closure $callback = null)
{
$interpreter = new Interpreter();
if ($this->unstrict) {
$interpreter->unstrict();
}
if (is_null($callback)) {
$results = array();
$callback = function (array $row) use(&$results) {
$results[] = $row;
};
} else {
$results = null;
}
$interpreter->addObserver($callback);
$lexer = new Lexer($this->config);
$lexer->parse($file, $interpreter);
return $results;
}
/**
* Extract data from de CSV and create new Location objects
*
* @param string $countryCode Country code
* @param string $pathname Pathname
*
* @return LocationInterface Root location
*/
private function loadLocationsFrom($countryCode, $pathname)
{
$countryInfo = $this->getCountryInfo($countryCode);
$country = $this->locationBuilder->addLocation($countryInfo[0], $countryInfo[1], $countryInfo[0], 'country');
$interpreter = new Interpreter();
$interpreter->unstrict();
$nbItems = 0;
$interpreter->addObserver(function (array $columns) use($country, &$nbItems) {
if (isset($columns[1], $columns[2], $columns[3], $columns[4], $columns[5], $columns[6])) {
$nbItems++;
$stateId = $this->normalizeId($columns[4]);
$state = $this->locationBuilder->addLocation($country->getId() . '_' . $stateId, $columns[3], $stateId, 'state', $country);
$provinceId = $this->normalizeId($columns[6]);
$province = $this->locationBuilder->addLocation($state->getId() . '_' . $provinceId, $columns[5], $provinceId, 'province', $state);
$cityId = $this->normalizeId($columns[2]);
$city = $this->locationBuilder->addLocation($province->getId() . '_' . $cityId, $columns[2], $cityId, 'city', $province);
$postalCodeId = $this->normalizeId($columns[1]);
$this->locationBuilder->addLocation($city->getId() . '_' . $postalCodeId, $columns[1], $postalCodeId, 'postalcode', $city);
}
});
$config = new LexerConfig();
$config->setDelimiter("\t");
$lexer = new Lexer($config);
$lexer->parse($pathname, $interpreter);
return $country;
}
/**
* @return Interpreter
*/
protected function getInterpreter()
{
$interpreter = new Interpreter();
$interpreter->unstrict();
return $interpreter;
}
/**
* CSV parser
*
* @param string $file Path of the file.
* @return array Returns the array from csv.
* @since 0.1.0
*/
public static function csv_parser($csv_file)
{
if (!is_file($csv_file)) {
return new WP_Error('error', 'The CSV file is not found.');
} elseif (!self::is_textfile($csv_file)) {
return new WP_Error('error', 'The file is not CSV.');
}
$csv = array();
/**
* Filter the CSV setting for the csv parser.
*
* @param array settings for the csv parser.
*/
$format = apply_filters('acsv_csv_format', array('from_charset' => 'UTF-8', 'to_charset' => 'UTF-8', 'delimiter' => ',', 'enclosure' => '"', 'escape' => '\\'));
$config = new LexerConfig();
$config->setFromCharset($format['from_charset']);
$config->setToCharset($format['to_charset']);
$config->setDelimiter($format['delimiter']);
$config->setEnclosure($format['enclosure']);
$config->setEscape($format['escape']);
$lexer = new Lexer($config);
$interpreter = new Interpreter();
$interpreter->unstrict();
$interpreter->addObserver(function (array $row) use(&$csv) {
$csv[] = $row;
});
$lexer->parse($csv_file, $interpreter);
return $csv;
}
/**
* Populate a country
*
* @param OutputInterface $output The output interface
* @param string $countryCode Country Code
* @param boolean $sourcePackageMustbeReloaded Source package must be reloaded
*
* @return CountryInterface|null Country populated if created
*/
public function populateCountry(OutputInterface $output, $countryCode, $sourcePackageMustbeReloaded)
{
$countryCode = strtoupper($countryCode);
$file = $this->downloadRemoteFileFromCountryCode($output, $countryCode, $sourcePackageMustbeReloaded);
$countryData = $this->getCountryArray()[$countryCode];
$country = $this->geoBuilder->addCountry($countryData[0], $countryData[1]);
$interpreter = new Interpreter();
$interpreter->unstrict();
$interpreter->addObserver(function (array $columns) use($country) {
$state = $this->geoBuilder->addState($country, $columns[4], $columns[3]);
$province = $this->geoBuilder->addProvince($state, $columns[6], $columns[5]);
$city = $this->geoBuilder->addCity($province, $columns[2]);
$this->geoBuilder->addPostalCode($city, $columns[1]);
});
$config = new LexerConfig();
$config->setDelimiter("\t");
$lexer = new Lexer($config);
$started = new DateTime();
$output->writeln('<header>[Geo]</header> <body>Starting file processing</body>');
$lexer->parse($file->getRealpath(), $interpreter);
$finished = new DateTime();
$elapsed = $finished->diff($started);
$output->writeln('<header>[Geo]</header> <body>File processed in ' . $elapsed->format('%s') . ' seconds</body>');
return $country;
}
