public function ClearUpVariables($data)
{
//amend values as I wish
$data['country_code'] = strtolower($data['country_code']);
//country_slug
if (!empty($data['country_name'])) {
$slug = new SlugClass();
$data['country_slug'] = $slug->slugify($data['country_name']);
}
return $data;
}
public function ClearUpVariables($data, $hour_stamp)
{
//amend values as I wish
$data['hour_stamp'] = $hour_stamp;
$data['hour_stamp_flight_number'] = str_replace(' ', '', $hour_stamp . $data['flight_number']);
//unique column
$data['scheduled_time'] = Carbon::createFromFormat('d.m. H:i', $data['scheduled_time']);
//airline code
preg_match("|^..|", $data['flight_number'], $output_array_3);
$data['airline_code'] = strtolower($output_array_3[0]);
if ($data['airline_code'] == "ez") {
$data['airline_code'] = "u2";
}
//easyjet not using their IATA code U2 but EZY - manual correction
//terminal
preg_match("|[1-9]|", $data['terminal'], $output_array_4);
$data['terminal'] = $output_array_4[0];
//destination
$data['destination'] = strtolower($data['destination']);
//keep all such data in lowercase (because Sqlite differentiate between Capitals and lower)
preg_match("|(.*)\\((.*)\\)|", $data['destination'], $output_array_1);
if (count($output_array_1) > 1) {
$CitySlashAirport = $output_array_1[1];
$data['country_code'] = $output_array_1[2];
} else {
$CitySlashAirport = null;
$data['country_code'] = 'xx';
}
unset($data['destination']);
//in case some strange codes comes
$country = Country::where('country_code', $data['country_code']);
if (count($country) == 0) {
$newcountry = new Country();
$newcountry->country_code = $data['country_code'];
$newcountry->country_name = 'Undefined';
$newcountry->save();
}
$output_array_2 = preg_split("|\\/|", $CitySlashAirport);
$data['city'] = $output_array_2[0];
if (empty($data['city'])) {
$data['city'] = 'empty';
}
if (count($output_array_2) > 1) {
$data['airport'] = $output_array_2[1];
} else {
$data['airport'] = null;
}
//city_slug
$slug = new SlugClass();
$data['city_slug'] = $slug->slugify($data['city']);
//find if this city_slug is in cities table
$city_slug_in_cities = City::where('city_slug', $data['city_slug'])->first();
if (empty($city_slug_in_cities)) {
$city_slug_bridge = Slug::where('city_slug_dirty', $data['city_slug'])->first();
if (!empty($city_slug_bridge)) {
$data['city_slug'] = $city_slug_bridge->city_slug_clean;
echo $data['city_slug'];
}
}
$data = array_map('trim', $data);
//trim values
return $data;
}
public function Geocoding()
{
//take addresses not yet scraped in the past or incorrectly scraped, geocode scraping, save to cities table
$addressesWithoutGeolocation = Address::whereNull('city_id')->get();
$GeoScraped = new GoogleMapsScraper();
foreach ($addressesWithoutGeolocation as $address) {
$gmaps_output = $GeoScraped->Geocode($address->raw_input);
//get gmaps_status
$dataArray['gmaps_status'] = null;
$dataArray = parent::ExtractWithXpath($gmaps_output, $this->GeoPatterns, 'text', 'first', 'xml');
//get address_components in array
$componentsArray = array();
$componentsArray = parent::ExtractWithXpath($gmaps_output, $this->AddressComponentsPattern, 'html', 'multiple', 'xml');
$cityArray['city_name'] = null;
$countryArray['country_code'] = null;
$countryArray['country_name'] = null;
//search country component in every address component, once found extract name and code
foreach ($componentsArray as $AddressComponent) {
if (preg_match("|type>country<|", $AddressComponent)) {
$AddressComponent = '<result>' . $AddressComponent . '</result>';
//get country_code, country_name
$countryArray = parent::ExtractWithXpath($AddressComponent, $this->CountryPatterns, 'text', 'first', 'xml');
} elseif (preg_match("|type>natural_feature<|", $AddressComponent)) {
$AddressComponent = '<result>' . $AddressComponent . '</result>';
//get country_code, country_name
$cityArray = parent::ExtractWithXpath($AddressComponent, $this->IslandPatterns, 'text', 'first', 'xml');
} elseif (preg_match("|type>locality<|", $AddressComponent)) {
$AddressComponent = '<result>' . $AddressComponent . '</result>';
//get country_code, country_name
$cityArray = parent::ExtractWithXpath($AddressComponent, $this->CityPatterns, 'text', 'first', 'xml');
}
}
if (empty($cityArray['city_name']) and !empty($cityArray['island'])) {
$cityArray['city_name'] = $cityArray['island'];
}
if (empty($countryArray['country_code'])) {
$countryArray['country_code'] = 'xx';
}
//if city doesn't exist, save it to cities
$city = City::firstOrNew(['city_name' => $cityArray['city_name'], 'country_code' => strtolower($countryArray['country_code'])]);
$city->gmaps_status = $dataArray['gmaps_status'];
$city->gmaps_output = $gmaps_output;
$city->country_name = $countryArray['country_name'];
//city_slug
$slug = new SlugClass();
$city->city_slug = $slug->slugify($cityArray['city_name']);
$city->save();
//save foreign key in addresses
$address->city_id = $city->id;
$address->save();
}
}
public function slugs(Request $request)
{
$fixFlightsCitiesJoins = new SlugClass();
$fixFlightsCitiesJoins->UpdateSlugTable($request);
$fixFlightsCitiesJoins->UpdateFlightsTable($request);
return redirect('scraper/slugs');
}
