public function substract($rest, $unit)
{
$lemmatizator = new Lemmatizator();
$norm_rest = str_replace(" - FORMS", "", $lemmatizator->makeUnitNormal($rest));
$norm_unit = str_replace(" - FORMS", "", $lemmatizator->makeUnitNormal($unit));
$rest_arr_normals = array_diff(explode(" ", $norm_rest), explode(" ", $norm_unit));
$rest_arr = array_combine(explode(" ", $norm_rest), explode(" ", $rest));
$new_rest = "";
foreach ($rest_arr_normals as $norm) {
$new_rest .= " " . $rest_arr[$norm];
}
return $new_rest;
}
public function combineCommonUnits($results, $units_set, $urls_count)
{
$lemmatizator = new Lemmatizator();
$combined_result = [];
$entries_sums = [];
//is neccessary for bodies counting
$zones = $this->allZones();
$queries = array_keys($results);
$counter = new Counter();
foreach ($zones as $zone) {
foreach ($units_set as $unit) {
foreach ($queries as $query) {
if (isset($results[$query][$zone][$unit])) {
$combined_result[$zone][$unit]['entries'][] = $results[$query][$zone][$unit];
//count sums for bodies counting
if ($zone != 'title') {
//titles are not included in bodies counting
foreach ($results[$query][$zone][$unit] as $url_num => $entry_count) {
$entries_sums[$unit][$query][$url_num] = isset($entries_sums[$unit][$query][$url_num]) ? $entries_sums[$unit][$query][$url_num] + $entry_count : $entry_count;
}
}
//get the averages array only for those sets which has been realy counted for entries
$combined_result[$zone][$unit]['average'][] = $counter->getAverage($results[$query][$zone][$unit]);
} else {
//make the false sets of entries for those which have not been realy counted for entries
$combined_result[$zone][$unit]['entries'][] = array_fill(0, $urls_count[$query], 0);
}
}
}
}
// correct the extra FORMs
//========================
//get the list of 'asis' units from units list
//use the fact that we need those which have not no FORMs form right after the asis form
$asis_to_correct_set = [];
foreach ($units_set as $key => $unit) {
if (strpos($unit, 'FORMS') === false && (!isset($units_set[$key + 1]) || strpos($units_set[$key + 1], 'FORMS') === false)) {
$asis_to_correct_set[] = $unit;
}
}
//get the list of FORMs units that should be corrected
//to do this proceed each from asis list with makeUnitNormal function
$forms_to_correct_set = [];
foreach ($asis_to_correct_set as $asis) {
$asis_normal = $lemmatizator->makeUnitNormal($asis);
$asis_normal .= " - FORMS";
$forms_to_correct_set[$asis_normal][] = $asis;
}
//make the array FORMs-unit => all it's asis units
foreach (array_keys($forms_to_correct_set) as $asis_normal) {
$more_key = array_search($asis_normal, $units_set);
array_unshift($forms_to_correct_set[$asis_normal], $units_set[$more_key - 1]);
}
//prepare factors and divider data for each FORMs-unit (with zone)
$correct_factors = [];
foreach (array_keys($forms_to_correct_set) as $form) {
foreach ($zones as $zone) {
$correct_factors[$zone][$form]['divider'] = count($combined_result[$zone][$form]['average']);
foreach ($forms_to_correct_set[$form] as $asis) {
$correct_factors[$zone][$form]['factors'][$asis] = $correct_factors[$zone][$form]['divider'] - count($combined_result[$zone][$asis]['average']);
}
}
}
//not the end of correctness procedure -> to be continued
//=======================================================
//start the combine averages procedure
foreach ($zones as $zone) {
foreach ($units_set as $unit) {
$entries_flattened = [];
foreach ($combined_result[$zone][$unit]['entries'] as $arr) {
$entries_flattened = array_merge($entries_flattened, $arr);
}
// file_put_contents(storage_path("tmp/results.txt"), print_r($combined_result['plain']['КОШКА - FORMS'], true),FILE_APPEND);//FOR TESTING !!!
$combined_result[$zone][$unit]['average'] = $counter->getAverage($combined_result[$zone][$unit]['average']);
$combined_result[$zone][$unit]['entries'] = $entries_flattened;
}
}
//end of combine averages procedure
//==================================
//continue the correctness procedure
//get the amendment for each zone-FORMSunit
foreach (array_keys($forms_to_correct_set) as $form) {
foreach ($zones as $zone) {
$dividend = 0;
foreach ($forms_to_correct_set[$form] as $asis) {
$summand = $correct_factors[$zone][$form]['factors'][$asis] * $combined_result[$zone][$asis]['average'];
$dividend += $summand;
}
$correct_factors[$zone][$form]['amendment'] = -1.0 * round($dividend / $correct_factors[$zone][$form]['divider'], 0, PHP_ROUND_HALF_UP);
}
}
//correct the averages in combined result
foreach ($correct_factors as $zone => $form_array) {
foreach ($form_array as $form => $factors_array) {
$combined_result[$zone][$form]['average'] += $factors_array['amendment'];
if ($combined_result[$zone][$form]['average'] < 0) {
$combined_result[$zone][$form]['average'] = 0;
}
}
}
//start the bodies count procedure
//================================
$bodies = [];
foreach ($entries_sums as $unit => $query_sums) {
$bodies[$unit] = [];
foreach ($query_sums as $query => $sums) {
$bodies[$unit] = array_merge($bodies[$unit], $sums);
}
$bodies[$unit] = $counter->getAverage($bodies[$unit]);
// $body_amendment = 0;
foreach ($zones as $zone) {
if ($zone != 'title') {
// $bodies[$unit] = $bodies[$unit];
$bodies[$unit] = $bodies[$unit] - $combined_result[$zone][$unit]['average'];
}
}
//get the average amendment from all zones
foreach ($correct_factors as $zone => $form_array) {
if (isset($correct_factors[$zone][$unit]) && $zone != 'title') {
$body_amendment = isset($body_amendment) ? $body_amendment += $correct_factors[$zone][$unit]['amendment'] : $correct_factors[$zone][$unit]['amendment'];
}
}
// $body_amendment = $body_amendment / 4.0;
if (isset($body_amendment)) {
$body_amendment = round($body_amendment, 0, PHP_ROUND_HALF_UP);
//average for 4 zones w/o title
$bodies[$unit] += $body_amendment;
if ($bodies[$unit] < 0) {
$bodies[$unit] = 0;
}
}
//correct the body amount with average amendment
}
return [$combined_result, $bodies];
}
public function handle_referencies(Request $request, Reference $reference, Lemmatizator $lemmatizator, Storekeeper $storekeeper)
{
$this->authorize('client-usage', $request->user());
$raw_data = $request->all();
// file_put_contents("/var/www/test.txt", print_r($raw_data,true));
// validates if there is any notEmpty queries in the set
// validates if the sample url is valid
// validates if project_id or new project name is present
if (isset($raw_data['ref'])) {
$raw_data['ref'] = $this->removeSecondScreenDuplicates($raw_data['ref']);
}
if (!empty($raw_data['ref'])) {
$raw_data['ref'] = $this->removeQueriesWithNoRefs($raw_data['ref']);
}
if (isset($raw_data['sample_url'])) {
$raw_data['sample_url'] = trim($raw_data['sample_url']);
}
$validator = Validator::make($raw_data, ['sample_url' => 'required_without:sample_html|url', 'sample_html' => 'required_without:sample_url|mimes:html,htm,htmls,odt,oth,ott,doc,dot,docx,dotx', 'ref' => 'required', 'query' => 'required_without:project_id', 'project_name' => 'required_without:project_id|max:20|alpha_num_dash_lat', 'project_id' => 'required_without:project_name']);
//validates if mentioned only one of url or local file for sample
$validator->after(function ($validator) {
if (Input::get('sample_url') != '' && Input::hasFile('sample_html')) {
$validator->errors()->add('sample_url', 'Choose smth one between remote url or local file');
}
});
//make $inputed_responce using ref_dirty data to fill form fields with already edited by user data if validation fails
if ($validator->fails()) {
$id = isset($raw_data['project_id']) ? $raw_data['project_id'] : null;
$inputed_responce = [];
foreach (Input::get('ref_dirty') as $query => $dirty_refs) {
if ($dirty_refs) {
$query = $query == '0' ? "" : $query;
$inputed_responce[$query] = $dirty_refs;
} else {
$inputed_responce = [];
break;
}
}
return redirect()->action('WebsiteController@referencies_handler', [$id])->withErrors($validator)->withInput(Input::except('ref', 'query', 'ref_dirty'))->with('inputed_responce', $inputed_responce);
}
if (isset($raw_data['last'])) {
$last = $raw_data['last'];
} else {
$last = null;
}
//get the project_id from INPUT or set the new one
if (isset($raw_data['project_id'])) {
$project_id = strip_tags($raw_data['project_id']);
} else {
// mkdir empty dirs which are necessary for query handling
// initialize project
$new_project = $storekeeper->initializeStoring($raw_data['project_name'], $raw_data['query']);
$project_id = $new_project->id;
}
//set exact flag to the project if it is checked
if (isset($raw_data['exact'])) {
$this_project = Project::find($project_id);
$this_project->exact = true;
$this_project->save();
}
//store additional queries to file at project dir
if (isset($raw_data['dop_queries'])) {
$raw_query['dop_queries'] = strip_tags($raw_data['dop_queries']);
$dop_queries_array = preg_split("/[,;\\.]*[\t\n\f\r]+\\s*/", trim($raw_data['dop_queries']), -1, PREG_SPLIT_NO_EMPTY);
} else {
$dop_queries_array = [];
}
//check if dop_queries duplicate main queries
foreach ($dop_queries_array as $num => $dop_query) {
$dop_query = $this->clean_query($dop_query);
if (in_array($dop_query, array_keys($raw_data['ref']))) {
unset($dop_queries_array[$num]);
}
}
if (!empty($dop_queries_array)) {
foreach ($dop_queries_array as $key => $raw_query) {
$dop_query = trim(array_keys($lemmatizator->splitToWords($raw_query))[0]);
$dop_queries_array[$key] = $dop_query;
}
$dop_queries = implode("\n", $dop_queries_array);
$project_dir = Project::where('id', $project_id)->first()->dir;
file_put_contents(storage_path("tmp/" . $project_dir . "/reports/dop_queries.txt"), $dop_queries);
}
// obtain an array of referencies instead of text from textarea
// text is splitted against different white-spaces and the commas
foreach ($raw_data['ref'] as $ref => $urls_text) {
//splitToWords ЗДЕСЬ используется только для "очистки" запроса
//меняется е на ё, убираются стоп-слова и т.д.
//splitToWords возвращает массив, в котором ключами являются тексты запросов, а значениями массивы составляющих их слов
//используем splitToWords для того, чтобы "очистка" была одинаковой на всех этапах обработки
$query = array_keys($lemmatizator->splitToWords($ref))[0];
//заменяем запрос на очищенный в переменной last
if ($query != $ref && !is_null($last)) {
$last[$query] = $last[$ref];
unset($last[$ref]);
}
//make array of urls instead of text
$data[$query] = preg_split("/[\\s,]+/", $urls_text);
}
$responce = [];
$refs_handled = [];
foreach ($data as $query => $urls_array) {
foreach ($urls_array as $key => $url) {
//remove empty strings if any
if ($url == "") {
unset($data[$query][$key]);
}
}
}
if ($raw_data['sample_url'] || Input::hasFile('sample_html')) {
if ($raw_data['sample_url']) {
$src_type = 'http';
$sample_url = Input::get('sample_url');
} else {
$src_type = 'local';
$sample_url = Input::file('sample_html')->getRealPath();
}
//get html of sample from file or from www
try {
$sample_content = $reference->getHtml($sample_url, 'sample', 0, $project_id, $src_type);
$sample_cleaned_content = $reference->removeGarbage($sample_content, $project_id, "sample_0.html");
} catch (\Exception $e) {
$mes = $e->getMessage();
$inputed_responce = [];
foreach (Input::get('ref') as $query => $refs) {
$query = $query == '0' ? "" : $query;
$inputed_responce[$query] = $refs;
}
return redirect()->action('WebsiteController@referencies_handler', [$project_id])->with('alert', "Cannot fetch html from the sample url: " . $sample_url . "\n" . $mes)->withInput(Input::except('ref', 'query'))->with('inputed_responce', $inputed_responce);
}
} else {
$sample_file_name = 'sample_0.html';
$sample_cleaned_content = '';
}
// return '<pre>'.print_r($last,true).'</pre>';
$this->dispatch(new HandleDataJob($sample_cleaned_content, $data, $project_id, $last));
return redirect()->action('UserController@show_user', [$request->user()->id])->with('info', "Your Project is in process. Gather report when it'll appear in reports column.");
}