public function makeZonesFilesArrays($nodes_arrays, $project)
{
$zones = $this->allZones();
$storekeeper = new Storekeeper();
$files_arrays = [];
//empty array for storing the filenames according to zones (key->(string)zone_name => value->array with the filenames set)
$paths = $this->getZonesPaths($zones, $project);
foreach ($nodes_arrays as $query_text => $url) {
if (!is_array($url)) {
// ????????? TODO: SAVE THE ERROR HERE
$storekeeper->storeErrors([$url], $project_id);
unset($nodes_arrays[$query_text]);
}
}
// make the empty arrays for filenames for each zone
foreach ($nodes_arrays as $query_text => $files) {
foreach ($files as $file => $nodes) {
$filename = str_replace(".html", ".txt", $file);
// add the filename to zone-filename array
foreach ($zones as $zone) {
$files_arrays[$zone][$query_text][] = $filename;
if (!in_array($zone, ['a', 'h1', 'title'])) {
file_put_contents($paths[$zone] . $filename, "");
// clean content of the storage file
}
}
//add the text content of each node to the corresponding ZONE file
$this->separatePlainFromFragment($nodes, $paths, $filename);
}
}
// make CLEANED files for each zone, put them to zone-dir, make the array of filenames
// (remove punctuation marks, remove stop-words, remove extra whitespaces)
$lemmatizator = new Lemmatizator();
foreach ($zones as $zone) {
foreach ($files_arrays[$zone] as $query => $files_array) {
foreach ($files_array as $filename) {
$content = file_get_contents($paths[$zone] . $filename);
$words[$zone][] = str_word_count($content, 0, "АаБбВвГгДдЕеЁёЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЫыЬьЭэЮюЯя*");
$short_content = $lemmatizator->makeTextShort($content);
file_put_contents($paths[$zone] . $filename, $short_content);
}
}
}
return [$files_arrays, $words];
}
