get_popular_terms($terms_limit);
$words_counter = 0;
$i = 0;
$terms_in_limit = array();
$lang_dir = api_get_path(SYS_LANG_PATH);
$arch_dir = api_get_path(SYS_ARCHIVE_PATH);
/**
* Code run
*/
foreach ($terms as $row) {
if ($words_counter > 10000) { break; }
$words = str_word_count(get_lang($row['term_name'],null,$orig_lang));
$words_counter += $words;
$terms_in_limit[$row['term_name']] = $i;
//echo "Term ".$row['term_name']." is '".get_lang($row['term_name'],null,$orig_lang)."' which means $words words
\n";
//if ($words_counter%1000 >= 0) {
//echo "Reached $words_counter words at term $i (".$row['term_name']." used ".$row['term_count']." times)...
\n";
//}
$i++;
}
//echo $words_counter.'
';
echo "Reached ".count($terms_in_limit)." terms for the $words_counter most-used words
\n";
echo "Scanning English files, trying to find these terms...
\n";
if (!is_dir($arch_dir.'/langstats')) {
mkdir($arch_dir.'/langstats');
mkdir($arch_dir.'/langstats/'.$orig_lang);
}
$list_files = scandir($lang_dir.'/'.$orig_lang);
$j = 1;
$terms_found = array();
$words_found = 0;
$global_var = array(); //keep the combination of all vars
$terms_in_limit = array_flip($terms_in_limit);
foreach ($list_files as $file) {
if (substr($file,0,1) == '.') {continue;}
//echo "'".substr($file,0,-8)."',
"; //print in a PHP array format
$vars = file($lang_dir.'/'.$orig_lang.'/'.$file);
$local_var = array();
$file_string = '0) {
//echo $var[1]."
";
if (in_array(substr($var[1],1),$terms_in_limit)) {
//echo "Var ".$var[1]." was in the limit
";
$local_var[$var[1]] = $line;
$file_string .= $line;
$terms_found[] = substr($var[1],1); //e.g. store Tools
$words_found += str_word_count(get_lang($var[1],null,$orig_lang));
} elseif (in_array(substr($var[1],5),$terms_in_limit)) {
//echo "Var ".$var[1]." was in the limit
";
$local_var[$var[1]] = $line;
$file_string .= $line;
$terms_found[] = substr($var[1],5); //e.g. store langTools
$words_found += str_word_count(get_lang(substr($var[1],5),null,$orig_lang));
} //else do not care
}
}
echo "Writing ".$arch_dir.'/langstats/'.$orig_lang.'/'.$file."
\n";
file_put_contents($arch_dir.'/langstats/'.$orig_lang.'/'.$file,$file_string);
$global_var += $local_var;
};
$terms_diff = count($global_var)-count($terms_in_limit);
echo count($global_var)." terms found in English files (summing up to $words_found words). Some terms ($terms_diff in this case) might have appeared in two different files
";
/**
* Display results
*/
echo "Difference between filtered and found in English:
";
//print_r($terms_found);
echo "".print_r(array_diff($terms_in_limit,$terms_found),1)."
";
echo "#";