How to remove elements from an associative array in PHP - php

Question

I am trying to read all the words in HTML documents locally. I have a loop which does it for me. I have created an array which holds the unwanted characters. I do not want those special unwanted characters to be in my word array. I have tried the following code but nothing changed.
$rii = new RecursiveIteratorIterator(new RecursiveDirectoryIterator('fulltext/course'));
$fulltext_course_files = array();
$unwantedChars = array('', ' ', '"', '!', '\'', '+', '%', '&', '/', '(', ')', '=',
'*', '.', ',', '?', '-', '_', ':', ';', '\\');
foreach ($rii as $f) {
if ($f->isDir()) {
continue;
} else {
$st = strip_tags(strtolower(file_get_contents($f)));
$swc = deleteStopWords(str_word_count($st, 1));
if (!in_array($st, $unwantedChars)) {
$fulltext_course_files[$f->getFilename()] = array_count_values($swc);
}
}
}
I still see dashes, empties ('') when I var_dump($arr);
array (size=230)
'4.html' =>
array (size=50)
'-' => int 7 ??
'cs' => int 1
'page' => int 1
'systems' => int 2
'programming' => int 1
'' => int 12 ??
'operating' => int 2
...
What can I do in order to remove the elements pointed with ??.
Edit 1
Better solution is preventing unwanted characters entering to the array as #David suggests. I have tried to change the if condition from
if (!in_array($st, $unwantedChars))
to
if (!in_array($f->getFilename(), $unwantedChars))
nothing changed. Unwanted keys still there.
Edit 2
I have tried the following also:
foreach ($fulltext_course_files as $key => $val) {
if (in_array($key, $unwantedChars)) {
unset($fulltext_course_files[$key] );
}
}
Again, no help!

Answer 1

You can use unset: http://php.net/manual/en/function.unset.php
unset($array['mykey']);

Answer 2

not sure what the $f->getFilename() does, but would it not be easier to test it against your characters?
if(!in_array($f->getFilename(), $unwantedChars) {
$fulltext_course_files[$f->getFilename()] = array_count_values($swc);
}

Answer 3

Instead of using in_array to search for unwanted characters, you could store them all in a string, and use strchr on it: it's basically equivalent to what you wrote, but with a string for storage rather than an array, which should be faster. That said...
My guess is that the unwanted characters still remaining in your final array are actually characters graphically similars to the normal punctuation chars, but with a different code point (the integer value corresponding to that character). Could it be that your document uses an encoding with several different dashes and double quote characters, like say, utf-8? If that's the case, you'll have a hard time filtering all the noise to keep only alphabet characters that way. However, if you use a white listing scheme (ie, check for good characters rather than bad ones), perhaps you'd be able to only keep those chars you're interested in. Luckily, there are functions to help you doing that: ctype_alpha for only alphabet, and ctype_alnum for alphanumeric ones. The Ctype extension they belong to is usually enabled in most php installation.
Here's a quick implementation:
function get_word_count($content){
$words = array();
$b = 0;
$len = strlen($content);
for ($i = 0; $i < $len; $i++){
$c = $content[$i];
if (!ctype_alnum($c)){
if ($b < $i){
$w = strtolower(substr($content, $b, $i - $b));
if (isset($words[$w]))
$words[$w]++;
else $words[$w] = 1;
}
$b = $i + 1;
}
}
return $words;
}
Beware that:
because it only accepts alphanumeric characters, you will not be able to index non english words.
even in that context, there are compound words you"d probably want to consider as one, for instance you're or step-wise. This function will not help you with that. If you need a more robust approach, I suggest that you look into existing natural language processing toolkits for PHP (your search engine of choice will report severals projects).