exchanging values of a variable, by values of an a

问题:

I have a code that compares the output with the values of the array, and only terminates the operation with words in the array:

First code(just a example)

$myVar = 'essa pizza é muito gostosa, que prato de bom sabor';
$myWords=array(
    array('sabor','gosto','delicia'),
    array('saborosa','gostosa','deliciosa'),
);

foreach($myWords as $words){
    shuffle($words); // randomize the subarray
    // pipe-together the words and return just one match
    if(preg_match('/\K\b(?:'.implode('|',$words).')\b/',$myVar,$out)){
        // generate "replace_pair" from matched word and a random remaining subarray word
        // replace and preserve the new sentence
        $myVar=strtr($myVar,[$out[0]=>current(array_diff($words,$out))]);
    }
}
echo $myVar;

---

My Question:

I have a second code, which is not for rand/shuffle(I do not want rand, I want precision in substitutions, I always change column 0 through 1), is to always exchange the values:

// wrong output: $myVar = "minha irmã alanné é not aquela blnode, elere é a bom plperito";
$myVar = "my sister alannis is not that blonde, here is a good place";
$myWords=array(array("is","é"),
    array("on","no"),
    array("that","aquela"),
    //array("blonde","loira"),
    //array("not","não"),
    array("sister","irmã"), 
    array("my","minha"),
    //array("nothing","nada"),
    array("myth","mito"),
    array("he","ele"),
    array("good","bom"),
    array("ace","perito"),
   // array("here","aqui"), //if [here] it does not exist, it is not to do replacement from the line he=ele = "elere" non-existent word  
); 
$replacements = array_combine(array_column($myWords,0),array_column($myWords,1));
$myVar = strtr($myVar,$replacements);
echo $myVar;
// expected output:  minha irmã alannis é not aquela blonde, here é a bom place
//  avoid replace words slice!

expected output: minha irmã alannis é not aquela blonde, here é a bom place

    //  avoid replace words slice! always check if the word exists in the array before making the substitution.

alanné, blnode, elere, plperito

it examines whether the output will be of real words, which exist in the array myWords, this avoids typing errors like:

that 4 words is not an existent words, a writing error. how do you do that for the second code?

   in short, the exchange must be made by a complete word / key, an existing word. and not create something strange using slices of keywords!

回答1:

My previous method was incredibly inefficient. I didn't realize how much data you were processing, but if we are upwards of 4000 lines, then efficiency is vital (I think I my brain was stuck thinking about strtr() related processing based on your previous question(s)). This is my new/improved solution which I expect to leave my previous solution in the dust.

Code: (Demo)

$myVar="My sister alannis Is not That blonde, here is a good place. I know Ariane is not MY SISTER!";
echo "$myVar\n";

$myWords=array(
    array("is","é"),
    array("on","no"),
    array("that","aquela"),
    array("sister","irmã"), 
    array("my","minha"),
    array("myth","mito"),
    array("he","ele"),
    array("good","bom"),
    array("ace","perito"),
    array("i","eu")  // notice I must be lowercase
);
$translations=array_combine(array_column($myWords,0),array_column($myWords,1));  // or skip this step and just declare $myWords as key-value pairs

// length sorting is not necessary
// preg_quote() and \Q\E are not used because dealing with words only (no danger of misinterpretation by regex)

$pattern='/\b(?>'.implode('|',array_keys($translations)).')\b/i';  // atomic group is slightly faster (no backtracking)
/* echo $pattern;
   makes: /\b(?>is|on|that|sister|my|myth|he|good|ace)\b/i
   demo: https://regex101.com/r/DXTtDf/1
*/
$translated=preg_replace_callback(
    $pattern,
    function($m)use($translations){  // bring $translations (lookup) array to function
        $encoding='UTF-8';  // default setting
        $key=mb_strtolower($m[0],$encoding);  // standardize keys' case for lookup accessibility
        if(ctype_lower($m[0])){ // treat as all lower
            return $translations[$m[0]];
        }elseif(mb_strlen($m[0],$encoding)>1 && ctype_upper($m[0])){  // treat as all uppercase
            return mb_strtoupper($translations[$key],$encoding);
        }else{  // treat as only first character uppercase
            return mb_strtoupper(mb_substr($translations[$key],0,1,$encoding),$encoding)  // uppercase first
                  .mb_substr($translations[$key],1,mb_strlen($translations[$key],$encoding)-1,$encoding);  // append remaining lowercase
        }
    },
    $myVar);

echo $translated;

Output:

My sister alannis Is not That blonde, here is a good place. I know Ariane is not MY SISTER!
Minha irmã alannis É not Aquela blonde, here é a bom place. Eu know Ariane é not MINHA IRMÃ!

This method:

• does only 1 pass through $myVar, not 1 pass for every subarray of $myWords.
• does not bother with sorting the lookup array ($myWords/$translations).
• does not bother with regex escaping (preg_quote()) or making pattern components literal (\Q..\E) because only words are being translated.
• uses word boundaries so that only complete word matches are replaced.
• uses an atomic group as a micro-optimization which maintains accuracy while denying backtracking.
• declares an $encoding value for stability / maintainability / re-usability.
• matches with case-insensitivity but replaces with case-sensitivity ...if the English match is:
  1. All lowercase, so is the replacement
  2. All uppercase (and larger than a single character), so is the replacement
  3. Capitalized (only first character of multi-character string), so is the replacement

回答2:

Unfortunately strtr() is the wrong tool for this job because it is "word boundary ignorant". To target whole words there is no simpler way that using a regex pattern with word boundaries.

Furthermore, to ensure that longer strings are match prior to shorter strings (strings that may exist inside other strings), you must sort $myWords by string length (descending / longest to shortest; using the multi-byte version only if necessary).

Once the array of words is sorted and converted to individual regex patterns, you can feed the arrays into the pattern and replace parameters of preg_replace().

Code (Demo)

$myVar = "my sister alannis is not that blonde, here is a good place";
$myWords=array(
    array("is","é"),
    array("on","no"),
    array("that","aquela"),
    array("sister","irmã"), 
    array("my","minha"),
    array("myth","mito"),
    array("he","ele"),
    array("good","bom"),
    array("ace","perito")
); 
usort($myWords,function($a,$b){return mb_strlen($b[0])<=>mb_strlen($a[0]);});  // sort subarrays by first column multibyte length
// remove mb_ if first column holds no multi-byte characters.  strlen() is much faster.

foreach($myWords as &$words){
    $words[0]='/\b'.$words[0].'\b/i';  // generate patterns using search word, word boundaries, and case-insensitivity
}

//var_export($myWords);
//var_export(array_column($myWords,0));
//var_export(array_column($myWords,1));

$myVar=preg_replace(array_column($myWords,0),array_column($myWords,1),$myVar);
echo $myVar;

Output:

minha irmã alannis é not aquela blonde, here é a bom place

---

What this doesn't do is appreciate the case of the matched substrings. I mean, my and My will both be replaced by minha.

To accommodate different casing, you will need to use preg_replace_callback().

Here is that consideration (which handles uppercase first letter words, not ALL CAPS words):

Code (Demo) <-- run this to see the original casing preserved after the replacement.

foreach($myWords as $words){
    $myVar=preg_replace_callback(
        $words[0],
        function($m)use($words){
            return ctype_upper(mb_substr($m[0],0,1))?
                mb_strtoupper(mb_substr($words[1],0,1)).mb_strtolower(mb_substr($words[1],1)):
                $words[1];
        },
        $myVar);
}
echo $myVar;