Online PHP editor | perf for SEdf7

<?php /** * Over-engineered solution to most capitalisation issues. * * Version 1.0 */ class str { /** * Words or abbreviations that should always be all uppercase */ const ALL_UPPERCASE = [ "UK", "VAT", ]; /** * Words or abbreviations that should always be all lowercase */ const ALL_LOWERCASE = [ "and", "as", "by", "in", "of", "or", "to", ]; /** * Honorifics that only contain vowels. * */ const CONSONANT_ONLY_HONORIFICS = [ # English "Mr", "Mrs", "Ms", "Dr", "Br", "Sr", "Fr", "Pr", "St", # Afrikaans "Mnr", ]; /** * Surname prefixes that should be lowercase, * unless not following another word (firstname). */ const SURNAME_PREFIXES = [ "de la", "de las", "van de", "van der", "vit de", "von", "van", "del", "der", ]; /** * Capitalises every (appropriate) word in a given string. * * @param string|null $string * * @return string|null */ public static function capitalise(?string $string): ?string { if(!$string){ return $string; } # Strip away multi-spaces $string = preg_replace("/\s{2,}/", " ", $string); # Ensure there is always a space after a comma $string = preg_replace("/,([^\s])/", ", $1", $string); # A word is anything separated by spaces or a dash $string = preg_replace_callback("/([^\s\-\.]+)/", function($matches){ # Make the word lowercase $word = mb_strtolower($matches[1]); # If the word needs to be all lowercase if(in_array($word, self::ALL_LOWERCASE)){ return strtolower($word); } # If the word needs to be all uppercase if(in_array(mb_strtoupper($word), self::ALL_UPPERCASE)){ return strtoupper($word); } # Create a version without diacritics $transliterator = \Transliterator::createFromRules(':: Any-Latin; :: Latin-ASCII; :: NFD; :: [:Nonspacing Mark:] Remove; :: Lower(); :: NFC;', \Transliterator::FORWARD); $ascii_word = $transliterator->transliterate($word); # If the word contains non-alpha characters (numbers, &, etc), with exceptions (comma, '), assume it's an abbreviation if(preg_match("/[^a-z,']/i", $ascii_word)){ return strtoupper($word); } # If the word doesn't contain any vowels, assume it's an abbreviation if(!preg_match("/[aeiouy]/i", $ascii_word)){ # Unless the word is an honorific if(!in_array(ucfirst($word), self::CONSONANT_ONLY_HONORIFICS)){ return strtoupper($word); } } # If the word contains two of the same vowel and is 3 characters or fewer, assume it's an abbreviation if(strlen($word) <= 3 && preg_match("/([aeiouy])\1/", $word)){ return strtoupper($word); } # Ensure O'Connor, L'Oreal, etc, are double capitalised, with exceptions (d') if(preg_match("/\b([a-z]')(\w+)\b/i", $word, $match)){ # Some prefixes (like d') are not capitalised if(in_array($match[1], ["d'"])){ return $match[1] . ucfirst($match[2]); } # Otherwise, everything is capitalised return strtoupper($match[1]) . ucfirst($match[2]); } # Otherwise, return the word with the first letter (only) capitalised return ucfirst($word); //The most common outcome }, $string); # Cater for the Mc prefix $pattern = "/(Mc)([b-df-hj-np-tv-z])/"; //Mc followed by a consonant $string = preg_replace_callback($pattern, function($matches){ return "Mc" . ucfirst($matches[2]); }, $string); # Cater for Roman numerals (need to be in all caps) $pattern = "/\b((?<![MDCLXVI])(?=[MDCLXVI])M{0,3}(?:C[MD]|D?C{0,3})(?:X[CL]|L?X{0,3})(?:I[XV]|V?I{0,3}))\b/i"; $string = preg_replace_callback($pattern, function($matches){ return strtoupper($matches[1]); }, $string); # Cater for surname prefixes (must be after the Roman numerals) $pattern = "/\b (".implode("|", self::SURNAME_PREFIXES).") \b/i"; //A surname prefix, bookended by words $string = preg_replace_callback($pattern, function($matches){ return strtolower(" {$matches[1]} "); }, $string); # Cater for ordinal numbers $pattern = "/\b(\d+(?:st|nd|rd|th))\b/i"; //A number suffixed with an ordinal $string = preg_replace_callback($pattern, function($matches){ return strtolower($matches[1]); }, $string); # And we're done return $string; } } $complicated_names = " DONALD MCDONALD SINEAD O'CONNOR JOHAN VAN ZYL OSCAR DE LA HOYA P.F. CHANG KFC ST. JOHN DR ZEUZ PROF. GREEN VAN DER BERG THE 3RD SÃO JOÃO DOS SANTOS KING HENRY VII KUJE'S HIGH,ROAD FLUG-HAFEN FLUGIG-O'DONNALD MARY O'CALLAHAN JOHN O'DONALD THE O'CALLAHAN-O'DONALD RESIDENCE 2ND NOVEMBER STREET The 15th king of scotland FCT MICHAEL VIVA GINA C.A. KOTOR DUTCH NAMES van der vaart van vollenhoven van 't zandt van het zand el hamdoie van der Rooi-van Velzen Zuidewijn - van rooien teggelen onder t boven guido op 't drooge friso van drooge Zuidewijn - van rooien teggelen onder t boven ZUID-HOLLAND 's hertogen-bosch De Rooi Van Zuidewijn van onder Van Der Wijk-Zeewuster de Vries-van der Leest Den Oudsten - van 't Veldt Hare Koninklijke Hoogheid Alexia Juliana Marcela Laurentien Prinses der Nederlanden, Prinses van Oranje-Nassau Hare Koninklijke Hoogheid Máxima, Prinses der Nederlanden, Prinses van Oranje-Nassau, Mevrouw van Amsberg van Lippe-Biesterfeld van Vollenhoven "; var_dump(str::capitalise($complicated_names));

Version	System time (s)	User time (s)	Memory (MiB)
8.5.0	0.012	0.089	26.10
8.4.15	0.001	0.001	14.05
8.4.14	0.018	0.151	21.21
8.4.13	0.008	0.100	23.07
8.4.12	0.010	0.119	25.93
8.4.11	0.014	0.102	26.21
8.4.10	0.010	0.133	21.70
8.4.9	0.014	0.126	25.76
8.4.8	0.011	0.111	24.43
8.4.7	0.011	0.119	22.51
8.4.6	0.010	0.126	23.91
8.4.5	0.012	0.123	21.66
8.4.4	0.013	0.126	21.52
8.4.3	0.016	0.143	22.34
8.4.2	0.010	0.120	21.47
8.4.1	0.007	0.109	23.25
8.3.28	0.015	0.090	24.32
8.3.27	0.014	0.141	20.51
8.3.26	0.012	0.113	20.92
8.3.25	0.008	0.115	22.32
8.3.24	0.014	0.118	20.31
8.3.23	0.012	0.123	20.20
8.3.22	0.012	0.108	22.31
8.3.21	0.013	0.118	20.52
8.3.20	0.010	0.135	20.40
8.3.19	0.013	0.142	22.64
8.3.18	0.008	0.114	21.98
8.3.17	0.007	0.126	20.70
8.3.16	0.023	0.156	21.99
8.3.15	0.003	0.115	24.01
8.3.14	0.017	0.159	20.34
8.3.13	0.010	0.106	22.02
8.3.12	0.010	0.113	22.48
8.3.11	0.010	0.111	21.94
8.3.10	0.007	0.109	20.41
8.3.9	0.020	0.150	26.77
8.3.8	0.003	0.143	20.23
8.3.7	0.010	0.219	20.31
8.3.6	0.020	0.262	20.48
8.3.5	0.010	0.231	23.71
8.3.4	0.010	0.134	25.65
8.3.3	0.007	0.120	22.67
8.3.2	0.000	0.096	25.23
8.3.1	0.016	0.052	25.58
8.3.0	0.017	0.071	26.16
8.2.29	0.015	0.116	24.10
8.2.28	0.014	0.127	22.41
8.2.27	0.007	0.129	20.64
8.2.26	0.007	0.120	23.76
8.2.25	0.007	0.106	20.52
8.2.24	0.003	0.142	20.88
8.2.23	0.003	0.112	22.58
8.2.22	0.010	0.106	24.06
8.2.21	0.003	0.114	26.77
8.2.20	0.010	0.135	20.39
8.2.19	0.013	0.223	20.51
8.2.18	0.010	0.264	25.92
8.2.17	0.003	0.128	22.66
8.2.16	0.013	0.114	23.95
8.2.15	0.010	0.073	25.66
8.2.14	0.007	0.071	25.75
8.2.13	0.013	0.085	26.16
8.2.12	0.010	0.059	23.76
8.2.11	0.010	0.064	23.91
8.2.10	0.003	0.072	22.94
8.1.33	0.012	0.112	19.96
8.1.32	0.013	0.121	19.81
8.1.31	0.003	0.112	22.11
8.1.30	0.010	0.105	21.99
8.1.29	0.010	0.133	30.84
8.1.28	0.013	0.199	25.92
8.1.27	0.007	0.060	25.78
8.1.26	0.007	0.062	26.35
8.1.25	0.007	0.061	28.09
8.1.24	0.003	0.072	22.85
8.1.23	0.003	0.072	24.02
8.1.9	0.008	0.102	21.63
7.4.30	0.006	0.102	20.79
7.4.27	0.005	0.089	20.79
7.3.33	0.008	0.103	20.62
7.2.34	0.005	0.094	20.72
7.1.33	0.010	0.095	21.20