PHP preg_split() - don't split spaces between ' '

Question

I have this string:

$string = "My name is Emma and i have a dillemma, what's the distance between 'New York' and 'Athene' ?";

I'm splitting this string by space and some operators(=,<,>,!=,>=,<=,<>) using this code:

$split = preg_split('/\s+|(,|[<>!]?=|<>?|>)/', $string, null, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);

For now the result of this splitting is this array:

Array
(
    [0] => My
    [1] => name
    [2] => is
    [3] => Emma
    [4] => and
    [5] => i
    [6] => have
    [7] => a
    [8] => dillemma
    [9] => ,
    [10] => what's
    [11] => the
    [12] => distance
    [13] => between
    [14] => 'New
    [15] => York'
    [16] => and
    [17] => 'Athene'
    [18] => ?
)

Now the only problem that i have is that i want the white spaces between '' to not be splitted but to remove '' after split, in this example above you can see 'New York' is splitted into:

[14] => 'New
[15] => York'

My desired outcome is:

[14] => New York

And also 'Athene', i want it to be:

[16] => Athene

So basicly the above array should look like this:

Array
(
    [0] => My
    [1] => name
    [2] => is
    [3] => Emma
    [4] => and
    [5] => i
    [6] => have
    [7] => a
    [8] => dillemma
    [9] => ,
    [10] => what's
    [11] => the
    [12] => distance
    [13] => between
    [14] => New York
    [15] => and
    [16] => Athena
    [17] => ?
)

And yes the distance between those two cities is 4,925 miles or 7925 kilometers :D

Thank you! :D

Answer 1

Regular Expression

(?:\'([^\']*[\'s]?)\'|\"([^\"]*)\")|[^\s,<>=!]+|(?:,|[<>!]?=|<>?|>)

You can see the matches here: https://regex101.com/r/LkHnHt/3

PHP Code

$text = "My name is Emma and i have a dillemma, what's the distance between 'New York' and 'Athene' ?";
preg_match_all('/(?:\'([^\']*[\'s]?)\'|\"([^\"]*)\")|[^\s,<>=!]+|(?:,|[<>!]?=|<>?|>)/', $text, $matches);
foreach (array_filter($matches[1]) as $k => $v)
    $matches[0][$k] = $v;

Results

Array
(
    [0] => My
    [1] => name
    [2] => is
    [3] => Emma
    [4] => and
    [5] => i
    [6] => have
    [7] => a
    [8] => dillemma
    [9] => ,
    [10] => what's
    [11] => the
    [12] => distance
    [13] => between
    [14] => New York pop
    [15] => and
    [16] => Athene
    [17] => ?
)

Array
(
    [0] => age
    [1] => <
    [2] => 21
    [3] => ,
    [4] => length
    [5] => >
    [6] => 10
    [7] => ,
    [8] => height
    [9] => <>
    [10] => 10
    [11] => ,
    [12] => width
    [13] => !=
    [14] => 100
    [15] => ,
    [16] => name
    [17] => =
    [18] => Emma Einarsson
    [19] => or
    [20] => it
    [21] => can
    [22] => be
    [23] => words
    [24] => time
    [25] => >=
    [26] => 10
    [27] => ,
    [28] => clouds
    [29] => <=
    [30] => 4
)

Pay attention, all the data that capture saved in the array $matches[0]

Answer 2

If I understand the question requirements (after reading the question and many comments), the only tricky bit is the preserving of the single-quoted substrings.

You want to isolate:

1. Single quote wrapped substrings that may contain spaces.
2. Words that may contain an apostrophe (single quote)
3. Numbers
4. Five specific operators: <, >!,=,?`

Pattern: ~\B'\K(?:[^']+)|\b[a-z']+\b|\d+|[<>!=?]+~i

Code with battery of tests (Demo)

$strings = [
    "age<21,length>10,height<>10,width!=100,name='Emma Einarsson' or it can be words time>=10,clouds<=4",
    "age < 21, length > 10, height <> 10, width != 100, name = 'Emma Einarsson' or it can be words time >= 10, clouds <= 4",
    "My name is Emma and i have a dillemma, what's the distance between 'New York' and 'Athene' ?",
    "'New York' and London at the start and end  with Paris and 'Los Angeles'"
];

foreach ($strings as $string) {
    var_export(preg_match_all("~\B'\K(?:[^']+)|\b[a-z']+\b|\d+|[<>!=?]+~i", $string, $out) ? $out[0] : 'fail');
    echo "\n";
}

Pattern Demo

Pattern Breakdown:

~                 #start of pattern delimiter
\B'\K(?:[^']+)    #match a single-quote not preceded by [a-zA-Z0-9_], then restart the fullstring match using (\K), then match one or more non-single quote characters
|                 #OR
\b[a-z']+\b       #match one or more letters and apostrophes 
|                 #OR
\d+               #match one or more digits
|                 #OR
[<>!=?]+          #match one or more of your listed operators/symbols
~                 #end of pattern delimiter
i                 #pattern modifier - make whole pattern case-insensitive

According to your sample input strings, you can technically remove the two \b (word boundary markers) from my pattern to improve pattern efficiency, but I left them in for maximum accuracy.