PHP正则表达式匹配多个选项

Question

I am attempting to write some code that looks for the following:

Yesterday
Last 7 Days
Last 30 Days
This Year
Last Year

I have the following regex:

/yesterday|(\d+)(?=\s+(\w+))|(\w+)(?=\s+(year))/i

using:

preg_match("/yesterday|(\d+)(?=\s+(\w+))|(\w+)(?=\s+(year))/i", $input, $output)

I get the following results using phpliveregex.com with the preg_match:

array(5
0   =>  Last
1   =>  
2   =>  
3   =>  Last
4   =>  Year
)
array(5
0   =>  This
1   =>  
2   =>  
3   =>  This
4   =>  year
)
array(1
0   =>  yesterday
)
array(3
0   =>  30
1   =>  30
2   =>  days
)
array(3
0   =>  7
1   =>  7
2   =>  days

My issue is with the 'Year' options and the fact that they have empty keys because I want to refer to $output[1] and $output[2] to get the interval and 'span' (days). Only a single string will be passed at a time so it will be one of the options listed above and not multiple options to look for at once.

If anyone can help me find the best solution to return 'yesterday' or ('7' and 'days') or ('30' and 'days') or ('This' and 'Year') or ('Last' and 'Year') I would appreciate it very much!

EDIT

This is my desired output:

'Yesterday'
$output[0] => 'Yesterday'

'Last 7 Days'
$output[0] => '7'
$output[1] => 'Days'

'Last 30 Days'
$output[0] => '30'
$output[1] => 'Days'

'This Year'
$output[0] => 'This'
$output[1] => 'Year'

'Last Year'
$output[0] => 'Last'
$output[1] => 'Year'

I am trying to capture the 'groups' necessary to process the rest of my code.

Answer 1

What you just described can be Achieved with the following Regex:

(yesterday|\d+(?=\s+\w+)|\w+(?=\s+year))\s*(\w*)$

Tested on Regex101.com Demo Here :

Answer 2

You can use the branch reset feature to avoid empty groups:

$text = <<<'EOD'
Yesterday
Last 7 Days
Last 30 Days
This Year
Last Year
EOD;

$pattern = '~\b(?|yesterday\b|\d+(?= (days\b))|\w+(?= (year\b)))~i';
if (preg_match_all($pattern, $text, $matches, PREG_SET_ORDER))
    print_r($matches);
// or preg_match without PREG_SET_ORDER if you test the strings one by one

pattern details:

\b
(?|                     # open the branch reset group
    yesterday \b        # when this branch succeeds the capture group is not defined
  |
    \d+ (?=[ ](days\b)) # in each branch the capture group
  |
    \w+ (?=[ ](year\b)) # has the same number 
)                       # (so there is only one capture group)

result:

Array
(
    [0] => Array
        (
            [0] => Yesterday
        )

    [1] => Array
        (
            [0] => 7
            [1] => Days
        )

    [2] => Array
        (
            [0] => 30
            [1] => Days
        )

    [3] => Array
        (
            [0] => This
            [1] => Year
        )

    [4] => Array
        (
            [0] => Last
            [1] => Year
        )

)

Note that when you build the branch reset, you must begin with alternatives that has no groups, then alternatives with one groups, then two groups, etc. otherwise you may obtain useless empty groups in the result.

Note too that the group 0 isn't really a capture group but it is the whole match.

Answer 3

You can use:

/((?:Last|This)\s+(?:\d+\s+Days|Year)|Yesterday)/

---

Matches:

MATCH 1
1.  [0-9]   `Yesterday`
MATCH 2
1.  [10-21] `Last 7 Days`
MATCH 3
1.  [22-34] `Last 30 Days`
MATCH 4
1.  [35-44] `This Year`
MATCH 5
1.  [45-54] `Last Year`

---

Regex Demo:

https://regex101.com/r/mA8jZ5/1

---

Regex Explanation:

/((?:Last|This)\s+(?:\d+\s+Days|Year)|Yesterday)/

1st Capturing group ((?:Last|This)\s+(?:\d+\s+Days|Year)|Yesterday)
    1st Alternative: (?:Last|This)\s+(?:\d+\s+Days|Year)
        (?:Last|This) Non-capturing group
            1st Alternative: Last
                Last matches the characters Last literally (case sensitive)
            2nd Alternative: This
                This matches the characters This literally (case sensitive)
        \s+ match any white space character [\r\n\t\f ]
            Quantifier: + Between one and unlimited times, as many times as possible, giving back as needed [greedy]
        (?:\d+\s+Days|Year) Non-capturing group
            1st Alternative: \d+\s+Days
                \d+ match a digit [0-9]
                    Quantifier: + Between one and unlimited times, as many times as possible, giving back as needed [greedy]
                \s+ match any white space character [\r\n\t\f ]
                    Quantifier: + Between one and unlimited times, as many times as possible, giving back as needed [greedy]
                Days matches the characters Days literally (case sensitive)
            2nd Alternative: Year
                Year matches the characters Year literally (case sensitive)
    2nd Alternative: Yesterday
        Yesterday matches the characters Yesterday literally (case sensitive)