php extract data from unstructured string using regex

Question

I need to extract data from unstructured string coming from a sms

The data I need to extract is the following

Code: This a 5 letter alphanumeric string it must contain at least one digit

Identity document: This is a numeric string between 5 and 8 characters, valid formats are:

V55555555
E55555555
55555
55 555
E55 555 555
55 555 555
5 555 555
555 555

The data I need to extract could be in any position in the string, I have normalized the string, replaced duplicate spaces by only one, and deleting anything that is not a space, number and letter

Samples

1. resuelvete 15C20 Pdero Perez c.i. V55.555.555,
2. Pedro Perez resuelvete 15c20 55 555 555,
3. 15c20 Resuelvete 555555 Pedro Perez,
4. Resuelvete 555555 Pedro Perez 15c20

For the code part I've tried this regex:

$regex = '/([a-zA-Z0-9]{5})/i';

I also tried this: $regex = '(?=.{5})(?=.*[A-Z])(?=.*[a-z])(?=.*\d)[a-zA-Z\d]';, that I see here but it's not working (I must say I don't totally understand this regex)

But it's not working, its returnig the first five characters of the string, I need it to return in this examples 15c20

For the Identity document part I've tried the following:

// This not work with spaces
$regex = "/(V|E)?(\d{5,8})/i";

// This not work without spaces
//This fail in first case returning only 7 digits instead of 8
// Also fails in cases  3 and 4,  does not match anything
$regex = "/(V|E)?(\d{1,2}? ?\d{3} ?\d{3})/i";

In the code, are there any restriction at all regarding the number of alphabet characters, the number numbers, the relative positions of alpha and number, etc? If you it will basically be impossible to distinguish this code from any other five-character string. — Mike Brant
– Mike Brant, Commented Jan 10, 2013 at 0:07
The code must be exactly five characters length and must contain at least one digit in any positions ex. 1azsd 125a1 az12a, also it must not be composed of only numbers — Cesar
– Cesar, Commented Jan 10, 2013 at 0:09
But, must the code contain at least one alphanumeric value? If not you will not be able to discern from one of the identity patterns. — Mike Brant
– Mike Brant, Commented Jan 10, 2013 at 0:20
@Cesar please see the edit in my answer. Although little it's an important one. — inhan
– inhan, Commented Jan 10, 2013 at 21:22

Mike Brant · Accepted Answer · 2013-01-10 00:43:28Z

This should work for the code part (note I am assuming there must be at least one alpha character here, otherwise you would not be able to tell the difference between code and identity for ##### case)

$code_pattern = '/\b(?=.*[\d].*\b)(?=.*[a-zA-Z].*\b)[a-zA-Z\d]{5}\b/';

Note that the (?=....) syntax is what is called a positive look-ahead. It is used to assert that upcoming values in that portion of the pattern match that pattern (without actually counting as characters within the match.

For the identity part, I would keep it simple (i.e. not look for a one regex fits all solution) and use an array of patterns with your preg_* function.

$identity_patterns = array(
    '/\b(V|E)[0-9]{8}\b/', // V########, E########
    '/\b[\d]{5}\b/', // #####
    '/\bE[\d]{2}\w[\d]{3}\w[\d]{3}\b/', // E## ### ###
    '/\b[\d]{1,3}\w[\d]{3}(\w[\d]{3})?\b/' // #{1,3} ### (###)?
);

It would certainly be possible to unify all of that into a single regex, but it would make it very difficult to read and modify if needed in the future.

inhan · Accepted Answer · 2013-01-10 21:21:05Z

[EDIT] I changed all \w's in the following regex pattern into [a-zA-Z0-9] because \w includes the underscore ( _ ) character as well, which I forgot at the time I wrote this answer.

For the code you can use something like

~\b(?=[a-zA-Z0-9]{5}\b)[a-zA-Z0-9]*\d[a-zA-Z0-9]*~

which can be broken into

\b                  # a word boundary (beginning of word in this case)
(?=                 # from here on...
    [a-zA-Z0-9]{5}  # 5 alphanumeric characters (a-z,A-Z,0-9)
    \b              # followed by a word boundary (end of word)
)
[a-zA-Z0-9]*        # 0 or more alphanumeric characters (a-z,A-Z,0-9)
\d                  # a decimal
[a-zA-Z0-9]*        # 0 or more alphanumeric characters (a-z,A-Z,0-9)

For the id document… If there are no possible groupings other than the examples you provided, you can use something like

'~(?<=\b|\bE|\bV)(?=[\d\ .]{5,10}\b)(
    \d{8}|
    \d{5}|
    \d{3}[\ .]\d{3}|
    \d{2}([\ .])\d{3}(\2\d{3})?|
    \d([\ .])\d{3}\4\d{3}
)~x'

and this says

(?<=                        # preceded by
    \b|                     # a word boundary (beginning of word) or
    \bE|                    # a word boundary (beginning of word) and E or
    \bV                     # a word boundary (beginning of word) and V
)
(?=                         # from here on...
    [\d\ .]{5,10}           # a group consisting of only decimals, spaces and dots 
    \b                      # followed by a word boundary (end of word)
)
(                           # either:
    \d{8}|                  # an 8 digit number
    \d{5}|                  # a 5 digit number
    \d{3}[\ .]\d{3}|        # a 3 digit number, space or dot, a 3 digit number
    \d{2}([\ .])\d{3}       # a 2 digit number, space or dot, a 3 digit number
    (                       # optionally...
        \2                  # previous sign (space or dot)
        \d{3}               # a 3 digit number
    )?| 
    \d([\ .])\d{3}\4\d{3}   # a 1 digit number, space or dot, a 3 digit number, previous sign, a 3 digit number
)

Collectives™ on Stack Overflow

php extract data from unstructured string using regex

2 Answers 2

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related