regex allows one character (it should not) why?

Question

Hello I am trying to create a regex that recognizes money and numbers being inputted. I have to allow numbers because I am expecting non-formatted numbers to be inputted programmatically and then I will format them myself. For some reason my regex is allowing a one letter character as a possible input.

[\$]?[0-9,]*\.[0-9][0-9]

I understand that my regex accepts the case where multiple commas are added and also needs two digit after the decimal point. I have had an idea of how to fix that already. I have narrowed it down to possibly the *\. as the problem

EDIT

I found the regex expression that worked [\$]?([0-9,])*[\.][0-9]{2} but I still don't know how or why it was failing in the first place

I am using the .formatCurrency() to format the input into a money format. It can be found here but it still allows me to use alpha characters so i have to further masked it using the $(this).inputmask('Regex', { regex: "[\$]?([0-9,])*[\.][0-9]{2}" }); where input mask is found here and $(this) is a reference to a input element of type text. My code would look something like this

<input type="text" id="123" data-Money="true">

 //in the script
 .find("input").each(function () {          
        if ($(this).attr("data-Money") == "true") {                            
            $(this).inputmask('Regex', { regex: "[\$]?([0-9,])*[\.][0-9]{2}" });
            $(this).on("blur", function () {
                $(this).formatCurrency();
            });

I hope this helps. I try creating a JSfiddle but Idk how to add external libraries/plugin/extension

The backslash in `\.` is probably getting "eaten" by whatever software you're using. You'll need to provide specifics of your environment, failing code, etc. — Peter Alfvin, Sep 05 '13 at 22:24
What tool or language are you using, what you are saying is impossible unless the tool you are using have something to say about it :) — Ibrahim Najjar, Sep 05 '13 at 22:37
@sreservoir What do you mean that I am wrong? Like my regex does not really do what I expect it to do? — Jack Thor, Sep 05 '13 at 22:38
The regex doesn't respect the user's locale. For example, the input string €4.655,00 is a valid currency value in parts of Europe - and that's why it's better to use locale support if your language supports it. We need more details about your environment; is this Linux shell, python, ...? — kmorris, Sep 05 '13 at 22:40
The dot and the dollar sign are not escaped in your regexp. You should rewrite `regex: "[\$]?([0-9,])*[\.][0-9]{2}" }` to `regex: "[\\$]?([0-9,])*[\\.][0-9]{2}" }` — rbernabe, Sep 05 '13 at 23:30
You appear to be using Robin Herbots' jquery.inputmask. If this is the case, you should mention it and add that as a tag on this question. — Rob Raisch, Sep 05 '13 at 23:46
@JackThor the regex functioned as described, but you incorrectly believed that that was the regex in your code. the iirc later-added context clarified this. — muhmuhten, Sep 15 '13 at 22:49

score 3 · Accepted Answer · answered Sep 05 '13 at 23:35

The "regular expression" you're using in your example script isn't a RegExp:

$(this).inputmask('Regex', { regex: "[\$]?([0-9,])*[\.][0-9]{2}" });

Rather, it's a String which contains a pattern which at some point is being converted into a true RegExp by your library using something along the lines of

var RE=!(value instanceof RegExp) ? new RegExp(value) : value;

Within Strings a backslash \ is used to represent special characters, like \n to represent a new-line. Adding a backslash to the beginning of a period, i.e. \., does nothing as there is no need to "escape" the period.

Thus, the RegExp being created from your String isn't seeing the backslash at all.

Instead of providing a String as your regular expression, use JavaScript's literal regular expression delimiters.

So rather than:

$(this).inputmask('Regex', { regex: "[\$]?([0-9,])*[\.][0-9]{2}" });

use

$(this).inputmask('Regex', { regex: /[\$]?([0-9,])*[\.][0-9]{2}/ });

And I believe your "regular expression" will perform as you expect.

(Note the use of forward slashes / to delimit your pattern, which JavaScript will use to provide a true RegExp.)

Did you find out about the `var RE=!(value instanceof RegExp) ? new RegExp(value) : value;` by reading the source code? and how come I can not tag you? — Jack Thor, Sep 05 '13 at 23:57
No, though I could if you wish. I'm pretty confident I'm right even though I _assumed_ this was the case because the value you provided is a String. — Rob Raisch, Sep 06 '13 at 00:01
Oh no it ok. I was just trying to figure out if this was general knowledge, or was something specific to the jquery.inputmask — Jack Thor, Sep 06 '13 at 00:20

score 2 · Answer 2 · edited May 23 '17 at 12:17

Firstly, you can replace '[0-9]' with '\d'. So we can rewrite your first regex a little more cleanly as

\$?[\d,]*\.\d\d

Breaking this down:

\$?       - A literal dollar sign, zero or one
[\d,]*    - Either a digit or a comma, zero or more
\.        - A literal dot, required
\d        - A digit, required
\d        - A digit, required

From this, we can see that the minimum legal string is \.\d\d, three characters long. The regex you gave will never validate against any one character string.

Looking at your second regex,

[\$]?     - A literal dollar sign, zero or one
([0-9,])* - Either a digit or a comma, subexpression for later use, zero or more
[\.]      - A literal dot, required
[0-9]{2}  - A digit, twice required

This has the exact same minimum matchable string as above - \.\d\d.

edit: As mentioned, depending on the language you may need to escape forward slashes to ensure they aren't misinterpretted by the language when processing the string.

Also, as an aside, the below regex is probably closer to what you need.

[A-Z]{3} ?(\d{0,3}(?:([,. ])\d{3}(?:\2\d{3})*)?)(?!\2)[,.](\d\d)\b

Explanation:

[A-Z]{3}       - Three letters; for an ISO currency code
 ?             - A space, zero or more; for readability
(              - Capture block; to catch the integer currency amount
  \d{0,3}        - A digit, between one and three; for the first digit block
  (?:            - Non capturing block (NC)
    ([,. ])        - A comma, dot or space; as a thousands delimiter
    \d{3}          - A digit, three; the first possible whole thousands
    (?:            - Non capturing block (NC) 
      \2             - Match 2; the captured thousands delimiter above
      \d{3}          - A digits, three
    )*             - The above group, zero or more, i.e. as many thousands as we want
  )?              - The above (NC) group, zero or one, ie. all whole thousands
)              - The above group, i.e everything before the decimal
[.,]           - A comma or dot, as a decimal delimiter
(\d{2})        - Capture, A digit, two; ie. the decimal portion
\b             - A word boundry; to ensure that we don't catch another
                 digit in the wrong place.

The negative lookahead was provided by an answer from John Kugelman in this question.

This correctly matches (matches enclosed in square brackets):

[AUD 1.00]
[USD 1,300,000.00]
[YEN 200 000.00]
I need [USD 1,000,000.00], all in non-sequential bills.

But not:

GBP 1.000
YEN 200,000

regex allows one character (it should not) why?

2 Answers2

Linked