# Forum: Ruby Scanning a string for decimal numbers

Announcement (2017-05-07): www.ruby-forum.com is now read-only since I unfortunately do not have the time to support and maintain the forum any more. Please see rubyonrails.org/community and ruby-lang.org/en/community for other Rails- und Ruby-related community platforms.
on 2006-02-05 00:20
```Hi all, how do you scan a string and avoid getting my decimal numbers
divided into 2 numbers.

Example:

a = "24,4 + 55,2"
a.scan! (/\d+/)
puts a

my output for a will be:
24
4
55
2

But I want to keep my decimal numbers intact like this:
24,4
55,2

How do I solve this problem without putting the numbers into seperate
strings?```
on 2006-02-05 00:38
```"24,4 + 55,2".scan /[\d,]+/

Kent```
on 2006-02-05 01:56
```Kent S. wrote:
>> a = "24,4 + 55,2"
>> 24,4
>> 55,2
>>
>>
>> How do I solve this problem without putting the numbers into seperate
>> strings?
>>
>>
>
>
try
a.scan!(/\d+,\d+/)
Ernie```
on 2006-02-05 04:48
```On Sun, 5 Feb 2006, Ernest Ellingson wrote:

>>>
>>> But I want to keep my decimal numbers intact like this:
> try
> a.scan!(/\d+,\d+/)
> Ernie

careful.  you'll kill negatives.

-a```
on 2006-02-05 13:48
```Jeppe J. wrote:
> 24
> strings?
Hi Jeppe,
you can use the following:

a.scan /[-+]?[0-9]*\,?[0-9]+/

which will take care of negative numbers as well.

HTH,
Antonio```
on 2006-02-05 14:04
```> a.scan /[-+]?[0-9]*\,?[0-9]+/

Or to be less verbose you can use \d in place of [0-9] :-)

HTH
Antonio```
on 2006-02-05 14:16
```On Feb 5, 2006, at 13:03, Antonio C. wrote:

>> a.scan /[-+]?[0-9]*\,?[0-9]+/
>
> Or to be less verbose you can use \d in place of [0-9] :-)

If you have Perl installed, the incantation

perldoc -q float

gives some regexps for numbers.

-- fxn```
on 2006-02-12 00:43
```Hi!

At Sun, 05 Feb 2006 11:43:52 +0000, Antonio C. wrote:
> a.scan /[-+]?[0-9]*\,?[0-9]+/

Shouldn't that rather be the following?

a.scan /[-+]?([1-9]\d*(\,[0-9]+)?)|(0(\,[0-9]+)?)/

Josef 'Jupp' Schugt```
on 2006-02-12 01:12
`Will that expression include both integers and decimal numbers?`
on 2006-02-12 01:29
```On 2/4/06, Jeppe J. <removed_email_address@domain.invalid> wrote:
> 24
> strings?
>
This should handle periods or commas as the separator.

a = "24,4 + 55,2 + 55 - 44,0"
=> "24,4 + 55,2 + 55 - 44,0"
a.scan /(\d+,?.?\d*)(?=\s|\$)/
=> [["24,4"], ["55,2"], ["55"], ["44,0"]]```
on 2006-02-12 01:49
```Nice, that was the thing I was looking for :)

on 2006-02-12 02:41
```>
> This should handle periods or commas as the separator.
>
> a = "24,4 + 55,2 + 55 - 44,0"
>    => "24,4 + 55,2 + 55 - 44,0"
> a.scan /(\d+,?.?\d*)(?=\s|\$)/
>    => [["24,4"], ["55,2"], ["55"], ["44,0"]]
>

Some problems here:
- signs are disregarded ("-24,4" becomes "24,4")
- Invalid numbers are accepted: eg. "24,.4" "24,." "24." "24,"
- "." should be escaped. As you used it here, it means "any character"
(except newline), so many invalid numbers are accepted (e.g. "24w"...)
- If something different from whitespace follows the number, it is not
or false accepted, e.g. "24.4." becomes "4." instead of "24.4"
- ...

Alexis.```
on 2006-02-12 03:32
```2006/2/12, Alexis R. <removed_email_address@domain.invalid>:
> Some problems here:
>
>
>

Let me see if I got it right then. I'll like to use periods only for my
decimal numbers. I also need normal integers so 24. being accepted won't
matter. Will this fix the problems you presented?:
/[-+]?(\d+\.?\d*)(?=\s|\$)/

I don't know if it takes care of the last problem, because I didn't
understand it.```
on 2006-02-12 03:40
```Seems I accidently got my text marked as a qoute in my last mail, so
I'll
just send it a again:

Let me see if I got it right then. I'll like to use periods only for my
decimal numbers. I also need normal integers so 24. being accepted won't
matter. Will this fix the problems you presented?:
/[-+]?(\d+\.?\d*)(?=\s|\$)/

I don't know if it takes care of the last problem, because I didn't
understand it.

on 2006-02-12 05:28
```Well, that's what I get for dashing off a quick e-mail before dinner.
The last problem Alexis mentioned is caused by the overly-specific
lookahead at the end.  Here's a version that fixes that:

irb(main):013:0> a = '24.5 + 24 + 24. + 24.4.'
=> "24.5 + 24 + 24. + 24.4."
irb(main):014:0> a.scan /[-+]?(\d+(?:\.\d+)?)(?=[^\d])/
=> [["24.5"], ["24"], ["24"], ["24.4"]]
irb(main):015:0>

One of the characters '-' or '+', optionally
Followed by at least one digit.
Followed by an optional group containing a period, and one or more
digits.
The capturing group ends when the next character is something other
than a digit.

The (?:) mess is there so that '24.' doesn't end up with the period on
the end.```
on 2006-02-12 22:41
```Yes that worked, but I intend to convert the digits of my array to
floats,
and I get a NoMethodError on to_f now when I do this:

digits[0] = digits[0].to_f

I don't understand that :-/

on 2006-02-12 23:53
```The scan process returns an array of arrays, so:
digits[0] is an Array containing '24.4'.
You could do:
digits.flatten!
just before digits[0], and get what you expect.```
on 2006-02-12 23:59
```ok, but I think but wouldn't this regex do the same for me?:

/[-+]?\d+\.?\d+/

Except that it will return an array containing my digit?

on 2006-02-13 00:15
`Yes, as long as the numbers are always at least two digits.`
on 2006-02-14 00:28
```Hi!

At Sun, 12 Feb 2006 08:12:47 +0900, Jeppe J. wrote:
>
> Will that expression include both integers and decimal numbers?

[-+]?([1-9]\d*(\,[0-9]+)?)|(0(\,[0-9]+)?)

has two parts:

[-+]?
([1-9]\d*(\,[0-9]+)?)|(0(\,[0-9]+)?)

The first one is an optional sign. The second one is an alternative
between to two cases:

[1-9]\d*(\,[0-9]+)?
0(\,[0-9]+)?

Let's first consider the first case

[1-9]\d*(\,[0-9]+)?

It has two parts, namely

[1-9]\d*
(\,[0-9])?

The first part by itself covers all integers larger than zero. The
overall expression additionally covers all floating point numbers
larger than 1.

Now the second case

0(\,[0-9]+)?

This one covers zero and all decimal numbers larger than 0 and smaller
than 1.

The regex I provided intentionally supports none of

[+-],\d+
[+-]0+\d+

You may as well use the shorter version

[-+]?(([1-9]\d*(\,\d+)?)|(0(\,\d+)?))

Wait a moment, I am not sure if that is correct. To be on the safe
side I'd rather use one of these where anything that follows the
optional sign has been put into another pair of parentheses:

[-+]?(([1-9]\d*(\,[0-9]+)?)|(0(\,[0-9]+)?))
[-+]?((([1-9]\d*(\,\d+)?)|(0(\,\d+)?)))

I am one of those guys who sometime run out of placeholders when doing
search and replace in vim (which has nine of them).

Josef 'Jupp' Schugt```
on 2006-02-14 20:09
```Thank you for clearing things, up for me, but could you explain what the
last part of the expression Wilson provided me with means?

it's (?=[^\d])

on 2006-02-14 21:46
```DÅ?a Utorok 14 FebruÃ¡r 2006 19:07 Jeppe J. napÃ­sal:
> Thank you for clearing things, up for me, but could you explain what the
> last part of the expression Wilson provided me with means?
>
> it's (?=[^\d])
>

That's a positive zero-width lookahead. I think. Gotta love regexspeak.

In English: look for a single character that's not a decimal digit, and
don't
include it in the match.

David V.```
on 2006-02-15 11:28
```David V. wrote:
> DÅ?a Utorok 14 FebruÃ¡r 2006 19:07 Jeppe J. napÃ­sal:
>> Thank you for clearing things, up for me, but could you explain what
>> the last part of the expression Wilson provided me with means?
>>
>> it's (?=[^\d])

These is equivalent (?=\D)

A negative lookahead might work, too: (?!\d)

> That's a positive zero-width lookahead. I think. Gotta love
> regexspeak.
>
> In English: look for a single character that's not a decimal digit,
> and don't include it in the match.

I'd go with this quite simple regexp

/[-+]?\d+(?:,\d+)?/

If numbers like "1," should be detected, too, then just change the "+"
in
the last group to "*".

If one wants to prevent to match numbers with leading zeros then it
becomes more complicated but it seems not be worth the effort in this
case.

Kind regards

robert```
This topic is locked and can not be replied to.