Extracting numbers from a string with a special structure with regular expressions

Question

I have a string with the structure

Resolution:  1200, Time: 16.255 (7.920 GFlop => 1487.23 MFlop/s, residual 0.007113, 500 iterations)

and I am trying to efficiently extract the floats from that string. As the stream of string does not only contain strings with this pattern but others as well. Is there a way to get the numbers from this string with the package re?

I am looking for a way like this:

import re

a = "Resolution:  1200, Time: 16.255 (7.920 GFlop => 1487.23 MFlop/s, residual 0.007113, 500 iterations)"

nbrs = re.match("Resolution:  \d, Time: \d (\d GFlop => \d MFlop/s, residual \d, \d iterations)"

where \d is an identifier for an arbitrary float? Or is the easiest way of doing this just to strip the string multiple times and check for specific contents?

codester_09 · Answer 1 · 2022-05-06T11:48:09.810

1

import re
s = 'Resolution:  1200, Time: 16.255 (7.920 GFlop => 1487.23 MFlop/s, residual 0.007113, 500 iterations)'

p = re.findall(r'\d+[\.]\d+|\d+',s)
print(p)

OUTPUT:

['1200', '16.255', '7.920', '1487.23', '0.007113', '500']

edited May 06 '22 at 11:48

answered May 06 '22 at 11:39

codester_09

5,622
2
5
27

Are you intentionally matching any character between digit groups with `.` or did you mean to match a literal period? – isaactfa May 06 '22 at 11:42
@isaactfa _" matching any character between digit groups with ."_ – codester_09 May 06 '22 at 11:44
@isaactfa `r'\d+.\d+'` this will always return `int` or `float`. – codester_09 May 06 '22 at 11:45
No it won't. It'll also match `12#34` for example. – isaactfa May 06 '22 at 11:46
@isaactfa Check now. – codester_09 May 06 '22 at 11:48
I mean a litteral period - the performance of my CPU is 1487230 Flop/s – DeerFreak May 06 '22 at 11:51
1

You could make it even simpler with `\d+(?:\.\d+)?`. – isaactfa May 06 '22 at 11:53
@isaactfa I am not good in `re` . – codester_09 May 06 '22 at 11:54
1

Yes, that would be much simpler. So checking whether this string matches the exact structure of my string could be done with this and an additional `re.split`, correct? – DeerFreak May 06 '22 at 11:54
@DeerFreak If you just want to get the numbers out of the string, the above solution does just that. If you want to do something else, you should be more specific in your question. – isaactfa May 06 '22 at 11:56
@Isaactfa yeah sorry. I was looking for a way to extract numbers from a stream of strings, where they have to exactly match said pattern. So if there is another string like "90234.12 °C on Core 12" I wouldn't want to have this pattern applied but return a None. I think this additional check can be done via `re.split`. – DeerFreak May 06 '22 at 11:59

score 1 · Answer 2 · answered May 06 '22 at 12:00

The following splits a string in parts according to numbers found in it. The parts always alternate between non-numerical ones and numerical ones. Many types of floats are accepted, following this SO answer.

In the result:

elements [0::2] are non-numerical;
elements [1::2] are either int or float.

# adapted from: https://stackoverflow.com/a/55592455/758174
float_pat = re.compile(r'([+-]?(?:\d+(?:[.]\d*)?(?:[eE][+-]?\d+)?|[.]\d+(?:[eE][+-]?\d+)?))')

def int_or_float(s):
    try:
        return int(s)
    except ValueError:
        return float(s)

def split_numerical(s):
    a = re.split(float_pat, s)
    a[1::2] = map(int_or_float, a[1::2])
    return a

On your string:

s = 'Resolution:  1200, Time: 16.255 (7.920 GFlop => 1487.23 MFlop/s, residual 0.007113, 500 iterations)'
>>> split_numerical(s)
['Resolution:  ',
 1200,
 ', Time: ',
 16.255,
 ' (',
 7.92,
 ' GFlop => ',
 1487.23,
 ' MFlop/s, residual ',
 0.007113,
 ', ',
 500,
 ' iterations)']

score 1 · Accepted Answer · answered May 06 '22 at 12:14

Given what you said in your comment to me, I believe a more appropriate solution for your problem might be this:

import re

s = 'Resolution:  1200, Time: 16.255 (7.920 GFlop => 1487.23 MFlop/s, residual 0.007113, 500 iterations)'

pattern = re.compile(r"Resolution:  (?P<resolution>\d+), Time: (?P<time>\d+\.\d+) \((?P<gflops>\d+\.\d+) GFlop => (?P<mflops>\d+\.\d+) MFlop/s, residual (?P<residual>\d+\.\d+), (?P<iterations>\d+) iterations\)")

m = pattern.match(s)

Because of the named capture groups, you can get each value individually:

m = pattern.match(s)
print(m.group('resolution')) # 1200
print(m.group('time')) # 16.255
print(m.group('gflops')) # 7.920
# ...

But it won't match any string that isn't formatted exactly like the one you supplied. For example:

assert pattern.match("90234.12 °C on Core 12") is None

Extracting numbers from a string with a special structure with regular expressions

3 Answers3