re.findall 行为怪异

Question

The source string is:

# Python 3.4.3
s = r'abc123d, hello 3.1415926, this is my book'

and here is my pattern:

pattern = r'-?[0-9]+(\\.[0-9]*)?|-?\\.[0-9]+'

however, re.search can give me correct result:

m = re.search(pattern, s)
print(m)  # output: <_sre.SRE_Match object; span=(3, 6), match='123'>

re.findall just dump out an empty list:

L = re.findall(pattern, s)
print(L)  # output: ['', '', '']

why can't re.findall give me the expected list:

['123', '3.1415926']

Answer 1

There are two things to note here:

• re.findall returns captured texts if the regex pattern contains capturing groups in it
• the r'\\.' part in your pattern matches two consecutive chars, \ and any char other than a newline.

See findall reference :

If one or more groups are present in the pattern, return a list of groups; this will be a list of tuples if the pattern has more than one group. Empty matches are included in the result unless they touch the beginning of another match.

Note that to make re.findall return just match values , you may usually

• remove redundant capturing groups (eg (a(b)c) -> abc )
• convert all capturing groups into non-capturing (that is, replace ( with (?: ) unless there are backreferences that refer to the group values in the pattern (then see below)
• use re.finditer instead ( [x.group() for x in re.finditer(pattern, s)] )

In your case, findall returned all captured texts that were empty because you have \\ within r'' string literal that tried to match a literal \ .

To match the numbers, you need to use

-?\d*\.?\d+

The regex matches:

• -? - Optional minus sign
• \d* - Optional digits
• \.? - Optional decimal separator
• \d+ - 1 or more digits.

See demo

Here is IDEONE demo :

import re
s = r'abc123d, hello 3.1415926, this is my book'
pattern = r'-?\d*\.?\d+'
L = re.findall(pattern, s)
print(L)

Answer 2

s = r'abc123d, hello 3.1415926, this is my book'
print re.findall(r'-?[0-9]+(?:\.[0-9]*)?|-?\.[0-9]+',s)

You dont need to escape twice when you are using raw mode .

Output: ['123', '3.1415926']

Also the return type will be a list of strings . If you want return type as integers and floats use map

import re,ast
s = r'abc123d, hello 3.1415926, this is my book'
print map(ast.literal_eval,re.findall(r'-?[0-9]+(?:\.[0-9]*)?|-?\.[0-9]+',s))

Output: [123, 3.1415926]

Answer 3

Just to explain why you think that search returned what you want and findall didn't?

search return a SRE_Match object that hold some information like:

• string : attribute contains the string that was passed to search function.
• re : REGEX object used in search function.
• groups() : list of string captured by the capturing groups inside the REGEX .
• group(index) : to retrieve the captured string by group using index > 0 .
• group(0) : return the string matched by the REGEX .

search stops when It found the first mach build the SRE_Match Object and returning it, check this code:

import re

s = r'abc123d'
pattern = r'-?[0-9]+(\.[0-9]*)?|-?\.[0-9]+'
m = re.search(pattern, s)
print(m.string)  # 'abc123d'
print(m.group(0))  # REGEX matched 123
print(m.groups())  # there is only one group in REGEX (\.[0-9]*) will  empy string tgis why it return (None,) 

s = ', hello 3.1415926, this is my book'
m2 = re.search(pattern, s)  # ', hello 3.1415926, this is my book'
print(m2.string)    # abc123d
print(m2.group(0))  # REGEX matched 3.1415926
print(m2.groups())  # the captured group has captured this part '.1415926'

findall behave differently because it doesn't just stop when It find the first mach it keeps extracting until the end of the text, but if the REGEX contains at least one capturing group the findall don't return the matched string but the captured string by the capturing groups:

import re
s = r'abc123d , hello 3.1415926, this is my book'
pattern = r'-?[0-9]+(\.[0-9]*)?|-?\.[0-9]+'
m = re.findall(pattern, s)
print(m)  # ['', '.1415926']

the first element is return when the first mach was found witch is '123' the capturing group captured only '' , but the second element was captured in the second match '3.1415926' the capturing group matched this part '.1415926' .

If you want to make the findall return matched string you should make all capturing groups () in your REGEX a non capturing groups (?:) :

import re
s = r'abc123d , hello 3.1415926, this is my book'
pattern = r'-?[0-9]+(?:\.[0-9]*)?|-?\.[0-9]+'
m = re.findall(pattern, s)
print(m)  # ['123', '3.1415926']