Python正则表达式拆分逗号或空格分隔的字符串

Question

I have shown an example of the input string and the output string I need to have. The numbers and strings can be in any order and those can have any number or any string (eg 'THIS' can be some other string)

I need following output

["'5'", "'THIS'", "'/,'", "'4.2560'", "'0.34000E-01'"]

for all following input strings:

""" 5,'THISMORETHAN4','/,',4.2560,0.34000E-01 """
""" 5,'THIS','/,',4.2560,0.34000E-01 """
"""5 , 'THIS' , '/,' , 4.2560 , 0.34000E-01 """
""" '5'  'THIS' '/,' '4.2560' '0.34000E-01' """
""" 5,'THIS','this','/,',4.2560,0.34000E-01 """
""" 5,'THIS','/,',4.2560,0.34000E-01 """

This is a continuation of previous question .

1. The strings can be comma separated or space separated. There may be or may not be spaces before or after a splitting comma.
2. sub strings in single quotes may have special characters within (eg '/,' as shown above)

Answer 1

As an improved version of Padraic Cunningham's solution from your previous question, the regex (["']).*?\\1(?<!\\\\["'])|[^\\r\\n\\t\\f ,]+ will capture all your fields.

The first part ( (["']).*?\\1(?<!\\\\["']) ) now also works with fields like 'asdf"' because the sourrounding quote characters have to be the same. It also works with escaped quotes because (?<!\\\\["']) asserts that there is no backslash before the second quote.

If the first part doesn't match (ie there is no string surrounded by quotes), the second part ( [^\\r\\n\\t\\f ,]+ ) matches everything that is not a whitespace or a comma. So it will ignore your delimiters but match everything else.

import re

rows = [""" 5,'THISMORE"THAN4','/,',4.2560,0.34000E-01 """,
        #              ^ added quote character here
        """ 5,'TH\\'IS','/,',4.2560,0.34000E-01 """,
        #          ^ added escaped quote here
        """5 , 'THIS' , '/,' , 4.2560 , 0.34000E-01 """,
        """ '5'  'THIS' '/,' '4.2560' '0.34000E-01' """,
        """ 5,'THIS','this','/,',4.2560,0.34000E-01 """,
        """ 5,'THIS','/,',4.2560,0.34000E-01 """]

pattern = re.compile(r'(["\']).*?\1(?<!\\["\'])|[^\r\n\t\f ,]+')

result = [[m.group(0).strip('"\'') for m in pattern.finditer(row)]
          for row in rows]

import pprint
pprint.pprint(result)

Prints:

[['5', 'THISMORE"THAN4', '/,', '4.2560', '0.34000E-01'],
 ['5', "TH\\'IS", '/,', '4.2560', '0.34000E-01'],
 ['5', 'THIS', '/,', '4.2560', '0.34000E-01'],
 ['5', 'THIS', '/,', '4.2560', '0.34000E-01'],
 ['5', 'THIS', 'this', '/,', '4.2560', '0.34000E-01'],
 ['5', 'THIS', '/,', '4.2560', '0.34000E-01']]

What will still be problematic are unquoted fields that contain spaces within a line that has comma separation. Therefore

'hello there, "I actually", have, 5, fields'

Will result in:

['hello','there','I actually','have','5','fields']

Do you have that in your data?

Answer 2

此正则表达式适用于所有测试用例

(\d)\W*\'([A-Z]{0,4})\w*\'.*(\/)\W*(\d*\.\d*)\W*(\d*\.\d*E-\d*)