regex repeated group capture

Question

Even though I've looked through multiple examples of how to capture a repeated group. I cant seem to be able to understand it correctly for my case to make it work.

use ((?:([a-z_0-9]*)\.)*)(\w*);

Regex example

As shown in the above example, I wish to capture the package name in segments so that use ieee.std_logic_1164.all; returns [IEEE, std_logic_1164, all] etc. Currently my regex returns [ieee.std_logic_1164., std_logic_1164, all] which is wrong.

I am missing something from the expression I've done. It seems that I can only capture the last iteration of the repeated group even though I've encapsulated my repeated section in another capturing group according to the specification.

Regards

Answer 1

Another option with the PyPi regex module using allcaptures() or capturesdict() and the capture groups with the same name, in this example name.

If you want the first item in the result list to be upper cased, you can use .upper()

\buse (?:(?<name>\w+)\.)*(?<name>\w+);

See a regex demo with a different engine selected, but showing the same idea.

import regex as re

s = ("library ieee;\n"
"use ieee.std_logic_1164.all;\n"
"use ieee.numeric_std.all;\n"
"use work.pk_avalon_mm_extif_defs.all;\n"
"use work.common_tb_pkg.all;\n"
"package common_register_interface_pkg is")

pattern = r"\buse (?:(?P<name>\w+)\.)*(?P<name>\w+);"
matches = re.finditer(pattern, s)

for matchNum, m1 in enumerate(matches, start=1):
    print(m1.allcaptures()[1])

for matchNum, m1 in enumerate(matches, start=1):
    print(m1.capturesdict()['name'])

Both will output

['ieee', 'std_logic_1164', 'all']
['ieee', 'numeric_std', 'all']
['work', 'pk_avalon_mm_extif_defs', 'all']
['work', 'common_tb_pkg', 'all']

Answer 2

You cannot have a variable number of capture groups. So either you must have a separate capture group for each chunk you want to match (foreseeing enough of them), or else you match them all in one capture group, and split them later.

With the first idea you'd still have to identify for each match, which capture groups resulted in an empty string, which means it should be ignored. This seems not give any benefit over the second idea.

Here is how the second idea would work:

s = """-------------------------------------------------------------------------------
library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;

use work.pk_avalon_mm_extif_defs.all;
use work.common_tb_pkg.all;

package common_register_interface_pkg is"""

results = [m.split(".") for m in re.findall(r"use ((?:(?:[a-z_0-9]*)\.)*\w*);", s)]

print(results)

Answer 3

I don't think your pattern needs groups. You can specify that the searched expression is preceded by use and followed by ; with lookarounds. Then just split your matches along . as suggested in a comment:

import re

text = """library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;

use work.pk_avalon_mm_extif_defs.all;
use work.common_tb_pkg.all;

package common_register_interface_pkg is"""

matches = re.findall(r"(?<=use )(?:\w+\.)+\w*(?=;)", text)
res = [m.split('.') for m in matches]

Output:

[['ieee', 'std_logic_1164', 'all'], ['ieee', 'numeric_std', 'all'], ['work', 'pk_avalon_mm_extif_defs', 'all'], ['work', 'common_tb_pkg', 'all']]

Answer 4

If you can install PyPi regex module then you can do this in a single regex:

(?:^use\s+|(?!^)\G\.)\K\w+

RegEx Demo

RegEx Details:

• (?:: Start non-capture group
  • ^: Match start position
  • use: Match use
  • \s+: Match 1+ whitespaces
  • |: OR
  • (?!^)\G: Start at the end position of the previous match
  • \.: Match a DOT
• ): End non-capture group
• \K: Reset match info
• \w+: Our match that is 1+ word characters