space dash as delimiter for group match

Question

I'm trying to greedy match string and mark it is as group until it encounter space dash ( -). So far I've got:

^ *-{2}([\w\d-]+)(=([ \w\d-]+))?(.*)|(.+)

and my test string is:

--match1=match2
--match1=match2 match2
--match1=match2 match2-match2
--mat-ch1=match2 match2
--match1=match2-match2 match2
--match1=match2 -lastmatch
--match1=match2 --lastmatch
--match1 -lastmatch
--match1 --lastmatch
lastormatch

Every each kind of match (match1, match2, lastmatch, lastormatch) should be grouped together if they are next to each other. This is working for every line, except 6 and 7. Basically '-' should act as delimiter when it comes to match2 group (but - surrounded by /w/d is fine as part of string inside group). I know, that lookahead/lookbehind should be used, but I can't get it right.

What I have comparing to what I need

regex101.com

You have not precised what exact group structure is expected for Cases 6 and 7. — Wiktor Stribiżew, Mar 14 '17 at 14:26
--(match1)=(match2)( -lastmatch) --(match1)=(match2)( --lastmatch) let's say that here bracket shows where group should be. — a_z, Mar 14 '17 at 14:28
Sorry, no idea what you need. Please add explanations to the question — Wiktor Stribiżew, Mar 14 '17 at 14:31
"should be grouped together, if they are ... marked as one group" Isn't that a tautology? — Jordan Running, Mar 14 '17 at 14:33
[What I have](https://i.stack.imgur.com/G2vdL.png) comparing to [what I need](http://i.imgur.com/XtKAf2R.png) — a_z, Mar 14 '17 at 14:35
What about: [`^ *(?:--)?([\w-]+)(?:=((?:[\w-]| \w)+))?(.*)`](https://regex101.com/r/lOyUzG/2) — trincot, Mar 14 '17 at 14:40

score 0 · Accepted Answer · answered Mar 14 '17 at 14:39

0

Not entirely sure what you're trying to achieve, but based on your "what I need" picture this might be sufficient:

^ *-{2}([\w\d-]+)(=(?:[\w\d-]| (?!-))+)?(.*)|(.+)

By accepting either [\w\d-] or spaces not followed by - instead of [ \w\d-] it seems to work for you test data.

But without understanding exactly what your goal is there might be issues with a different data set.

answered Mar 14 '17 at 14:39

rvalvik

1,559
11
15

The idea is to recognise parameters that are passed to the bash script, since I don't want to use getopts. `^ *-{2}([\w\d-]+)(=((?:[\w\d-]| (?!-))+))?(.*)|(.+)` (one bracket more) worked exactly as I wanted. – a_z Mar 14 '17 at 14:44
@a_z Great! Glad you got it sorted. If you are looking to capture the different arguments then perhaps something along the lines of `(?:--?((?:[\w\d-]| (?!-))+)=?((?:[\w\d-]| (?![-]))+)?)|^([^-]+)$` could be relevant - https://regex101.com/r/NVgqNe/1 – rvalvik Mar 14 '17 at 15:12
Idea is to match first potential option +optional input (like `-v` or `--log-file=logfile`) convert it to `verbose=true` or `log_file=logfile` inside bash, shift initial match from input array and iterate trough rest of input array. I'm not quite done with interpreting single dash options (like -q or -qwer with optional input) just yet, but I'm getting there :). http://stackoverflow.com/questions/192249/how-do-i-parse-command-line-arguments-in-bash is a good post, but every answer is missing a bit or code is huge. – a_z Mar 14 '17 at 15:32

score 0 · Answer 2 · answered Mar 14 '17 at 15:10

0

^ (?:--)([\w-]+)(?:=((?:[\w-]| \w)+))?|(.) Is exactly what I need. Combined answers of @trincot and @rvalvik.

answered Mar 14 '17 at 15:10

a_z

133
1
6

space dash as delimiter for group match

2 Answers2