Bash regex doesn't match the pattern with \w and $

Question

I'm very new to Bash. Can someone tell me what am I doing wrong. I tried the following regex pattern online and it works fine. But Bash script below refuses to match.

I need to extract the value of status: from the last line of a text block like so:

res=$(cat <<-EOM
Conducting pre-submission checks for my.zip and initiating connection to the Apple notary service...
Submission ID received
  id: 573ebf8c-5434-4820-86c6-46adf01f81a9
Successfully uploaded file
  id: 573ebf8c-5434-4820-86c6-46adf01f81a9
  path: path-to.zip
Waiting for processing to complete.
Current status: Invalid.....Processing complete
  id: 573ebf8c-5434-4820-86c6-46adf01f81a9
  status: Invalid
EOM
)

# Get status -- need word "invalid" from this line:  status: Invalid
status=""
regex='status:\s*(\w+)$'
[[ $res =~ $regex ]] &&
  status=${BASH_REMATCH[1]}

echo "stat=$status"

For me it says `stat=Invalid`. Why do you think it's not working? — Arkadiusz Drabczyk, Apr 27 '23 at 19:41
I'm running it right now and it's getting an empty string... interesting — c00000fd, Apr 27 '23 at 19:43
You're sure you run it with Bash? ZSH is now the default shell on Mac. I have `GNU bash, version 5.0.18(1)-release (x86_64-slackware-linux-gnu)` — Arkadiusz Drabczyk, Apr 27 '23 at 19:43
Hmm, with ZSH I get another error: `./so-status-regex.sh:18: read-only variable: status` — Arkadiusz Drabczyk, Apr 27 '23 at 19:44
@ArkadiuszDrabczyk: Oh, it could be ZSH. I'm saving it in a .sh file and just double-clicking it in the Finder. — c00000fd, Apr 27 '23 at 19:45
The default Bash on MacOS does not support PCRE character classes. A Posix regex would be `'status:[[:blank:]]*([[:word:]]+)$'` Try that. Posix character classes [HERE](https://www.regular-expressions.info/posixbrackets.html) — dawg, Apr 27 '23 at 20:13
Another issue: For `zsh` the variable `$status` is read-only. — dawg, Apr 27 '23 at 20:22

Gilles Quénot · Accepted Answer · 2023-04-27T20:49:08.077

1

warning the PCRE like syntax \w is not supported in your version of bash.

What I would do:

#!/usr/bin/env bash

# let's KISS: Keep It Simple Stupid
res='
Conducting pre-submission checks for my.zip and initiating connection to the Apple notary service...
Submission ID received
  id: 573ebf8c-5434-4820-86c6-46adf01f81a9
Successfully uploaded file
  id: 573ebf8c-5434-4820-86c6-46adf01f81a9
  path: path-to.zip
Waiting for processing to complete.
Current status: Invalid.....Processing complete
  id: 573ebf8c-5434-4820-86c6-46adf01f81a9
  status: Invalid
'

# Get status -- need this one:  status: Invalid
status=""
regex='status:[[:space:]]*([[:word:]]+)'
[[ $res =~ $regex ]] && status=${BASH_REMATCH[1]}

echo "stat=$status"

Output

stat=Invalid

The regular expression matches as follows:

Node	Explanation
`status:`	'status:'
`[[:space:]]*`	any character of: whitespace characters (like \s) (0 or more times (matching the most amount possible))
`(`	group and capture to \1:
`[[:word:]]+`	any character of: alphanumeric and underscore characters (like \w) (1 or more times (matching the most amount possible))
`)`	end of \1

For some doc about POSIX character classes, see POSIX spec's

edited Apr 27 '23 at 20:49

answered Apr 27 '23 at 19:40

Gilles Quénot

173,512
41
224
223

1

Hmm. Interesting. It still doesn't work. It's very puzzling... – c00000fd Apr 27 '23 at 19:47
Show output of `echo $ZSH_VERSION`. – Cyrus Apr 27 '23 at 19:51
Do you have bash in /bin? Replace the shebang with `#!/usr/bin/env bash` – Arkadiusz Drabczyk Apr 27 '23 at 19:51
@Cyrus: I'm getting `5.9` – c00000fd Apr 27 '23 at 19:52
@ArkadiuszDrabczyk: Yes, I'm using `#!/usr/bin/env bash` – c00000fd Apr 27 '23 at 19:52
@c00000fd: How do you run your script? – Cyrus Apr 27 '23 at 19:53
@Cyrus: Double-click in the Finder. – c00000fd Apr 27 '23 at 19:53
@GillesQuénot: I tried your script one-to-one and got `stat=` like mine. I have no idea why it doesn;t work – c00000fd Apr 27 '23 at 19:54
1

Post edited, the regex was wrong for your version of bash. Tested on FreeBSD with close to same than MacOs – Gilles Quénot Apr 27 '23 at 19:55
@GillesQuénot: OK, now you got it! Thanks. I'm wondering, why is it using its own version of regex syntax? – c00000fd Apr 27 '23 at 19:56
And is there a cheat-sheet for its regex? – c00000fd Apr 27 '23 at 19:57
You need a very recent bash to be able to use PCRE syntax `\s` and `\w` – Gilles Quénot Apr 27 '23 at 19:58
Really! Wow. macOS has some old version installed by default? Anyway, where do you get the regex syntax for it from? – c00000fd Apr 27 '23 at 19:59
From my mind ?! =) `GNU bash, version 5.1.16` here and `FreeBSD` have `GNU bash, version 5.2.15` – Gilles Quénot Apr 27 '23 at 20:00
:) OK, thanks. One last question, I see that if I want to match a GUID, like `573ebf8c-5434-4820-86c6-46adf01f81a9`, I can use `[[:xdigit:]]` to match hex digits, but how do I also match a dash? In other words, what shall I use instead of `([0-9A-Fa-f\-]+)`? – c00000fd Apr 27 '23 at 20:04
1

Use this : `[[:xdigit:]-]` Check `man tr` – Gilles Quénot Apr 27 '23 at 20:09
For `zsh` the variable `$status` is read-only. – dawg Apr 27 '23 at 20:26
OP explicitly used bash in code + tags dudes ;) – Gilles Quénot Apr 27 '23 at 20:27

Bash regex doesn't match the pattern with \w and $

1 Answers1

Output

The regular expression matches as follows: