Extended Regular Expressions (EREs): How do I include pattern but exclude specific superset of pattern in matches?

Question

The man page for the pgrep command indicates that the pattern "specifies an Extended Regular Expression for matching against the process names or command lines". Apparently, Extended Regular Expressions are an extension of the regular expressions used in the original UNIX grep command.

I need to be able to match the string 'myProcess' but exclude the string 'test_myProcess' using Extended Regular Expressions. Initially, I assumed that Extended Regular Expressions would support the negative lookbehind assertion, which led me to the following regular expression: (?<!test_)(myProcess). This apparently works for other regular expression tools (I tried it on the Regex101 website).

However, when I try it with pgrep, I get the following error: Invalid preceding regular expression. After looking at some Extended Regular Expression pages, it appears that the negative lookbehind assertion doesn't exist for this type of regular expression (or at least I was unable to find any reference to it). I need to know how to match the string 'myProcess' but exclude the string `'test_myProcess' using Extended Regular Expressions.

Edit: Examples of text I want to match (or not match) Match desired:

myProcess
/usr/local/bin/myProcess.py
/usr/bin/myProcess

Match NOT desired:

test_myProcess
/usr/local/bin/test_myProcess.py
/usr/bin/test_myProcess

Right now, I am toying with the idea of running pgrep twice, once matching 'myProcess' and once matching 'test_myProcess' and taking the difference of the two results as the correct answer. However, I am definitely interested in alternative solutions.

Edit: Note that the process name itself will not include 'myProcess'. It will simply appear in the command-line for that process. The process name will most likely be python2.7. And, 'myProcess' is not guaranteed to occupy either the beginning or the end of the string. Also, as pointed out, I want to match the string if 'myProcess' is present but not if 'test_myProcess' is also present.

That looks like a great tool. Unfortunately, I am not allowed to install anything on the target system so need to find a workaround using EREs if possible — tintedFrantic, Jan 17 '17 at 20:14
GNU grep supports Perl Compatible Regular Expressions (PCRE). Use the -P option; e.g. grep -P "(?<test_)myProcess" — JRFerguson, Jan 17 '17 at 20:31
I have edited my answer to include a better example of exactly what I am trying to do. Let me know if I can clarify further. — tintedFrantic, Jan 17 '17 at 20:36
That is a good point. In this case, I would NOT want it to report a match if test_myProcess is present. — tintedFrantic, Jan 17 '17 at 20:43
The comment above by @JRFerguson about grep -P seems promising... What I could do instead of pgrep -f is ps auxf | grep -P '(?<test_)myProcess'. But, I suppose that would still match if both were there. Is there a regular expression that would not match if both are present? — tintedFrantic, Jan 17 '17 at 20:46
@don_crissti If I could, I would upvote your last comment. The only thing I changed was to add a trick to not select the grep commands themselves. So, my final solution is ps auxf | grep -E [m]yProcess | grep -vE test_[m]yProcess. Thanks so much. — tintedFrantic, Jan 17 '17 at 21:00
Some background: Why does my regular expression work in X but not in Y? — Gilles 'SO- stop being evil', Jan 17 '17 at 23:50

score 2 · Accepted Answer · edited Jan 17 '17 at 21:39

2

After quite a bit of discussion with @don_crissti, the final solution was

ps auxf | grep -E '[m]yProcess' | grep -vE 'test_[m]yProcess'

This will first get all the lines containing 'myProcess' and then match all lines NOT containing the lines with 'test_myProcess' (the -v inverts the match). The -E coupled with the brackets simply serves to prevent matching the two grep processes that will be listed in the ps output. This turned out to be much simpler than trying to find one regex to rule them all. Thank you all for your help.

edited Jan 17 '17 at 21:39

Stéphane Chazelas

544,893

answered Jan 17 '17 at 21:10

tintedFrantic

39

You could've also used ps auxf | grep -P (?<!test_)myProcess since you were trying to use perl-compatible regular expressions, not mere extended regular expressions. – Joe Sewell Dec 17 '20 at 20:00

score 1 · Answer 2 · answered Jan 17 '17 at 21:45

1

(?<!) is a perl regexp operator. The specification for the standard ERE can be found here.

Sounds like here you want to find processes whose argument list either starts with myProcess or contains /myProcess.

So:

pgrep -f '(^|/)myProcess'

answered Jan 17 '17 at 21:45

Stéphane Chazelas

544,893

score 0 · Answer 3 · edited Jan 17 '17 at 21:38

0

So, you want to pgrep the processes whose name starts with "myProcess"? Then use the regex ^myProcess:

pgrep '^myProcess'

In a regex, the ^ character means that what follows has to be found at the beginning of the string.

Example:

$ pgrep -a app 
3267 nm-applet
3280 /usr/bin/python /usr/share/system-config-printer/applet.py
3297 /usr/lib/mate-panel/wnck-applet
3306 /usr/lib/mate-panel/clock-applet

$ pgrep -a '^app'
3280 /usr/bin/python /usr/share/system-config-printer/applet.py

edited Jan 17 '17 at 21:38

Stéphane Chazelas

544,893

answered Jan 17 '17 at 20:46

xhienne

17,793
2
53
69

Unfortunately the process name does not necessarily begin that way. In fact, the process name is python2.7, but the command-line is different depending on where the script I am running is installed. It will have 'myProcess' somewhere in the command-line (I use pgrep -f) – tintedFrantic Jan 17 '17 at 20:48
@tintedFrantic Can you edit your question and give a real world example please? You example is misleading. – xhienne Jan 17 '17 at 20:49
@don_crissti I will edit the question title. Thank you. – tintedFrantic Jan 17 '17 at 21:01

Extended Regular Expressions (EREs): How do I include pattern but exclude specific superset of pattern in matches?

3 Answers3