Create custom-collections from txt file [v1.3, final]

pjft

This is the regular expression of things I allow/exclude:

[^A-Za-z0-9-[]()\s']

So, we should allow:

letters (lower and upper case)
numbers
spaces
'

I believe none of these would pose problems with any file system. That was the main criteria.

cyperghost

@meleu As whitespaces are working ~~(even if they are not allowed!)~~ I would suggest following

if [[ "$line" =~ ^[^\/]+$ ]] && [[ -n "$line" ]]; then

remove the whitespace in the regex but check $line if empty or not.

Maybe pjft can tell more ;)
Goodnight mate

EDIT: Was to late
So we should allow whitespaces ;)

pjft

@cyperghost I think in concept everything should work as long as it's accepted by the file system. I just wanted to play it safe and create a minimum common denominator set of characters.

cyperghost

@pjft I think so but as I'm not a C++ coder I can't exactly tell. Thank you for taking part!

The touch command is a smooth way to leave files unchanged so this script seems to be save even if we update with new collections. The exclusion that @meleu made ... the two slashes are correct. But I wan't avoid that LF entries create custom-.cfg files therefore a check of $line parameter is needed.

I will update first post and it's codes as @meleu agrees to changes - maybe he gots a better idea ;)

meleu

@pjft said in Create custom-collections from txt file:

This is the regular expression of things I allow/exclude:

[^A-Za-z0-9-[]()\s']

So, we should allow:

letters (lower and upper case)

numbers

spaces

'

Are you saying that this is what currently works with your custom collections feature? Or is it your personal opinion?

meleu

@cyperghost said in Create custom-collections from txt file:

@meleu As whitespaces are working ~~(even if they are not allowed!)~~ I would suggest following
if [[ "$line" =~ ^[^\/]+$ ]] && [[ -n "$line" ]]; then
remove the whitespace in the regex but check $line if empty or not.

looks like you didn't test that... :-) that regex don't match empty strings. Try it and see.

prompt$ [[ "" =~ ^[^\/]+$ ]] && echo valid || echo invalid
invalid

Translating that regex:

^ - start of the line
[^\ /] - any character that is NOT a space '\ ' or a slash '/'.
+ - one or more occurences (here is the trick for not allowing empty strings)
$ - end of the line

pjft

@meleu those are the checks I implemented for when the user can type a collection name.

cyperghost

@meleu You did forget the timeshift ;)
I updated first posting with script!
It works flawless...
Whitespaces in filenames of collections were never the problem

@pjft I think the simple RegEx method inside ES is okay and covers all BE/AE usecases

meleu

@cyperghost just to match the same limits @pjft imposed to what the user can type for a collection name, I suggest this regex in the script: ^[][()\ \'A-Za-z0-9-]+$.

Then that if inside the while becomes this:

if [[ "$line" =~ ^[][()\ \'A-Za-z0-9-]+$ ]]; then

@pjft out of curiosity: is the underscore _ invalid?

pjft

@meleu I suppose it is, by omission, not deliberately. As I said, I tried to get a regular expression that would match the majority of names people would try to enter - it was not extensive, but had the main concern of not allowing people to enter characters that would fail miserably in some OSs ('/', '*', '?', etc).

Underscores could certainly be added. My recommendation is that if you manage to try it out and confirm it works via your script in ES, then by all means keep it. As I mentioned, these checks are only used at text entry time, so they shouldn't affect in any way how ES would work if you create them elsewhere. But if they fail, don't hold me against it :)

cyperghost

@meleu I updated!
You restrict underscore with this but well ... I think that enough

Thx for your help and in creating generic solutions ;)

@pjft

But if they fail, don't hold me against it :)

Always ;)

cyperghost

@meleu @pjft
I updated previous post to give tributes.
Well I think there are other characters that can be included. But I think space and "-" and () are good way.
The underscore works btw ;)
What do you both think.... Is v1.3 with RegEx ^[][()\ \'A-Za-z0-9-]+$ final? or should it be extended. For my personal usage it's enough

We can add \_ and then the underscore works, what's next? #,&,!,?,~.....

pjft

@cyperghost No, please stay away from most of the characters you just mentioned, as some of them are special characters in different file systems (~, ?, etc).

I'd rather keep the REGEXP exactly the same for compatibility purposes. I don't like to give people tools for them to easily shoot themselves in the foot :)

cyperghost

Okay... I update script (hopefully last time) and add the underscore.

meleu

@cyperghost said in Create custom-collections from txt file [v1.3, final]:

We can add _and then the underscore works

There's no need to scape underscore. Then the regex is ^[][()\ \'_A-Za-z0-9-]+$

w0lv3r1nix

Hi

Really nice idea ! Thanks

Suggestion :
Instead of using if/else, just sanitize the string with the regex used in the if:
sline=${$line//^[][()\ '_A-Za-z0-9-]/}

cyperghost

@meleu updated
Can we discuss the RegEx things on the bash script thread?
It's a very powerfull feature indeed ;)

@w0lv3r1nix
Well that's also a suggestions there a many ways to skin a cat ;)
As I'm still learning I'm thankfull for every little bash lesson.
But I think the if/else clause is easier for code readings and beginners like me.
See if you stumple after years on your piece of code.... you would ask... what was my intentions there.

With if you have a expression with a clear result ... no if without then and maybe an else >> good result ;)

meleu

@cyperghost said in Create custom-collections from txt file [v1.3, final]:

Can we discuss the RegEx things on the bash script thread?

Sure. Go ahead. :-)

meleu

@cyperghost said in Create custom-collections from txt file [v1.3, final]:

TRIM CR from windows files

Just a small (bad) joke:

cyperghost

@meleu We just trim CR ;)
No need for getting mad ... But well we live in WINDOWS world ;)

Sorry mate - I'm currently hard working :(
So no time for asking bash ;)