On Thursday 07 November 2002 9:38 am, Jean-Marc Lasgouttes wrote:
> >>>>> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
>
> Lars> We used to have code in the xfroms FileDialog that made a bit
> Lars> simpler for beginners, in that it allowed simple glob patterns
> Lars> as well.
>
> Lars> Typically we just rewrote "*" -> ".*" and "." -> "\."
>
> Lars> I don't agree on the "long time" argument.
>
> You can get used to regexps in a short time if you _want_ and _need_
> to use them. We are talking about people who do not care about them,
> and for whom this will only be an annoyance. I do not think that the
> interface should be complicated for most people, just for the benefit
> of a handful of power users.
>
> Let's admit it, even for people who like to use regexps, most of the
> time searches in bibliography do not require them. This is just a
> nifty gadget.
Sure, I wan't proposing to change the interface, just on what happens behind
the scenes.
If the user wants to search for a 'simple' string then we would need to pass
search_expr through an RE first to ensure that all special chars are escaped.
That's all.
Reading the boost::regex docs:
// All characters are literals except: .|*?+(){}[]^$\
// These characters are literals when preceded by a "\".
This sed expression works:
aleem-> search_expr='. | * ? + ( ) { } [ ] ^ $ \'
aleem-> echo "${search_expr}"
. | * ? + ( ) { } [ ] ^ $ \
aleem-> echo "${search_expr}" | \
> sed 's;\([]\\\.\[\|\*\?\+\(\)\{\}\^\$]\);\\\1;g'
\. \| \* \? \+ \( \) \{ \} \[ \] \^ \$ \\
I suppose we can't do the equivalent in 1 line with boost, but I'm sure it
woun't take 40 lines either ;-)
Angus
