parse thru txt file with elements separated by {} brakets - linux

Question

Is there a way to parse thru a txt file that includes elements seperated by {}s
Here is a sample from the file:
virtual vs_devtnet_80 {
snat automap
pool pool_devtnet_80
destination 167.69.107.41:http
ip protocol tcp
profiles {
profile_http_health {}
tcp-lan-optimized {}
}
}
virtual vs_devdpp_4430 {
snat automap
pool pool_devdpp_5430
destination 167.69.107.31:https
ip protocol tcp
persist devdpp
profiles tcp-lan-optimized {}
}
virtual vs_devwww30_80 {
snat automap
pool pool_devwww30_80
destination 167.69.107.46:http
ip protocol tcp
profiles {
profile_http_health {}
tcp-lan-optimized {}
}
}
As you can see, the elements are separated, but with {}
Any help would be gladly appreciated. I was trying to use grep but it only returns one line...
I would like to be able to search by the top most element, for example searh.sh virtual vs_devtnet_80, and have it return the entire blob..furthermore perhaps be able to search for botht eh top layer and one of its sub layers, for example search.sh virtual vs_devtnet_80 pool which would return pool_devtnet_80

Answer 1

Something like:
cat .tmp | sed '/.*{/ ! { s/.*//g}'
this wont solve it completely, but i think it does something similar to what you want

Answer 2

Look at JSON pasers, they're written in all sort of languages, syntax looks similar enough to give you some ideas on how to tackle this.
Basically what you need to do is have a recursive function that calls itself whenever it encounters a '{' and returns the content whenever it encounters a '}'.
I wrote an article on a lisp-like parser that actually does just that. Check it out here for inspiration: http://www.codeproject.com/KB/linq/TinyLisp.aspx
Rgds Gert-Jan

Answer 3

This is Tcl syntax, so you can set up a mechanism to run it as a Tcl script that creates a data structure of itself.
#!/usr/bin/env tclsh
# create a safe slave interpreter
set slave [interp create -safe]
# set up the slave to parse the input file
$slave eval {
proc virtual {subkey body} {
eval $body
}
proc unknown {cmd args} {
upvar 1 subkey subkey ;# access the 'subkey' variable from the 'virtual' proc
if {[llength $args] == 1} {set args [lindex $args 0]}
dict set ::data $subkey $cmd $args
}
set data [dict create]
}
$slave expose source
# now do the parsing in the slave
$slave eval source [lindex $argv 0]
# fetch the data structure
set data [$slave eval set data]
# and do stuff with it.
dict for {subkey subdict} $data {
puts "$subkey destination = [dict get $subdict destination]"
}
And then, parser.tcl input_file outputs
vs_devaetnet_80 destination = 167.69.107.41:http
vs_devdpp_4430 destination = 167.69.107.31:https
vs_devwww30_80 destination = 167.69.107.46:http

Answer 4

This should give you the 2nd 'blob', for instance:
sed -n '/virtual vs_devdpp_4430.*/,/^}$/p' filename.txt
and pipe through grep -oP '(?<=pool ).*' to get what follows pool in that blob.

Answer 5

I ended up having to create a recursive function that first used strpos to find the search variable within the entire config file. Then using a recursive function pop'd on and pop'd off brackets to return the searched variable's entire body.