Parse a text file and replace text

[arsenik]

I want to parse a C file and replace the following example line:

function(x(a,b,c),y,x(d,e,f))

with:

function(b,y,e)

I have a regular expression as:

x\([^,],[^,],[^\)]) to parse the needed variables

How would I go about doing this in a bash shell script?

[snova]

Look into sed. I don't know exactly how.

Alternatively, why not use a language with regexp features built in, such as Perl?

Also, why?

[arsenik]

I stated what I needed in the first message. I am a programmer, its for a C program I created. I could use perl, sed, or awk. That's why I'm here asking for help :)

[geirha]

Once you've found the proper replacement regex ...


$ echo 'function(x(a,b,c),y,x(d,e,f))' | sed -r 's/\<function\(\w+\(\w+,(\w+),\w+\),(\w+),\w+\(\w+,(\w+ ),\w+\)\)/function(\1,\2,\3)/'
function(b,y,e)


Run it on the file


sed -r 's/\<function\(\w+\(\w+,(\w+),\w+\),(\w+),\w+\(\w+,(\w+ ),\w+\)\)/function(\1,\2,\3)/' file.c > newfile.c
# Alternatively add the -i option to make the replacement in place
sed -r -i~ 's/\<function\(\w+\(\w+,(\w+),\w+\),(\w+),\w+\(\w+,(\w+ ),\w+\)\)/function(\1,\2,\3)/' file.c

[arsenik]

Thanks I appreciate the help. I understand everything but the \w+ 's . What do they represent in regex?

[ghostdog74]

no need regexp to complicate things, just use fields/delimiters


# echo "function(x(a,b,c),y,x(d,e,f))" |awk -F"," '{print "function("$2","$4","$6")"}'
function(b,y,e)

[geirha]

\w+ should be equivalent to [a-zA-Z0-9_]+ or something close to that. I never quite remember what exactly it matches, but in the case of source code, it should match functions and variable names nicely.