mavericknik has asked for the wisdom of the Perl Monks concerning the following question:

Hello monks, I have a question about modifying a file I have on. It is a verilog netlist, so the lines are so: 
 DFFX1 k0_reg_184  ( .D(key_184 ), .CLK(clk), .QN(n18736) );
DFFX1 k0_reg_183  ( .D(key_183 ), .CLK(clk), .Q(k0_183 ), .QN(n65993)
         );

[download]
So some cells are contained in one line, whereas some cells span 2 lines(2nd one above). I'd like to move cells spanning 2 lines into one. I know the lines not containing cells will not have a capital X. So if i search for lines that dont match "X", I get all lines that should be appended to the line above it. How would I do this for a very large(~600000) line file? I plan to modify the netlist after this to make the changes I require.

Replies are listed 'Best First'.
Re: File modification on the fly (Verilog)
by toolic (Bishop) on Aug 02, 2016 at 21:13 UTC
    Parsing Verilog files can be tricky, and you should probably use a parser: Verilog-Perl. But, this might work for simple files: 
    use warnings;
use strict;

# Change input record separator to parse whole Verilog lines
local $/ = ';';
while (<DATA>) {
    s/\s+/ /g;    # Convert all whitespace to single-space
    print "$_\n";
}

__DATA__
 DFFX1 k0_reg_184  ( .D(key_184 ), .CLK(clk), .QN(n18736) );
DFFX1 k0_reg_183  ( .D(key_183 ), .CLK(clk), .Q(k0_183 ), .QN(n65993)
         );

    [download]

    Output: 

     DFFX1 k0_reg_184 ( .D(key_184 ), .CLK(clk), .QN(n18736) );
 DFFX1 k0_reg_183 ( .D(key_183 ), .CLK(clk), .Q(k0_183 ), .QN(n65993) 
+);

    [download]
[reply]
[d/l]
[select]
      Thank you so much for your help, I works perfectly. If I understand your script correctly, you are setting the line delimiter to a ";" and replacing all multiple spaces to single spaces? I have parsed the netlist using verilog-perl, but the time requirement seems to increase exponentially with number of cells. For older netlists with ~5k cells, the script takes about 12 hours. With my new netlist of about 450000 cells, I'm estimating a runtime of about 200 days which is unacceptable so I'm looking at alternative approaches.
[reply]
        Thank you so much for your help
        You're welcome.
        If I understand your script correctly, you are setting the line delimiter to a ";"
        Yes.
        and replacing all multiple spaces to single spaces?
        \s refers to all types of whitespace; refer to perlre. The key here is that \s includes the newline character (\n). So, \s+ replaces multiple adjacent spaces, tabs and newlines. s///g changes all whitespace on a line.
        For older netlists with ~5k cells, the script takes about 12 hours.
        That is a surprisingly long time. Can you post your code for that?
[reply]

        Depending on how your program works you can maybe gain some speed from using auto flush $| = 1;. OUTPUT_AUTOFLUSH

[reply]
[d/l]

Back to Seekers of Perl Wisdom