perl.md

grep, sed, and awk replacement

• Grep lines: perl -ne 'print $ARGV . ":" . $. . $_ if /pattern/' filename (or use unless)
• Recursive grep with ack (perl compatible regexps): ack pattern
• Recursive grep: find cache-write/ -type f -exec perl -ne 'print $ARGV . ":" . $. . $_ if /pattern/' {} +
• Modify lines: perl -i -pe 's/pattern/replacement/g' filename
• Delete lines: perl -i -ne 'print if not /bar/' filename
• Recursively modify files: ack -l --print0 pattern | xargs -r0 perl -i -pe 's/pattern/replacement/g'
• Print nth (9th in example) column, count starts from 0: perl -lane 'if (/pattern/) { print $F[8] }' ./file

hints

• Use ack instead of grep or find + perl, since it has better syntax than find+perl, and perl-compatible regexps, so you can test them.
• Few problems are reported by default, use strict and warnings modules while writing/debugging Perl code.

Misc

• booleans: "false" is anything of undef, 0, "0", "", and "true" is everything else. Note that this can lead to subtle error if you do while($line) where line is "0".
• processing stdin: use <>. This operator is a call to readline() wrapped. You can do e.g. my $line = <> Or while(<>) {…} is a "for-each-input-line" loop. And here map m{Directory.+trunk/(.+?)/}, <> we do replacement similarly for each input line.
• @F is the current line as array of columns, $_ is the line as string.
• qw(foo bar) makes an array of strings ('foo', 'bar').
• String replacement: inside a while(<>) loop use if (/pattern/)
• String replacement delimiter changed: given a if (/pattern/), add m to change delimiter, e.g. if (m{pattern})
• Set-builder notation: my @arr = map m{Directory.+trunk/(.+?)/}, <> here we map the function over stdin, and store result in arr
• Printing array without assigning to variable: you may need to explicitly order the code print ((map m{Directory.+trunk/(.+?)/}, <>)[0]).
• Note that some ways of splitting a string on newlines leave the newline inside the split string. Use split /\n/, $mystr or if you want a list of files in a dir, then glob().
• print /.*/ returns just 1: read Matching in list context in perldoc perlop. Basically, it would by default print matched groups in regexp or just 1 for success if none groups were used. So you can make it work with either print /(.*)/ or print /.*/g.
• string to number conversion: implicit. Just use a number-related op and conversion will happen automatically. E.g.: perl -e 'print "5.45" + 0.1'
• An example of summing up human-readable sizes from a file, represented by lines like 1 KiB, 3M, 4.7T, etc: perl -lane 'BEGIN{use Number::Bytes::Human qw(parse_bytes); my $sz = 0}; $sz += (parse_bytes $_); END{print $sz}' ./1.txt