-e means execute this ... -n ... for each line in the input ... -p ... and print the result
s/...// replaces anything that matches with the empty string. The /s on the end makes . match newlines.
^ matches the beginning of a line
^(foo|bar) matches either foo or bar at the beginning of the line
(?<!foo)bar is a negative lookbehind assertion, it matches bar if not preceded by foo, so (?<!SQL: ).* matches any text not preceded by SQL:
So the whole regex means "match either SQL: at the beginning of a line, or any line not beginning with SQL: and replace it with the empty string". All we're left with is the SQL queries.
And if I've counted correctly, the bit inside 'single quotes' is one character shorter than the two vim commands and the newline between them, although I'll not mind if you consider the -pne to be cheating :-)
perl -pne 's/^(SQL: |(?<!SQL: ).*)//s'-emeans execute this ...-n... for each line in the input ...-p... and print the results/...//replaces anything that matches with the empty string. The/son the end makes.match newlines.^matches the beginning of a line^(foo|bar)matches either foo or bar at the beginning of the line(?<!foo)baris a negative lookbehind assertion, it matchesbarif not preceded byfoo, so(?<!SQL: ).*matches any text not preceded bySQL:So the whole regex means "match either
SQL:at the beginning of a line, or any line not beginning withSQL:and replace it with the empty string". All we're left with is the SQL queries.And if I've counted correctly, the bit inside 'single quotes' is one character shorter than the two vim commands and the newline between them, although I'll not mind if you consider the -pne to be cheating :-)
Yay, and here it is: the shorter way. David, have you ever played regex golf? :)