Case Munging

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

On a Linux mailing list near me a user asked:

  I have a list of company names all in upper case; one per line.
  Can I use a combination of sed, tr, and other tools to automate
  converting them all to mixed case (i.e., initial letter uppercase,
  all others lower case)? There must be a shell script already
  written to do this, but not in the reference books I have.
[download]

I immediately blasted out a knee jerk reply:

   perl -ne 'print ucfirst lc;' FILENAME
   will do it.
[download]

And then proceeded to read the answers others had provided.

Some suggestions caused me to wonder where the virtue of Laziness is hiding:

   perl -pe 'tr/A-Z/a-z/; s/(\S+)/\u$1/g;' <filename>
[download]

Others reminded me to think before posting

   My sed-fu was deficient I guess (or my sed was) so I fell back to p
+erl:

   If you want word-case
   perl -pe 's/ \b (\w) ([^\s]+) \b /\1\L\2/gx'

   # ONE COMPANY  -> One Company (rather than One company)
[download]

Oh, yes, multi word company names. How ... almost all the blathering time.

So I went back to redo my original and came up with:

   perl -i -ne 's/(\b\w)/{uc $1}/eg; print;' <FILENAME>
[download]

Which I'm content with. But...can you come up with something better?

There is one nit I have with that solution. RUN4LIFE translates to Run4life rather than Run4Life. My reading of `man perlre` doesn't turn up a \ code for alpha only. The systems I have available to me at the moment don't support [:class:] for trying that out.

Be Appropriate && Follow Your Curiosity

Comment on Case Munging Select or Download Code

Replies are listed 'Best First'.
Re: Case Munging by johngg (Canon) on Apr 16, 2009 at 21:52 UTC
You could split on anything that wasn't an uppercase letter (`[^A-Z]`), capturing what you `split` on, do your `ucfirst lc` in a map then concatenate the results with join. `$ cat companies WOOLWORTHS CAP GEMINI AVIVA AMERADA HESS RUN4LIFE $ perl -pi.bak -e '$_ = join q{}, map { ucfirst lc } split m{([^A-Z]+) +};' companies $ cat companies Woolworths Cap Gemini Aviva Amerada Hess Run4Life $` [download] I hope this is useful. Cheers, JohnGG	[reply] [d/l] [select]
Re: Case Munging by almut (Canon) on Apr 16, 2009 at 21:22 UTC
My reading of `man perlre` doesn't turn up a \ code for alpha only. The systems I have available to me at the moment don't support `[:class:]` I think you should be able to use the character class `[^\W\d_]` as a substitute (which should support locales and unicode — as opposed to homebrewn ranges like `[A-Za-z]`).	[reply] [d/l] [select]
Re: Case Munging by morgon (Priest) on Apr 16, 2009 at 21:52 UTC
What about this: `perl -i.old -pe 's/(?<![a-z])(\w)/uc $1/eg' <FILENAME>` [download] After this runs you have all "translated" names in <FILENAME>. The old version of the file is kept as "<FILENAME>.old". And it translates run4life to Run4Life as requested. And in case the regex is unclear: The first part is a negative lookahead (see perldoc perlre).	[reply] [d/l]
Re^2: Case Munging by morgon (Priest) on Apr 17, 2009 at 00:15 UTC
It just occurs to me that you said the file holds the names in upper cases (my previous post assumed lower-case). So use this instead: `perl -i.old -pe 's/(?<=[A-Z])(\w)/lc $1/eg' <FILENAME>` [download] And the regex contains a (this time positive) lookbehind-assertion (and not lookahead as I wrote above). Sorry for the confusion.	[reply] [d/l]
Re^3: Case Munging by AnomalousMonk (Archbishop) on Apr 17, 2009 at 02:16 UTC
FWIW, the following may give a slight improvement in speed, simplicity or generality (I will not bother to put it into file-processing form): `>perl -wMstrict -le "for (@ARGV) { s{ (?<= [[:upper:]]) ([[:upper:]]+) }{\L$1}xmsg; print; } " "XYZZY" "RUN4LIFE" "RUN42LIFE CO" "GENERAL WIDGET CO., INTL" "MI-GO BRAIN CYLINDERS, IPTY." Xyzzy Run4Life Run42Life Co General Widget Co., Intl Mi-Go Brain Cylinders, Ipty.` [download] Supposed advantages: The `([[:upper:]]+)` quantified capture grabs as much as possible of whatever is to be lower-cased (rather than capturing a character at a time); The use of the `\L` interpolation modifier eliminates the need for evaluation in the regex (i.e., no `//e` regex modifier); The use of `[[:upper:]]` and `\L` bring locale fully into play (I think).	[reply] [d/l] [select]
Re^3: Case Munging by mikeraz (Friar) on Apr 17, 2009 at 02:38 UTC
Good clarification for the archives. I'd mentally prefaced `lc <INPUT>` since that was a necessary step to get to the end state. Be Appropriate && Follow Your Curiosity	[reply]
Re^2: Case Munging by mikeraz (Friar) on Apr 17, 2009 at 00:20 UTC
I like that, just the kind of thing I was fumbling towards. Be Appropriate && Follow Your Curiosity	[reply]
Re: Case Munging by Bloodnok (Vicar) on Apr 17, 2009 at 14:17 UTC
Hmmm , An exercise in sed(1) for the interested reader, I'm game ... `sed 's,.,\L&,; s,.,\u&,; s,[ 0-9][^ ],\U&,g' file ^ ^ ^ \| \| \| \| \| \| \| \| Capitalise any non-space char after a + space or a numeric \| Uppercase the first char Lower case the whole name` [download] `echo " FIRST SECONF THRID RUN4LIFE" \| sed 's,.,\L&,; s,.,\u&,; s,[ 0-9][^ ],\U&,g' First Seconf Thrid Run4Life` [download] A user level that continues to overstate my experience :-))	[reply] [d/l] [select]