paulwratt/tr-replacement

A replacement for `tr`. A superset of coreutils `tr` "a filter to translate characters". One with RegEx.

tr-replacement

A replacement for tr. A superset of coreutils tr "a filter to translate characters". One with RegEx.

Compatibility

The commands tr2 and tr3 will always be a superset of tr which means it is a drop-in replacement at the command level AND the source file level. If you build tr2 yourself, you can substitute tr with a link to tr2, or replace the command /usr/bin/tr.

If you are building coreutils package you can replace src/tr.c with the renamed file tr2.c. Or you can usetr3.c if you always want regular expressions available. Either will then build as tr command as part of coreutils.

Detecting

You will always be able to tell the difference be an original tr and one replaced with tr2, as -2 switch will not return an error code. You can always test for the presence of /usr/bin/tr2 or tr -2.

I you want to detect the presence of regular expressions, you can use tr -3. This will be (again) a superset of tr2, as tr3. This just allows system admins to decide just how much bloat they want with there coreutils. On some systems it does not make sense to run a tr command that requires extra resources needed for regular expressions.

Differences

There is no difference, it is a superset, so all the command line switches available in tr are also available in tr2 and tr3. Ok, so what is this superset of switches:

-2              errors if not tr2/tr3 (has tr extensions)
-3              errors if not tr3 (has regular expressions)
-a <n>[,<m>]    only action on characters, discard the rest.
-n <n>[,<m>]    number of replacements.
-m <n>[,<m>]    begin and end of replacements.
-o <n>[,<m>]    omit characters before input.
-y <n>[,<m>]    yeild characters after output.
-V [<char>]     only output text not changed, optionally use <char> for those that have changed
-Q <string>     query <string> for search characters instead of SET1
-S <string>     substitutue <string> for replaced characters instead of SET2
-Z              use repeating "background" replacement of <string> with `-S` instead of SET2
-R              reverse initial input string
-r              reverse final output string
-W              also write the characters excluded from `-a` in final output

The switches with n and m work the same as {n,m} does in regular expresions, so:
tr2 -a 10 a A will only apply matches from characters 10 onwards, discard the rest.
tr2 -aW 1,10 a A will only apply matches to the first 10 characters, dont discard the rest.
tr2 -n 5 -d a will stop deleting charcater a after the 5th match.
tr2 -n 2,5 a b will replace at least 2, but not more than 5 characters a with character b.
tr2 -m 5 -d B will delete only from the 5th match of character B.
tr2 -m 2,5 a A will replace only from the 2nd to the 5th match of characters a with character A.

However the switches with n and m also have a - action not found in regular expresions, so:
tr2 -a -10 a A will only apply matches to the last 10 characters, discard the rest.
tr2 -a 2,-2 a A will only apply matches from to the string line 2 characters indented from each end, discard the rest.
tr2 -n -5 -d a will only delete the last 5 matches of charcater a.
tr2 -n 2,-5 a B will replace the last 5 matches, but stop after 2 matches of characters a with character B.
tr2 -n -7,5 a b will replace the last 7 matches, but not more than 5 characters a with character b.
tr2 -m -5,-2 a A will replace only from the last 5 matches to the last 2nd match of characters a with character A.

Also with n and m you can use [:even:] and [:odd:] to manipulate the switch input and output. the same can be done with th when added to a number, therefore:
tr2 -a [:even:] a A will only apply matches on even numbered chacaters, discard the rest.
tr2 -a 3th a A will only apply matches every 3rd chacater, discard the rest.
tr2 -Wa [:odd:] b B will only apply matches on odd numbered chacaters, dont discard the rest.
tr2 -n 5th -d a will only delete the 5th match of charcater a.
tr2 -m [:odd:] -d a will only delete every odd numbered match of charcater a.

That covers just about anything I can think of for the basic extension to tr. The following allows tr to manipulated input an output text string lines before and after actions taken, using regular expression matching to decide when those actions are taken, and wheather or not to output the original line (ala sed) or no line at all (ala grep) when there are not pattern matches.

RegEx

The command tr3 takes tr2 and adds grep compatible regular expression command line switches, and a way to detect its presence:

-2              errors if not tr2/tr3 (has tr extensions)
-3              errors if not tr3 (has regular expressions)
-e <pattern>    action only on regular expression pattern found
-E              pattern is an extended regular expression (ERE)
-F              pattern is a set of newline seperated fixed strings
-G              pattern is an basic regular expression (BRE)
-P              pattern is an Perl regular expression (PCRE)
-f <filename>   get pattern from file
-i              ignore case
-v              invert pattern match
-w              force pattern to macth a whole word
-x              force pattern to match a whole line
-z              a line ends with 0 byte, not newline
-A              match the pattern after everything else, otherwise on `-a` match
-B              match the pattern before everything else, otherwise on `-a` match
-D              dont drop line output if no match found (SED style)

see man grep for some regular expression usage information, or do an internet search for examples.