trst - kind of like sed
trst is a program to read input streams, replacing certain patterns with certain other patterns. It is modelled loosely on the classic sed program, but it aims to be easier to use while remaining as powerful in most use cases.
tr is for TRanslate, TRansfigure, TRansform, TRansmute. st is for STrings or STreams.
"sed" is Spanish for thirst; "trst" isn't too far from how a native Spanish speaker might pronounce thirst.
When using sed, have you ever got your regular expressions right quickly, only to spend far too much time getting escaping -- both for the shell and for sed itself -- right? I have.
trst attempts to alleviate that pain through a variety of convenient features.
-
Easy escaping and easy to read positional arguments
By default when '%' starts an expression the expression is escaped. (You can choose a different character for this if you wish.)
Try this:
echo '$dont[dothis] = 123;' | ./trst.py all '%$' '[a-zA-Z0-9]+' '%[' '[a-zA-Z][a-zA-Z0-9]*' '%]' "\1\2\3'\4'\5"Now try to do it in sed. Modulo tyops, did it work right for you the first time with sed? It didn't for me, but with trst it did.
-
The --spaces argument:
Try this:
echo 'first second third' | ./trst.py --verbose --spaces all first second third '\3 \2 \1'Or this:
echo 'first second third' | ./trst.py --verbose --spaces all first second third '\3 \2 \1' -
The --debug argument:
Of course, sometimes debugging your regular expressions really is the hard part. If so, try this:
echo 'hello world' | ./trst.py --debug all 'h[ea]llo' world 'wow dude'Then try this:
echo 'world hello' | ./trst.py --debug all 'h[ea]llo' world 'wow dude'
First and most important: trst's command line arguments and flags are
experimental and subject to change. If you use it in a script your
script could break later. In particular, the all argument is likely
to go.
Second: It's certainly buggy. I don't know how well it will work with characters not included in ASCII. Nor do I know how it will deal with different encodings. Getting this in order is important, but in the meantime, please don't use it on important files you don't have backed up or which aren't in version control.
Third: traditional sed is part of an interesting conceptual lineage which can be traced back at least as far as The Standard Text Editor and which extends to vi and its many clones and extensions. trst discards this lineage entirely. There are cool things you can do with sed that you'll probably never be able to trst.
Easy prototyping. Getting the command line parsing right is tricky, not because libraries to do it are deficient (quite the opposite in fact) but rather because it isn't really clear at all what the command line arguments should be. After all, if there's one thing git has taught us, it's that getting a coherent command line interface right is paramount, and you can worry about getting internal structures right later. Right? Right. Anyway, doing it in Python means that I -- and hopefully you too -- can experiment easily.
Once I'm reasonably happy with how it works, rewriting it in a faster language is something I'm open to.
- Improve arguments and their semantics
- Create a test suite
- Fix bugs
- Come up with a ridiculous logo.
Usage ideas:
trst [general args] [search args] [args]
Questions:
-
What sorts of subcommands make sense?
Possibilities: replace-all, replace-any. If none, assume replace-all?
-
What sorts of general args make sense?
Apply more than once in a given line -- cf sed's /g -- by default?
-
How to avoid or mitigate shell quoting annoyances?
-
What symbols -- ie, more or less anything you get by hitting shift + number -- are available for commandeering?
Keep in mind that the symbols easily available on your keyboard might not be so easily available on someone else's, especially when that someone else lives in a different country.