Only last instance of repeated pattern gets coloured

[TheophileMot]

I tried this example from the docs:

Color in yellow file permissions with read rights for everyone: ls -l | colout '.(r.-){3}' yellow normal

and noticed that in a permission string like -rw-r--r--, only the last three characters were yellow, when I think we would expect the last nine characters (i.e., everything following the initial -) to be yellow.

For a simpler example:

echo test | colout '(.){4}' red

Here the final t is red, even though the match is really all four characters of test.

[tomasohara]

This could be by design. If you use outer parentheses, it works better:

$ echo test | colout '(.){4}' red | hexdump --canonical
00000000  74 65 73 1b 5b 31 3b 33  31 6d 74 1b 5b 30 6d 0a  |tes.[1;31mt.[0m.|
00000010

$ echo test | colout '((.){4})' red | hexdump --canonical
00000000  1b 5b 31 3b 33 31 6d 74  65 73 74 1b 5b 30 6d 1b  |.[1;31mtest.[0m.|
00000010  5b 31 3b 33 31 6d 74 1b  5b 30 6d 0a              |[1;31mt.[0m.|
0000001c

This reveals a separate issue, namely extraneous text added (i.e., "[test]t"). I'm in the course of documenting that.

[TheophileMot]

@tomasohara Thanks for looking into this. I had already tried your idea of using outer parentheses, but unfortunately the extraneous text made it unusable for what I needed to do. I don't think there's an issue for this, but it is mentioned on the main page under Limitations ("Don't use nested groups").

[nojhan]

This seem tricky to fix, given that groups have two meanings for colout (one for regexp, one for coloring).

[TheophileMot]

@nojhan How about parsing the input string to find the outer groups? For example, given ((abc)(def)gh)(ijkl), you could store outer_groups = [1, 4], and then iterate over that list instead of all groups.

[nojhan]

I unfortunately don't have time for this in a near future, but I am open to merging a pull request that would do that.

[TheophileMot]

@nojhan I thought about it and came up with an idea that's even simpler than parsing the input string: just let the user do it on the command line through another argument. This has the additional advantage of generally addressing your above point that "groups have two meanings for colout (one for regexp, one for coloring)", because now the user can decide which is which.

I made a pull request (#117) but I don't know Python so it will surely need tweaking; I used a global variable when I got stuck, but otherwise it seems to work as expected.