Detect ansi clear lines and add ansi timestamps #1128
Conversation
|
<3<3<3 |
|
For further context, currently Buildkite can't show line timestamps like here:
This change will allow us to show timestamps on each of those lines. |
|
It needs some more testing, I suspect there are some more escape codes that we need to handle. There are some edge cases that I don't handle too. |
| // loop through line breaks and add prefix | ||
| for l := len(data); offset < l; { | ||
| // find either a newline or an escape char | ||
| next := bytes.IndexAny(data[offset:], "\n\x1b") |
There was a problem hiding this comment.
It would be nice to standardise on hex \x1b or octal \033 for escape chars in the codebase/docs/etc.
Currently there’s a bit of both in our codebase. Octal is a ridiculous radix. So this \x1b is good. Basically ignore this comment.
| for offset < len(data) { | ||
| next := bytes.IndexRune(data[offset:], '\n') | ||
| // loop through line breaks and add prefix | ||
| for l := len(data); offset < l; { |
There was a problem hiding this comment.
Would a bufio.Scanner be appropriate here, with a custom split function detecting new line or escape?
There was a problem hiding this comment.
Programs that need more control over error handling or large tokens, or must run sequential scans on a reader, should use bufio.Reader instead.
There was a problem hiding this comment.
bufio.Scanner fails on long lines, it's a pain.
|
Can we merge and see how it goes? |
|
I’ll test it now. |
|
I fixed the last remaining bug in the UI for this, so I'm keen to get this merged in. |
Currently the ANSI timestamps don't show up certain output (like docker) that uses creative ANSI control characters for cursor movements.
This detects things like
\x1b[Kwhich clears the current line and includes the ANSI timestamp technology.