EOF must be sent twice on stdin if no other input is sent

[willcassella]

If you run bat and then immediately press the EOF sequence (Ctrl-D), bat prints the filename 'STDIN' but continues to read from it. You have to press it again to actually close the file.

[sharkdp]

Hm, interesting. Thank you very much for reporting this.

I guess this is related to the first-line syntax detection.

[mziter]

I have found and resolved issue locally. However, I am not sure how to exit immeditately as process::exit(0) will exit immediately and not do any destructuring, etc...

Other two options are 1) Wrap new function in a result, which seems idiomatic, but would require a lot of downstream changes I am sure. 2) Wrap in a new ContentType that would basically get passed on and ignored at some future point.

Would like your thoughts...

[mziter]

Sorry, just saw this referenced already. :) Will put some comments on pull request.

[sharkdp]

closed via #492 by @reidwagner

[sharkdp]

Fixed in v0.10.0.

[reidwagner]

This has been closed - but there was a change in expected behavior, where we no longer get a header and footer for empty input files.

Trying to bring back that behavior while keeping the error logic here seemed to get complex. I think I have a simpler solution all around.

Instead of what's implemented in #492 (that could be reverted), we simply don't call print_file_ranges() if our reader.first_line is empty. That appears to give all the desired functionality, + it prints the STDIN header and footer when we press CTRL-D, which seems more consistent in the first place. e.g in print_file():

if !reader.first_line.is_empty() {
    self.print_file_ranges(printer, writer, reader, &self.config.line_ranges)?;
}

I think maybe the check that first_line.is_empty() could also be removed from read_line since the condition above would work for empty files.

[sharkdp]

Instead of what's implemented in #492 (that could be reverted), we simply don't call print_file_ranges() if our reader.first_line is empty. That appears to give all the desired functionality, + it prints the STDIN header and footer when we press CTRL-D, which seems more consistent in the first place

Sounds great!

I think maybe the check that first_line.is_empty() could also be removed from read_line since the condition above would work for empty files.

I can't tell, at the moment. We have to make sure to correctly handle (1) empty files (2) files with a single line (3) files with a single line and no \n (4) files with multiple lines (6) files with other encodings (mainly UTF-16 for now). I hope that most of this should be captured by tests, but it might be worth to double-check.

Edit: I have added some additional tests in #502 (not 100% related to this issue)

[reidwagner]

So I opened #504 with the simple change proposed + reverting #492, in case you want to get the fix in soon. As for removing the code from read_line I can double check that before making the change (either added to #504 or another PR).