cksum: even more fixes #6929
cksum: even more fixes #6929
Conversation
6bbbb87 to
31de97a
Compare
|
GNU testsuite comparison: |
|
a few jobs are failing |
42997a2 to
f36fe46
Compare
|
GNU testsuite comparison: |
f36fe46 to
219a91d
Compare
|
GNU testsuite comparison: |
219a91d to
00d1730
Compare
|
GNU testsuite comparison: |
eda6659 to
f836b08
Compare
I've fixed what I could. |
|
GNU testsuite comparison: |
5f22dad to
5e92491
Compare
|
GNU testsuite comparison: |
5e92491 to
6873fe2
Compare
|
GNU testsuite comparison: |
| let (filename_to_check_unescaped, prefix) = unescape_filename(filename); | ||
| let real_filename_to_check = os_str_from_bytes(&filename_to_check_unescaped)?; |
There was a problem hiding this comment.
I suggest to rename filename_to_check_unescaped to unescaped_filename_to_check to match the naming of real_filename_to_check. And I think the _to_check suffices can be dropped in both cases.
c410c28 to
e7fa8dd
Compare
…of the expected digest
…ore and introduce LineFormat struct
e7fa8dd to
958222a
Compare
|
Thanks :) |
|
@RenjiSann btw, you will see that the fuzzer finds some issues: to run it locally: |
Indeed. Many tests fail because the cksum installed in the CI is older than ours (probably GNU coreutils 9.1 which is Ubuntu's default), which does not support --base64 and --raw. Another issue is that --length validity should be checked before the --check argument. |
Fixes #6653
This PR changes the way the regex to parse a checksum line is chosen. Now, every line gets to choose the regex it matches with, instead of trying to match with the regex that matched the first line (except for non-algo based line which should follow the flavor of the first one).
While doing this, I've decided to get rid of the
regex::Capturewhich was being carried around everywhere, and create aLineInfostruct that holds all the parsed data.It adds
lazy_staticas a direct dependency touu_core(it was an indirect dependency until now). I allows to not recompile the regexes again and again.EDIT: I have added some code to rework the way the expected checksum is decoded, this PR now fixes #6576 and #6614 as well.