Support recursion across all archive types #47
Conversation
Generate changelog in
|
There was a problem hiding this comment.
Some minor feedback and PR needs to be deconflicted, but on the whole the structure is much nicer and looks to make sense.
In the abstract may have a bit of feedback on the WalkerProvider construction, but nothing that's fundamental/major enough to derail the current approach.
pkg/archive/paths_test.go
Outdated
| @@ -96,3 +105,21 @@ func mustZipFileReader(t *testing.T, path string) (*zip.Reader, func(*testing.T) | |||
| assert.NoError(t, file.Close()) | |||
| } | |||
| } | |||
| func mustTarGZReader(t *testing.T, path string) (*tar.Reader, func(*testing.T)) { | |||
There was a problem hiding this comment.
nit: add newline between end of previous declaration and start of this one.
pkg/archive/paths.go
Outdated
|
|
||
| // FileWalkFn is called by a WalkFn on each file contained in an archive. | ||
| type FileWalkFn func(ctx context.Context, path string, size int64, contents io.Reader) (proceed bool, err error) | ||
|
|
||
| // WalkZipFiles walks the files in the provided *zip.Reader, calling walkFn on each header. |
There was a problem hiding this comment.
Function can now be private (or even inlined into the one calling location) -- previously this was exposed since it was referenced directly, but no more need to do that with this refactor.
pkg/archive/paths.go
Outdated
| return err | ||
| } | ||
| // WalkTarFiles walks the files in the provided *tar.Reader, calling walkFn on each header. | ||
| func WalkTarFiles(ctx context.Context, r *tar.Reader, walkFn FileWalkFn) (err error) { |
There was a problem hiding this comment.
Same comment as for zip (can either be unexported or possibly even inlined to one calling location)
glynternet
force-pushed
the
gh/generic-archive-recursion
branch
from
d22b3b3
to
c325f6e
Compare
|
Needs changelog and conflict resolution |
glynternet
force-pushed
the
gh/generic-archive-recursion
branch
from
277b990
to
c4afd39
Compare
glynternet
force-pushed
the
gh/generic-archive-recursion
branch
from
c4afd39
to
b446ca8
Compare
Recursion supported across all archive types by making
crawl.Log4jIdentifiertype no longer holds any concepts of specific archives or archive types and then reusing pretty much the same approach that was used before.Instead, the following fields have been added to it:
When a file or path is encountered, the
ParseArchiveFormatfunction is used to determine if the file is an archive, and the archive type is passed along to theArchiveWalkersfunction to retrieve anarchive.WalkerProvider.A new
archive.WalkerProvidershould support creating aWalkFnfrom either a file or a reader. This is handy because during the first walk, we have access to files on disk and so do not have to load anything into memory. When we are recursing into archives we do not have access to disk as everything is currently in memory, so aWalkerProvidermust also provide aFromReadermethod to create one from anio.Reader. Reading into memory is only required by thearchive.ZipWalkersprovider currently, which still has a configurable memory limit, as it did before this PR (this situation will be improved in #40).