oss/syft
2
0
Fork 0
mirror of https://github.com/anchore/syft.git synced 2026-10-31 14:36:33 +01:00
Code Issues Packages Projects Releases Wiki Activity

Use file indexer directly when scanning with file source (#3333)

Browse Source 
* Use file indexer when scanning with file source

Prevents filesystem walks when scanning a single file, to
optimise memory & scan times in case the scanned file
lives in a directory containing many files.

Signed-off-by: adammcclenaghan <adam@mcclenaghan.co.uk>

* Create filetree resolver

Shared behaviour for resolving indexed filetrees.

Signed-off-by: adammcclenaghan <adam@mcclenaghan.co.uk>

---------

Signed-off-by: adammcclenaghan <adam@mcclenaghan.co.uk>
This commit is contained in:
Adam McClenaghan
2024-11-22 11:53:53 -05:00
committed by GitHub
parent 8abd97a5bf
commit 21df38798e
8 changed files with 2247 additions and 1609 deletions
Download Patch File Download Diff File Expand all files Collapse all files
syft/source/filesource/file_source.go
+9 -36
View File
@@ -3,7 +3,6 @@ package filesource
import (
"crypto"
"fmt"
"io/fs"
"os"
"path"
"path/filepath"
@@ -36,7 +35,7 @@ type fileSource struct {
id artifact.ID
digestForVersion string
config Config
resolver *fileresolver.Directory
resolver file.Resolver
mutex *sync.Mutex
closer func() error
digests []file.Digest
@@ -165,48 +164,22 @@ func (s fileSource) FileResolver(_ source.Scope) (file.Resolver, error) {
return nil, err
}
var res *fileresolver.Directory
if isArchiveAnalysis {
// this is an analysis of an archive file... we should scan the directory where the archive contents
res, err = fileresolver.NewFromDirectory(s.analysisPath, "", exclusionFunctions...)
if err != nil {
return nil, fmt.Errorf("unable to create directory resolver: %w", err)
}
} else {
// this is an analysis of a single file. We want to ultimately scan the directory that the file is in, but we
// don't want to include any other files except this the given file.
exclusionFunctions = append([]fileresolver.PathIndexVisitor{
// note: we should exclude these kinds of paths first before considering any other user-provided exclusions
func(_, p string, _ os.FileInfo, _ error) error {
if p == absParentDir {
// this is the root directory... always include it
return nil
}
if filepath.Dir(p) != absParentDir {
// we are no longer in the root directory containing the single file we want to scan...
// we should skip the directory this path resides in entirely!
return fs.SkipDir
}
if filepath.Base(p) != filepath.Base(s.config.Path) {
// we're in the root directory, but this is not the file we want to scan...
// we should selectively skip this file (not the directory we're in).
return fileresolver.ErrSkipPath
}
return nil
},
}, exclusionFunctions...)
res, err = fileresolver.NewFromDirectory(absParentDir, absParentDir, exclusionFunctions...)
res, err := fileresolver.NewFromDirectory(s.analysisPath, "", exclusionFunctions...)
if err != nil {
return nil, fmt.Errorf("unable to create directory resolver: %w", err)
}
s.resolver = res
return s.resolver, nil
}
// This is analysis of a single file. Use file indexer.
res, err := fileresolver.NewFromFile(absParentDir, s.analysisPath, exclusionFunctions...)
if err != nil {
return nil, fmt.Errorf("unable to create file resolver: %w", err)
}
s.resolver = res
return s.resolver, nil
}