Add package compression type

TODO: Handle the path substitution in a separate field
Adds GZip support and allows the compiler to switch compression type.

It also adds Concurrency as a compiler attribute (not consumed yet)

Refers to #33

go.mod

@@ -15,6 +15,7 @@ require (
 	github.com/go-yaml/yaml v2.1.0+incompatible // indirect
 	github.com/hashicorp/go-version v1.2.0
 	github.com/jinzhu/copier v0.0.0-20180308034124-7e38e58719c3
+	github.com/klauspost/pgzip v1.2.1
 	github.com/kyokomi/emoji v2.1.0+incompatible
 	github.com/logrusorgru/aurora v0.0.0-20190417123914-21d75270181e
 	github.com/marcsauter/single v0.0.0-20181104081128-f8bf46f26ec0

pkg/compiler/artifact.go

@@ -17,6 +17,7 @@ package compiler
 
 import (
 	"archive/tar"
+	"bufio"
 	"io"
 	"io/ioutil"
 	"os"
@@ -24,16 +25,24 @@ import (
 	"path/filepath"
 	"regexp"
 
+	gzip "github.com/klauspost/pgzip"
+
 	//"strconv"
 	"strings"
 	"sync"
 
+	"github.com/mudler/luet/pkg/helpers"
 	. "github.com/mudler/luet/pkg/logger"
 	"github.com/mudler/luet/pkg/solver"
-	yaml "gopkg.in/yaml.v2"
-
-	"github.com/mudler/luet/pkg/helpers"
 	"github.com/pkg/errors"
+	yaml "gopkg.in/yaml.v2"
+)
+
+type CompressionImplementation string
+
+const (
+	None CompressionImplementation = "none" // e.g. tar for standard packages
+	GZip CompressionImplementation = "gzip"
 )
 
 type ArtifactIndex []Artifact
@@ -57,10 +66,11 @@ type PackageArtifact struct {
 	CompileSpec     *LuetCompilationSpec      `json:"compilationspec"`
 	Checksums       Checksums                 `json:"checksums"`
 	SourceAssertion solver.PackagesAssertions `json:"-"`
+	CompressionType CompressionImplementation `json:"compression"`
 }
 
 func NewPackageArtifact(path string) Artifact {
-	return &PackageArtifact{Path: path, Dependencies: []*PackageArtifact{}, Checksums: Checksums{}}
+	return &PackageArtifact{Path: path, Dependencies: []*PackageArtifact{}, Checksums: Checksums{}, CompressionType: None}
 }
 
 func NewPackageArtifactFromYaml(data []byte) (Artifact, error) {
@@ -73,6 +83,10 @@ func NewPackageArtifactFromYaml(data []byte) (Artifact, error) {
 	return p, err
 }
 
+func (a *PackageArtifact) SetCompressionType(t CompressionImplementation) {
+	a.CompressionType = t
+}
+
 func (a *PackageArtifact) Hash() error {
 	return a.Checksums.Generate(a)
 }
@@ -174,13 +188,85 @@ func (a *PackageArtifact) SetPath(p string) {
 }
 
 // Compress Archives and compress (TODO) to the artifact path
-func (a *PackageArtifact) Compress(src string) error {
-	return helpers.Tar(src, a.Path)
+func (a *PackageArtifact) Compress(src string, concurrency int) error {
+	switch a.CompressionType {
+	case None:
+		return helpers.Tar(src, a.Path)
+
+	case GZip:
+		err := helpers.Tar(src, a.Path)
+		if err != nil {
+			return err
+		}
+		original, err := os.Open(a.Path)
+		if err != nil {
+			return err
+		}
+		defer original.Close()
+
+		gzipfile := a.Path + ".gz"
+		bufferedReader := bufio.NewReader(original)
+
+		// Open a file for writing.
+		dst, err := os.Create(gzipfile)
+		if err != nil {
+			return err
+		}
+		// Create gzip writer.
+		w := gzip.NewWriter(dst)
+		w.SetConcurrency(concurrency, 10)
+		defer w.Close()
+		defer dst.Close()
+		_, err = io.Copy(w, bufferedReader)
+		if err != nil {
+			return err
+		}
+		w.Close()
+		os.RemoveAll(a.Path) // Remove original
+		a.Path = gzipfile
+	}
+	return errors.New("Compression type must be supplied")
 }
 
 // Unpack Untar and decompress (TODO) to the given path
 func (a *PackageArtifact) Unpack(dst string, keepPerms bool) error {
-	return helpers.Untar(a.GetPath(), dst, keepPerms)
+	switch a.CompressionType {
+	case None:
+		return helpers.Untar(a.GetPath(), dst, keepPerms)
+
+	case GZip:
+		// Create the uncompressed archive
+		archive, err := os.Create(a.GetPath() + ".tar")
+		if err != nil {
+			return err
+		}
+		defer os.RemoveAll(a.GetPath() + ".tar")
+
+		original, err := os.Open(a.Path)
+		if err != nil {
+			return err
+		}
+		defer original.Close()
+
+		bufferedReader := bufio.NewReader(original)
+		r, err := gzip.NewReader(bufferedReader)
+		if err != nil {
+			return err
+		}
+		defer r.Close()
+
+		_, err = io.Copy(archive, r)
+		if err != nil {
+			return err
+		}
+
+		err = helpers.Untar(a.GetPath()+".tar", dst, keepPerms)
+		if err != nil {
+			return err
+		}
+
+	}
+	return errors.New("Compression type must be supplied")
 }
 
 func (a *PackageArtifact) FileList() ([]string, error) {
@@ -242,7 +328,7 @@ func worker(i int, wg *sync.WaitGroup, s <-chan CopyJob) {
 }
 
 // ExtractArtifactFromDelta extracts deltas from ArtifactLayer from an image in tar format
-func ExtractArtifactFromDelta(src, dst string, layers []ArtifactLayer, concurrency int, keepPerms bool, includes []string) (Artifact, error) {
+func ExtractArtifactFromDelta(src, dst string, layers []ArtifactLayer, concurrency int, keepPerms bool, includes []string, t CompressionImplementation) (Artifact, error) {
 
 	archive, err := ioutil.TempDir(os.TempDir(), "archive")
 	if err != nil {
@@ -306,10 +392,11 @@ func ExtractArtifactFromDelta(src, dst string, layers []ArtifactLayer, concurren
 
 	close(toCopy)
 	wg.Wait()
-
-	err = helpers.Tar(archive, dst)
+	a := NewPackageArtifact(dst)
+	a.SetCompressionType(t)
+	err = a.Compress(archive, concurrency)
 	if err != nil {
 		return nil, errors.Wrap(err, "Error met while creating package archive")
 	}
-	return NewPackageArtifact(dst), nil
+	return a, nil
 }

pkg/compiler/compiler.go

@@ -21,6 +21,7 @@ import (
 	"os"
 	"path/filepath"
 	"regexp"
+	"runtime"
 	"strings"
 	"sync"
 
@@ -40,6 +41,8 @@ type LuetCompiler struct {
 	Database           pkg.PackageDatabase
 	ImageRepository    string
 	PullFirst, KeepImg bool
+	Concurrency        int
+	CompressionType    CompressionImplementation
 }
 
 func NewLuetCompiler(backend CompilerBackend, db pkg.PackageDatabase) Compiler {
@@ -52,10 +55,16 @@ func NewLuetCompiler(backend CompilerBackend, db pkg.PackageDatabase) Compiler {
 		Database:        db,
 		ImageRepository: "luet/cache",
 		PullFirst:       true,
+		CompressionType: None,
 		KeepImg:         true,
+		Concurrency:     runtime.NumCPU(),
 	}
 }
 
+func (cs *LuetCompiler) SetConcurrency(i int) {
+	cs.Concurrency = i
+}
+
 func (cs *LuetCompiler) compilerWorker(i int, wg *sync.WaitGroup, cspecs chan CompilationSpec, a *[]Artifact, m *sync.Mutex, concurrency int, keepPermissions bool, errors chan error) {
 	defer wg.Done()
 
@@ -70,6 +79,7 @@ func (cs *LuetCompiler) compilerWorker(i int, wg *sync.WaitGroup, cspecs chan Co
 		m.Unlock()
 	}
 }
+
 func (cs *LuetCompiler) CompileWithReverseDeps(concurrency int, keepPermissions bool, ps CompilationSpecs) ([]Artifact, []error) {
 	artifacts, err := cs.CompileParallel(concurrency, keepPermissions, ps)
 	if len(err) != 0 {
@@ -340,7 +350,8 @@ func (cs *LuetCompiler) compileWithImage(image, buildertaggedImage, packageImage
 			cs.stripIncludesFromRootfs(p.GetIncludes(), rootfs)
 		}
 		artifact = NewPackageArtifact(p.Rel(p.GetPackage().GetFingerPrint() + ".package.tar"))
-		err = artifact.Compress(rootfs)
+		artifact.SetCompressionType(cs.CompressionType)
+		err = artifact.Compress(rootfs, concurrency)
 		if err != nil {
 			return nil, errors.Wrap(err, "Error met while creating package archive")
 		}
@@ -349,10 +360,11 @@ func (cs *LuetCompiler) compileWithImage(image, buildertaggedImage, packageImage
 	} else {
 		Info(pkgTag, "Generating delta")
 
-		artifact, err = ExtractArtifactFromDelta(rootfs, p.Rel(p.GetPackage().GetFingerPrint()+".package.tar"), diffs, concurrency, keepPermissions, p.GetIncludes())
+		artifact, err = ExtractArtifactFromDelta(rootfs, p.Rel(p.GetPackage().GetFingerPrint()+".package.tar"), diffs, concurrency, keepPermissions, p.GetIncludes(), cs.CompressionType)
 		if err != nil {
 			return nil, errors.Wrap(err, "Could not generate deltas")
 		}
+
 		artifact.SetCompileSpec(p)
 	}
 
@@ -365,7 +377,7 @@ func (cs *LuetCompiler) compileWithImage(image, buildertaggedImage, packageImage
 	return artifact, nil
 }
 
-func (cs *LuetCompiler) packageFromImage(p CompilationSpec, tag string, keepPermissions, keepImg bool) (Artifact, error) {
+func (cs *LuetCompiler) packageFromImage(p CompilationSpec, tag string, keepPermissions, keepImg bool, concurrency int) (Artifact, error) {
 	pkgTag := ":package:  " + p.GetPackage().GetName()
 
 	Info(pkgTag, "   🍩 Build starts 🔨 🔨 🔨 ")
@@ -405,8 +417,9 @@ func (cs *LuetCompiler) packageFromImage(p CompilationSpec, tag string, keepPerm
 	}
 	artifact := NewPackageArtifact(p.Rel(p.GetPackage().GetFingerPrint() + ".package.tar"))
 	artifact.SetCompileSpec(p)
+	artifact.SetCompressionType(cs.CompressionType)
 
-	err = artifact.Compress(rootfs)
+	err = artifact.Compress(rootfs, concurrency)
 	if err != nil {
 		return nil, errors.Wrap(err, "Error met while creating package archive")
 	}
@@ -480,7 +493,7 @@ func (cs *LuetCompiler) compile(concurrency int, keepPermissions bool, p Compila
 	// Treat last case (easier) first. The image is provided and we just compute a plain dockerfile with the images listed as above
 	if p.GetImage() != "" {
 		if p.ImageUnpack() { // If it is just an entire image, create a package from it
-			return cs.packageFromImage(p, "", keepPermissions, cs.KeepImg)
+			return cs.packageFromImage(p, "", keepPermissions, cs.KeepImg, concurrency)
 		}
 
 		return cs.compileWithImage(p.GetImage(), "", "", concurrency, keepPermissions, cs.KeepImg, p)
@@ -527,7 +540,7 @@ func (cs *LuetCompiler) compile(concurrency int, keepPermissions bool, p Compila
 					return nil, errors.New("No image defined for package: " + assertion.Package.GetName())
 				}
 				Info(pkgTag, ":whale: Sourcing package from image", compileSpec.GetImage())
-				artifact, err := cs.packageFromImage(compileSpec, currentPackageImageHash, keepPermissions, cs.KeepImg)
+				artifact, err := cs.packageFromImage(compileSpec, currentPackageImageHash, keepPermissions, cs.KeepImg, concurrency)
 				if err != nil {
 					return nil, errors.Wrap(err, "Failed compiling "+compileSpec.GetPackage().GetName())
 				}

pkg/compiler/interface.go

@@ -25,7 +25,7 @@ type Compiler interface {
 	CompileParallel(concurrency int, keepPermissions bool, ps CompilationSpecs) ([]Artifact, []error)
 	CompileWithReverseDeps(concurrency int, keepPermissions bool, ps CompilationSpecs) ([]Artifact, []error)
 	ComputeDepTree(p CompilationSpec) (solver.PackagesAssertions, error)
-
+	SetConcurrency(i int)
 	FromPackage(pkg.Package) (CompilationSpec, error)
 
 	SetBackend(CompilerBackend)
@@ -63,7 +63,8 @@ type Artifact interface {
 	GetCompileSpec() CompilationSpec
 	WriteYaml(dst string) error
 	Unpack(dst string, keepPerms bool) error
-	Compress(src string) error
+	Compress(src string, concurrency int) error
+	SetCompressionType(t CompressionImplementation)
 	FileList() ([]string, error)
 	Hash() error
 	Verify() error