tar: create and validate a manifest from a tar stream

This commit contains added features to go-mtree that allows
user to create an mtree spec with '-T' option when specifying
a tar archive. Users can also validate an mtree spec against
a tar archive with an mtree spec. Also for the test archive,
there is a mixture of files, and folders (empty & non-empty),
and symlinks (broken & unbroken).

Signed-off-by: Stephen Chung <schung@redhat.com>

check.go

@@ -10,6 +10,8 @@ import (
 type Result struct {
 	// list of any failures in the Check
 	Failures []Failure `json:"failures"`
+	Missing  []Entry
+	Extra    []Entry
 }
 
 // Failure of a particular keyword for a path
@@ -90,3 +92,99 @@ func Check(root string, dh *DirectoryHierarchy, keywords []string) (*Result, err
 	}
 	return &result, nil
 }
+
+// TarCheck is the tar equivalent of checking a file hierarchy spec against a tar stream to
+// determine if files have been changed.
+func TarCheck(tarDH, dh *DirectoryHierarchy, keywords []string) (*Result, error) {
+	var result Result
+	var err error
+	var tarRoot *Entry
+
+	for _, e := range tarDH.Entries {
+		if e.Name == "." {
+			tarRoot = &e
+			break
+		}
+	}
+	tarRoot.Next = &Entry{
+		Name: "seen",
+		Type: CommentType,
+	}
+	curDir := tarRoot
+	creator := dhCreator{DH: dh}
+	sort.Sort(byPos(creator.DH.Entries))
+
+	var outOfTree bool
+	for i, e := range creator.DH.Entries {
+		switch e.Type {
+		case SpecialType:
+			if e.Name == "/set" {
+				creator.curSet = &creator.DH.Entries[i]
+			} else if e.Name == "/unset" {
+				creator.curSet = nil
+			}
+		case RelativeType, FullType:
+			if outOfTree {
+				return &result, fmt.Errorf("No parent node from %s", e.Path())
+			}
+			// TODO: handle the case where "." is not the first Entry to be found
+			tarEntry := curDir.Descend(e.Name)
+			if tarEntry == nil {
+				result.Missing = append(result.Missing, e)
+				continue
+			}
+
+			tarEntry.Next = &Entry{
+				Type: CommentType,
+				Name: "seen",
+			}
+
+			// expected values from file hierarchy spec
+			var kvs KeyVals
+			if creator.curSet != nil {
+				kvs = MergeSet(creator.curSet.Keywords, e.Keywords)
+			} else {
+				kvs = NewKeyVals(e.Keywords)
+			}
+
+			// actual
+			var tarkvs KeyVals
+			if tarEntry.Set != nil {
+				tarkvs = MergeSet(tarEntry.Set.Keywords, tarEntry.Keywords)
+			} else {
+				tarkvs = NewKeyVals(tarEntry.Keywords)
+			}
+
+			for _, kv := range kvs {
+				if _, ok := KeywordFuncs[kv.Keyword()]; !ok {
+					return nil, fmt.Errorf("Unknown keyword %q for file %q", kv.Keyword(), e.Path())
+				}
+				if keywords != nil && !inSlice(kv.Keyword(), keywords) {
+					continue
+				}
+				if tarkv := tarkvs.Has(kv.Keyword()); tarkv != emptyKV {
+					if string(tarkv) != string(kv) {
+						failure := Failure{Path: tarEntry.Path(), Keyword: kv.Keyword(), Expected: kv.Value(), Got: tarkv.Value()}
+						result.Failures = append(result.Failures, failure)
+					}
+				}
+			}
+			// Step into a directory
+			if tarEntry.Prev != nil {
+				curDir = tarEntry
+			}
+		case DotDotType:
+			if outOfTree {
+				return &result, fmt.Errorf("No parent node.")
+			}
+			curDir = curDir.Ascend()
+			if curDir == nil {
+				outOfTree = true
+			}
+		}
+	}
+	result.Extra = filter(tarRoot, func(e *Entry) bool {
+		return e.Next == nil
+	})
+	return &result, err
+}

cmd/gomtree/main.go

@@ -5,6 +5,8 @@ import (
 	"encoding/json"
 	"flag"
 	"fmt"
+	"io"
+	"io/ioutil"
 	"log"
 	"os"
 	"strings"
@@ -20,6 +22,7 @@ var (
 	flUseKeywords  = flag.String("k", "", "Use the specified (delimited by comma or space) keywords as the current set of keywords")
 	flListKeywords = flag.Bool("list-keywords", false, "List the keywords available")
 	flResultFormat = flag.String("result-format", "bsd", "output the validation results using the given format (bsd, json, path)")
+	flTar          = flag.String("T", "", "use tar archive to create or validate a directory hierarchy spec")
 )
 
 var formats = map[string]func(*mtree.Result) string{
@@ -123,9 +126,43 @@ func main() {
 		rootPath = *flPath
 	}
 
+	// -T <tar file>
+	var tdh *mtree.DirectoryHierarchy
+	if *flTar != "" {
+		fh, err := os.Open(*flTar)
+		if err != nil {
+			log.Println(err)
+			isErr = true
+			return
+		}
+		ts := mtree.NewTarStreamer(fh, currentKeywords)
+
+		if _, err := io.Copy(ioutil.Discard, ts); err != nil && err != io.EOF {
+			log.Println(err)
+			isErr = true
+			return
+		}
+		if err := ts.Close(); err != nil {
+			log.Println(err)
+			isErr = true
+			return
+		}
+		defer fh.Close()
+		tdh, err = ts.Hierarchy()
+		if err != nil {
+			log.Println(err)
+			isErr = true
+			return
+		}
+	}
 	// -c
 	if *flCreate {
 		// create a directory hierarchy
+		// with a tar stream
+		if tdh != nil {
+			tdh.WriteTo(os.Stdout)
+		} else {
+			// with a root directory
 			dh, err := mtree.Walk(rootPath, nil, currentKeywords)
 			if err != nil {
 				log.Println(err)
@@ -133,9 +170,16 @@ func main() {
 				return
 			}
 			dh.WriteTo(os.Stdout)
-	} else if dh != nil {
+		}
+	} else if tdh != nil || dh != nil {
+		var res *mtree.Result
+		var err error
 		// else this is a validation
-		res, err := mtree.Check(rootPath, dh, currentKeywords)
+		if *flTar != "" {
+			res, err = mtree.TarCheck(tdh, dh, currentKeywords)
+		} else {
+			res, err = mtree.Check(rootPath, dh, currentKeywords)
+		}
 		if err != nil {
 			log.Println(err)
 			isErr = true
@@ -148,6 +192,23 @@ func main() {
 				log.Println(err)
 				isErr = true
 				return
+		if res != nil {
+			if len(res.Failures) > 0 {
+				defer os.Exit(1)
+				for _, failure := range res.Failures {
+					fmt.Println(failure)
+				}
+			}
+			if len(res.Extra) > 0 {
+				defer os.Exit(1)
+				for _, extra := range res.Extra {
+					fmt.Printf("%s extra\n", extra.Path())
+				}
+			}
+			if len(res.Missing) > 0 {
+				defer os.Exit(1)
+				for _, missing := range res.Missing {
+					fmt.Printf("%s missing\n", missing.Path())
 				}
 			}
 		} else {
@@ -156,6 +217,7 @@ func main() {
 			defer os.Exit(1)
 			return
 		}
+	}
 }
 
 func splitKeywordsArg(str string) []string {

entry.go

@@ -25,6 +25,29 @@ type Entry struct {
 	Type       EntryType
 }
 
+// Descend searches thru an Entry's children to find the Entry associated with
+// `filename`. Directories are stored at the end of an Entry's children so do a
+// traverse backwards. If you descend to a "."
+func (e Entry) Descend(filename string) *Entry {
+	if filename == "." || filename == "" {
+		return &e
+	}
+	numChildren := len(e.Children)
+	for i := range e.Children {
+		c := e.Children[numChildren-1-i]
+		if c.Name == filename {
+			return c
+		}
+	}
+	return nil
+}
+
+// Ascend gets the parent of an Entry. Serves mainly to maintain readability
+// when traversing up and down an Entry tree
+func (e Entry) Ascend() *Entry {
+	return e.Parent
+}
+
 // Path provides the full path of the file, despite RelativeType or FullType
 func (e Entry) Path() string {
 	if e.Parent == nil || e.Type == FullType {
@@ -43,7 +66,6 @@ func (e Entry) String() string {
 	if e.Type == DotDotType {
 		return e.Name
 	}
-	// TODO(vbatts) if type is RelativeType and a keyword of not type=dir
 	if e.Type == SpecialType || e.Type == FullType || inSlice("type=dir", e.Keywords) {
 		return fmt.Sprintf("%s %s", e.Name, strings.Join(e.Keywords, " "))
 	}

parse.go

@@ -81,6 +81,7 @@ func ParseSpec(r io.Reader) (*DirectoryHierarchy, error) {
 				e.Type = RelativeType
 			}
 			e.Keywords = f[1:]
+			// TODO: gather keywords if using tar stream
 			e.Parent = creator.curDir
 			for i := range e.Keywords {
 				kv := KeyVal(e.Keywords[i])

tar.go

@@ -4,19 +4,19 @@ import (
 	"archive/tar"
 	"io"
 	"io/ioutil"
-	"log"
 	"os"
 	"path/filepath"
 	"strings"
 )
 
-// Streamer interface that wraps an io.ReadCloser with a function that will
-// return it's Hierarchy
+// Streamer creates a file hierarchy out of a tar stream
 type Streamer interface {
 	io.ReadCloser
 	Hierarchy() (*DirectoryHierarchy, error)
 }
 
+var tarDefaultSetKeywords = []string{"type=file", "flags=none", "mode=0664"}
+
 // NewTarStreamer streams a tar archive and creates a file hierarchy based off
 // of the tar metadata headers
 func NewTarStreamer(r io.Reader, keywords []string) Streamer {
@@ -46,15 +46,28 @@ type tarStream struct {
 func (ts *tarStream) readHeaders() {
 	// We have to start with the directory we're in, and anything beyond these
 	// items is determined at the time a tar is extracted.
-	e := Entry{
-		Name:     ".",
-		Keywords: []string{"size=0", "type=dir"},
+	rootComment := Entry{
+		Raw:  "# .",
+		Type: CommentType,
 	}
-	ts.creator.curDir = &e
+	root := Entry{
+		Name: ".",
+		Type: RelativeType,
+		Prev: &rootComment,
+		Set: &Entry{
+			Name: "meta-set",
+			Type: SpecialType,
+		},
+	}
+	metadataEntries := signatureEntries("<user specified tar archive>")
+	for _, e := range metadataEntries {
+		e.Pos = len(ts.creator.DH.Entries)
 		ts.creator.DH.Entries = append(ts.creator.DH.Entries, e)
+	}
 	for {
 		hdr, err := ts.tarReader.Next()
 		if err != nil {
+			flatten(&root, ts)
 			ts.pipeReader.CloseWithError(err)
 			return
 		}
@@ -84,17 +97,30 @@ func (ts *tarStream) readHeaders() {
 		// Alright, it's either file or directory
 		e := Entry{
 			Name: filepath.Base(hdr.Name),
-			Pos:  len(ts.creator.DH.Entries),
 			Type: RelativeType,
 		}
+
 		// now collect keywords on the file
 		for _, keyword := range ts.keywords {
 			if keyFunc, ok := KeywordFuncs[keyword]; ok {
+				// We can't extract directories on to disk, so "size" keyword
+				// is irrelevant for now
+				if hdr.FileInfo().IsDir() && keyword == "size" {
+					continue
+				}
+
+				if string(hdr.Typeflag) == string('1') {
+					// TODO: get number of hardlinks for a file
+				}
 				val, err := keyFunc(hdr.Name, hdr.FileInfo(), tmpFile)
 				if err != nil {
 					ts.setErr(err)
 				}
+
+				// for good measure, check that we actually get a value for a keyword
+				if val != "" {
 					e.Keywords = append(e.Keywords, val)
+				}
 
 				// don't forget to reset the reader
 				if _, err := tmpFile.Seek(0, 0); err != nil {
@@ -105,39 +131,40 @@ func (ts *tarStream) readHeaders() {
 				}
 			}
 		}
+
+		// collect meta-set keywords for a directory so that we can build the
+		// actual sets in `flatten`
+		if hdr.FileInfo().IsDir() {
+			s := Entry{
+				Name: "meta-set",
+				Type: SpecialType,
+			}
+			for _, setKW := range SetKeywords {
+				if keyFunc, ok := KeywordFuncs[setKW]; ok {
+					val, err := keyFunc(hdr.Name, hdr.FileInfo(), tmpFile)
+					if err != nil {
+						ts.setErr(err)
+					}
+					if val != "" {
+						s.Keywords = append(s.Keywords, val)
+					}
+					if _, err := tmpFile.Seek(0, 0); err != nil {
 						tmpFile.Close()
 						os.Remove(tmpFile.Name())
+						ts.pipeReader.CloseWithError(err)
+					}
+				}
+			}
+			if filepath.Dir(filepath.Clean(hdr.Name)) == "." {
+				root.Set = &s
+			} else {
+				e.Set = &s
+			}
+		}
+		populateTree(&root, &e, hdr, ts)
 
-		// compare directories, to determine parent of the current entry
-		cd := compareDir(filepath.Dir(hdr.Name), ts.creator.curDir.Path())
-		switch {
-		case cd == sameDir:
-			e.Parent = ts.creator.curDir
-			if e.Parent != nil {
-				e.Parent.Children = append(e.Parent.Children, &e)
-			}
-		case cd == parentDir:
-			e.Parent = ts.creator.curDir.Parent
-			if e.Parent != nil {
-				e.Parent.Children = append(e.Parent.Children, &e)
-			}
-		}
-
-		if hdr.FileInfo().IsDir() {
-			ts.creator.curDir = &e
-		}
-		// TODO getting the parent child relationship of these entries!
-		if hdr.FileInfo().IsDir() {
-			log.Println(strings.Split(hdr.Name, "/"), strings.Split(ts.creator.curDir.Path(), "/"))
-		}
-
-		ts.creator.DH.Entries = append(ts.creator.DH.Entries, e)
-
-		// Now is the wacky part of building out the entries. Since we can not
-		// control how the archive was assembled, can only take in the order given.
-		// Using `/set` will be tough. Hopefully i can do the directory stepping
-		// with relative paths, but even then I may get a new directory, and not
-		// the files first, but its directories first. :-\
+		tmpFile.Close()
+		os.Remove(tmpFile.Name())
 	}
 }
 
@@ -150,6 +177,167 @@ const (
 	parentDir
 )
 
+// populateTree creates a file tree hierarchy using an Entry's Parent and
+// Children fields. When examining the Entry e to insert in the tree, we
+// determine if the path to that Entry exists yet. If it does, insert it in the
+// appropriate position in the tree. If not, create a path with "placeholder"
+// directories, and then insert the Entry. populateTree does not consider
+// symbolic links yet.
+func populateTree(root, e *Entry, hdr *tar.Header, ts *tarStream) {
+	isDir := hdr.FileInfo().IsDir()
+	wd := filepath.Clean(hdr.Name)
+
+	if !isDir {
+		// If entry is a file, we only want the directory it's in.
+		wd = filepath.Dir(wd)
+	}
+	if filepath.Dir(wd) == "." {
+		if isDir {
+			root.Keywords = e.Keywords
+		} else {
+			root.Children = append([]*Entry{e}, root.Children...)
+			e.Parent = root
+		}
+		return
+	}
+
+	dirNames := strings.Split(wd, "/")
+	parent := root
+	for _, name := range dirNames[1:] {
+		if node := parent.Descend(name); node == nil {
+			// Entry for directory doesn't exist in tree relative to root
+			var newEntry *Entry
+			if isDir {
+				newEntry = e
+			} else {
+				newEntry = &Entry{
+					Name: name,
+					Type: RelativeType,
+				}
+			}
+			newEntry.Parent = parent
+			parent.Children = append(parent.Children, newEntry)
+			parent = newEntry
+		} else {
+			// Entry for directory exists in tree, just keep going
+			parent = node
+		}
+	}
+	if !isDir {
+		parent.Children = append([]*Entry{e}, parent.Children...)
+		e.Parent = parent
+	} else {
+		commentEntry := Entry{
+			Raw:  "# " + e.Path(),
+			Type: CommentType,
+		}
+		e.Prev = &commentEntry
+	}
+}
+
+// After constructing the tree from the tar stream, we want to "flatten" this
+// tree by appending Entry's into ts.creator.DH.Entries in an appropriate
+// manner to simplify writing the output with ts.creator.DH.WriteTo
+// root: the "head" of the sub-tree to flatten
+// ts  : tarStream to keep track of Entry's
+func flatten(root *Entry, ts *tarStream) {
+	if root.Prev != nil {
+		// root.Prev != nil implies root is a directory
+		ts.creator.DH.Entries = append(ts.creator.DH.Entries,
+			Entry{
+				Type: BlankType,
+				Pos:  len(ts.creator.DH.Entries),
+			})
+		root.Prev.Pos = len(ts.creator.DH.Entries)
+		ts.creator.DH.Entries = append(ts.creator.DH.Entries, *root.Prev)
+
+		// Check if we need a new set
+		if ts.creator.curSet == nil {
+			ts.creator.curSet = &Entry{
+				Type:     SpecialType,
+				Name:     "/set",
+				Keywords: keywordSelector(append(tarDefaultSetKeywords, root.Set.Keywords...), ts.keywords),
+				Pos:      len(ts.creator.DH.Entries),
+			}
+			ts.creator.DH.Entries = append(ts.creator.DH.Entries, *ts.creator.curSet)
+		} else {
+			needNewSet := false
+			for _, k := range root.Set.Keywords {
+				if !inSlice(k, ts.creator.curSet.Keywords) {
+					needNewSet = true
+					break
+				}
+			}
+			if needNewSet {
+				ts.creator.curSet = &Entry{
+					Name:     "/set",
+					Type:     SpecialType,
+					Pos:      len(ts.creator.DH.Entries),
+					Keywords: keywordSelector(append(tarDefaultSetKeywords, root.Set.Keywords...), ts.keywords),
+				}
+				ts.creator.DH.Entries = append(ts.creator.DH.Entries, *ts.creator.curSet)
+			}
+		}
+	}
+	root.Set = ts.creator.curSet
+	root.Keywords = setDifference(root.Keywords, ts.creator.curSet.Keywords)
+	root.Pos = len(ts.creator.DH.Entries)
+	ts.creator.DH.Entries = append(ts.creator.DH.Entries, *root)
+
+	for _, c := range root.Children {
+		flatten(c, ts)
+	}
+
+	if root.Prev != nil {
+		// Show a comment when stepping out
+		root.Prev.Pos = len(ts.creator.DH.Entries)
+		ts.creator.DH.Entries = append(ts.creator.DH.Entries, *root.Prev)
+		dotEntry := Entry{
+			Type: DotDotType,
+			Name: "..",
+			Pos:  len(ts.creator.DH.Entries),
+		}
+		ts.creator.DH.Entries = append(ts.creator.DH.Entries, dotEntry)
+	}
+}
+
+// filter takes in a pointer to an Entry, and returns a slice of Entry's that
+// satisfy the predicate p
+func filter(root *Entry, p func(*Entry) bool) []Entry {
+	var validEntrys []Entry
+	if len(root.Children) > 0 || root.Prev != nil {
+		for _, c := range root.Children {
+			// if an Entry is a directory, filter the directory
+			if c.Prev != nil {
+				validEntrys = append(validEntrys, filter(c, p)...)
+			}
+			if p(c) {
+				if c.Prev == nil {
+					// prepend directories
+					validEntrys = append([]Entry{*c}, validEntrys...)
+				} else {
+					validEntrys = append(validEntrys, *c)
+				}
+			}
+		}
+		return validEntrys
+	}
+	return nil
+}
+
+func setDifference(this, that []string) []string {
+	if len(this) == 0 {
+		return that
+	}
+	diff := []string{}
+	for _, kv := range this {
+		if !inSlice(kv, that) {
+			diff = append(diff, kv)
+		}
+	}
+	return diff
+}
+
 func compareDir(curDir, prevDir string) relationship {
 	curDir = filepath.Clean(curDir)
 	prevDir = filepath.Clean(prevDir)

tar_test.go

@@ -3,6 +3,7 @@ package mtree
 import (
 	"archive/tar"
 	"bytes"
+	"fmt"
 	"io"
 	"io/ioutil"
 	"os"
@@ -45,6 +46,14 @@ func TestTar(t *testing.T) {
 		buf := bytes.NewBuffer(data)
 		str := NewTarStreamer(buf, append(DefaultKeywords, "sha1"))
 	*/
+	/*
+		// open empty folder and check size.
+		fh, err := os.Open("./testdata/empty")
+		if err != nil {
+			t.Fatal(err)
+		}
+		log.Println(fh.Stat())
+		fh.Close() */
 	fh, err := os.Open("./testdata/test.tar")
 	if err != nil {
 		t.Fatal(err)
@@ -59,23 +68,71 @@ func TestTar(t *testing.T) {
 	}
 	defer fh.Close()
 
-	/*
-		fi, err := fh.Stat()
+	// get DirectoryHierarcy struct from walking the tar archive
+	tdh, err := str.Hierarchy()
 	if err != nil {
 		t.Fatal(err)
 	}
-		if i != fi.Size() {
-			t.Errorf("expected length %d; got %d", fi.Size(), i)
-		}
-	*/
-	dh, err := str.Hierarchy()
-	if err != nil {
-		t.Fatal(err)
-	}
-	if dh == nil {
+	if tdh == nil {
 		t.Fatal("expected a DirectoryHierarchy struct, but got nil")
 	}
-	//dh.WriteTo(os.Stdout)
+
+	fh, err = os.Create("./testdata/test.mtree")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.Remove("./testdata/test.mtree")
+
+	// put output of tar walk into test.mtree
+	_, err = tdh.WriteTo(fh)
+	if err != nil {
+		t.Fatal(err)
+	}
+	fh.Close()
+
+	// now simulate gomtree -T testdata/test.tar -f testdata/test.mtree
+	fh, err = os.Open("./testdata/test.mtree")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer fh.Close()
+
+	dh, err := ParseSpec(fh)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	res, err := TarCheck(tdh, dh, append(DefaultKeywords, "sha1"))
+
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// print any failures, and then call t.Fatal once all failures/extra/missing
+	// are outputted
+	if res != nil {
+		errors := ""
+		switch {
+		case len(res.Failures) > 0:
+			for _, f := range res.Failures {
+				fmt.Printf("%s\n", f)
+			}
+			errors += "Keyword validation errors\n"
+		case len(res.Missing) > 0:
+			for _, m := range res.Missing {
+				fmt.Printf("Missing file: %s\n", m.Path())
+			}
+			errors += "Missing files not expected for this test\n"
+		case len(res.Extra) > 0:
+			for _, e := range res.Extra {
+				fmt.Printf("Extra file: %s\n", e.Path())
+			}
+			errors += "Extra files not expected for this test\n"
+		}
+		if errors != "" {
+			t.Fatal(errors)
+		}
+	}
 }
 
 // minimal tar archive stream that mimics what is in ./testdata/test.tar