github.com/noqcks/syft@v0.0.0-20230920222752-a9e2c4e288e5/syft/pkg/cataloger/alpm/parse_alpm_db.go (about)

     1  package alpm
     2  
     3  import (
     4  	"bufio"
     5  	"compress/gzip"
     6  	"fmt"
     7  	"io"
     8  	"path/filepath"
     9  	"strconv"
    10  	"strings"
    11  	"time"
    12  
    13  	"github.com/mitchellh/mapstructure"
    14  	"github.com/vbatts/go-mtree"
    15  
    16  	"github.com/anchore/syft/syft/artifact"
    17  	"github.com/anchore/syft/syft/file"
    18  	"github.com/anchore/syft/syft/pkg"
    19  	"github.com/anchore/syft/syft/pkg/cataloger/generic"
    20  )
    21  
    22  var _ generic.Parser = parseAlpmDB
    23  
    24  var (
    25  	ignoredFiles = map[string]bool{
    26  		"/set":       true,
    27  		".BUILDINFO": true,
    28  		".PKGINFO":   true,
    29  		"":           true,
    30  	}
    31  )
    32  
    33  type parsedData struct {
    34  	Licenses         string `mapstructure:"license"`
    35  	pkg.AlpmMetadata `mapstructure:",squash"`
    36  }
    37  
    38  func parseAlpmDB(resolver file.Resolver, env *generic.Environment, reader file.LocationReadCloser) ([]pkg.Package, []artifact.Relationship, error) {
    39  	data, err := parseAlpmDBEntry(reader)
    40  	if err != nil {
    41  		return nil, nil, err
    42  	}
    43  
    44  	base := filepath.Dir(reader.RealPath)
    45  	r, err := getFileReader(filepath.Join(base, "mtree"), resolver)
    46  	if err != nil {
    47  		return nil, nil, err
    48  	}
    49  
    50  	pkgFiles, err := parseMtree(r)
    51  	if err != nil {
    52  		return nil, nil, err
    53  	}
    54  
    55  	// replace the files found the pacman database with the files from the mtree These contain more metadata and
    56  	// thus more useful.
    57  	// TODO: probably want to use MTREE and PKGINFO here
    58  	data.Files = pkgFiles
    59  
    60  	// We only really do this to get any backup database entries from the files database
    61  	files := filepath.Join(base, "files")
    62  	_, err = getFileReader(files, resolver)
    63  	if err != nil {
    64  		return nil, nil, err
    65  	}
    66  	filesMetadata, err := parseAlpmDBEntry(reader)
    67  	if err != nil {
    68  		return nil, nil, err
    69  	} else if filesMetadata != nil {
    70  		data.Backup = filesMetadata.Backup
    71  	}
    72  
    73  	if data.Package == "" {
    74  		return nil, nil, nil
    75  	}
    76  
    77  	return []pkg.Package{
    78  		newPackage(
    79  			data,
    80  			env.LinuxRelease,
    81  			reader.Location.WithAnnotation(pkg.EvidenceAnnotationKey, pkg.PrimaryEvidenceAnnotation),
    82  		),
    83  	}, nil, nil
    84  }
    85  
    86  func parseAlpmDBEntry(reader io.Reader) (*parsedData, error) {
    87  	scanner := newScanner(reader)
    88  	metadata, err := parseDatabase(scanner)
    89  	if err != nil {
    90  		return nil, err
    91  	}
    92  	return metadata, nil
    93  }
    94  
    95  func newScanner(reader io.Reader) *bufio.Scanner {
    96  	// This is taken from the apk parser
    97  	// https://github.com/anchore/syft/blob/v0.47.0/syft/pkg/cataloger/apkdb/parse_apk_db.go#L37
    98  	const maxScannerCapacity = 1024 * 1024
    99  	bufScan := make([]byte, maxScannerCapacity)
   100  	scanner := bufio.NewScanner(reader)
   101  	scanner.Buffer(bufScan, maxScannerCapacity)
   102  	onDoubleLF := func(data []byte, atEOF bool) (advance int, token []byte, err error) {
   103  		for i := 0; i < len(data); i++ {
   104  			if i > 0 && data[i-1] == '\n' && data[i] == '\n' {
   105  				return i + 1, data[:i-1], nil
   106  			}
   107  		}
   108  		if !atEOF {
   109  			return 0, nil, nil
   110  		}
   111  		// deliver the last token (which could be an empty string)
   112  		return 0, data, bufio.ErrFinalToken
   113  	}
   114  
   115  	scanner.Split(onDoubleLF)
   116  	return scanner
   117  }
   118  
   119  func getFileReader(path string, resolver file.Resolver) (io.Reader, error) {
   120  	locs, err := resolver.FilesByPath(path)
   121  	if err != nil {
   122  		return nil, err
   123  	}
   124  
   125  	if len(locs) == 0 {
   126  		return nil, fmt.Errorf("could not find file: %s", path)
   127  	}
   128  	// TODO: Should we maybe check if we found the file
   129  	dbContentReader, err := resolver.FileContentsByLocation(locs[0])
   130  	if err != nil {
   131  		return nil, err
   132  	}
   133  	return dbContentReader, nil
   134  }
   135  
   136  func parseDatabase(b *bufio.Scanner) (*parsedData, error) {
   137  	var err error
   138  	pkgFields := make(map[string]interface{})
   139  	for b.Scan() {
   140  		fields := strings.SplitN(b.Text(), "\n", 2)
   141  
   142  		// End of File
   143  		if len(fields) == 1 {
   144  			break
   145  		}
   146  
   147  		// The alpm database surrounds the keys with %.
   148  		key := strings.ReplaceAll(fields[0], "%", "")
   149  		key = strings.ToLower(key)
   150  		value := strings.TrimSpace(fields[1])
   151  
   152  		switch key {
   153  		case "files":
   154  			var files []map[string]string
   155  			for _, f := range strings.Split(value, "\n") {
   156  				path := fmt.Sprintf("/%s", f)
   157  				if ok := ignoredFiles[path]; !ok {
   158  					files = append(files, map[string]string{"path": path})
   159  				}
   160  			}
   161  			pkgFields[key] = files
   162  		case "backup":
   163  			var backup []map[string]interface{}
   164  			for _, f := range strings.Split(value, "\n") {
   165  				fields := strings.SplitN(f, "\t", 2)
   166  				path := fmt.Sprintf("/%s", fields[0])
   167  				if ok := ignoredFiles[path]; !ok {
   168  					backup = append(backup, map[string]interface{}{
   169  						"path": path,
   170  						"digests": []file.Digest{{
   171  							Algorithm: "md5",
   172  							Value:     fields[1],
   173  						}}})
   174  				}
   175  			}
   176  			pkgFields[key] = backup
   177  		case "reason":
   178  			fallthrough
   179  		case "size":
   180  			pkgFields[key], err = strconv.ParseInt(value, 10, 64)
   181  			if err != nil {
   182  				return nil, fmt.Errorf("failed to parse %s to integer", value)
   183  			}
   184  		default:
   185  			pkgFields[key] = value
   186  		}
   187  	}
   188  
   189  	return parsePkgFiles(pkgFields)
   190  }
   191  
   192  func parsePkgFiles(pkgFields map[string]interface{}) (*parsedData, error) {
   193  	var entry parsedData
   194  	if err := mapstructure.Decode(pkgFields, &entry); err != nil {
   195  		return nil, fmt.Errorf("unable to parse ALPM metadata: %w", err)
   196  	}
   197  
   198  	if entry.Backup == nil {
   199  		entry.Backup = make([]pkg.AlpmFileRecord, 0)
   200  	}
   201  
   202  	if entry.Package == "" && len(entry.Files) == 0 && len(entry.Backup) == 0 {
   203  		return nil, nil
   204  	}
   205  	return &entry, nil
   206  }
   207  
   208  func parseMtree(r io.Reader) ([]pkg.AlpmFileRecord, error) {
   209  	var err error
   210  	var entries []pkg.AlpmFileRecord
   211  
   212  	r, err = gzip.NewReader(r)
   213  	if err != nil {
   214  		return nil, err
   215  	}
   216  	specDh, err := mtree.ParseSpec(r)
   217  	if err != nil {
   218  		return nil, err
   219  	}
   220  	for _, f := range specDh.Entries {
   221  		var entry pkg.AlpmFileRecord
   222  		entry.Digests = make([]file.Digest, 0)
   223  		fileFields := make(map[string]interface{})
   224  		if ok := ignoredFiles[f.Name]; ok {
   225  			continue
   226  		}
   227  		path := fmt.Sprintf("/%s", f.Name)
   228  		fileFields["path"] = path
   229  		for _, kv := range f.Keywords {
   230  			kw := string(kv.Keyword())
   231  			switch kw {
   232  			case "time":
   233  				// All unix timestamps have a .0 suffixs.
   234  				v := strings.Split(kv.Value(), ".")
   235  				i, _ := strconv.ParseInt(v[0], 10, 64)
   236  				tm := time.Unix(i, 0)
   237  				fileFields[kw] = tm
   238  			case "sha256digest":
   239  				entry.Digests = append(entry.Digests, file.Digest{
   240  					Algorithm: "sha256",
   241  					Value:     kv.Value(),
   242  				})
   243  			case "md5digest":
   244  				entry.Digests = append(entry.Digests, file.Digest{
   245  					Algorithm: "md5",
   246  					Value:     kv.Value(),
   247  				})
   248  			default:
   249  				fileFields[kw] = kv.Value()
   250  			}
   251  		}
   252  		if err := mapstructure.Decode(fileFields, &entry); err != nil {
   253  			return nil, fmt.Errorf("unable to parse ALPM mtree data: %w", err)
   254  		}
   255  		entries = append(entries, entry)
   256  	}
   257  	return entries, nil
   258  }