packer.go

Documentation: github.com/vbatts/tar-split/tar/storage

     1  package storage
     2  
     3  import (
     4  	"encoding/json"
     5  	"errors"
     6  	"io"
     7  	"path/filepath"
     8  	"unicode/utf8"
     9  )
    10  
    11  // ErrDuplicatePath occurs when a tar archive has more than one entry for the
    12  // same file path
    13  var ErrDuplicatePath = errors.New("duplicates of file paths not supported")
    14  
    15  // Packer describes the methods to pack Entries to a storage destination
    16  type Packer interface {
    17  	// AddEntry packs the Entry and returns its position
    18  	AddEntry(e Entry) (int, error)
    19  }
    20  
    21  // Unpacker describes the methods to read Entries from a source
    22  type Unpacker interface {
    23  	// Next returns the next Entry being unpacked, or error, until io.EOF
    24  	Next() (*Entry, error)
    25  }
    26  
    27  type jsonUnpacker struct {
    28  	seen seenNames
    29  	dec  *json.Decoder
    30  }
    31  
    32  func (jup *jsonUnpacker) Next() (*Entry, error) {
    33  	var e Entry
    34  	err := jup.dec.Decode(&e)
    35  	if err != nil {
    36  		return nil, err
    37  	}
    38  
    39  	// check for dup name
    40  	if e.Type == FileType {
    41  		cName := filepath.Clean(e.GetName())
    42  		if _, ok := jup.seen[cName]; ok {
    43  			return nil, ErrDuplicatePath
    44  		}
    45  		jup.seen[cName] = struct{}{}
    46  	}
    47  
    48  	return &e, err
    49  }
    50  
    51  // NewJSONUnpacker provides an Unpacker that reads Entries (SegmentType and
    52  // FileType) as a json document.
    53  //
    54  // Each Entry read are expected to be delimited by new line.
    55  func NewJSONUnpacker(r io.Reader) Unpacker {
    56  	return &jsonUnpacker{
    57  		dec:  json.NewDecoder(r),
    58  		seen: seenNames{},
    59  	}
    60  }
    61  
    62  type jsonPacker struct {
    63  	w    io.Writer
    64  	e    *json.Encoder
    65  	pos  int
    66  	seen seenNames
    67  }
    68  
    69  type seenNames map[string]struct{}
    70  
    71  func (jp *jsonPacker) AddEntry(e Entry) (int, error) {
    72  	// if Name is not valid utf8, switch it to raw first.
    73  	if e.Name != "" {
    74  		if !utf8.ValidString(e.Name) {
    75  			e.NameRaw = []byte(e.Name)
    76  			e.Name = ""
    77  		}
    78  	}
    79  
    80  	// check early for dup name
    81  	if e.Type == FileType {
    82  		cName := filepath.Clean(e.GetName())
    83  		if _, ok := jp.seen[cName]; ok {
    84  			return -1, ErrDuplicatePath
    85  		}
    86  		jp.seen[cName] = struct{}{}
    87  	}
    88  
    89  	e.Position = jp.pos
    90  	err := jp.e.Encode(e)
    91  	if err != nil {
    92  		return -1, err
    93  	}
    94  
    95  	// made it this far, increment now
    96  	jp.pos++
    97  	return e.Position, nil
    98  }
    99  
   100  // NewJSONPacker provides a Packer that writes each Entry (SegmentType and
   101  // FileType) as a json document.
   102  //
   103  // The Entries are delimited by new line.
   104  func NewJSONPacker(w io.Writer) Packer {
   105  	return &jsonPacker{
   106  		w:    w,
   107  		e:    json.NewEncoder(w),
   108  		seen: seenNames{},
   109  	}
   110  }
   111
View as plain text