header.js 10.8 KB
// parse a 512-byte header block to a data object, or vice-versa
// If the data won't fit nicely in a simple header, then generate
// the appropriate extended header file, and return that.

module.exports = TarHeader

var tar = require("../tar.js")
  , fields = tar.fields
  , fieldOffs = tar.fieldOffs
  , fieldEnds = tar.fieldEnds
  , fieldSize = tar.fieldSize
  , numeric = tar.numeric
  , assert = require("assert").ok
  , space = " ".charCodeAt(0)
  , slash = "/".charCodeAt(0)
  , bslash = process.platform === "win32" ? "\\".charCodeAt(0) : null

function TarHeader (block) {
  if (!(this instanceof TarHeader)) return new TarHeader(block)
  if (block) this.decode(block)
}

TarHeader.prototype =
  { decode : decode
  , encode: encode
  , calcSum: calcSum
  , checkSum: checkSum
  }

TarHeader.parseNumeric = parseNumeric
TarHeader.encode = encode
TarHeader.decode = decode

// note that this will only do the normal ustar header, not any kind
// of extended posix header file.  If something doesn't fit comfortably,
// then it will set obj.needExtended = true, and set the block to
// the closest approximation.
function encode (obj) {
  if (!obj && !(this instanceof TarHeader)) throw new Error(
    "encode must be called on a TarHeader, or supplied an object")

  obj = obj || this
  var block = obj.block = new Buffer(512)

  // if the object has a "prefix", then that's actually an extension of
  // the path field.
  if (obj.prefix) {
    // console.error("%% header encoding, got a prefix", obj.prefix)
    obj.path = obj.prefix + "/" + obj.path
    // console.error("%% header encoding, prefixed path", obj.path)
    obj.prefix = ""
  }

  obj.needExtended = false

  if (obj.mode) {
    if (typeof obj.mode === "string") obj.mode = parseInt(obj.mode, 8)
    obj.mode = obj.mode & 0777
  }

  for (var f = 0; fields[f] !== null; f ++) {
    var field = fields[f]
      , off = fieldOffs[f]
      , end = fieldEnds[f]
      , ret

    switch (field) {
      case "cksum":
        // special, done below, after all the others
        break

      case "prefix":
        // special, this is an extension of the "path" field.
        // console.error("%% header encoding, skip prefix later")
        break

      case "type":
        // convert from long name to a single char.
        var type = obj.type || "0"
        if (type.length > 1) {
          type = tar.types[obj.type]
          if (!type) type = "0"
        }
        writeText(block, off, end, type)
        break

      case "path":
        // uses the "prefix" field if > 100 bytes, but <= 255
        var pathLen = Buffer.byteLength(obj.path)
          , pathFSize = fieldSize[fields.path]
          , prefFSize = fieldSize[fields.prefix]

        // paths between 100 and 255 should use the prefix field.
        // longer than 255
        if (pathLen > pathFSize &&
            pathLen <= pathFSize + prefFSize) {
          // need to find a slash somewhere in the middle so that
          // path and prefix both fit in their respective fields
          var searchStart = pathLen - 1 - pathFSize
            , searchEnd = prefFSize
            , found = false
            , pathBuf = new Buffer(obj.path)

          for ( var s = searchStart
              ; (s <= searchEnd)
              ; s ++ ) {
            if (pathBuf[s] === slash || pathBuf[s] === bslash) {
              found = s
              break
            }
          }

          if (found !== false) {
            prefix = pathBuf.slice(0, found).toString("utf8")
            path = pathBuf.slice(found + 1).toString("utf8")

            ret = writeText(block, off, end, path)
            off = fieldOffs[fields.prefix]
            end = fieldEnds[fields.prefix]
            // console.error("%% header writing prefix", off, end, prefix)
            ret = writeText(block, off, end, prefix) || ret
            break
          }
        }

        // paths less than 100 chars don't need a prefix
        // and paths longer than 255 need an extended header and will fail
        // on old implementations no matter what we do here.
        // Null out the prefix, and fallthrough to default.
        // console.error("%% header writing no prefix")
        var poff = fieldOffs[fields.prefix]
          , pend = fieldEnds[fields.prefix]
        writeText(block, poff, pend, "")
        // fallthrough

      // all other fields are numeric or text
      default:
        ret = numeric[field]
            ? writeNumeric(block, off, end, obj[field])
            : writeText(block, off, end, obj[field] || "")
        break
    }
    obj.needExtended = obj.needExtended || ret
  }

  var off = fieldOffs[fields.cksum]
    , end = fieldEnds[fields.cksum]

  writeNumeric(block, off, end, calcSum.call(this, block))

  return block
}

// if it's a negative number, or greater than will fit,
// then use write256.
var MAXNUM = { 12: 077777777777
             , 11: 07777777777
             , 8 : 07777777
             , 7 : 0777777 }
function writeNumeric (block, off, end, num) {
  var writeLen = end - off
    , maxNum = MAXNUM[writeLen] || 0

  num = num || 0
  // console.error("  numeric", num)

  if (num instanceof Date ||
      Object.prototype.toString.call(num) === "[object Date]") {
    num = num.getTime() / 1000
  }

  if (num > maxNum || num < 0) {
    write256(block, off, end, num)
    // need an extended header if negative or too big.
    return true
  }

  // god, tar is so annoying
  // if the string is small enough, you should put a space
  // between the octal string and the \0, but if it doesn't
  // fit, then don't.
  var numStr = Math.floor(num).toString(8)
  if (num < MAXNUM[writeLen - 1]) numStr += " "

  // pad with "0" chars
  if (numStr.length < writeLen) {
    numStr = (new Array(writeLen - numStr.length).join("0")) + numStr
  }

  if (numStr.length !== writeLen - 1) {
    throw new Error("invalid length: " + JSON.stringify(numStr) + "\n" +
                    "expected: "+writeLen)
  }
  block.write(numStr, off, writeLen, "utf8")
  block[end - 1] = 0
}

function write256 (block, off, end, num) {
  var buf = block.slice(off, end)
  var positive = num >= 0
  buf[0] = positive ? 0x80 : 0xFF

  // get the number as a base-256 tuple
  if (!positive) num *= -1
  var tuple = []
  do {
    var n = num % 256
    tuple.push(n)
    num = (num - n) / 256
  } while (num)

  var bytes = tuple.length

  var fill = buf.length - bytes
  for (var i = 1; i < fill; i ++) {
    buf[i] = positive ? 0 : 0xFF
  }

  // tuple is a base256 number, with [0] as the *least* significant byte
  // if it's negative, then we need to flip all the bits once we hit the
  // first non-zero bit.  The 2's-complement is (0x100 - n), and the 1's-
  // complement is (0xFF - n).
  var zero = true
  for (i = bytes; i > 0; i --) {
    var byte = tuple[bytes - i]
    if (positive) buf[fill + i] = byte
    else if (zero && byte === 0) buf[fill + i] = 0
    else if (zero) {
      zero = false
      buf[fill + i] = 0x100 - byte
    } else buf[fill + i] = 0xFF - byte
  }
}

function writeText (block, off, end, str) {
  // strings are written as utf8, then padded with \0
  var strLen = Buffer.byteLength(str)
    , writeLen = Math.min(strLen, end - off)
    // non-ascii fields need extended headers
    // long fields get truncated
    , needExtended = strLen !== str.length || strLen > writeLen

  // write the string, and null-pad
  if (writeLen > 0) block.write(str, off, writeLen, "utf8")
  for (var i = off + writeLen; i < end; i ++) block[i] = 0

  return needExtended
}

function calcSum (block) {
  block = block || this.block
  assert(Buffer.isBuffer(block) && block.length === 512)

  if (!block) throw new Error("Need block to checksum")

  // now figure out what it would be if the cksum was "        "
  var sum = 0
    , start = fieldOffs[fields.cksum]
    , end = fieldEnds[fields.cksum]

  for (var i = 0; i < fieldOffs[fields.cksum]; i ++) {
    sum += block[i]
  }

  for (var i = start; i < end; i ++) {
    sum += space
  }

  for (var i = end; i < 512; i ++) {
    sum += block[i]
  }

  return sum
}


function checkSum (block) {
  var sum = calcSum.call(this, block)
  block = block || this.block

  var cksum = block.slice(fieldOffs[fields.cksum], fieldEnds[fields.cksum])
  cksum = parseNumeric(cksum)

  return cksum === sum
}

function decode (block) {
  block = block || this.block
  assert(Buffer.isBuffer(block) && block.length === 512)

  this.block = block
  this.cksumValid = this.checkSum()

  var prefix = null

  // slice off each field.
  for (var f = 0; fields[f] !== null; f ++) {
    var field = fields[f]
      , val = block.slice(fieldOffs[f], fieldEnds[f])

    switch (field) {
      case "ustar":
        // if not ustar, then everything after that is just padding.
        if (val.toString() !== "ustar\0") {
          this.ustar = false
          return
        } else {
          // console.error("ustar:", val, val.toString())
          this.ustar = val.toString()
        }
        break

      // prefix is special, since it might signal the xstar header
      case "prefix":
        var atime = parseNumeric(val.slice(131, 131 + 12))
          , ctime = parseNumeric(val.slice(131 + 12, 131 + 12 + 12))
        if ((val[130] === 0 || val[130] === space) &&
            typeof atime === "number" &&
            typeof ctime === "number" &&
            val[131 + 12] === space &&
            val[131 + 12 + 12] === space) {
          this.atime = atime
          this.ctime = ctime
          val = val.slice(0, 130)
        }
        prefix = val.toString("utf8").replace(/\0+$/, "")
        // console.error("%% header reading prefix", prefix)
        break

      // all other fields are null-padding text
      // or a number.
      default:
        if (numeric[field]) {
          this[field] = parseNumeric(val)
        } else {
          this[field] = val.toString("utf8").replace(/\0+$/, "")
        }
        break
    }
  }

  // if we got a prefix, then prepend it to the path.
  if (prefix) {
    this.path = prefix + "/" + this.path
    // console.error("%% header got a prefix", this.path)
  }
}

function parse256 (buf) {
  // first byte MUST be either 80 or FF
  // 80 for positive, FF for 2's comp
  var positive
  if (buf[0] === 0x80) positive = true
  else if (buf[0] === 0xFF) positive = false
  else return null

  // build up a base-256 tuple from the least sig to the highest
  var zero = false
    , tuple = []
  for (var i = buf.length - 1; i > 0; i --) {
    var byte = buf[i]
    if (positive) tuple.push(byte)
    else if (zero && byte === 0) tuple.push(0)
    else if (zero) {
      zero = false
      tuple.push(0x100 - byte)
    } else tuple.push(0xFF - byte)
  }

  for (var sum = 0, i = 0, l = tuple.length; i < l; i ++) {
    sum += tuple[i] * Math.pow(256, i)
  }

  return positive ? sum : -1 * sum
}

function parseNumeric (f) {
  if (f[0] & 0x80) return parse256(f)

  var str = f.toString("utf8").split("\0")[0].trim()
    , res = parseInt(str, 8)

  return isNaN(res) ? null : res
}