Göran Krampe / Blimp

  import md5, os, osproc, parseopt2, strutils, parsecfg, streams, lapp, subexes, tables

  # blimp is a little utility program for handling large files

  # in git repositories. Its inspired by git-fat and s3annex

  # but doesn't rely on S3 for storage - it uses rsync like git-fat.
  # It is a single binary without any dependencies.

  #

  # Manual use:
  #
  # Use "blimp d mybigfile" to deflate it before commit.
  # Use "blimp i mybigfile" to inflate it back to original size.

  #

  # When deflated the file only has:

  #   "blimphash:" <filename> + <an md5sum>

  # ...inside it.

  #

  # The file is copied over to a local dir:
  #  <blimpstore>/<filename>-<md5sum>

  #

  # Configuration is in these locations in order:

  #

  #   ./.blimp.conf

  #   <gitroot>/.blimp.conf

  #   ~/<blimpstore>/.blimp.conf

  #   ~/.blimp.conf

  #
  # This way you can have settings per directory, per git clone,
  # per store and per user. A default blimpstore with a commented .blimp.conf
  # is created in ~/blimpstore if you run blimp and no .blimp.conf is found.

  
  const 
    versionMajor* = 0

    versionMinor* = 3
    versionPatch* = 0

    versionAsString* = $versionMajor & "." & $versionMinor & "." & $versionPatch

  var

    blimpStore, remoteBlimpStore, uploadCommandFormat, downloadCommandFormat, deleteCommandFormat, rsyncPassword, blimpVersion: string = nil

    homeDir, currentDir, gitRootDir: string

    verbose, stdio, onAllDeflated, onAllFiltered: bool

    stdinContent: string = nil

  let
    defaultConfig = """
  [rsync]

  # Set your local blimpstore directory. You can use %home%, %cwd% and %gitroot% in paths, works cross platform.

  # Example:
  #   # Place it inside the git clone
  #   blimpstore = "%gitroot%/.git/blimpstore"
  #   # Place it in current working directory (not very useful)
  #   blimpstore = "%cwd%/blimpstore"
  #
  # Default:
  #   # Place it in the users home directory
  #   blimpstore = "%home%/blimpstore"
  
  # Set this to your remote rsync location

  remote = "blimpuser@some-rsync-server.com::blimpstore"

  # Set this to your rsync password, it will be written out as a password-file called .blimp.pass on every rsync.

  password = "some-good-rsync-password-for-blimpuser"

  # The following three formats should not need editing.
  # $1 is the blimp filename, $2 is remote location and $3 is the local blimpstore directory set above.
  # NOTE: The password-file .blimp.pass will be created by blimp on every command, do not remove that option.

  upload = "rsync --password-file $3/.blimp.pass -avzP $3/$1 $2/"
  download = "rsync --password-file $3/.blimp.pass -avzP $2/$1 $3/"

  # This deletes a single file from destination, that is already deleted in source. Yeah... insane! But it works.

  delete = "rsync --password-file $3/.blimp.pass -dv --delete --existing --ignore-existing --include '$1' --exclude '*' $3/ $2"

  
  [blimp]
  # Minimal version, otherwise stop

  # version = """ & versionAsString

  
  proc cmd(cmd: string): string =

    try:

      let tup = execCmdEx(cmd)

      if tup[1] == 0:
        result = strip(tup[0])
      else:
        result = nil
    except:
      result = nil
  

  # Find git root dir or nil
  proc gitRoot(): string =
    cmd("git rev-parse --show-toplevel")
  
  # Git config
  proc gitConfigSet(key, val: string) =
    discard cmd("git config " & $key & " " & $val)
  
  proc gitConfigGet(key: string): string =
    cmd("git config --get " & $key)
  
  # Set blimp filter
  proc setBlimpFilter() =
    gitConfigSet("filter.blimp.clean", "\"blimp -s d %f\"")
    gitConfigSet("filter.blimp.smudge", "\"blimp -s i %f\"")
  
  
  # Ensure the filter is set
  proc ensureBlimpFilter() =
    if gitConfigGet("filter.blimp.clean") != "":
      setBlimpFilter()
  
  

  # Simple expansion of %home%, %cwd% and %gitroot%
  proc expandDirs(templ: string): string =
    result = templ.replace("%home%", homeDir)
    result = result.replace("%cwd%", currentDir)
    if result.contains("%gitroot%"):
      if gitRootDir.isNil: quit("Not in a git clone, can not expand %gitroot% in '" & templ & "'") 
      result = result.replace("%gitroot%", gitRootDir)
  

  # Load a blimp.conf file

  proc parseConfFile(filename: string) =

    var f = newFileStream(filename, fmRead)
    if f != nil:

      if verbose: echo "Reading config: " & filename

      var p: CfgParser  

      open(p, f, filename)
      while true:
        var e = next(p)
        case e.kind
        of cfgEof: 
          break
        of cfgSectionStart:
          continue # Ignore
        of cfgKeyValuePair:

          case e.key

          of "blimpstore":

            if blimpStore.isNil: blimpStore = expandDirs(e.value)

          of "remote":

            if remoteBlimpStore.isNil: remoteBlimpStore = expandDirs(e.value)

          of "password":
            if rsyncPassword.isNil: rsyncPassword = e.value

          of "upload":

            if uploadCommandFormat.isNil: uploadCommandFormat = expandDirs(e.value)

          of "download":

            if downloadCommandFormat.isNil: downloadCommandFormat = expandDirs(e.value)

          of "delete":

            if deleteCommandFormat.isNil: deleteCommandFormat = expandDirs(e.value)

          of "version":
            if blimpVersion.isNil: blimpVersion = e.value

          else:
            quit("Unknown configuration: " & e.key)
        of cfgOption:
          quit("Unknown configuration: " & e.key)
        of cfgError:
          quit("Parsing " & filename & ": " & e.msg)
      close(p)
  

  # Trivial helper to enable verbose
  proc run(cmd: string): auto =
    if verbose: echo(cmd)
    execCmd(cmd)

  # Every rsync command, make sure we have a password file
  proc rsyncRun(cmd: string): auto =

    if not rsyncPassword.isNil:
      writeFile(blimpStore / ".blimp.pass", rsyncPassword)
      if execCmd("chmod 600 " & blimpStore / ".blimp.pass") != 0:
        quit("Failed to chmod 600 " & blimpStore / ".blimp.pass")

    run(cmd)
  

  # Upload a file to the remote master blimpStore
  proc uploadFile(blimpFilename: string) =
    if remoteBlimpStore.isNil:

      echo("Remote blimpstore not set in configuration file, skipping uploading content:\n\t" & blimpFilename)

      return

    let errorCode = rsyncRun(format(uploadCommandFormat, blimpFilename, remoteBlimpStore, blimpStore))

    if errorCode != 0:
      quit("Something went wrong uploading " & blimpFilename & " to " & remoteBlimpStore, 2)

  # Download a file to the remote master blimpStore
  proc downloadFile(blimpFilename: string) =
    if remoteBlimpStore.isNil:
      quit("Remote blimpstore not set in configuration file, can not download content:\n\t" & blimpFilename)

    let errorCode = rsyncRun(format(downloadCommandFormat, blimpFilename, remoteBlimpStore, blimpStore))

    if errorCode != 0:
      quit("Something went wrong downloading " & blimpFilename & " from " & remoteBlimpStore, 3)

  # Delete a file from the remote master blimpStore
  proc remoteDeleteFile(blimpFilename: string) =
    if remoteBlimpStore.isNil:
      return

    let errorCode = rsyncRun(format(deleteCommandFormat, blimpFilename, remoteBlimpStore, blimpStore))

    if errorCode != 0:
      quit("Something went wrong deleting " & blimpFilename & " from " & remoteBlimpStore, 3)

  
  # Copy content to blimpStore, no upload yet.
  proc copyToBlimpStore(filename, blimpFilename: string) =
    if not existsFile(blimpStore / blimpFilename):

      if stdio:
        try:
          writeFile(blimpStore / blimpFilename, stdinContent)
        except:
          quit("Failed writing file: " & blimpStore / blimpFilename & " from stdin", 1)
      else:
        copyFile(filename, blimpStore / blimpFilename)

      uploadFile(blimpFilename)

  
  # Copy content from blimpStore, and downloading first if needed

  proc copyFromBlimpStore(blimpFilename, filename: string) =

    if not existsFile(blimpStore / blimpFilename):
      downloadFile(blimpFilename)

    if stdio:
      try:
        var content = readFile(blimpStore / blimpFilename)
        write(stdout, content)
      except:
        quit("Failed reading file: " & blimpStore / blimpFilename & " to stdout", 1)
    else:
      copyFile(blimpStore / blimpFilename, filename)

  # Delete from blimpStore and remote.
  proc deleteFromBlimpStore(blimpFilename, filename: string) =
    if existsFile(blimpStore / blimpFilename):
      removeFile(blimpStore / blimpFilename)
    remoteDeleteFile(blimpFilename)
  

  proc blimpFileNameFromString(line: string): string =

    let hashline = split(strip(line), ':')
    if hashline[0] == "blimphash":

      result = hashline[1]
    else:
      result = nil
  

  # Pick out blimpFilename (filename & "-" & hash)
  proc blimpFileName(filename: string): string =
    if stdio:
      blimpFileNameFromString(stdinContent)
    else:
      var hashfile: File
      if not open(hashfile, filename):
        quit("Failed opening file: " & filename, 4)
      blimpFileNameFromString(string(readLine(hashfile)))  
  

  # Get hash and compute blimpFilename
  proc computeBlimpFilename(filename: string): string =

    var content: string
    try:
      content = readFile(filename)
    except:
      quit("Failed opening file: " & filename, 1)
    let hash = getMD5(content)

    result = extractFilename(filename) & "-" & hash

  # Copy original file to blimpStore and replace with hash stub in git.

  proc deflate(filename: string) =

    if verbose: echo "Deflating " & filename
    var blimpFilename = blimpFilename(filename)
    if not blimpFilename.isNil:
      echo("\t" & filename & " is already deflated, skipping.")
    else:
      blimpFilename = computeBlimpFilename(filename)
      copyToBlimpStore(filename, blimpFilename)

      if stdio:

        write(stdout, "blimphash:" & blimpFilename)

      else:

        writeFile(filename, "blimphash:" & blimpFilename)

      if verbose: echo("\t" & filename & " deflated.")

  # Iterator over all deflated files in the git clone
  iterator allDeflated() =
    let filenames = cmd("git ls-files " & gitRootDir).split('\l')
    for fn in filenames:
      if not blimpFilename(fn).isNil:
        yield fn

        
  # Iterator over all files matching the blimp filter in the git clone
  iterator allFiltered() =
    let lines = cmd("git ls-files | git check-attr --stdin filter").split('\l')
    for line in lines:
      let status = line.split(':')
      if strip(status[2]) == "blimp":
        yield status[0]

  # Parse out hash from hash stub and copy back original content from blimpStore.

  proc inflate(filename: string) =

    if verbose: echo "Inflating " & filename

    let blimpFilename = blimpFilename(filename)
    if blimpFilename.isNil:

      echo("\t" & filename & " is not deflated, skipping.")

    else:
      copyFromBlimpStore(blimpfilename, filename)

      if verbose: echo("\t" & filename & " inflated.")

  
  # Inflates file first (if deflated) and then removes current content for it,
  # both locally and in remote.
  proc remove(filename: string) =
    var blimpFilename = blimpFilename(filename)
    if not blimpFilename.isNil:
      copyFromBlimpStore(blimpfilename, filename)

    else:

      blimpFilename = computeBlimpFilename(filename)
    deleteFromBlimpStore(blimpfilename, filename)
    echo("\t" & filename & " content removed from blimpstore locally and remotely.")

  
  proc setupBlimpStore() =
    try:
      if not existsDir(blimpStore):
        createDir(blimpStore)
    except:
      quit("Could not create " & blimpStore & " directory.", 1)
  
    try:
      if not existsFile(blimpStore / ".blimp.conf"):
        writeFile(blimpStore / ".blimp.conf", defaultConfig)
    except:
      quit("Could not create .blimp.conf config file in " & blimpStore & " directory.", 1)
  

  proc `$`(x: string): string =
    if x.isNil: "nil" else: x
  

  proc dumpConfig() =
    echo "\nDump of configuration:"
    echo "\tblimpStore: " & blimpStore
    echo "\tremoteBlimpStore: " & remoteBlimpStore
    echo "\tuploadCommandFormat: " & uploadCommandFormat
    echo "\tdownloadCommandFormat: " & downloadCommandFormat
    echo "\tdeleteCommandFormat: " & deleteCommandFormat

    echo "\trsyncPassword: " & $rsyncPassword

    echo "\tblimpVersion: " & $blimpVersion

    echo "\n"

  let synopsis = """

    blimp [options] <command> <filenames...>

      -h,--help                Show this
      --version                Show version of blimp

      -v,--verbose             Verbosity, only works without -s

      -i,--init                Set blimp filter in git config
      ----------

      <command>   (string)     (d)eflate, (i)nflate, remove

      -a,--all                 Operate on all deflated files in clone

      -f,--filter              Operate on all files matching blimp filter

      ----------
      -s,--stdio               If given, use stdin/stdout for content.
      <filenames> (string...)  One or more filepaths to inflate/deflate
  """
  let help = """

    blimp is a little utility program for handling large files
    in git repositories. Its inspired by git-fat and s3annex
    but doesn't rely on S3 for storage - it uses rsync like git-fat.

    It is a single binary without any dependencies. Its not as advanced
    as git-fat but basically does the same thing.

  
    Manual use:
  

    Use "blimp d mybigfile" to deflate a file, typically before commit.
    Use "blimp i mybigfile" to inflate it back to original content.

  
    When deflated the file only has:

      "blimphash:" <filename> "-" <md5sum>

    ...inside it.
  

    Deflate also copies the real content to your local blimpstore:

      <blimpstore>/<filename>-<md5sum>

    ...and if configured also uploads it to "remote", using rsync.

  
    Configuration is in these locations in order:
   
      ./.blimp.conf

      <gitroot>/.blimp.conf

      ~/<blimpstore>/.blimp.conf
      ~/.blimp.conf
  
    This way you can have settings per directory, per git clone,
    per store and per user. A default blimpstore with a commented .blimp.conf
    is created in ~/blimpstore if you run blimp and no .blimp.conf is found.
  
    Edit ~/blimpstore/.blimp.conf (or in another location) and set a proper
    remote and the proper rsync password. This ensures that its also properly
    synced with a master rsync repository that is typically shared.
  
    Inflate will bring back the original content by copying from

    your local blimpstore, and if its not there, first download from the remote.

    Use this whenever you need to work/edit the big file - in order to get
    its real content.
  
    The filenames given are all processed. If -s is used content is processed via
    stdin/stdout and only one filename can be passed. This is used when running blimp

    via a git filter (smudge/clean), see below.

  
    The remove command (no single character shortcut) will remove the file(s) content
    both from the local blimpstore and from the remote. This only removes
    the current content version, not older versions. The file itself is first
    inflated, if needed, and not deleted. This only "unblimps" the file.

    
    In order to have blimp work automatically you can:
    
    * Create a .gitattributes file with lines like:

      *.png filter=blimp binary

      
    * Configure blimp as a filter by running:
      git config filter.blimp.clean "blimp -s d %f"
      git config filter.blimp.smudge "blimp -s i %f"
  
    The above git config can be done by running "blimp --init".
  
    When the above is done (per clone) git will automatically run blimp deflate
    just before committing and blimp inflate when operations are done.
    
    This means that if you clone a git repository that already has a .gitattributes
    file in it that uses the blimp filter, then you should do:
    
      blimp --init inflate --all
    
    This will configure the blimp filter and also find and inflate all deflated
    files throughout the clone.

  """

  ################################ main #####################################

  # Set some dirs
  homeDir = getHomeDir()
  homeDir = homeDir[0.. -2] # Not sure why it keeps a trailing "/" on Linux
  currentDir = getCurrentDir()
  gitRootDir = gitRoot()

  # Using lapp to get args, on parsing failure this will show usage automatically

  var args = parse(synopsis)

  verbose = args["verbose"].asBool

  stdio = args["stdio"].asBool

  onAllDeflated = args["all"].asBool
  onAllFiltered = args["filter"].asBool

  
  # Can't do verbose with -s, that messes up stdout,
  # read in all of stdin once and for all
  if stdio:
    verbose = false
    try:
      stdinContent = readAll(stdin)
      close(stdin)
    except:
      quit("Failed reading stdin", 1)
  

  
  # Parse configuration files, may shadow and override each other

  parseConfFile(currentDir / ".blimp.conf")

  if not gitRootDir.isNil and gitRootDir != currentDir:

    parseConfFile(gitRootDir / ".blimp.conf")

  # If we haven't gotten a blimpstore yet, we set a default one
  if blimpStore.isNil:

    blimpStore = homeDir / "blimpstore"

  if existsDir(blimpStore):
    parseConfFile(blimpStore / ".blimp.conf")

  parseConfFile(homeDir / ".blimp.conf")

  # Let's just show what we have :)

  if verbose: dumpConfig()

  # These two are special, they short out

  if args.showHelp: quit(synopsis & help)

  if args.showVersion: quit("blimp version: " & versionAsString)

  # Check blimpVersion
  if not blimpVersion.isNil and blimpVersion != versionAsString:
    quit("Wrong version of blimp, configuration wants: " & blimpVersion)
  

  # Should we install the filter?
  if args["init"].asBool:
    ensureBlimpFilter()
    if verbose: echo("Installed blimp filter")
  

  let command = args["command"].asString

  var filenames: seq[PValue]
  if args.hasKey("filenames"):
    filenames = args["filenames"].asSeq

  
  # Make sure the local blimpstore is setup.
  setupBlimpStore()
  

  # Do the deed

  if command != "":
    if command == "d" or command == "deflate":

      if onAllFiltered:
        for fn in allFiltered():
          deflate(fn)
      else:
        for fn in filenames:
          deflate(fn.asString)

    elif command == "i" or command == "inflate":

      if onAllDeflated:

        for fn in allDeflated():
          inflate(fn)

      elif onAllFiltered:
        for fn in allFiltered():
          inflate(fn)

      else:
        for fn in filenames:
          inflate(fn.asString)
    elif command == "remove":
      for fn in filenames:
        remove(fn.asString)
    else:

      quit("Unknown command, only (d)eflate, (i)inflate or remove are valid.", 6)

  
  # All good
  quit(0)