added Comm Build System

24 files changed, 2502 insertions, 0 deletions

diff --git a/build/autoconf/mozconfig-find b/build/autoconf/mozconfig-find
new file mode 100755
index 0000000..97dd90c
--- /dev/null
+++ b/build/autoconf/mozconfig-find
@@ -0,0 +1,76 @@
+#! /bin/sh
+#
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+# mozconfigfind - Loads options from .mozconfig onto configure's
+#    command-line. The .mozconfig file is searched for in the 
+#    order:
+#       If $MOZCONFIG is set, use that.
+#       If one of $TOPSRCDIR/.mozconfig or $TOPSRCDIR/mozconfig exists, use it.
+#       If both exist, or if various legacy locations contain a mozconfig, error.
+#       Otherwise, use the default build options.
+#
+topsrcdir=$1
+
+abspath() {
+  if uname -s | grep -q MINGW; then
+    # We have no way to figure out whether we're in gmake or pymake right
+    # now. gmake gives us Unix-style paths while pymake gives us Windows-style
+    # paths, so attempt to handle both.
+    regexes='^\([A-Za-z]:\|\\\\\|\/\) ^\/'
+  else
+    regexes='^\/'
+  fi
+
+  for regex in $regexes; do
+    if echo $1 | grep -q $regex; then
+      echo $1
+      return
+    fi
+  done
+
+  # If we're at this point, we have a relative path
+  echo `pwd`/$1
+}
+
+if [ -n "$MOZCONFIG" ] && ! [ -f "$MOZCONFIG" ]; then
+  echo "Specified MOZCONFIG \"$MOZCONFIG\" does not exist!" 1>&2
+  exit 1
+fi
+
+if [ -n "$MOZ_MYCONFIG" ]; then
+  echo "Your environment currently has the MOZ_MYCONFIG variable set to \"$MOZ_MYCONFIG\". MOZ_MYCONFIG is no longer supported. Please use MOZCONFIG instead." 1>&2
+  exit 1
+fi
+
+if [ -z "$MOZCONFIG" ] && [ -f "$topsrcdir/.mozconfig" ] && [ -f "$topsrcdir/mozconfig" ]; then
+  echo "Both \$topsrcdir/.mozconfig and \$topsrcdir/mozconfig are supported, but you must choose only one. Please remove the other." 1>&2
+  exit 1
+fi
+
+for _config in "$MOZCONFIG" \
+               "$topsrcdir/.mozconfig" \
+               "$topsrcdir/mozconfig"
+do
+  if test -f "$_config"; then
+    abspath $_config
+    exit 0
+  fi
+done
+
+# We used to support a number of other implicit .mozconfig locations. We now
+# detect if we were about to use any of these locations and issue an error if we
+# find any.
+for _config in "$topsrcdir/mozconfig.sh" \
+               "$topsrcdir/myconfig.sh" \
+               "$HOME/.mozconfig" \
+               "$HOME/.mozconfig.sh" \
+               "$HOME/.mozmyconfig.sh"
+do
+  if test -f "$_config"; then
+    echo "You currently have a mozconfig at \"$_config\". This implicit location is no longer supported. Please move it to $topsrcdir/.mozconfig or specify it explicitly via \$MOZCONFIG." 1>&2
+    exit 1
+  fi
+done
diff --git a/build/autoconf/mozconfig2client-mk b/build/autoconf/mozconfig2client-mk
new file mode 100755
index 0000000..aaf8de1
--- /dev/null
+++ b/build/autoconf/mozconfig2client-mk
@@ -0,0 +1,76 @@
+#! /bin/sh
+#
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+# mozconfig2client-mk - Translates .mozconfig into options for client.mk.
+#    Prints defines to stdout.
+#
+# See mozconfig2configure for more details
+
+print_header() {
+  cat <<EOF
+# gmake
+# This file is automatically generated for client.mk.
+# Do not edit. Edit $FOUND_MOZCONFIG instead.
+
+EOF
+}
+
+ac_add_options() {
+  for _opt
+  do
+    case "$_opt" in
+    --target=*)
+      echo $_opt | sed s/--target/CONFIG_GUESS/
+      ;;
+    *)
+      echo "# $_opt is used by configure (not client.mk)"
+      ;;
+    esac
+  done
+}
+
+ac_add_app_options() {
+  echo "# $* is used by configure (not client.mk)"
+}
+
+mk_add_options() {
+  for _opt
+  do
+    # Escape shell characters, space, tab, dollar, quote, backslash,
+    # and substitute '@<word>@' with '$(<word>)'.
+    _opt=`echo "$_opt" | sed -e 's/\([\"\\]\)/\\\\\1/g; s/@\([^@]*\)@/\$(\1)/g;'`
+    echo $_opt;
+  done
+}
+
+# Main
+#--------------------------------------------------
+
+scriptdir=`dirname $0`
+topsrcdir=$1
+
+# If the path changes, configure should be rerun
+echo "# PATH=$PATH"
+
+# If FOUND_MOZCONFIG isn't set, look for it and make sure the script doesn't error out
+isfoundset=${FOUND_MOZCONFIG+yes}
+if [ -z $isfoundset ]; then
+  FOUND_MOZCONFIG=`$scriptdir/mozconfig-find $topsrcdir`
+  if [ $? -ne 0 ]; then
+    echo '$(error Fix above errors before continuing.)'
+  else
+    isfoundset=yes
+  fi
+fi
+
+if [ -n $isfoundset ]; then
+  if [ "$FOUND_MOZCONFIG" ]
+  then
+    print_header
+    . "$FOUND_MOZCONFIG"
+    echo "FOUND_MOZCONFIG := $FOUND_MOZCONFIG"
+  fi
+fi
diff --git a/build/dumbmake-dependencies b/build/dumbmake-dependencies
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/build/dumbmake-dependencies
diff --git a/build/mach_bootstrap.py b/build/mach_bootstrap.py
new file mode 100644
index 0000000..896d028
--- /dev/null
+++ b/build/mach_bootstrap.py
@@ -0,0 +1,14 @@
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+from __future__ import unicode_literals
+
+import os, sys
+
+def bootstrap(topsrcdir, mozilla_dir=None):
+    if mozilla_dir is None:
+        mozilla_dir = os.path.join(topsrcdir, 'platform')
+    sys.path[0:0] = [mozilla_dir]
+    import build.mach_bootstrap
+    return build.mach_bootstrap.bootstrap(topsrcdir, mozilla_dir)
diff --git a/build/pymake/make.py b/build/pymake/make.py
new file mode 100755
index 0000000..99d4839
--- /dev/null
+++ b/build/pymake/make.py
@@ -0,0 +1,40 @@
+#!/usr/bin/env python
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+# This is a wrapper around mozilla-central's pymake. If that isn't found then
+# this uses client.py to pull it in.
+
+import os
+import sys
+import subprocess
+import shlex
+
+def getpath(relpath):
+    thisdir = os.path.dirname(__file__)
+    return os.path.abspath(os.path.join(thisdir, *relpath))
+
+PYMAKE = getpath(["..", "..", "platform", "build", "pymake", "make.py"])
+
+def main(args):
+    if 'TINDERBOX_OUTPUT' in os.environ:
+        # When building on mozilla build slaves, execute mozmake instead. Until bug
+        # 978211, this is the easiest, albeit hackish, way to do this.
+        mozmake = os.path.join(os.path.dirname(__file__), '..', '..',
+            'mozmake.exe')
+        if os.path.exists(mozmake):
+            cmd = [mozmake]
+            cmd.extend(sys.argv[1:])
+            shell = os.environ.get('SHELL')
+            if shell and not shell.lower().endswith('.exe'):
+                cmd += ['SHELL=%s.exe' % shell]
+            sys.exit(subprocess.call(cmd))
+
+    if not os.path.exists(PYMAKE):
+        raise Exception("Pymake not found")
+
+    subprocess.check_call([sys.executable, PYMAKE] + args)
+
+if __name__ == "__main__":
+    main(sys.argv[1:])
diff --git a/build/pypng/check-sync-exceptions b/build/pypng/check-sync-exceptions
new file mode 100644
index 0000000..b326f7c
--- /dev/null
+++ b/build/pypng/check-sync-exceptions
@@ -0,0 +1,3 @@
+# Nothing in this directory needs to be in sync with mozilla
+# The contents are used only in c-c
+*
\ No newline at end of file
diff --git a/build/pypng/exnumpy.py b/build/pypng/exnumpy.py
new file mode 100644
index 0000000..82daf0a
--- /dev/null
+++ b/build/pypng/exnumpy.py
@@ -0,0 +1,128 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/exnumpy.py $
+# $Rev: 126 $
+
+# Numpy example.
+# Original code created by Mel Raab, modified by David Jones.
+
+'''
+  Example code integrating RGB PNG files, PyPNG and NumPy
+  (abstracted from Mel Raab's functioning code)
+'''
+
+# http://www.python.org/doc/2.4.4/lib/module-itertools.html
+import itertools
+
+import numpy
+import png
+
+
+''' If you have a PNG file for an RGB image,
+    and want to create a numpy array of data from it.
+'''
+# Read the file "picture.png" from the current directory.  The `Reader`
+# class can take a filename, a file-like object, or the byte data
+# directly; this suggests alternatives such as using urllib to read
+# an image from the internet:
+# png.Reader(file=urllib.urlopen('http://www.libpng.org/pub/png/PngSuite/basn2c16.png'))
+pngReader=png.Reader(filename='picture.png')
+# Tuple unpacking, using multiple assignment, is very useful for the
+# result of asDirect (and other methods).
+# See
+# http://docs.python.org/tutorial/introduction.html#first-steps-towards-programming
+row_count, column_count, pngdata, meta = pngReader.asDirect()
+bitdepth=meta['bitdepth']
+plane_count=meta['planes']
+
+# Make sure we're dealing with RGB files
+assert plane_count == 3
+
+''' Boxed row flat pixel:
+      list([R,G,B, R,G,B, R,G,B],
+           [R,G,B, R,G,B, R,G,B])
+    Array dimensions for this example:  (2,9)
+
+    Create `image_2d` as a two-dimensional NumPy array by stacking a
+    sequence of 1-dimensional arrays (rows).
+    The NumPy array mimics PyPNG's (boxed row flat pixel) representation;
+    it will have dimensions ``(row_count,column_count*plane_count)``.
+'''
+# The use of ``numpy.uint16``, below, is to convert each row to a NumPy
+# array with data type ``numpy.uint16``.  This is a feature of NumPy,
+# discussed further in 
+# http://docs.scipy.org/doc/numpy/user/basics.types.html .
+# You can use avoid the explicit conversion with
+# ``numpy.vstack(pngdata)``, but then NumPy will pick the array's data
+# type; in practice it seems to pick ``numpy.int32``, which is large enough
+# to hold any pixel value for any PNG image but uses 4 bytes per value when
+# 1 or 2 would be enough.
+# --- extract 001 start
+image_2d = numpy.vstack(itertools.imap(numpy.uint16, pngdata))
+# --- extract 001 end
+# Do not be tempted to use ``numpy.asarray``; when passed an iterator
+# (`pngdata` is often an iterator) it will attempt to create a size 1
+# array with the iterator as its only element.
+# An alternative to the above is to create the target array of the right
+# shape, then populate it row by row:
+if 0:
+    image_2d = numpy.zeros((row_count,plane_count*column_count),
+                           dtype=numpy.uint16)
+    for row_index, one_boxed_row_flat_pixels in enumerate(pngdata):
+        image_2d[row_index,:]=one_boxed_row_flat_pixels
+
+del pngReader
+del pngdata
+
+
+''' Reconfigure for easier referencing, similar to
+        Boxed row boxed pixel:
+            list([ (R,G,B), (R,G,B), (R,G,B) ],
+                 [ (R,G,B), (R,G,B), (R,G,B) ])
+    Array dimensions for this example:  (2,3,3)
+
+    ``image_3d`` will contain the image as a three-dimensional numpy
+    array, having dimensions ``(row_count,column_count,plane_count)``.
+'''
+# --- extract 002 start
+image_3d = numpy.reshape(image_2d,
+                         (row_count,column_count,plane_count))
+# --- extract 002 end
+
+
+''' ============= '''
+
+''' Convert NumPy image_3d array to PNG image file.
+
+    If the data is three-dimensional, as it is above, the best thing
+    to do is reshape it into a two-dimensional array with a shape of
+    ``(row_count, column_count*plane_count)``.  Because a
+    two-dimensional numpy array is an iterator, it can be passed
+    directly to the ``png.Writer.write`` method.
+'''
+
+row_count, column_count, plane_count = image_3d.shape
+assert plane_count==3
+
+pngfile = open('picture_out.png', 'wb')
+try:
+    # This example assumes that you have 16-bit pixel values in the data
+    # array (that's what the ``bitdepth=16`` argument is for).
+    # If you don't, then the resulting PNG file will likely be
+    # very dark.  Hey, it's only an example.
+    pngWriter = png.Writer(column_count, row_count,
+                           greyscale=False,
+                           alpha=False,
+                           bitdepth=16)
+    # As of 2009-04-13 passing a numpy array that has an element type
+    # that is a numpy integer type (for example, the `image_3d` array has an
+    # element type of ``numpy.uint16``) generates a deprecation warning.
+    # This is probably a bug in numpy; it may go away in the future.
+    # The code still works despite the warning.
+    # See http://code.google.com/p/pypng/issues/detail?id=44
+# --- extract 003 start
+    pngWriter.write(pngfile,
+                    numpy.reshape(image_3d, (-1, column_count*plane_count)))
+# --- extract 003 end
+finally:
+    pngfile.close()
+
diff --git a/build/pypng/iccp.py b/build/pypng/iccp.py
new file mode 100644
index 0000000..190db73
--- /dev/null
+++ b/build/pypng/iccp.py
@@ -0,0 +1,537 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/iccp.py $
+# $Rev: 182 $
+
+# iccp
+#
+# International Color Consortium Profile
+#
+# Tools for manipulating ICC profiles.
+#
+# An ICC profile can be extracted from a PNG image (iCCP chunk).
+#
+#
+# Non-standard ICCP tags.
+#
+# Apple use some (widespread but) non-standard tags.  These can be
+# displayed in Apple's ColorSync Utility.
+# - 'vcgt' (Video Card Gamma Tag).  Table to load into video
+#    card LUT to apply gamma.
+# - 'ndin' Apple display native information.
+# - 'dscm' Apple multi-localized description strings.
+# - 'mmod' Apple display make and model information.
+# 
+
+# References
+#
+# [ICC 2001] ICC Specification ICC.1:2001-04 (Profile version 2.4.0)
+# [ICC 2004] ICC Specification ICC.1:2004-10 (Profile version 4.2.0.0)
+
+import struct
+
+import png
+
+class FormatError(Exception):
+    pass
+
+class Profile:
+    """An International Color Consortium Profile (ICC Profile)."""
+
+    def __init__(self):
+        self.rawtagtable = None
+        self.rawtagdict = {}
+        self.d = dict()
+
+    def fromFile(self, inp, name='<unknown>'):
+
+        # See [ICC 2004]
+        profile = inp.read(128)
+        if len(profile) < 128:
+            raise FormatError("ICC Profile is too short.")
+        size, = struct.unpack('>L', profile[:4])
+        profile += inp.read(d['size'] - len(profile))
+        return self.fromString(profile, name)
+
+    def fromString(self, profile, name='<unknown>'):
+        self.d = dict()
+        d = self.d
+        if len(profile) < 128:
+            raise FormatError("ICC Profile is too short.")
+        d.update(
+          zip(['size', 'preferredCMM', 'version',
+               'profileclass', 'colourspace', 'pcs'],
+              struct.unpack('>L4sL4s4s4s', profile[:24])))
+        if len(profile) < d['size']:
+            warnings.warn(
+              'Profile size declared to be %d, but only got %d bytes' %
+                (d['size'], len(profile)))
+        d['version'] = '%08x' % d['version']
+        d['created'] = readICCdatetime(profile[24:36])
+        d.update(
+          zip(['acsp', 'platform', 'flag', 'manufacturer', 'model'],
+              struct.unpack('>4s4s3L', profile[36:56])))
+        if d['acsp'] != 'acsp':
+            warnings.warn('acsp field not present (not an ICC Profile?).')
+        d['deviceattributes'] = profile[56:64]
+        d['intent'], = struct.unpack('>L', profile[64:68])
+        d['pcsilluminant'] = readICCXYZNumber(profile[68:80])
+        d['creator'] = profile[80:84]
+        d['id'] = profile[84:100]
+        ntags, = struct.unpack('>L', profile[128:132])
+        d['ntags'] = ntags
+        fmt = '4s2L' * ntags
+        # tag table
+        tt = struct.unpack('>' + fmt, profile[132:132+12*ntags])
+        tt = group(tt, 3)
+
+        # Could (should) detect 2 or more tags having the same sig.  But
+        # we don't.  Two or more tags with the same sig is illegal per
+        # the ICC spec.
+        
+        # Convert (sig,offset,size) triples into (sig,value) pairs.
+        rawtag = map(lambda x: (x[0], profile[x[1]:x[1]+x[2]]), tt)
+        self.rawtagtable = rawtag
+        self.rawtagdict = dict(rawtag)
+        tag = dict()
+        # Interpret the tags whose types we know about
+        for sig, v in rawtag:
+            if sig in tag:
+                warnings.warn("Duplicate tag %r found.  Ignoring." % sig)
+                continue
+            v = ICCdecode(v)
+            if v is not None:
+                tag[sig] = v
+        self.tag = tag
+        return self
+
+    def greyInput(self):
+        """Adjust ``self.d`` dictionary for greyscale input device.
+        ``profileclass`` is 'scnr', ``colourspace`` is 'GRAY', ``pcs``
+        is 'XYZ '.
+        """
+
+        self.d.update(dict(profileclass='scnr',
+          colourspace='GRAY', pcs='XYZ '))
+        return self
+
+    def maybeAddDefaults(self):
+        if self.rawtagdict:
+            return
+        self._addTags(
+          cprt='Copyright unknown.',
+          desc='created by $URL: http://pypng.googlecode.com/svn/trunk/code/iccp.py $ $Rev: 182 $',
+          wtpt=D50(),
+          )
+
+    def addTags(self, **k):
+        self.maybeAddDefaults()
+        self._addTags(**k)
+
+    def _addTags(self, **k):
+        """Helper for :meth:`addTags`."""
+
+        for tag, thing in k.items():
+            if not isinstance(thing, (tuple, list)):
+                thing = (thing,)
+            typetag = defaulttagtype[tag]
+            self.rawtagdict[tag] = encode(typetag, *thing)
+        return self
+
+    def write(self, out):
+        """Write ICC Profile to the file."""
+
+        if not self.rawtagtable:
+            self.rawtagtable = self.rawtagdict.items()
+        tags = tagblock(self.rawtagtable)
+        self.writeHeader(out, 128 + len(tags))
+        out.write(tags)
+        out.flush()
+
+        return self
+
+    def writeHeader(self, out, size=999):
+        """Add default values to the instance's `d` dictionary, then
+        write a header out onto the file stream.  The size of the
+        profile must be specified using the `size` argument.
+        """
+
+        def defaultkey(d, key, value):
+            """Add ``[key]==value`` to the dictionary `d`, but only if
+            it does not have that key already.
+            """
+
+            if key in d:
+                return
+            d[key] = value
+
+        z = '\x00' * 4
+        defaults = dict(preferredCMM=z,
+                        version='02000000',
+                        profileclass=z,
+                        colourspace=z,
+                        pcs='XYZ ',
+                        created=writeICCdatetime(),
+                        acsp='acsp',
+                        platform=z,
+                        flag=0,
+                        manufacturer=z,
+                        model=0,
+                        deviceattributes=0,
+                        intent=0,
+                        pcsilluminant=encodefuns()['XYZ'](*D50()),
+                        creator=z,
+                        )
+        for k,v in defaults.items():
+            defaultkey(self.d, k, v)
+
+        hl = map(self.d.__getitem__,
+                 ['preferredCMM', 'version', 'profileclass', 'colourspace',
+                  'pcs', 'created', 'acsp', 'platform', 'flag',
+                  'manufacturer', 'model', 'deviceattributes', 'intent',
+                  'pcsilluminant', 'creator'])
+        # Convert to struct.pack input
+        hl[1] = int(hl[1], 16)
+
+        out.write(struct.pack('>L4sL4s4s4s12s4s4sL4sLQL12s4s', size, *hl))
+        out.write('\x00' * 44)
+        return self
+
+def encodefuns():
+    """Returns a dictionary mapping ICC type signature sig to encoding
+    function.  Each function returns a string comprising the content of
+    the encoded value.  To form the full value, the type sig and the 4
+    zero bytes should be prefixed (8 bytes).
+    """
+
+    def desc(ascii):
+        """Return textDescription type [ICC 2001] 6.5.17.  The ASCII part is
+        filled in with the string `ascii`, the Unicode and ScriptCode parts
+        are empty."""
+
+        ascii += '\x00'
+        l = len(ascii)
+
+        return struct.pack('>L%ds2LHB67s' % l,
+                           l, ascii, 0, 0, 0, 0, '')
+
+    def text(ascii):
+        """Return textType [ICC 2001] 6.5.18."""
+
+        return ascii + '\x00'
+
+    def curv(f=None, n=256):
+        """Return a curveType, [ICC 2001] 6.5.3.  If no arguments are
+        supplied then a TRC for a linear response is generated (no entries).
+        If an argument is supplied and it is a number (for *f* to be a
+        number it  means that ``float(f)==f``) then a TRC for that
+        gamma value is generated.
+        Otherwise `f` is assumed to be a function that maps [0.0, 1.0] to
+        [0.0, 1.0]; an `n` element table is generated for it.
+        """
+
+        if f is None:
+            return struct.pack('>L',  0)
+        try:
+            if float(f) == f:
+                return struct.pack('>LH', 1, int(round(f*2**8)))
+        except (TypeError, ValueError):
+            pass
+        assert n >= 2
+        table = []
+        M = float(n-1)
+        for i in range(n):
+            x = i/M
+            table.append(int(round(f(x) * 65535)))
+        return struct.pack('>L%dH' % n, n, *table)
+
+    def XYZ(*l):
+        return struct.pack('>3l', *map(fs15f16, l))
+
+    return locals()
+
+# Tag type defaults.
+# Most tags can only have one or a few tag types.
+# When encoding, we associate a default tag type with each tag so that
+# the encoding is implicit.
+defaulttagtype=dict(
+  A2B0='mft1',
+  A2B1='mft1',
+  A2B2='mft1',
+  bXYZ='XYZ',
+  bTRC='curv',
+  B2A0='mft1',
+  B2A1='mft1',
+  B2A2='mft1',
+  calt='dtim',
+  targ='text',
+  chad='sf32',
+  chrm='chrm',
+  cprt='desc',
+  crdi='crdi',
+  dmnd='desc',
+  dmdd='desc',
+  devs='',
+  gamt='mft1',
+  kTRC='curv',
+  gXYZ='XYZ',
+  gTRC='curv',
+  lumi='XYZ',
+  meas='',
+  bkpt='XYZ',
+  wtpt='XYZ',
+  ncol='',
+  ncl2='',
+  resp='',
+  pre0='mft1',
+  pre1='mft1',
+  pre2='mft1',
+  desc='desc',
+  pseq='',
+  psd0='data',
+  psd1='data',
+  psd2='data',
+  psd3='data',
+  ps2s='data',
+  ps2i='data',
+  rXYZ='XYZ',
+  rTRC='curv',
+  scrd='desc',
+  scrn='',
+  tech='sig',
+  bfd='',
+  vued='desc',
+  view='view',
+)
+
+def encode(tsig, *l):
+    """Encode a Python value as an ICC type.  `tsig` is the type
+    signature to (the first 4 bytes of the encoded value, see [ICC 2004]
+    section 10.
+    """
+
+    fun = encodefuns()
+    if tsig not in fun:
+        raise "No encoder for type %r." % tsig
+    v = fun[tsig](*l)
+    # Padd tsig out with spaces.
+    tsig = (tsig + '   ')[:4]
+    return tsig + '\x00'*4 + v
+
+def tagblock(tag):
+    """`tag` should be a list of (*signature*, *element*) pairs, where
+    *signature* (the key) is a length 4 string, and *element* is the
+    content of the tag element (another string).
+    
+    The entire tag block (consisting of first a table and then the
+    element data) is constructed and returned as a string.
+    """
+
+    n = len(tag)
+    tablelen = 12*n
+
+    # Build the tag table in two parts.  A list of 12-byte tags, and a
+    # string of element data.  Offset is the offset from the start of
+    # the profile to the start of the element data (so the offset for
+    # the next element is this offset plus the length of the element
+    # string so far).
+    offset = 128 + tablelen + 4
+    # The table.  As a string.
+    table = ''
+    # The element data
+    element = ''
+    for k,v in tag:
+        table += struct.pack('>4s2L', k, offset + len(element), len(v))
+        element += v
+    return struct.pack('>L', n) + table + element
+
+def iccp(out, inp):
+    profile = Profile().fromString(*profileFromPNG(inp))
+    print >>out, profile.d
+    print >>out, map(lambda x: x[0], profile.rawtagtable)
+    print >>out, profile.tag
+
+def profileFromPNG(inp):
+    """Extract profile from PNG file.  Return (*profile*, *name*)
+    pair."""
+    r = png.Reader(file=inp)
+    _,chunk = r.chunk('iCCP')
+    i = chunk.index('\x00')
+    name = chunk[:i]
+    compression = chunk[i+1]
+    assert compression == chr(0)
+    profile = chunk[i+2:].decode('zlib')
+    return profile, name
+
+def iccpout(out, inp):
+    """Extract ICC Profile from PNG file `inp` and write it to
+    the file `out`."""
+
+    out.write(profileFromPNG(inp)[0])
+
+def fs15f16(x):
+    """Convert float to ICC s15Fixed16Number (as a Python ``int``)."""
+
+    return int(round(x * 2**16))
+
+def D50():
+    """Return D50 illuminant as an (X,Y,Z) triple."""
+
+    # See [ICC 2001] A.1
+    return (0.9642, 1.0000, 0.8249)
+
+
+def writeICCdatetime(t=None):
+    """`t` should be a gmtime tuple (as returned from
+    ``time.gmtime()``).  If not supplied, the current time will be used.
+    Return an ICC dateTimeNumber in a 12 byte string.
+    """
+
+    import time
+    if t is None:
+        t = time.gmtime()
+    return struct.pack('>6H', *t[:6])
+
+def readICCdatetime(s):
+    """Convert from 12 byte ICC representation of dateTimeNumber to
+    ISO8601 string. See [ICC 2004] 5.1.1"""
+
+    return '%04d-%02d-%02dT%02d:%02d:%02dZ' % struct.unpack('>6H', s)
+
+def readICCXYZNumber(s):
+    """Convert from 12 byte ICC representation of XYZNumber to (x,y,z)
+    triple of floats.  See [ICC 2004] 5.1.11"""
+
+    return s15f16l(s)
+
+def s15f16l(s):
+    """Convert sequence of ICC s15Fixed16 to list of float."""
+    # Note: As long as float has at least 32 bits of mantissa, all
+    # values are preserved.
+    n = len(s)//4
+    t = struct.unpack('>%dl' % n, s)
+    return map((2**-16).__mul__, t)
+
+# Several types and their byte encodings are defined by [ICC 2004]
+# section 10.  When encoded, a value begins with a 4 byte type
+# signature.  We use the same 4 byte type signature in the names of the
+# Python functions that decode the type into a Pythonic representation.
+
+def ICCdecode(s):
+    """Take an ICC encoded tag, and dispatch on its type signature
+    (first 4 bytes) to decode it into a Python value.  Pair (*sig*,
+    *value*) is returned, where *sig* is a 4 byte string, and *value* is
+    some Python value determined by the content and type.
+    """
+
+    sig = s[0:4].strip()
+    f=dict(text=RDtext,
+           XYZ=RDXYZ,
+           curv=RDcurv,
+           vcgt=RDvcgt,
+           sf32=RDsf32,
+           )
+    if sig not in f:
+        return None
+    return (sig, f[sig](s))
+
+def RDXYZ(s):
+    """Convert ICC XYZType to rank 1 array of trimulus values."""
+
+    # See [ICC 2001] 6.5.26
+    assert s[0:4] == 'XYZ '
+    return readICCXYZNumber(s[8:])
+
+def RDsf32(s):
+    """Convert ICC s15Fixed16ArrayType to list of float."""
+    # See [ICC 2004] 10.18
+    assert s[0:4] == 'sf32'
+    return s15f16l(s[8:])
+
+def RDmluc(s):
+    """Convert ICC multiLocalizedUnicodeType.  This types encodes
+    several strings together with a language/country code for each
+    string.  A list of (*lc*, *string*) pairs is returned where *lc* is
+    the 4 byte language/country code, and *string* is the string
+    corresponding to that code.  It seems unlikely that the same
+    language/country code will appear more than once with different
+    strings, but the ICC standard does not prohibit it."""
+    # See [ICC 2004] 10.13
+    assert s[0:4] == 'mluc'
+    n,sz = struct.unpack('>2L', s[8:16])
+    assert sz == 12
+    record = []
+    for i in range(n):
+        lc,l,o = struct.unpack('4s2L', s[16+12*n:28+12*n])
+        record.append(lc, s[o:o+l])
+    # How are strings encoded?
+    return record
+
+def RDtext(s):
+    """Convert ICC textType to Python string."""
+    # Note: type not specified or used in [ICC 2004], only in older
+    # [ICC 2001].
+    # See [ICC 2001] 6.5.18
+    assert s[0:4] == 'text'
+    return s[8:-1]
+
+def RDcurv(s):
+    """Convert ICC curveType."""
+    # See [ICC 2001] 6.5.3
+    assert s[0:4] == 'curv'
+    count, = struct.unpack('>L', s[8:12])
+    if count == 0:
+        return dict(gamma=1)
+    table = struct.unpack('>%dH' % count, s[12:])
+    if count == 1:
+        return dict(gamma=table[0]*2**-8)
+    return table
+
+def RDvcgt(s):
+    """Convert Apple CMVideoCardGammaType."""
+    # See
+    # http://developer.apple.com/documentation/GraphicsImaging/Reference/ColorSync_Manager/Reference/reference.html#//apple_ref/c/tdef/CMVideoCardGammaType
+    assert s[0:4] == 'vcgt'
+    tagtype, = struct.unpack('>L', s[8:12])
+    if tagtype != 0:
+        return s[8:]
+    if tagtype == 0:
+        # Table.
+        channels,count,size = struct.unpack('>3H', s[12:18])
+        if size == 1:
+            fmt = 'B'
+        elif size == 2:
+            fmt = 'H'
+        else:
+            return s[8:]
+        l = len(s[18:])//size
+        t = struct.unpack('>%d%s' % (l, fmt), s[18:])
+        t = group(t, count)
+        return size, t
+    return s[8:]
+
+
+def group(s, n):
+    # See
+    # http://www.python.org/doc/2.6/library/functions.html#zip
+    return zip(*[iter(s)]*n)
+
+
+def main(argv=None):
+    import sys
+    from getopt import getopt
+    if argv is None:
+        argv = sys.argv
+    argv = argv[1:]
+    opt,arg = getopt(argv, 'o:')
+    if len(arg) > 0:
+        inp = open(arg[0], 'rb')
+    else:
+        inp = sys.stdin
+    for o,v in opt:
+        if o == '-o':
+            f = open(v, 'wb')
+            return iccpout(f, inp)
+    return iccp(sys.stdout, inp)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/mkiccp.py b/build/pypng/mkiccp.py
new file mode 100644
index 0000000..08e8df6
--- /dev/null
+++ b/build/pypng/mkiccp.py
@@ -0,0 +1,45 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/mkiccp.py $
+# $Rev: 182 $
+# Make ICC Profile
+
+# References
+#
+# [ICC 2001] ICC Specification ICC.1:2001-04 (Profile version 2.4.0)
+# [ICC 2004] ICC Specification ICC.1:2004-10 (Profile version 4.2.0.0)
+
+import struct
+
+# Local module.
+import iccp
+
+def black(m):
+    """Return a function that maps all values from [0.0,m] to 0, and maps
+    the range [m,1.0] into [0.0, 1.0] linearly.
+    """
+
+    m = float(m)
+
+    def f(x):
+        if x <= m:
+            return 0.0
+        return (x-m)/(1.0-m)
+    return f
+
+# For monochrome input the required tags are (See [ICC 2001] 6.3.1.1):
+# profileDescription [ICC 2001] 6.4.32
+# grayTRC [ICC 2001] 6.4.19
+# mediaWhitePoint [ICC 2001] 6.4.25
+# copyright [ICC 2001] 6.4.13
+
+def agreyprofile(out):
+    it = iccp.Profile().greyInput()
+    it.addTags(kTRC=black(0.07))
+    it.write(out)
+
+def main():
+    import sys
+    agreyprofile(sys.stdout)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pdsimgtopng b/build/pypng/pdsimgtopng
new file mode 100644
index 0000000..975db93
--- /dev/null
+++ b/build/pypng/pdsimgtopng
@@ -0,0 +1,99 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pdsimgtopng $
+# $Rev: 154 $
+# PDS Image to PNG
+
+import re
+import struct
+
+import png
+
+class FormatError(Exception):
+    pass
+
+def pdskey(s, k):
+    """Lookup key `k` in string `s`.  Returns value (as a string), or
+    raises exception if not found.
+    """
+
+    assert re.match(r' *\^?[:\w]+$', k)
+    safere = '^' + re.escape(k) +r' *= *(\w+)'
+    m = re.search(safere, s, re.MULTILINE)
+    if not m:
+        raise FormatError("Can't find %s." % k)
+    return m.group(1)
+
+def img(inp):
+    """Open the PDS IMG file `inp` and return (*pixels*, *info*).
+    *pixels* is an iterator over the rows, *info* is the information
+    dictionary.
+    """
+
+    err = __import__('sys').stderr
+
+    consumed = 1024
+
+    s = inp.read(consumed)
+    record_type = pdskey(s, 'RECORD_TYPE')
+    if record_type != 'FIXED_LENGTH':
+        raise FormatError(
+          "Can only deal with FIXED_LENGTH record type (found %s)" %
+            record_type)
+    record_bytes = int(pdskey(s,'RECORD_BYTES'))
+    file_records = int(pdskey(s, 'FILE_RECORDS'))
+    label_records = int(pdskey(s, 'LABEL_RECORDS'))
+    remaining = label_records * record_bytes - consumed
+    s += inp.read(remaining)
+    consumed += remaining
+
+    image_pointer = int(pdskey(s, '^IMAGE'))
+    # "^IMAGE" locates a record.  Records are numbered starting from 1.
+    image_index = image_pointer - 1
+    image_offset = image_index * record_bytes
+    gap = image_offset - consumed
+    assert gap >= 0
+    if gap:
+        inp.read(gap)
+    # This assumes there is only one OBJECT in the file, and it is the
+    # IMAGE.
+    height = int(pdskey(s, '  LINES'))
+    width = int(pdskey(s, '  LINE_SAMPLES'))
+    sample_type = pdskey(s, '  SAMPLE_TYPE')
+    sample_bits = int(pdskey(s, '  SAMPLE_BITS'))
+    # For Messenger MDIS, SAMPLE_BITS is reported as 16, but only values
+    # from 0 ot 4095 are used.
+    bitdepth = 12
+    if sample_type == 'MSB_UNSIGNED_INTEGER':
+        fmt = '>H'
+    else:
+        raise 'Unknown sample type: %s.' % sample_type
+    sample_bytes = (1,2)[bitdepth > 8]
+    row_bytes = sample_bytes * width
+    fmt = fmt[:1] + str(width) + fmt[1:]
+    def rowiter():
+        for y in range(height):
+            yield struct.unpack(fmt, inp.read(row_bytes))
+    info = dict(greyscale=True, alpha=False, bitdepth=bitdepth,
+      size=(width,height), gamma=1.0)
+    return rowiter(), info
+
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    argv = argv[1:]
+    arg = argv
+    if len(arg) >= 1:
+        f = open(arg[0], 'rb')
+    else:
+        f = sys.stdin
+    pixels,info = img(f)
+    w = png.Writer(**info)
+    w.write(sys.stdout, pixels)
+
+if __name__ == '__main__':
+    main()
+
+
diff --git a/build/pypng/pipasgrey b/build/pypng/pipasgrey
new file mode 100644
index 0000000..2b3727f
--- /dev/null
+++ b/build/pypng/pipasgrey
@@ -0,0 +1,73 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipasgrey $
+# $Rev: 187 $
+
+# pipasgrey
+
+# Convert image to grey (L, or LA), but only if that involves no colour
+# change.
+
+def asgrey(out, inp, quiet=False):
+    """Convert image to greyscale, but only when no colour change.  This
+    works by using the input G channel (green) as the output L channel
+    (luminance) and checking that every pixel is grey as we go.  A non-grey
+    pixel will raise an error, but if `quiet` is true then the grey pixel
+    check is suppressed.
+    """
+
+    from array import array
+
+    import png
+
+    r = png.Reader(file=inp)
+    _,_,pixels,info = r.asDirect()
+    if info['greyscale']:
+        w = png.Writer(**info)
+        return w.write(out, pixels)
+    planes = info['planes']
+    targetplanes = planes - 2
+    alpha = info['alpha']
+    width = info['size'][0]
+    typecode = 'BH'[info['bitdepth'] > 8]
+    # Values per target row
+    vpr = width * (targetplanes)
+    def iterasgrey():
+        for i,row in enumerate(pixels):
+            row = array(typecode, row)
+            targetrow = array(typecode, [0]*vpr)
+            # Copy G (and possibly A) channel.
+            green = row[0::planes]
+            if alpha:
+                targetrow[0::2] = green
+                targetrow[1::2] = row[3::4]
+            else:
+                targetrow = green
+            # Check R and B channel match.
+            if not quiet and (
+              green != row[0::planes] or green != row[2::planes]):
+                raise ValueError('Row %i contains non-grey pixel.' % i)
+            yield targetrow
+    info['greyscale'] = True
+    del info['planes']
+    w = png.Writer(**info)
+    w.write(out, iterasgrey())
+
+def main(argv=None):
+    from getopt import getopt
+    import sys
+    if argv is None:
+        argv = sys.argv
+    argv = argv[1:]
+    opt,argv = getopt(argv, 'q')
+    quiet = False
+    for o,v in opt:
+        if o == '-q':
+            quiet = True
+    if len(argv) > 0:
+        f = open(argv[0], 'rb')
+    else:
+        f = sys.stdin
+    return asgrey(sys.stdout, f, quiet)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pipcat b/build/pypng/pipcat
new file mode 100644
index 0000000..e0d0805
--- /dev/null
+++ b/build/pypng/pipcat
@@ -0,0 +1,44 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipcat $
+# $Rev: 77 $
+
+# http://www.python.org/doc/2.4.4/lib/module-itertools.html
+import itertools
+import sys
+
+import png
+
+def cat(out, l):
+    """Concatenate the list of images.  All input images must be same
+    height and have the same number of channels.  They are concatenated
+    left-to-right.  `out` is the (open file) destination for the
+    output image.  `l` should be a list of open files (the input
+    image files).
+    """
+
+    l = map(lambda f: png.Reader(file=f), l)
+    # Ewgh, side effects.
+    map(lambda r: r.preamble(), l)
+    # The reference height; from the first image.
+    height = l[0].height
+    # The total target width
+    width = 0
+    for i,r in enumerate(l):
+        if r.height != height:
+            raise Error('Image %d, height %d, does not match %d.' %
+              (i, r.height, height))
+        width += r.width
+    pixel,info = zip(*map(lambda r: r.asDirect()[2:4], l))
+    tinfo = dict(info[0])
+    del tinfo['size']
+    w = png.Writer(width, height, **tinfo)
+    def itercat():
+        for row in itertools.izip(*pixel):
+            yield itertools.chain(*row)
+    w.write(out, itercat())
+
+def main(argv):
+    return cat(sys.stdout, map(lambda n: open(n, 'rb'), argv[1:]))
+
+if __name__ == '__main__':
+    main(sys.argv)
diff --git a/build/pypng/pipcolours b/build/pypng/pipcolours
new file mode 100644
index 0000000..7c76df8
--- /dev/null
+++ b/build/pypng/pipcolours
@@ -0,0 +1,56 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipcolours $
+# $Rev: 96 $
+
+# pipcolours - extract all colours present in source image.
+
+def colours(out, inp):
+    import itertools
+    import png
+
+    r = png.Reader(file=inp)
+    _,_,pixels,info = r.asDirect()
+    planes = info['planes']
+    col = set()
+    for row in pixels:
+        # Ewgh, side effects on col
+        map(col.add, png.group(row, planes))
+    col,planes = channel_reduce(col, planes)
+    col = list(col)
+    col.sort()
+    col = list(itertools.chain(*col))
+    width = len(col)//planes
+    greyscale = planes in (1,2)
+    alpha = planes in (2,4)
+    bitdepth = info['bitdepth']
+    w = png.Writer(width, 1,
+        bitdepth=bitdepth, greyscale=greyscale, alpha=alpha)
+    w.write(out, [col])
+
+def channel_reduce(col, planes):
+    """Attempt to reduce the number of channels in the set of
+    colours."""
+    if planes >= 3:
+        def isgrey(c):
+            return c[0] == c[1] == c[2]
+        if min(map(isgrey, col)) == True:
+            # Every colour is grey.
+            col = set(map(lambda x: x[0::3], col))
+            planes -= 2
+    return col,planes
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+
+    argv = argv[1:]
+    if len(argv) > 0:
+        f = open(argv[0], 'rb')
+    else:
+        f = sys.stdin
+    return colours(sys.stdout, f)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pipcomposite b/build/pypng/pipcomposite
new file mode 100644
index 0000000..21dd283
--- /dev/null
+++ b/build/pypng/pipcomposite
@@ -0,0 +1,121 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipcomposite $
+# $Rev: 208 $
+# pipcomposite
+# Image alpha compositing.
+
+"""
+pipcomposite [--background #rrggbb] file.png
+
+Composite an image onto a background and output the result.  The
+background colour is specified with an HTML-style triple (3, 6, or 12
+hex digits), and defaults to black (#000).
+
+The output PNG has no alpha channel.
+
+It is valid for the input to have no alpha channel, but it doesn't
+make much sense: the output will equal the input.
+"""
+
+import sys
+
+def composite(out, inp, background):
+    import png
+
+    p = png.Reader(file=inp)
+    w,h,pixel,info = p.asRGBA()
+
+    outinfo = dict(info)
+    outinfo['alpha'] = False
+    outinfo['planes'] -= 1
+    outinfo['interlace'] = 0
+
+    # Convert to tuple and normalise to same range as source.
+    background = rgbhex(background)
+    maxval = float(2**info['bitdepth'] - 1)
+    background = map(lambda x: int(0.5 + x*maxval/65535.0),
+                     background)
+    # Repeat background so that it's a whole row of sample values.
+    background *= w
+
+    def iterrow():
+        for row in pixel:
+            # Remove alpha from row, then create a list with one alpha
+            # entry _per channel value_.
+            # Squirrel the alpha channel away (and normalise it).
+            t = map(lambda x: x/maxval, row[3::4])
+            row = list(row)
+            del row[3::4]
+            alpha = row[:]
+            for i in range(3):
+                alpha[i::3] = t
+            assert len(alpha) == len(row) == len(background)
+            yield map(lambda a,v,b: int(0.5 + a*v + (1.0-a)*b),
+                      alpha, row, background)
+
+    w = png.Writer(**outinfo)
+    w.write(out, iterrow())
+
+def rgbhex(s):
+    """Take an HTML style string of the form "#rrggbb" and return a
+    colour (R,G,B) triple.  Following the initial '#' there can be 3, 6,
+    or 12 digits (for 4-, 8- or 16- bits per channel).  In all cases the
+    values are expanded to a full 16-bit range, so the returned values
+    are all in range(65536).
+    """
+
+    assert s[0] == '#'
+    s = s[1:]
+    assert len(s) in (3,6,12)
+
+    # Create a target list of length 12, and expand the string s to make
+    # it length 12.
+    l = ['z']*12
+    if len(s) == 3:
+        for i in range(4):
+            l[i::4] = s
+    if len(s) == 6:
+        for i in range(2):
+            l[i::4] = s[i::2]
+            l[i+2::4] = s[i::2]
+    if len(s) == 12:
+        l[:] = s
+    s = ''.join(l)
+    return map(lambda x: int(x, 16), (s[:4], s[4:8], s[8:]))
+
+class Usage(Exception):
+    pass
+
+def main(argv=None):
+    import getopt
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+
+    argv = argv[1:]
+
+    try:
+        try:
+            opt,arg = getopt.getopt(argv, '',
+                                    ['background='])
+        except getopt.error, msg:
+            raise Usage(msg)
+        background = '#000'
+        for o,v in opt:
+            if o in ['--background']:
+                background = v
+    except Usage, err:
+        print >>sys.stderr, __doc__
+        print >>sys.stderr, str(err)
+        return 2
+
+    if len(arg) > 0:
+        f = open(arg[0], 'rb')
+    else:
+        f = sys.stdin
+    return composite(sys.stdout, f, background)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pipdither b/build/pypng/pipdither
new file mode 100644
index 0000000..c14c76c
--- /dev/null
+++ b/build/pypng/pipdither
@@ -0,0 +1,181 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipdither $
+# $Rev: 150 $
+
+# pipdither
+# Error Diffusing image dithering.
+# Now with serpentine scanning.
+
+# See http://www.efg2.com/Lab/Library/ImageProcessing/DHALF.TXT
+
+# http://www.python.org/doc/2.4.4/lib/module-bisect.html
+from bisect import bisect_left
+
+import png
+
+def dither(out, inp,
+  bitdepth=1, linear=False, defaultgamma=1.0, targetgamma=None,
+  cutoff=0.75):
+    """Dither the input PNG `inp` into an image with a smaller bit depth
+    and write the result image onto `out`.  `bitdepth` specifies the bit
+    depth of the new image.
+    
+    Normally the source image gamma is honoured (the image is
+    converted into a linear light space before being dithered), but
+    if the `linear` argument is true then the image is treated as
+    being linear already: no gamma conversion is done (this is
+    quicker, and if you don't care much about accuracy, it won't
+    matter much).
+    
+    Images with no gamma indication (no ``gAMA`` chunk) are normally
+    treated as linear (gamma = 1.0), but often it can be better
+    to assume a different gamma value: For example continuous tone
+    photographs intended for presentation on the web often carry
+    an implicit assumption of being encoded with a gamma of about
+    0.45 (because that's what you get if you just "blat the pixels"
+    onto a PC framebuffer), so ``defaultgamma=0.45`` might be a
+    good idea.  `defaultgamma` does not override a gamma value
+    specified in the file itself: It is only used when the file
+    does not specify a gamma.
+
+    If you (pointlessly) specify both `linear` and `defaultgamma`,
+    `linear` wins.
+
+    The gamma of the output image is, by default, the same as the input
+    image.  The `targetgamma` argument can be used to specify a
+    different gamma for the output image.  This effectively recodes the
+    image to a different gamma, dithering as we go.  The gamma specified
+    is the exponent used to encode the output file (and appears in the
+    output PNG's ``gAMA`` chunk); it is usually less than 1.
+
+    """
+
+    # Encoding is what happened when the PNG was made (and also what
+    # happens when we output the PNG).  Decoding is what we do to the
+    # source PNG in order to process it.
+
+    # The dithering algorithm is not completely general; it
+    # can only do bit depth reduction, not arbitrary palette changes.
+    import operator
+    maxval = 2**bitdepth - 1
+    r = png.Reader(file=inp)
+    # If image gamma is 1 or gamma is not present and we are assuming a
+    # value of 1, then it is faster to pass a maxval parameter to
+    # asFloat (the multiplications get combined).  With gamma, we have
+    # to have the pixel values from 0.0 to 1.0 (as long as we are doing
+    # gamma correction here).
+    # Slightly annoyingly, we don't know the image gamma until we've
+    # called asFloat().
+    _,_,pixels,info = r.asDirect()
+    planes = info['planes']
+    assert planes == 1
+    width = info['size'][0]
+    sourcemaxval = 2**info['bitdepth'] - 1
+    if linear:
+        gamma = 1
+    else:
+        gamma = info.get('gamma') or defaultgamma
+    # Convert gamma from encoding gamma to the required power for
+    # decoding.
+    decode = 1.0/gamma
+    # Build a lookup table for decoding; convert from pixel values to linear
+    # space:
+    sourcef = 1.0/sourcemaxval
+    incode = map(sourcef.__mul__, range(sourcemaxval+1))
+    if decode != 1.0:
+        incode = map(decode.__rpow__, incode)
+    # Could be different, later on.  targetdecode is the assumed gamma
+    # that is going to be used to decoding the target PNG.  It is the
+    # reciprocal of the exponent that we use to encode the target PNG.
+    # This is the value that we need to build our table that we use for
+    # converting from linear to target colour space.
+    if targetgamma is None:
+        targetdecode = decode
+    else:
+        targetdecode = 1.0/targetgamma
+    # The table we use for encoding (creating the target PNG), still
+    # maps from pixel value to linear space, but we use it inverted, by
+    # searching through it with bisect.
+    targetf = 1.0/maxval
+    outcode = map(targetf.__mul__, range(maxval+1))
+    if targetdecode != 1.0:
+        outcode = map(targetdecode.__rpow__, outcode)
+    # The table used for choosing output codes.  These values represent
+    # the cutoff points between two adjacent output codes.
+    choosecode = zip(outcode[1:], outcode)
+    p = cutoff
+    choosecode = map(lambda x: x[0]*p+x[1]*(1.0-p), choosecode)
+    def iterdither():
+        # Errors diffused downwards (into next row)
+        ed = [0.0]*width
+        flipped = False
+        for row in pixels:
+            row = map(incode.__getitem__, row)
+            row = map(operator.add, ed, row)
+            if flipped:
+                row = row[::-1]
+            targetrow = [0] * width
+            for i,v in enumerate(row):
+                # Clamp.  Necessary because previously added errors may take
+                # v out of range.
+                v = max(0.0, min(v, 1.0))
+                # `it` will be the index of the chosen target colour;
+                it = bisect_left(choosecode, v)
+                t = outcode[it]
+                targetrow[i] = it
+                # err is the error that needs distributing.
+                err = v - t
+                # Sierra "Filter Lite" distributes          * 2
+                # as per this diagram.                    1 1
+                ef = err/2.0
+                # :todo: consider making rows one wider at each end and
+                # removing "if"s
+                if i+1 < width:
+                    row[i+1] += ef
+                ef /= 2.0
+                ed[i] = ef
+                if i:
+                    ed[i-1] += ef
+            if flipped:
+                ed = ed[::-1]
+                targetrow = targetrow[::-1]
+            yield targetrow
+            flipped = not flipped
+    info['bitdepth'] = bitdepth
+    info['gamma'] = 1.0/targetdecode
+    w = png.Writer(**info)
+    w.write(out, iterdither())
+
+
+def main(argv=None):
+    # http://www.python.org/doc/2.4.4/lib/module-getopt.html
+    from getopt import getopt
+    import sys
+    if argv is None:
+        argv = sys.argv
+    opt,argv = getopt(argv[1:], 'b:c:g:lo:')
+    k = {}
+    for o,v in opt:
+        if o == '-b':
+            k['bitdepth'] = int(v)
+        if o == '-c':
+            k['cutoff'] = float(v)
+        if o == '-g':
+            k['defaultgamma'] = float(v)
+        if o == '-l':
+            k['linear'] = True
+        if o == '-o':
+            k['targetgamma'] = float(v)
+        if o == '-?':
+            print >>sys.stderr, "pipdither [-b bits] [-c cutoff] [-g assumed-gamma] [-l] [in.png]"
+
+    if len(argv) > 0:
+        f = open(argv[0], 'rb')
+    else:
+        f = sys.stdin
+
+    return dither(sys.stdout, f, **k)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/piprgb b/build/pypng/piprgb
new file mode 100644
index 0000000..fbe1082
--- /dev/null
+++ b/build/pypng/piprgb
@@ -0,0 +1,36 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/piprgb $
+# $Rev: 131 $
+# piprgb
+#
+# Convert input image to RGB or RGBA format.  Output will be colour type
+# 2 or 6, and will not have a tRNS chunk.
+
+import png
+
+def rgb(out, inp):
+    """Convert to RGB/RGBA."""
+
+    r = png.Reader(file=inp)
+    r.preamble()
+    if r.alpha or r.trns:
+        get = r.asRGBA
+    else:
+        get = r.asRGB
+    pixels,info = get()[2:4]
+    w = png.Writer(**info)
+    w.write(out, pixels)
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    if len(argv) > 1:
+        f = open(argv[1], 'rb')
+    else:
+        f = sys.stdin
+    return rgb(sys.stdout, f)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pipscalez b/build/pypng/pipscalez
new file mode 100644
index 0000000..c60762d
--- /dev/null
+++ b/build/pypng/pipscalez
@@ -0,0 +1,53 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipscalez $
+# $Rev: 131 $
+
+# pipscalez
+# Enlarge an image by an integer factor horizontally and vertically.
+
+def rescale(inp, out, xf, yf):
+    from array import array
+    import png
+
+    r = png.Reader(file=inp)
+    _,_,pixels,meta = r.asDirect()
+    typecode = 'BH'[meta['bitdepth'] > 8]
+    planes = meta['planes']
+    # We are going to use meta in the call to Writer, so expand the
+    # size.
+    x,y = meta['size']
+    x *= xf
+    y *= yf
+    meta['size'] = (x,y)
+    del x
+    del y
+    # Values per row, target row.
+    vpr = meta['size'][0] * planes
+    def iterscale():
+        for row in pixels:
+            bigrow = array(typecode, [0]*vpr)
+            row = array(typecode, row)
+            for c in range(planes):
+                channel = row[c::planes]
+                for i in range(xf):
+                    bigrow[i*planes+c::xf*planes] = channel
+            for _ in range(yf):
+                yield bigrow
+    w = png.Writer(**meta)
+    w.write(out, iterscale())
+
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    xf = int(argv[1])
+    if len(argv) > 2:
+        yf = int(argv[2])
+    else:
+        yf = xf
+    return rescale(sys.stdin, sys.stdout, xf, yf)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pipstack b/build/pypng/pipstack
new file mode 100644
index 0000000..5523670
--- /dev/null
+++ b/build/pypng/pipstack
@@ -0,0 +1,127 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipstack $
+# $Rev: 190 $
+
+# pipstack
+# Combine input PNG files into a multi-channel output PNG.
+
+"""
+pipstack file1.png [file2.png ...]
+
+pipstack can be used to combine 3 greyscale PNG files into a colour, RGB,
+PNG file.  In fact it is slightly more general than that.  The number of
+channels in the output PNG is equal to the sum of the numbers of
+channels in the input images.  It is an error if this sum exceeds 4 (the
+maximum number of channels in a PNG image is 4, for an RGBA image).  The
+output colour model corresponds to the number of channels: 1 -
+greyscale; 2 - greyscale+alpha; 3 - RGB; 4 - RGB+alpha.
+
+In this way it is possible to combine 3 greyscale PNG files into an RGB
+PNG (a common expected use) as well as more esoteric options: rgb.png +
+grey.png = rgba.png; grey.png + grey.png = greyalpha.png.
+
+Color Profile, Gamma, and so on.
+
+[This is not implemented yet]
+
+If an input has an ICC Profile (``iCCP`` chunk) then the output will
+have an ICC Profile, but only if it is possible to combine all the input
+ICC Profiles.  It is possible to combine all the input ICC Profiles
+only when: they all use the same Profile Connection Space; the PCS white
+point is the same (specified in the header; should always be D50);
+possibly some other things I haven't thought of yet.
+
+If some of the inputs have a ``gAMA`` chunk (specifying gamma) and
+an output ICC Profile is being generated, then the gamma information
+will be incorporated into the ICC Profile.
+
+When the output is an RGB colour type and the output ICC Profile is
+synthesized, it is necessary to supply colorant tags (``rXYZ`` and so
+on).  These are taken from ``sRGB``.
+
+If the input images have ``gAMA`` chunks and no input image has an ICC
+Profile then the output image will have a ``gAMA`` chunk, but only if
+all the ``gAMA`` chunks specify the same value.  Otherwise a warning
+will be emitted and no ``gAMA`` chunk.  It is possible to add or replace
+a ``gAMA`` chunk using the ``pipchunk`` tool.
+
+gAMA, pHYs, iCCP, sRGB, tIME, any other chunks.
+"""
+
+class Error(Exception):
+    pass
+
+def stack(out, inp):
+    """Stack the input PNG files into a single output PNG."""
+
+    from array import array
+    import itertools
+    # Local module
+    import png
+
+    if len(inp) < 1:
+        raise Error("Required input is missing.")
+
+    l = map(png.Reader, inp)
+    # Let data be a list of (pixel,info) pairs.
+    data = map(lambda p: p.asDirect()[2:], l)
+    totalchannels = sum(map(lambda x: x[1]['planes'], data))
+
+    if not (0 < totalchannels <= 4):
+        raise Error("Too many channels in input.")
+    alpha = totalchannels in (2,4)
+    greyscale = totalchannels in (1,2)
+    bitdepth = []
+    for b in map(lambda x: x[1]['bitdepth'], data):
+        try:
+            if b == int(b):
+                bitdepth.append(b)
+                continue
+        except (TypeError, ValueError):
+            pass
+        # Assume a tuple.
+        bitdepth += b
+    # Currently, fail unless all bitdepths equal.
+    if len(set(bitdepth)) > 1:
+        raise NotImplemented("Cannot cope when bitdepths differ - sorry!")
+    bitdepth = bitdepth[0]
+    arraytype = 'BH'[bitdepth > 8]
+    size = map(lambda x: x[1]['size'], data)
+    # Currently, fail unless all images same size.
+    if len(set(size)) > 1:
+        raise NotImplemented("Cannot cope when sizes differ - sorry!")
+    size = size[0]
+    # Values per row
+    vpr = totalchannels * size[0]
+    def iterstack():
+        # the izip call creates an iterator that yields the next row
+        # from all the input images combined into a tuple.
+        for irow in itertools.izip(*map(lambda x: x[0], data)):
+            row = array(arraytype, [0]*vpr)
+            # output channel
+            och = 0
+            for i,arow in enumerate(irow):
+                # ensure incoming row is an array
+                arow = array(arraytype, arow)
+                n = data[i][1]['planes']
+                for j in range(n):
+                    row[och::totalchannels] = arow[j::n]
+                    och += 1
+            yield row
+    w = png.Writer(size[0], size[1],
+      greyscale=greyscale, alpha=alpha, bitdepth=bitdepth)
+    w.write(out, iterstack())
+
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    argv = argv[1:]
+    arg = argv[:]
+    return stack(sys.stdout, arg)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pipwindow b/build/pypng/pipwindow
new file mode 100644
index 0000000..2f8c7a2
--- /dev/null
+++ b/build/pypng/pipwindow
@@ -0,0 +1,67 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pipwindow $
+# $Rev: 173 $
+
+# pipwindow
+# Tool to crop/expand an image to a rectangular window.  Come the
+# revolution this tool will allow the image and the window to be placed
+# arbitrarily (in particular the window can be bigger than the picture
+# and/or overlap it only partially) and the image can be OpenGL style
+# border/repeat effects (repeat, mirrored repeat, clamp, fixed
+# background colour, background colour from source file).  For now it
+# only acts as crop.  The window must be no greater than the image in
+# both x and y.
+
+def window(tl, br, inp, out):
+    """Place a window onto the image and cut-out the resulting
+    rectangle.  The window is an axis aligned rectangle opposite corners
+    at *tl* and *br* (each being an (x,y) pair). *inp* specifies the
+    input file which should be a PNG image.
+    """
+
+    import png
+
+    r = png.Reader(file=inp)
+    x,y,pixels,meta = r.asDirect()
+    if not (0 <= tl[0] < br[0] <= x):
+        raise NotImplementedError()
+    if not (0 <= tl[1] < br[1] <= y):
+        raise NotImplementedError()
+    # Compute left and right bounds for each row
+    l = tl[0] * meta['planes']
+    r = br[0] * meta['planes']
+    def itercrop():
+        """An iterator to perform the crop."""
+
+        for i,row in enumerate(pixels):
+            if i < tl[1]:
+                continue
+            if i >= br[1]:
+                # Same as "raise StopIteration"
+                return
+            yield row[l:r]
+    meta['size'] = (br[0]-tl[0], br[1]-tl[1])
+    w = png.Writer(**meta)
+    w.write(out, itercrop())
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    argv = argv[1:]
+
+    tl = (0,0)
+    br = tuple(map(int, argv[:2]))
+    if len(argv) >= 4:
+        tl = br
+        br = tuple(map(int, argv[2:4]))
+    if len(argv) in (2, 4):
+        f = sys.stdin
+    else:
+        f = open(argv[-1], 'rb')
+
+    return window(tl, br, f, sys.stdout)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/plan9topng.py b/build/pypng/plan9topng.py
new file mode 100644
index 0000000..4600b4c
--- /dev/null
+++ b/build/pypng/plan9topng.py
@@ -0,0 +1,293 @@
+#!/usr/bin/env python
+# $Rev: 184 $
+# $URL: http://pypng.googlecode.com/svn/trunk/code/plan9topng.py $
+
+# Imported from //depot/prj/plan9topam/master/code/plan9topam.py#4 on
+# 2009-06-15.
+
+"""Command line tool to convert from Plan 9 image format to PNG format.
+
+Plan 9 image format description:
+http://plan9.bell-labs.com/magic/man2html/6/image
+"""
+
+# http://www.python.org/doc/2.3.5/lib/module-itertools.html
+import itertools
+# http://www.python.org/doc/2.3.5/lib/module-re.html
+import re
+# http://www.python.org/doc/2.3.5/lib/module-sys.html
+import sys
+
+def block(s, n):
+    # See http://www.python.org/doc/2.6.2/library/functions.html#zip
+    return zip(*[iter(s)]*n)
+
+def convert(f, output=sys.stdout) :
+  """Convert Plan 9 file to PNG format.  Works with either uncompressed
+  or compressed files.
+  """
+
+  r = f.read(11)
+  if r == 'compressed\n' :
+    png(output, *decompress(f))
+  else :
+    png(output, *glue(f, r))
+
+
+def glue(f, r) :
+  """Return (metadata, stream) pair where `r` is the initial portion of
+  the metadata that has already been read from the stream `f`.
+  """
+
+  r = r + f.read(60-len(r))
+  return (r, f)
+
+def meta(r) :
+  """Convert 60 character string `r`, the metadata from an image file.
+  Returns a 5-tuple (*chan*,*minx*,*miny*,*limx*,*limy*).  5-tuples may
+  settle into lists in transit.
+  
+  As per http://plan9.bell-labs.com/magic/man2html/6/image the metadata
+  comprises 5 words separated by blanks.  As it happens each word starts
+  at an index that is a multiple of 12, but this routine does not care
+  about that."""
+
+  r = r.split()
+  # :todo: raise FormatError
+  assert len(r) == 5
+  r = [r[0]] + map(int, r[1:])
+  return r
+
+def bitdepthof(pixel) :
+    """Return the bitdepth for a Plan9 pixel format string."""
+
+    maxd = 0
+    for c in re.findall(r'[a-z]\d*', pixel) :
+        if c[0] != 'x':
+            maxd = max(maxd, int(c[1:]))
+    return maxd
+
+def maxvalof(pixel):
+  """Return the netpbm MAXVAL for a Plan9 pixel format string."""
+
+  bitdepth = bitdepthof(pixel)
+  return (2**bitdepth)-1
+
+def pixmeta(metadata, f) :
+    """Convert (uncompressed) Plan 9 image file to pair of (*metadata*,
+    *pixels*).  This is intended to be used by PyPNG format.  *metadata*
+    is the metadata returned in a dictionary, *pixels* is an iterator that
+    yields each row in boxed row flat pixel format.
+
+    `f`, the input file, should be cued up to the start of the image data.
+    """
+
+    chan,minx,miny,limx,limy = metadata
+    rows = limy - miny
+    width = limx - minx
+    nchans = len(re.findall('[a-wyz]', chan))
+    alpha = 'a' in chan
+    # Iverson's convention for the win!
+    ncolour = nchans - alpha
+    greyscale = ncolour == 1
+    bitdepth = bitdepthof(chan)
+    maxval = 2**bitdepth - 1
+    # PNG style metadata
+    meta=dict(size=(width,rows), bitdepth=bitdepthof(chan),
+      greyscale=greyscale, alpha=alpha, planes=nchans)
+
+    return itertools.imap(lambda x: itertools.chain(*x),
+      block(unpack(f, rows, width, chan, maxval), width)), meta
+
+def png(out, metadata, f):
+    """Convert to PNG format.  `metadata` should be a Plan9 5-tuple; `f`
+    the input file (see :meth:`pixmeta`).
+    """
+
+    import png
+
+    pixels,meta = pixmeta(metadata, f)
+    p = png.Writer(**meta)
+    p.write(out, pixels)
+
+def spam():
+  """Not really spam, but old PAM code, which is in limbo."""
+
+  if nchans == 3 or nchans == 1 :
+    # PGM (P5) or PPM (P6) format.
+    output.write('P%d\n%d %d %d\n' % (5+(nchans==3), width, rows, maxval))
+  else :
+    # PAM format.
+    output.write("""P7
+WIDTH %d
+HEIGHT %d
+DEPTH %d
+MAXVAL %d
+""" % (width, rows, nchans, maxval))
+
+def unpack(f, rows, width, pixel, maxval) :
+  """Unpack `f` into pixels.  Assumes the pixel format is such that the depth
+  is either a multiple or a divisor of 8.
+  `f` is assumed to be an iterator that returns blocks of input such
+  that each block contains a whole number of pixels.  An iterator is
+  returned that yields each pixel as an n-tuple.  `pixel` describes the
+  pixel format using the Plan9 syntax ("k8", "r8g8b8", and so on).
+  """
+
+  def mask(w) :
+    """An integer, to be used as a mask, with bottom `w` bits set to 1."""
+
+    return (1 << w)-1
+
+  def deblock(f, depth, width) :
+    """A "packer" used to convert multiple bytes into single pixels.
+    `depth` is the pixel depth in bits (>= 8), `width` is the row width in
+    pixels.
+    """
+
+    w = depth // 8
+    i = 0
+    for block in f :
+      for i in range(len(block)//w) :
+        p = block[w*i:w*(i+1)]
+        i += w
+        # Convert p to little-endian integer, x
+        x = 0
+        s = 1 # scale
+        for j in p :
+          x += s * ord(j)
+          s <<= 8
+        yield x
+
+  def bitfunge(f, depth, width) :
+    """A "packer" used to convert single bytes into multiple pixels.
+    Depth is the pixel depth (< 8), width is the row width in pixels.
+    """
+
+    for block in f :
+      col = 0
+      for i in block :
+        x = ord(i)
+        for j in range(8/depth) :
+          yield x >> (8 - depth)
+          col += 1
+          if col == width :
+            # A row-end forces a new byte even if we haven't consumed
+            # all of the current byte.  Effectively rows are bit-padded
+            # to make a whole number of bytes.
+            col = 0
+            break
+          x <<= depth
+
+  # number of bits in each channel
+  chan = map(int, re.findall(r'\d+', pixel))
+  # type of each channel
+  type = re.findall('[a-z]', pixel)
+
+  depth = sum(chan)
+
+  # According to the value of depth pick a "packer" that either gathers
+  # multiple bytes into a single pixel (for depth >= 8) or split bytes
+  # into several pixels (for depth < 8)
+  if depth >= 8 :
+    # 
+    assert depth % 8 == 0
+    packer = deblock
+  else :
+    assert 8 % depth == 0
+    packer = bitfunge
+
+  for x in packer(f, depth, width) :
+    # x is the pixel as an unsigned integer
+    o = []
+    # This is a bit yucky.  Extract each channel from the _most_
+    # significant part of x.
+    for j in range(len(chan)) :
+      v = (x >> (depth - chan[j])) & mask(chan[j])
+      x <<= chan[j]
+      if type[j] != 'x' :
+        # scale to maxval
+        v = v * float(maxval) / mask(chan[j])
+        v = int(v+0.5)
+        o.append(v)
+    yield o
+
+
+def decompress(f) :
+  """Decompress a Plan 9 image file.  Assumes f is already cued past the
+  initial 'compressed\n' string.
+  """
+
+  r = meta(f.read(60))
+  return r, decomprest(f, r[4])
+
+
+def decomprest(f, rows) :
+  """Iterator that decompresses the rest of a file once the metadata
+  have been consumed."""
+
+  row = 0
+  while row < rows :
+    row,o = deblock(f)
+    yield o
+
+
+def deblock(f) :
+  """Decompress a single block from a compressed Plan 9 image file.
+  Each block starts with 2 decimal strings of 12 bytes each.  Yields a
+  sequence of (row, data) pairs where row is the total number of rows
+  processed according to the file format and data is the decompressed
+  data for a set of rows."""
+
+  row = int(f.read(12))
+  size = int(f.read(12))
+  if not (0 <= size <= 6000) :
+    raise 'block has invalid size; not a Plan 9 image file?'
+
+  # Since each block is at most 6000 bytes we may as well read it all in
+  # one go.
+  d = f.read(size)
+  i = 0
+  o = []
+
+  while i < size :
+    x = ord(d[i])
+    i += 1
+    if x & 0x80 :
+      x = (x & 0x7f) + 1
+      lit = d[i:i+x]
+      i += x
+      o.extend(lit)
+      continue
+    # x's high-order bit is 0
+    l = (x >> 2) + 3
+    # Offset is made from bottom 2 bits of x and all 8 bits of next
+    # byte.  http://plan9.bell-labs.com/magic/man2html/6/image doesn't
+    # say whether x's 2 bits are most signiificant or least significant.
+    # But it is clear from inspecting a random file,
+    # http://plan9.bell-labs.com/sources/plan9/sys/games/lib/sokoban/images/cargo.bit
+    # that x's 2 bit are most significant.
+    # 
+    offset = (x & 3) << 8
+    offset |= ord(d[i])
+    i += 1
+    # Note: complement operator neatly maps (0 to 1023) to (-1 to
+    # -1024).  Adding len(o) gives a (non-negative) offset into o from
+    # which to start indexing.
+    offset = ~offset + len(o)
+    if offset < 0 :
+      raise 'byte offset indexes off the begininning of the output buffer; not a Plan 9 image file?'
+    for j in range(l) :
+      o.append(o[offset+j])
+  return row,''.join(o)
+
+def main(argv=None) :
+  if argv is None :
+    argv = sys.argv
+  if len(sys.argv) <= 1 :
+    return convert(sys.stdin)
+  else :
+    return convert(open(argv[1], 'rb'))
+
+if __name__ == '__main__' :
+  sys.exit(main())
diff --git a/build/pypng/pngchunk b/build/pypng/pngchunk
new file mode 100644
index 0000000..b00e4b1
--- /dev/null
+++ b/build/pypng/pngchunk
@@ -0,0 +1,172 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pngchunk $
+# $Rev: 156 $
+# pngchunk
+# Chunk editing/extraction tool.
+
+import struct
+import warnings
+
+# Local module.
+import png
+
+"""
+pngchunk [--gamma g] [--iccprofile file] [--sigbit b] [-c cHNK!] [-c cHNK:foo] [-c cHNK<file]
+
+The ``-c`` option is used to add or remove chunks.  A chunk is specified
+by its 4 byte chunk type.  If this is followed by a ``!`` then that
+chunk is removed from the PNG file.  If the chunk type is followed by a
+``:`` then the chunk is replaced with the contents of the rest of the
+argument (this is probably only useful if the content is mostly ASCII,
+otherwise it's a pain to quote the contents, otherwise see...).  A ``<``
+can be used to take the contents of the chunk from the named file.
+"""
+
+
+def chunk(out, inp, l):
+    """Process the input PNG file to the output, chunk by chunk.  Chunks
+    can be inserted, removed, replaced, or sometimes edited.  Generally, 
+    chunks are not inspected, so pixel data (in the ``IDAT`` chunks)
+    cannot be modified.  `l` should be a list of (*chunktype*,
+    *content*) pairs.  *chunktype* is usually the type of the PNG chunk,
+    specified as a 4-byte Python string, and *content* is the chunk's
+    content, also as a string; if *content* is ``None`` then *all*
+    chunks of that type will be removed.
+
+    This function *knows* about certain chunk types and will
+    automatically convert from Python friendly representations to
+    string-of-bytes.
+
+    chunktype
+    'gamma'     'gAMA'  float
+    'sigbit'    'sBIT'  int, or tuple of length 1,2 or 3
+
+    Note that the length of the strings used to identify *friendly*
+    chunk types is greater than 4, hence they cannot be confused with
+    canonical chunk types.
+
+    Chunk types, if specified using the 4-byte syntax, need not be
+    official PNG chunks at all.  Non-standard chunks can be created.
+    """
+
+    def canonical(p):
+        """Take a pair (*chunktype*, *content*), and return canonical
+        representation (*chunktype*, *content*) where `chunktype` is the
+        4-byte PNG chunk type and `content` is a string.
+        """
+
+        t,v = p
+        if len(t) == 4:
+            return t,v
+        if t == 'gamma':
+            t = 'gAMA'
+            v = int(round(1e5*v))
+            v = struct.pack('>I', v)
+        elif t == 'sigbit':
+            t = 'sBIT'
+            try:
+                v[0]
+            except TypeError:
+                v = (v,)
+            v = struct.pack('%dB' % len(v), *v)
+        elif t == 'iccprofile':
+            t = 'iCCP'
+            # http://www.w3.org/TR/PNG/#11iCCP
+            v = 'a color profile\x00\x00' + v.encode('zip')
+        else:
+            warnings.warn('Unknown chunk type %r' % t)
+        return t[:4],v
+
+    l = map(canonical, l)
+    # Some chunks automagically replace ones that are present in the
+    # source PNG.  There can only be one of each of these chunk types.
+    # Create a 'replace' dictionary to record these chunks.
+    add = []
+    delete = set()
+    replacing = set(['gAMA', 'sBIT', 'PLTE', 'tRNS', 'sPLT', 'IHDR'])
+    replace = dict()
+    for t,v in l:
+        if v is None:
+            delete.add(t)
+        elif t in replacing:
+            replace[t] = v
+        else:
+            add.append((t,v))
+    del l
+    r = png.Reader(file=inp)
+    chunks = r.chunks()
+    def iterchunks():
+        for t,v in chunks:
+            if t in delete:
+                continue
+            if t in replace:
+                yield t,replace[t]
+                del replace[t]
+                continue
+            if t == 'IDAT' and replace:
+                # Insert into the output any chunks that are on the
+                # replace list.  We haven't output them yet, because we
+                # didn't see an original chunk of the same type to
+                # replace.  Thus the "replace" is actually an "insert".
+                for u,w in replace.items():
+                    yield u,w
+                    del replace[u]
+            if t == 'IDAT' and add:
+                for item in add:
+                    yield item
+                del add[:]
+            yield t,v
+    return png.write_chunks(out, iterchunks())
+
+class Usage(Exception):
+    pass
+
+def main(argv=None):
+    import getopt
+    import re
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+
+    argv = argv[1:]
+
+    try:
+        try:
+            opt,arg = getopt.getopt(argv, 'c:',
+                                    ['gamma=', 'iccprofile=', 'sigbit='])
+        except getopt.error, msg:
+            raise Usage(msg)
+        k = []
+        for o,v in opt:
+            if o in ['--gamma']:
+                k.append(('gamma', float(v)))
+            if o in ['--sigbit']:
+                k.append(('sigbit', int(v)))
+            if o in ['--iccprofile']:
+                k.append(('iccprofile', open(v, 'rb').read()))
+            if o in ['-c']:
+                type = v[:4]
+                if not re.match('[a-zA-Z]{4}', type):
+                    raise Usage('Chunk type must consist of 4 letters.')
+                if v[4] == '!':
+                    k.append((type, None))
+                if v[4] == ':':
+                    k.append((type, v[5:]))
+                if v[4] == '<':
+                    k.append((type, open(v[5:], 'rb').read()))
+    except Usage, err:
+        print >>sys.stderr, (
+          "usage: pngchunk [--gamma d.dd] [--sigbit b] [-c cHNK! | -c cHNK:text-string]")
+        print >>sys.stderr, err.message
+        return 2
+
+    if len(arg) > 0:
+        f = open(arg[0], 'rb')
+    else:
+        f = sys.stdin
+    return chunk(sys.stdout, f, k)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pnghist b/build/pypng/pnghist
new file mode 100644
index 0000000..4fbbd0a
--- /dev/null
+++ b/build/pypng/pnghist
@@ -0,0 +1,79 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pnghist $
+# $Rev: 153 $
+# PNG Histogram
+# Only really works on grayscale images.
+
+from array import array
+import getopt
+
+import png
+
+def decidemax(level):
+    """Given an array of levels, decide the maximum value to use for the
+    histogram.  This is normally chosen to be a bit bigger than the 99th
+    percentile, but if the 100th percentile is not much more (within a
+    factor of 2) then the 100th percentile is chosen.
+    """
+
+    truemax = max(level)
+    sl = level[:]
+    sl.sort(reverse=True)
+    i99 = int(round(len(level)*0.01))
+    if truemax <= 2*sl[i99]:
+        return truemax
+    return 1.05*sl[i99]
+
+def hist(out, inp, verbose=None):
+    """Open the PNG file `inp` and generate a histogram."""
+
+    r = png.Reader(file=inp)
+    x,y,pixels,info = r.asDirect()
+    bitdepth = info['bitdepth']
+    level = [0]*2**bitdepth
+    for row in pixels:
+        for v in row:
+            level[v] += 1
+    maxlevel = decidemax(level)
+
+    h = 100
+    outbitdepth = 8
+    outmaxval = 2**outbitdepth - 1
+    def genrow():
+        for y in range(h):
+            y = h-y-1
+            # :todo: vary typecode according to outbitdepth
+            row = array('B', [0]*len(level))
+            fl = y*maxlevel/float(h)
+            ce = (y+1)*maxlevel/float(h)
+            for x in range(len(row)):
+                if level[x] <= fl:
+                    # Relies on row being initialised to all 0
+                    continue
+                if level[x] >= ce:
+                    row[x] = outmaxval
+                    continue
+                frac = (level[x] - fl)/(ce - fl)
+                row[x] = int(round(outmaxval*frac))
+            yield row
+    w = png.Writer(len(level), h, gamma=1.0,
+      greyscale=True, alpha=False, bitdepth=outbitdepth)
+    w.write(out, genrow())
+    if verbose: print >>verbose, level
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    argv = argv[1:]
+    opt,arg = getopt.getopt(argv, '')
+
+    if len(arg) < 1:
+        f = sys.stdin
+    else:
+        f = open(arg[0])
+    hist(sys.stdout, f)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/pnglsch b/build/pypng/pnglsch
new file mode 100644
index 0000000..d10d238
--- /dev/null
+++ b/build/pypng/pnglsch
@@ -0,0 +1,31 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/pnglsch $
+# $Rev: 107 $
+# pnglsch
+# PNG List Chunks
+
+import png
+
+def list(out, inp):
+    r = png.Reader(file=inp)
+    for t,v in r.chunks():
+        add = ''
+        if len(v) <= 28:
+            add = ' ' + v.encode('hex')
+        print >>out, "%s %10d%s" % (t, len(v), add)
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    arg = argv[1:]
+
+    if len(arg) > 0:
+        f = open(arg[0], 'rb')
+    else:
+        f = sys.stdin
+    return list(sys.stdout, f)
+
+if __name__ == '__main__':
+    main()
diff --git a/build/pypng/texttopng b/build/pypng/texttopng
new file mode 100644
index 0000000..ab0c690
--- /dev/null
+++ b/build/pypng/texttopng
@@ -0,0 +1,151 @@
+#!/usr/bin/env python
+# $URL: http://pypng.googlecode.com/svn/trunk/code/texttopng $
+# $Rev: 132 $
+# Script to renders text as a PNG image.
+
+from array import array
+import itertools
+
+font = {
+  32: '0000000000000000',
+  33: '0010101010001000',
+  34: '0028280000000000',
+  35: '0000287c287c2800',
+  36: '00103c5038147810',
+  37: '0000644810244c00',
+  38: '0020502054483400',
+  39: '0010100000000000',
+  40: '0008101010101008',
+  41: '0020101010101020',
+  42: '0010543838541000',
+  43: '000010107c101000',
+  44: '0000000000301020',
+  45: '000000007c000000',
+  46: '0000000000303000',
+  47: '0000040810204000',
+  48: '0038445454443800',
+  49: '0008180808080800',
+  50: '0038043840407c00',
+  51: '003c041804043800',
+  52: '00081828487c0800',
+  53: '0078407804047800',
+  54: '0038407844443800',
+  55: '007c040810101000',
+  56: '0038443844443800',
+  57: '0038443c04040400',
+  58: '0000303000303000',
+  59: '0000303000301020',
+  60: '0004081020100804',
+  61: '0000007c007c0000',
+  62: '0040201008102040',
+  63: '0038440810001000',
+  64: '00384c545c403800',
+  65: '0038447c44444400',
+  66: '0078447844447800',
+  67: '0038444040443800',
+  68: '0070484444487000',
+  69: '007c407840407c00',
+  70: '007c407840404000',
+  71: '003844405c443c00',
+  72: '0044447c44444400',
+  73: '0038101010103800',
+  74: '003c040404443800',
+  75: '0044487048444400',
+  76: '0040404040407c00',
+  77: '006c545444444400',
+  78: '004464544c444400',
+  79: '0038444444443800',
+  80: '0078447840404000',
+  81: '0038444444443c02',
+  82: '0078447844444400',
+  83: '0038403804047800',
+  84: '007c101010101000',
+  85: '0044444444443c00',
+  86: '0044444444281000',
+  87: '0044445454543800',
+  88: '0042241818244200',
+  89: '0044443810101000',
+  90: '007c081020407c00',
+  91: '0038202020202038',
+  92: '0000402010080400',
+  93: '0038080808080838',
+  94: '0010284400000000',
+  95: '000000000000fe00',
+  96: '0040200000000000',
+  97: '000038043c443c00',
+  98: '0040784444447800',
+  99: '0000384040403800',
+  100: '00043c4444443c00',
+  101: '000038447c403c00',
+  102: '0018203820202000',
+  103: '00003c44443c0438',
+  104: '0040784444444400',
+  105: '0010003010101000',
+  106: '0010003010101020',
+  107: '0040404870484400',
+  108: '0030101010101000',
+  109: '0000385454444400',
+  110: '0000784444444400',
+  111: '0000384444443800',
+  112: '0000784444784040',
+  113: '00003c44443c0406',
+  114: '00001c2020202000',
+  115: '00003c4038047800',
+  116: '0020203820201800',
+  117: '0000444444443c00',
+  118: '0000444444281000',
+  119: '0000444454543800',
+  120: '0000442810284400',
+  121: '00004444443c0438',
+  122: '00007c0810207c00',
+  123: '0018202060202018',
+  124: '0010101000101010',
+  125: '003008080c080830',
+  126: '0020540800000000',
+}
+
+def char(i):
+    """Get image data for the character `i` (a one character string).
+    Returned as a list of rows.  Each row is a tuple containing the
+    packed pixels.
+    """
+
+    i = ord(i)
+    if i not in font:
+        return [(0,)]*8
+    return map(lambda row: (ord(row),), font[i].decode('hex'))
+
+def texttoraster(m):
+    """Convert string *m* to a raster image (by rendering it using the
+    font in *font*).  A triple of (*width*, *height*, *pixels*) is
+    returned; *pixels* is in boxed row packed pixel format.
+    """
+
+    # Assumes monospaced font.
+    x = 8*len(m)
+    y = 8
+    return x,y,itertools.imap(lambda row: itertools.chain(*row),
+                              zip(*map(char, m)))
+
+
+def render(message, out):
+    import png
+
+    x,y,pixels = texttoraster(message)
+    w = png.Writer(x, y, greyscale=True, bitdepth=1)
+    w.write_packed(out, pixels)
+    out.flush()
+
+def main(argv=None):
+    import sys
+
+    if argv is None:
+        argv = sys.argv
+    if len(argv) > 1:
+        message = argv[1]
+    else:
+        message = sys.stdin.read()
+    render(message, sys.stdout)
+
+if __name__ == '__main__':
+    main()