gluemncbig

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""Usage: gluemncbig [-q] [--verbose] [--many] [-v <vars>] -o <outfile> <files>

 -v <vars>  comma-separated list of variable names or glob patterns
 --many     many tiles: assemble only along x in memory; less efficient
            on some filesystems, but opens fewer files simultaneously and
            uses less memory
 -q         suppress progress messages
 --verbose  report variables

All files must have the same variables.
Each variable (or 1 record of it) must fit in memory.
With --many, only a row of tiles along x must fit in memory.

Examples:

gluemncbig -o ptr.nc mnc_*/ptr_tave.*.nc
gluemncbig -o BIO.nc -v 'BIO_*' mnc_*/ptr_tave.*.nc
"""

# NetCDF reader/writer module modified from pupynere,
# https://bitbucket.org/robertodealmeida/pupynere/
# to allow delayed reading/writing of variable data.
# MIT license

moduledoc = u"""
NetCDF reader/writer module.

This module is used to read and create NetCDF files. NetCDF files are
accessed through the `netcdf_file` object. Data written to and from NetCDF
files are contained in `netcdf_variable` objects. Attributes are given
as member variables of the `netcdf_file` and `netcdf_variable` objects.

Notes
-----
NetCDF files are a self-describing binary data format. The file contains
metadata that describes the dimensions and variables in the file. More
details about NetCDF files can be found `here
<http://www.unidata.ucar.edu/software/netcdf/docs/netcdf.html>`_. There
are three main sections to a NetCDF data structure:

1. Dimensions
2. Variables
3. Attributes

The dimensions section records the name and length of each dimension used
by the variables. The variables would then indicate which dimensions it
uses and any attributes such as data units, along with containing the data
values for the variable. It is good practice to include a
variable that is the same name as a dimension to provide the values for
that axes. Lastly, the attributes section would contain additional
information such as the name of the file creator or the instrument used to
collect the data.

When writing data to a NetCDF file, there is often the need to indicate the
'record dimension'. A record dimension is the unbounded dimension for a
variable. For example, a temperature variable may have dimensions of
latitude, longitude and time. If one wants to add more temperature data to
the NetCDF file as time progresses, then the temperature variable should
have the time dimension flagged as the record dimension.

This module implements the Scientific.IO.NetCDF API to read and create
NetCDF files. The same API is also used in the PyNIO and pynetcdf
modules, allowing these modules to be used interchangeably when working
with NetCDF files. The major advantage of this module over other
modules is that it doesn't require the code to be linked to the NetCDF
libraries.

In addition, the NetCDF file header contains the position of the data in
the file, so access can be done in an efficient manner without loading
unnecessary data into memory. It uses the ``mmap`` module to create
Numpy arrays mapped to the data on disk, for the same purpose.

Examples
--------

To create a NetCDF file:

    >>> f = netcdf_file('simple.nc', 'w')
    >>> f.history = 'Created for a test'
    >>> f.location = u'北京'
    >>> f.createDimension('time', 10)
    >>> time = f.createVariable('time', 'i', ('time',))
    >>> time[:] = range(10)
    >>> time.units = u'µs since 2008-01-01'
    >>> f.close()

Note the assignment of ``range(10)`` to ``time[:]``.  Exposing the slice
of the time variable allows for the data to be set in the object, rather
than letting ``range(10)`` overwrite the ``time`` variable.

To read the NetCDF file we just created:

    >>> f = netcdf_file('simple.nc', 'r')
    >>> print f.history
    Created for a test
    >>> print f.location
    北京
    >>> time = f.variables['time']
    >>> print time.units
    µs since 2008-01-01
    >>> print time.shape
    (10,)
    >>> print time[-1]
    9
    >>> f.close()

TODO:
 * properly implement ``_FillValue``.
 * implement Jeff Whitaker's patch for masked variables.
 * fix character variables.
 * implement PAGESIZE for Python 2.6?
"""

__all__ = ['netcdf_file']


from operator import mul
try:
    from collections import OrderedDict
except ImportError:
    OrderedDict = dict
from mmap import mmap, ACCESS_READ

import numpy as np
from numpy import fromstring, ndarray, dtype, empty, array, asarray
from numpy import little_endian as LITTLE_ENDIAN

import sys

if sys.version_info[0] >= 3:
    def asbytes(s):
        if isinstance(s, bytes):
            return s
        return str(s).encode('latin1')

    def asstr(s):
        if isinstance(s, bytes):
            return s.decode('latin1')
        return str(s)
else:
    asbytes = str
    asstr = str


ABSENT       = asbytes('\x00\x00\x00\x00\x00\x00\x00\x00')
ZERO         = asbytes('\x00\x00\x00\x00')
NC_BYTE      = asbytes('\x00\x00\x00\x01')
NC_CHAR      = asbytes('\x00\x00\x00\x02')
NC_SHORT     = asbytes('\x00\x00\x00\x03')
NC_INT       = asbytes('\x00\x00\x00\x04')
NC_FLOAT     = asbytes('\x00\x00\x00\x05')
NC_DOUBLE    = asbytes('\x00\x00\x00\x06')
NC_DIMENSION = asbytes('\x00\x00\x00\n')
NC_VARIABLE  = asbytes('\x00\x00\x00\x0b')
NC_ATTRIBUTE = asbytes('\x00\x00\x00\x0c')


TYPEMAP = { NC_BYTE:   dtype(np.byte),
            NC_CHAR:   dtype('c'),
            NC_SHORT:  dtype(np.int16).newbyteorder('>'),
            NC_INT:    dtype(np.int32).newbyteorder('>'),
            NC_FLOAT:  dtype(np.float32).newbyteorder('>'),
            NC_DOUBLE: dtype(np.float64).newbyteorder('>'),
            }

REVERSE = { dtype(np.byte):    NC_BYTE,
            dtype('c'):        NC_CHAR,
            dtype(np.int16):   NC_SHORT,
            dtype(np.int32):   NC_INT,
            dtype(np.int64):   NC_INT,  # will be converted to int32
            dtype(np.float32): NC_FLOAT,
            dtype(np.float64): NC_DOUBLE,
            }


class NetCDFError(Exception):
    pass


class unmapped_array(object):
    def __init__(self, shape, dtype_):
        self.shape = shape
        self.dtype = dtype(dtype_)

    @property
    def itemsize(self):
        return self.dtype.itemsize

    @property
    def size(self):
        return reduce(mul, self.shape, 1)

    @property
    def nbytes(self):
        return self.size * self.itemsize

    def __len__(self):
        return self.shape[0]

    def __getitem__(self, indx):
        raise NetCDFError('netcdf_file: delay is True, use read_var or read_recvar to read variable data')

    def __setitem__(self, indx, val):
        raise NetCDFError('netcdf_file: delay is True, use write_var or write_recvar to assign variable data')


class netcdf_file(object):
    """
    A file object for NetCDF data.

    A `netcdf_file` object has two standard attributes: `dimensions` and
    `variables`. The values of both are dictionaries, mapping dimension
    names to their associated lengths and variable names to variables,
    respectively. Application programs should never modify these
    dictionaries.

    All other attributes correspond to global attributes defined in the
    NetCDF file. Global file attributes are created by assigning to an
    attribute of the `netcdf_file` object.

    If delay is True, variable data is only read/written on demand.
    In this case, if mode="w", write_metadata() needs to be called after
    all dimensions, variables and attributes have been defined, but before
    any variable data is written.

    Parameters
    ----------
    filename : string or file-like
        string -> filename
    mode : {'r', 'w'}, optional
        read-write mode, default is 'r'
    mmap : None or bool, optional
        Whether to mmap `filename` when reading.  Default is True
        when `filename` is a file name, False when `filename` is a
        file-like object
    delay : bool, optional
        Whether to delay reading of variable data.  Default is False.
        This is an alternative to mmap for more efficient reading.
    version : {1, 2}, optional
        version of netcdf to read / write, where 1 means *Classic
        format* and 2 means *64-bit offset format*.  Default is 1.  See
        `here <http://www.unidata.ucar.edu/software/netcdf/docs/netcdf/Which-Format.html>`_
        for more info.

    """
    def __init__(self, filename, mode='r', mmap=None, version=1, delay=False):
        """Initialize netcdf_file from fileobj (str or file-like)."""
        if delay:
            if mmap is None:
                mmap = False
            else:
                raise ValueError('Cannot delay variables for mmap')
        if hasattr(filename, 'seek'):  # file-like
            self.fp = filename
            self.filename = 'None'
            if mmap is None:
                mmap = False
            elif mmap and not hasattr(filename, 'fileno'):
                raise ValueError('Cannot use file object for mmap')
        else:  # string?
            self.filename = filename
            self.fp = open(self.filename, '%sb' % mode)
            if mmap is None:
                mmap = True
        self.use_mmap = mmap
        self.version_byte = version
        self.delay = delay

        if not mode in 'rw':
            raise ValueError("Mode must be either 'r' or 'w'.")
        self.mode = mode

        self.dimensions = OrderedDict()
        self.variables = OrderedDict()

        self._dims = []
        self._recs = 0
        self._recsize = 0

        self._mapped = False
        self._begins = OrderedDict()

        self._attributes = OrderedDict()

        if mode == 'r':
            self._read()

    def __setattr__(self, attr, value):
        # Store user defined attributes in a separate dict,
        # so we can save them to file later.
        try:
            self._attributes[attr] = value
        except AttributeError:
            pass
        self.__dict__[attr] = value

    def close(self):
        """Closes the NetCDF file."""
        try:
            closed = self.fp.closed
        except AttributeError:
            pass
        else:
            if not closed:
                try:
                    self.flush()
                finally:
                    self.fp.close()
    __del__ = close

    def createDimension(self, name, length):
        """
        Adds a dimension to the Dimension section of the NetCDF data structure.

        Note that this function merely adds a new dimension that the variables can
        reference.  The values for the dimension, if desired, should be added as
        a variable using `createVariable`, referring to this dimension.

        Parameters
        ----------
        name : str
            Name of the dimension (Eg, 'lat' or 'time').
        length : int
            Length of the dimension.

        See Also
        --------
        createVariable

        """
        self.dimensions[name] = length
        self._dims.append(name)

    def createVariable(self, name, type, dimensions):
        """
        Create an empty variable for the `netcdf_file` object, specifying its data
        type and the dimensions it uses.

        Parameters
        ----------
        name : str
            Name of the new variable.
        type : dtype or str
            Data type of the variable.
        dimensions : sequence of str
            List of the dimension names used by the variable, in the desired order.

        Returns
        -------
        variable : netcdf_variable
            The newly created ``netcdf_variable`` object.
            This object has also been added to the `netcdf_file` object as well.

        See Also
        --------
        createDimension

        Notes
        -----
        Any dimensions to be used by the variable should already exist in the
        NetCDF data structure or should be created by `createDimension` prior to
        creating the NetCDF variable.

        """
        shape = tuple([self.dimensions[dim] for dim in dimensions])
        shape_ = tuple([dim or 0 for dim in shape])  # replace None with 0 for numpy

        if not isinstance(type, dtype): type = dtype(type)
        if type.newbyteorder('=') not in REVERSE:
            raise ValueError("NetCDF 3 does not support type %s" % type)

        if self.delay:
            data = unmapped_array(shape_, type)
        else:
            data = empty(shape_, type)
        self.variables[name] = netcdf_variable(data, type, shape, dimensions)
        return self.variables[name]

    def flush(self):
        """
        Perform a sync-to-disk flush if the `netcdf_file` object is in write mode.

        See Also
        --------
        sync : Identical function

        """
        if getattr(self, 'mode', None) is 'w':
            if self.delay:
                if not self._mapped:
                    self._map()
                self.update_numrecs(self._recs)
            else:
                self._write()
    sync = flush

    def write_metadata(self):
        '''This needs to be called before assigning any data to variables!'''
        if self.delay:
            self._map()
        else:
            raise UserWarning('write_metadata is void unless delay is True')

    def _map(self):
        self.fp.seek(0)
        self.fp.write(asbytes('CDF'))
        self.fp.write(array(self.version_byte, '>b').tostring())

        # Write headers
        self._write_numrecs()
        self._write_dim_array()
        self._write_gatt_array()
        self._map_var_array()
        self._mapped = True

    def _map_var_array(self):
        if self.variables:
            self.fp.write(NC_VARIABLE)
            self._pack_int(len(self.variables))

            # Separate record variables from non-record ones, keep order
            nonrec_vars = [ k for k,v in self.variables.items() if not v.isrec ]
            rec_vars = [ k for k,v in self.variables.items() if v.isrec ]

            # Set the metadata for all variables.
            for name in nonrec_vars + rec_vars:
                self._map_var_metadata(name)

            # Now that we have the metadata, we know the vsize of
            # each variable, so we can calculate their position in the file

            pos0 = pos = self.fp.tell()

            # set file pointers for all variables.
            for name in nonrec_vars:
                var = self.variables[name]
                # Set begin in file header.
                self.fp.seek(var._begin)
                self._pack_begin(pos)
                self._begins[name] = pos
                pos += var._vsize

            recstart = pos

            for name in rec_vars:
                var = self.variables[name]
                # Set begin in file header.
                self.fp.seek(var._begin)
                self._pack_begin(pos)
                self._begins[name] = pos
                pos += var._vsize

            self.__dict__['_recsize'] = pos - recstart

            # first var
            self.fp.seek(pos0)
        else:
            self.fp.write(ABSENT)

    def _map_var_metadata(self, name):
        var = self.variables[name]

        self._pack_string(name)
        self._pack_int(len(var.dimensions))
        for dimname in var.dimensions:
            dimid = self._dims.index(dimname)
            self._pack_int(dimid)

        self._write_att_array(var._attributes)

        nc_type = REVERSE[var.dtype.newbyteorder('=')]
        self.fp.write(asbytes(nc_type))

        if not var.isrec:
            vsize = var.data.size * var.data.itemsize
            vsize += -vsize % 4
        else:  # record variable
            if 1:  #var.data.shape[0]:
                size = reduce(mul, var.data.shape[1:], 1)
                vsize = size * var.data.itemsize
            else:
                vsize = 0
            rec_vars = len([var for var in self.variables.values()
                    if var.isrec])
            if rec_vars > 1:
                vsize += -vsize % 4
        self.variables[name].__dict__['_vsize'] = vsize
        self._pack_int(vsize)

        # Pack a bogus begin, and set the real value later.
        self.variables[name].__dict__['_begin'] = self.fp.tell()
        self._pack_begin(0)

    @property
    def numrecs(self):
        return self._recs

    @property
    def attributes(self):
        return self._attributes

    @property
    def begins(self):
        return [(name,pos,self.variables[name].isrec) for name,pos in self._begins.items()]

    def write_var(self, name, val, j0=None, iY=-2):
        if not self.delay:
            raise NetCDFError('netcdf_file: delay is False, need to assign to variables')
        if not self._mapped:
            raise NetCDFError('netcdf_file: need to call write_metadata first')
        pos = self._begins[name]
        var = self.variables[name]
        count = var.data.size * var.data.itemsize
        if j0 is not None:
            if iY == 0:
                idxs = [()]
            else:
                idxs = np.ndindex(*var.data.shape[:iY])
            for idx in idxs:
                ii = idx + (j0,) + len(var.data.shape[iY+1:])*(0,)
                offset = np.ravel_multi_index(ii, var.data.shape)*var.data.itemsize
                end = offset + val[idx].size*var.data.itemsize
                if end > count:
                    raise NetCDFError('array too large: {} > {}'.format(end, count))
                self.fp.seek(pos+offset)
                np.asanyarray(val[idx], var.data.dtype.newbyteorder('>')).tofile(self.fp)
        else:
            end = val.size*var.data.itemsize
            if end != count:
                raise NetCDFError('array too large: {} > {}'.format(end, count))
            self.fp.seek(pos)
            np.asanyarray(val, var.data.dtype.newbyteorder('>')).tofile(self.fp)
            # pad
            self.fp.write(asbytes('0') * (var._vsize - count))

    def write_recvar(self, name, rec, val, j0=None, iY=-2):
        if not self.delay:
            raise NetCDFError('netcdf_file: delay is False, need to assign to variables')
        if not self._mapped:
            raise NetCDFError('netcdf_file: need to call write_metadata first')
        pos = self._begins[name] + rec*self._recsize
        var = self.variables[name]
        count = reduce(mul, var.data.shape[1:], 1) * var.data.itemsize
        if j0 is not None:
            if iY == 0:
                idxs = [()]
            else:
                idxs = np.ndindex(*var.data.shape[1:iY])
            for idx in idxs:
                ii = idx + (j0,) + len(var.data.shape[iY+1:])*(0,)
                offset = np.ravel_multi_index(ii, var.data.shape[1:])*var.data.itemsize
                end = offset + val[idx].size*var.data.itemsize
                if end > count:
                    raise NetCDFError('array too large: {} > {}'.format(end, count))
                self.fp.seek(pos+offset)
                np.asanyarray(val[idx], var.data.dtype.newbyteorder('>')).tofile(self.fp)
        else:
            end = val.size*var.data.itemsize
            if end != count:
                raise ValueError('netcdf_file.write_recvar: array too large')
            self.fp.seek(pos)
            np.asanyarray(val, var.data.dtype.newbyteorder('>')).tofile(self.fp)
            # pad
            self.fp.write(asbytes('0') * (var._vsize - count))
        if rec >= self._recs:
            self.__dict__['_recs'] = rec + 1

    def _write(self):
        self.fp.seek(0)
        self.fp.write(asbytes('CDF'))
        self.fp.write(array(self.version_byte, '>b').tostring())

        # Write headers and data.
        self._write_numrecs()
        self._write_dim_array()
        self._write_gatt_array()
        self._write_var_array()

    def _write_numrecs(self):
        # Get highest record count from all record variables.
        for var in self.variables.values():
            if var.isrec and len(var.data) > self._recs:
                self.__dict__['_recs'] = len(var.data)
        self.__dict__['_numrecs_begin'] = self.fp.tell()
        self._pack_int(self._recs)

    def update_numrecs(self, numrecs):
        self.__dict__['_recs'] = numrecs
        self.fp.seek(self._numrecs_begin)
        self._pack_int(numrecs)

    def _write_dim_array(self):
        if self.dimensions:
            self.fp.write(NC_DIMENSION)
            self._pack_int(len(self.dimensions))
            for name in self._dims:
                self._pack_string(name)
                length = self.dimensions[name]
                self._pack_int(length or 0)  # replace None with 0 for record dimension
        else:
            self.fp.write(ABSENT)

    def _write_gatt_array(self):
        self._write_att_array(self._attributes)

    def _write_att_array(self, attributes):
        if attributes:
            self.fp.write(NC_ATTRIBUTE)
            self._pack_int(len(attributes))
            for name, values in attributes.items():
                self._pack_string(name)
                self._write_values(values)
        else:
            self.fp.write(ABSENT)

    def _write_var_array(self):
        if self.variables:
            self.fp.write(NC_VARIABLE)
            self._pack_int(len(self.variables))

#            # Sort variables non-recs first, then recs. We use a DSU
#            # since some people use pupynere with Python 2.3.x.
#            deco = [ (v._shape and not v.isrec, k) for (k, v) in self.variables.items() ]
#            deco.sort()
#            variables = [ k for (unused, k) in deco ][::-1]
            # Separate record variables from non-record ones, keep order
            nonrec_vars = [ k for k,v in self.variables.items() if not v.isrec ]
            rec_vars = [ k for k,v in self.variables.items() if v.isrec ]
            variables = nonrec_vars + rec_vars

            # Set the metadata for all variables.
            for name in variables:
                self._write_var_metadata(name)
            # Now that we have the metadata, we know the vsize of
            # each record variable, so we can calculate recsize.
            self.__dict__['_recsize'] = sum([
                    var._vsize for var in self.variables.values()
                    if var.isrec])
            # Set the data for all variables.
            for name in variables:
                self._write_var_data(name)
        else:
            self.fp.write(ABSENT)

    def _write_var_metadata(self, name):
        var = self.variables[name]

        self._pack_string(name)
        self._pack_int(len(var.dimensions))
        for dimname in var.dimensions:
            dimid = self._dims.index(dimname)
            self._pack_int(dimid)

        self._write_att_array(var._attributes)

        nc_type = REVERSE[var.dtype.newbyteorder('=')]
        self.fp.write(asbytes(nc_type))

        if not var.isrec:
            vsize = var.data.size * var.data.itemsize
            vsize += -vsize % 4
        else:  # record variable
            try:
                vsize = var.data[0].size * var.data.itemsize
            except IndexError:
                vsize = 0
            rec_vars = len([var for var in self.variables.values()
                    if var.isrec])
            if rec_vars > 1:
                vsize += -vsize % 4
        self.variables[name].__dict__['_vsize'] = vsize
        self._pack_int(vsize)

        # Pack a bogus begin, and set the real value later.
        self.variables[name].__dict__['_begin'] = self.fp.tell()
        self._pack_begin(0)

    def _write_var_data(self, name):
        var = self.variables[name]

        # Set begin in file header.
        the_beguine = self.fp.tell()
        self.fp.seek(var._begin)
        self._pack_begin(the_beguine)
        self.fp.seek(the_beguine)

        # Write data.
        if not var.isrec:
            if (var.data.dtype.byteorder == '<' or 
                    (var.data.dtype.byteorder == '=' and LITTLE_ENDIAN)):
                var.data = var.data.byteswap()
            self.fp.write(var.data.tostring())
            count = var.data.size * var.data.itemsize
            self.fp.write(asbytes('0') * (var._vsize - count))
        else:  # record variable
            # Handle rec vars with shape[0] < nrecs.
            if self._recs > len(var.data):
                shape = (self._recs,) + var.data.shape[1:]
                var.data.resize(shape)

            pos0 = pos = self.fp.tell()
            for rec in var.data:
                if (rec.dtype.byteorder == '<' or 
                        (rec.dtype.byteorder == '=' and LITTLE_ENDIAN)):
                    rec = rec.byteswap()
                self.fp.write(rec.tostring())
                # Padding
                count = rec.size * rec.itemsize
                self.fp.write(asbytes('0') * (var._vsize - count))
                pos += self._recsize
                self.fp.seek(pos)
            self.fp.seek(pos0 + var._vsize)

    def _write_values(self, values):
        if hasattr(values, 'dtype'):
            nc_type = REVERSE[values.dtype.newbyteorder('=')]
        else:
            types = [
                    (int, NC_INT),
                    (long, NC_INT),
                    (float, NC_FLOAT),
                    (basestring, NC_CHAR),
                    ]
            try:
                sample = values[0]
            except (IndexError, TypeError):
                sample = values
            if isinstance(sample, unicode):
                if not isinstance(values, unicode):
                    raise ValueError("NetCDF requires that text be encoded as UTF-8")
                values = values.encode('utf-8')
            for class_, nc_type in types:
                if isinstance(sample, class_): break

        values = asarray(values, TYPEMAP[nc_type])

        self.fp.write(asbytes(nc_type))

        if values.dtype.char == 'S':
            nelems = values.itemsize
        else:
            nelems = values.size
        self._pack_int(nelems)

        if not values.shape and (values.dtype.byteorder == '<' or
                (values.dtype.byteorder == '=' and LITTLE_ENDIAN)):
            values = values.byteswap()
        self.fp.write(values.tostring())
        count = values.size * values.itemsize
        self.fp.write(asbytes('0') * (-count % 4))  # pad

    def _read(self):
        # Check magic bytes and version
        magic = self.fp.read(3)
        if not magic == asbytes('CDF'):
            raise TypeError("Error: %s is not a valid NetCDF 3 file" %
                            self.filename)
        self.__dict__['version_byte'] = fromstring(self.fp.read(1), '>b')[0]

        # Read file headers and set data.
        self._read_numrecs()
        self._read_dim_array()
        self._read_gatt_array()
        self._read_var_array()

    def _read_numrecs(self):
        self.__dict__['_recs'] = self._unpack_int()

    def _read_dim_array(self):
        header = self.fp.read(4)
        if not header in [ZERO, NC_DIMENSION]:
            raise ValueError("Unexpected header.")
        count = self._unpack_int()

        for dim in range(count):
            name = asstr(self._unpack_string())
            length = self._unpack_int() or None  # None for record dimension
            self.dimensions[name] = length
            self._dims.append(name)  # preserve order

    def _read_gatt_array(self):
        for k, v in self._read_att_array().items():
            self.__setattr__(k, v)

    def _read_att_array(self):
        header = self.fp.read(4)
        if not header in [ZERO, NC_ATTRIBUTE]:
            raise ValueError("Unexpected header.")
        count = self._unpack_int()

        attributes = OrderedDict()
        for attr in range(count):
            name = asstr(self._unpack_string())
            attributes[name] = self._read_values()
        return attributes

    def _read_var_array(self):
        header = self.fp.read(4)
        if not header in [ZERO, NC_VARIABLE]:
            raise ValueError("Unexpected header.")

        begin = 0
        dtypes = {'names': [], 'formats': []}
        rec_vars = []
        count = self._unpack_int()
        for var in range(count):
            name, dimensions, shape, attributes, type, begin_, vsize = self._read_var()
            # http://www.unidata.ucar.edu/software/netcdf/docs/netcdf.html
            # Note that vsize is the product of the dimension lengths
            # (omitting the record dimension) and the number of bytes
            # per value (determined from the type), increased to the
            # next multiple of 4, for each variable. If a record
            # variable, this is the amount of space per record. The
            # netCDF "record size" is calculated as the sum of the
            # vsize's of all the record variables.
            #
            # The vsize field is actually redundant, because its value
            # may be computed from other information in the header. The
            # 32-bit vsize field is not large enough to contain the size
            # of variables that require more than 2^32 - 4 bytes, so
            # 2^32 - 1 is used in the vsize field for such variables.
            if shape and shape[0] is None:  # record variable
                rec_vars.append(name)
                # The netCDF "record size" is calculated as the sum of
                # the vsize's of all the record variables.
                self.__dict__['_recsize'] += vsize
                if begin == 0: begin = begin_
                dtypes['names'].append(name)
                dtypes['formats'].append(str(shape[1:]) + '>' + type.char)

                # Handle padding with a virtual variable.
                if type.char in 'bch':
                    actual_size = reduce(mul, (1,) + shape[1:]) * type.itemsize
                    padding = -actual_size % 4
                    if padding:
                        dtypes['names'].append('_padding_%d' % var)
                        dtypes['formats'].append('(%d,)>b' % padding)

                # Data will be set later.
                if self.delay:
                    self._begins[name] = begin_
                    data = unmapped_array((self._recs,)+shape[1:], type)
                else:
                    data = None
            else:  # not a record variable
                # Calculate size to avoid problems with vsize (above)
                a_size = reduce(mul, shape, 1) * type.itemsize
                pos = self.fp.tell()
                if self.use_mmap:
                    mm = mmap(self.fp.fileno(), begin_+a_size, access=ACCESS_READ)
                    data = ndarray.__new__(ndarray, shape, dtype=type,
                            buffer=mm, offset=begin_, order=0)
                elif self.delay:
                    self._begins[name] = begin_
                    data = unmapped_array(shape, type)
                else:
                    self.fp.seek(begin_)
                    data = fromstring(self.fp.read(a_size), type)
                    data.shape = shape
                self.fp.seek(pos)

            # Add variable.
            self.variables[name] = netcdf_variable(data, type, shape, dimensions, attributes)

        if rec_vars and not self.delay:
            dtypes['formats'] = [f.replace('()', '').replace(' ', '') for f in dtypes['formats']]
            # Remove padding when only one record variable.
            if len(rec_vars) == 1:
                dtypes['names'] = dtypes['names'][:1]
                dtypes['formats'] = dtypes['formats'][:1]

            # Build rec array.
            pos = self.fp.tell()
            if self.use_mmap:
                mm = mmap(self.fp.fileno(), begin+self._recs*self._recsize, access=ACCESS_READ)
                rec_array = ndarray.__new__(ndarray, (self._recs,), dtype=dtypes,
                        buffer=mm, offset=begin, order=0)
            else:
                self.fp.seek(begin)
                rec_array = fromstring(self.fp.read(self._recs*self._recsize), dtype=dtypes)
                rec_array.shape = (self._recs,)
            self.fp.seek(pos)

            for var in rec_vars:
                self.variables[var].__dict__['data'] = rec_array[var]

    def read_var(self, name):
        var = self.variables[name]
        pos = self._begins[name]
        self.fp.seek(pos)
        data = fromstring(self.fp.read(var.data.nbytes), dtype=var.data.dtype)
        data.shape = var.data.shape
        return data

    def read_recvar(self, name, rec):
        var = self.variables[name]
        pos = self._begins[name] + rec*self._recsize
        self.fp.seek(pos)
        count = reduce(mul, var.data.shape[1:], 1) * var.data.itemsize
        data = fromstring(self.fp.read(count), dtype=var.data.dtype)
        data.shape = var.data.shape[1:]
        return data

    def _read_var(self):
        name = asstr(self._unpack_string())
        dimensions = []
        shape = []
        dims = self._unpack_int()

        for i in range(dims):
            dimid = self._unpack_int()
            dimname = self._dims[dimid]
            dimensions.append(dimname)
            dim = self.dimensions[dimname]
            shape.append(dim)
        dimensions = tuple(dimensions)
        shape = tuple(shape)

        attributes = self._read_att_array()
        nc_type = self.fp.read(4)
        vsize = self._unpack_int()
        begin = [self._unpack_int, self._unpack_int64][self.version_byte-1]()
        type = TYPEMAP[nc_type]

        return name, dimensions, shape, attributes, type, begin, vsize

    def _read_values(self):
        nc_type = self.fp.read(4)
        n = self._unpack_int()

        type = TYPEMAP[nc_type]

        count = n*type.itemsize
        values = self.fp.read(int(count))
        self.fp.read(-count % 4)  # read padding

        if type.char is not 'c':
            values = fromstring(values, type)
            if values.shape == (1,): values = values[0]
        else:
            ## text values are encoded via UTF-8, per NetCDF standard
            values = values.rstrip(asbytes('\x00')).decode('utf-8', 'replace')
        return values

    def _pack_begin(self, begin):
        if self.version_byte == 1:
            self._pack_int(begin)
        elif self.version_byte == 2:
            self._pack_int64(begin)

    def _pack_int(self, value):
        self.fp.write(array(value, '>i').tostring())
    _pack_int32 = _pack_int

    def _unpack_int(self):
        return int(fromstring(self.fp.read(4), '>i')[0])
    _unpack_int32 = _unpack_int

    def _pack_int64(self, value):
        self.fp.write(array(value, '>q').tostring())

    def _unpack_int64(self):
        return fromstring(self.fp.read(8), '>q')[0]

    def _pack_string(self, s):
        count = len(s)
        self._pack_int(count)
        self.fp.write(asbytes(s))
        self.fp.write(asbytes('0') * (-count % 4))  # pad

    def _unpack_string(self):
        count = self._unpack_int()
        s = self.fp.read(count).rstrip(asbytes('\x00'))
        self.fp.read(-count % 4)  # read padding
        return s


class netcdf_variable(object):
    """
    A data object for the `netcdf` module.

    `netcdf_variable` objects are constructed by calling the method
    `netcdf_file.createVariable` on the `netcdf_file` object. `netcdf_variable`
    objects behave much like array objects defined in numpy, except that their
    data resides in a file. Data is read by indexing and written by assigning
    to an indexed subset; the entire array can be accessed by the index ``[:]``
    or (for scalars) by using the methods `getValue` and `assignValue`.
    `netcdf_variable` objects also have attribute `shape` with the same meaning
    as for arrays, but the shape cannot be modified. There is another read-only
    attribute `dimensions`, whose value is the tuple of dimension names.

    All other attributes correspond to variable attributes defined in
    the NetCDF file. Variable attributes are created by assigning to an
    attribute of the `netcdf_variable` object.

    Parameters
    ----------
    data : array_like
        The data array that holds the values for the variable.
        Typically, this is initialized as empty, but with the proper shape.
    type: numpy dtype
        Desired data-type for the data array.
    shape : sequence of ints
        The shape of the array.  This should match the lengths of the
        variable's dimensions.
    dimensions : sequence of strings
        The names of the dimensions used by the variable.  Must be in the
        same order of the dimension lengths given by `shape`.
    attributes : dict, optional
        Attribute values (any type) keyed by string names.  These attributes
        become attributes for the netcdf_variable object.


    Attributes
    ----------
    dimensions : list of str
        List of names of dimensions used by the variable object.
    isrec, shape
        Properties

    See also
    --------
    isrec, shape

    """
    def __init__(self, data, type, shape, dimensions, attributes=None):
        self.data = data
        self.dtype = type
        self._shape = shape
        self.dimensions = dimensions

        self._attributes = attributes or OrderedDict()
        for k, v in self._attributes.items():
            self.__dict__[k] = v

    def __setattr__(self, attr, value):
        # Store user defined attributes in a separate dict,
        # so we can save them to file later.
        try:
            self._attributes[attr] = value
        except AttributeError:
            pass
        self.__dict__[attr] = value

    @property
    def attributes(self):
        return self._attributes

    def isrec(self):
        """Returns whether the variable has a record dimension or not.

        A record dimension is a dimension along which additional data could be
        easily appended in the netcdf data structure without much rewriting of
        the data file. This attribute is a read-only property of the
        `netcdf_variable`.

        """
        return self.data.shape and not self._shape[0]
    isrec = property(isrec)

    def shape(self):
        """Returns the shape tuple of the data variable.

        This is a read-only attribute and can not be modified in the
        same manner of other numpy arrays.
        """
        return self.data.shape
    shape = property(shape)

    def getValue(self):
        """
        Retrieve a scalar value from a `netcdf_variable` of length one.

        Raises
        ------
        ValueError
            If the netcdf variable is an array of length greater than one,
            this exception will be raised.

        """
        return self.data.item()

    def assignValue(self, value):
        """
        Assign a scalar value to a `netcdf_variable` of length one.

        Parameters
        ----------
        value : scalar
            Scalar value (of compatible type) to assign to a length-one netcdf
            variable. This value will be written to file.

        Raises
        ------
        ValueError
            If the input is not a scalar, or if the destination is not a length-one
            netcdf variable.

        """
        if not self.data.flags.writeable:
            # Work-around for a bug in NumPy.  Calling itemset() on a read-only
            # memory-mapped array causes a seg. fault.
            # See NumPy ticket #1622, and SciPy ticket #1202.
            # This check for `writeable` can be removed when the oldest version
            # of numpy still supported by scipy contains the fix for #1622.
            raise RuntimeError("variable is not writeable")

        self.data.itemset(value)

    def typecode(self):
        """
        Return the typecode of the variable.

        Returns
        -------
        typecode : char
            The character typecode of the variable (eg, 'i' for int).

        """
        return self.dtype.char

    def itemsize(self):
        """
        Return the itemsize of the variable.

        Returns
        -------
        itemsize : int
            The element size of the variable (eg, 8 for float64).

        """
        return self.dtype.itemsize

    def __getitem__(self, index):
        return self.data[index]

    def __setitem__(self, index, data):
        # Expand data for record vars?
        if self.isrec:
            if isinstance(index, tuple):
                rec_index = index[0]
            else:
                rec_index = index
            if isinstance(rec_index, slice):
                recs = (rec_index.start or 0) + len(data)
            else:
                recs = rec_index + 1
            if recs > len(self.data):
                shape = (recs,) + self._shape[1:]
                self.data.resize(shape)
        self.data[index] = data


NetCDFFile = netcdf_file
NetCDFVariable = netcdf_variable

# End of NetCDF reader/writer module


if __name__ == '__main__':
    import sys
    import os
    import errno
    import re
    import glob
    import fnmatch
    from getopt import gnu_getopt as getopt
    try:
        from collections import OrderedDict
    except ImportError:
        OrderedDict = dict
    import numpy as np

    tilepatt = re.compile(r'\.t([0-9]{3}[0-9]*)\.nc$')
    iterpatt = re.compile(r'(\.[0-9]{10})$')

    # parse command-line arguments
    try:
        optlist,args = getopt(sys.argv[1:], 'qo:v:', ['many', 'verbose'])
        opts = dict(optlist)
        assert '-o' in opts
        fnames = args
        assert len(fnames) > 0
    except (ValueError, AssertionError):
        sys.exit(__doc__)

    outname = opts.get('-o')
    progress = '-q' not in opts
    verbose = '--verbose' in opts
    manytiles = '--many' in opts
    tname = 'T'

    # turn into list of compiled regular expressions
    varpatt = opts.get('-v', '').split(',')
    varpatt = [ re.compile(fnmatch.translate(patt.strip())) for patt in varpatt ]

    readopts = dict(delay=True)
    writeopts = dict(delay=True)

    if len(fnames) == 1 and sum(s in fnames[0] for s in '*?[]'):
        globpatt = fnames[0]
        fnames = glob.glob(globpatt)
        if len(fnames) == 0:
            raise IOError('No files matching {0}', globpatt)

    fnames.sort()

    # Get list of iterations
    itfiles = {}
    # files without iteration number
    files0 = {}
    for fname in fnames:
        tn = tilepatt.search(fname).group(1)
        base = tilepatt.sub('', fname)
        m = iterpatt.search(base)
        if m:
            it = m.group(1)
            itfiles.setdefault(it, {})[tn] = fname
        else:
            files0[tn] = fname

    files0 = [v for k,v in sorted(files0.items())]

    its = itfiles.keys()
    its.sort()
    for it in its:
        itfiles[it] = [v for k,v in sorted(itfiles[it].items())]

    filess = [ itfiles[it] for it in its ]
    if len(files0) and len(filess):
        sys.exit('Both files with and without iteration number present.  Aborting.')
    elif len(filess):
        files0 = filess[0]
    else:
        filess = [files0]

    if verbose:
        print 'Files to be read:'
        for files in filess:
            print files[0], '... (%d)' % (len(files))

    nc = netcdf_file(files0[0], 'r', **readopts)
    gatt = OrderedDict(nc.attributes)
    #gatt['tile_number'] = 1
    #gatt['bi'] = 1
    #gatt['bj'] = 1
    del gatt['tile_number']
    del gatt['bi']
    del gatt['bj']
    for k in gatt.keys():
        if k.startswith('exch2_'):
            del gatt[k]

    sNx = gatt['sNx']
    sNy = gatt['sNy']
    Nx = gatt['Nx']
    Ny = gatt['Ny']
    ntx = gatt['nSx']*gatt['nPx']
    nty = gatt['nSy']*gatt['nPy']
    ntiles = ntx*nty

    for it in its:
        if len(itfiles[it]) != ntiles:
            raise ValueError('Error: found %d tiles for iteration %s, need %d' % (
                             len(itfiles[it]), it[1:], ntiles))

    Xslice = []
    Yslice = []
    for tn in range(ntx*nty):
        bj,bi = divmod(tn, ntx)
        ie = sNx*(bi+1-ntx) or None
        je = sNy*(bj+1-nty) or None
        Xslice.append(slice(sNx*bi, ie))
        Yslice.append(slice(sNy*bj, je))

    dims = OrderedDict()
    for k,n in nc.dimensions.items():
        if k[0] == 'X':
            n += Nx - sNx
        elif k[0] == 'Y':
            n += Ny - sNy
        dims[k] = n

    print 'Tiled dimensions:', ' '.join([k for k in nc.dimensions if k[0] in 'XY'])

    havetime = tname in dims
    if havetime:
        assert dims[tname] is None
        nrec = nc.numrecs
        print 'Record dimension:', tname
    else:
        assert len(its) <= 1
        nrec = None

    if verbose:
        print 'Variables:'

    varprops = OrderedDict()
    for name,var in nc.variables.items():
        if name in dims or sum(patt.search(name) is not None for patt in varpatt) or len(varpatt) == 0:
            varprops[name] = {}
            varprops[name]['dtype'] = var.data.dtype
            varprops[name]['dimensions'] = var.dimensions
            varprops[name]['ncattrs'] = OrderedDict(var.attributes)
            iX = None
            iY = None
            for i,dim in enumerate(var.dimensions):
                if dim[0] == 'X':
                    iX = i
                elif dim[0] == 'Y':
                    iY = i
            varprops[name]['iX'] = iX
            varprops[name]['iY'] = iY
            dimstrs = len(var.dimensions)*[':']
            if tname in var.dimensions: dimstrs[list(var.dimensions).index(tname)] = tname
            if iX is not None: dimstrs[iX] = var.dimensions[iX]
            if iY is not None: dimstrs[iY] = var.dimensions[iY]
            if verbose: print '%s %s(%s)' % (var.typecode(), name, ','.join(dimstrs))

    ######################################################################
    # create global netcdf file
    ncout = netcdf_file(outname, 'w', **writeopts)

    try:
        # global attributes
        for name,att in gatt.items():
            setattr(ncout, name, att)

        # create dimensions
        for name,n in dims.items():
            ncout.createDimension(name, n)

        # create variables with attributes
        vars = {}
        for name,var in varprops.items():
            dtype_ = np.dtype(var['dtype']).newbyteorder('>')
        #    if verbose: print 'Creating variable', name, dtype_, var['dimensions']
            vars[name] = ncout.createVariable(name, dtype_, var['dimensions'])
            for attname,att in var['ncattrs'].items():
                setattr(vars[name], attname, att)

        ncout.write_metadata()

        if manytiles:
            # open only files along x-slice of tiles simultaneously
            # will have to jump around output file a bit...

            # easier this way; have to reopen many files anyway
            nc.close()

            nyslice = sNy

            # sort tiles into x-slices
            iterslices = []
            for fnames in filess:
                myslicefiles = [[] for _ in range(nty)]
                for fname in fnames:
                    m = tilepatt.search(fname)
                    if m is None:
                        sys.exit('Could not extract tile number from file name: {0}'.format(fname))
                    tn = int(m.group(1)) - 1
                    bj,bi = divmod(tn, ntx)
                    myslicefiles[bj].append(fname)
                iterslices.append(myslicefiles)

            irec = 0
            for tileslices in iterslices:
                for bj in range(nty):
                    if len(tileslices[bj]) != ntx:
                        raise ValueError('found %d tiles for bj = %d, need %d' % (
                                         len(tileslices[bj]), bj+1, ntx))
                    # open files
                    ncs = []
                    for fname in tileslices[bj]:
                        try:
                            nc = netcdf_file(fname, 'r', **readopts)
                        except IOError as e:
                            if e.errno == errno.EMFILE:
                                sys.exit('ERROR: Too many open files.  Ask developer to improve script ;)')
                            raise
                        ncs.append(nc)

                    if irec == 0:
                        # assemble non-record variable data
                        if bj == 0 and progress and not verbose:
                            sys.stderr.write('Writing non-record variables\n')
                        indstrings = {}
                        for name,pos,isrec in ncout.begins:
                            if not isrec:
                                if verbose: print name
                                prop = varprops[name]
                                vardims = prop['dimensions']
                                var = vars[name]
                                indx = len(vardims)*[slice(None)]
                                iX = prop['iX']
                                iY = prop['iY']
                                shape = list(var.shape)
                                if iY is not None:
                                    # this is needed for V and vorticity point fields that have Ny+1
                                    shape[iY] = shape[iY] - Ny + nyslice
                                    j0 = bj*nyslice
                                else:
                                    j0 = None
                                data = np.empty(shape, var.data.dtype.newbyteorder('>'))
                                for nc in ncs:
                                    tn = nc.tile_number - 1
                                    if iX is not None: indx[iX] = Xslice[tn]
                                    data[tuple(indx)] = nc.read_var(name)

                                ncout.write_var(name, data, j0, iY)
                                del data
                            else:  # isrec
                                if not havetime:
                                    raise ValueError('record variables, but no time')

                    # assemble record variable data
                    if havetime:
                        # any of the ncs
                        nrec = nc.numrecs
                        if bj == 0 and progress and not verbose:
                            sys.stderr.write('Writing {0} records: '.format(nrec))
                        for irecin in range(nrec):
                            if bj == 0 and progress and not verbose:
                                sys.stderr.write('.')
                            for name,pos,isrec in ncout.begins:
                                if isrec:
                                    if verbose: print irec+irecin, name
                                    prop = varprops[name]
                                    vardims = prop['dimensions'][1:]
                                    var = vars[name]
                                    indx = len(vardims)*[slice(None)]
                                    iX = prop['iX']
                                    iY = prop['iY']
                                    shape = list(var.data.shape)
                                    if iY is not None:
                                        shape[iY] = shape[iY] - Ny + nyslice
                                        j0 = bj*nyslice
                                    else:
                                        j0 = None
                                    data = np.empty(shape[1:], var.data.dtype.newbyteorder('>'))
                                    for nc in ncs:
                                        tn = nc.tile_number - 1
                                        if iX is not None: indx[iX-1] = Xslice[tn]
                                        data[tuple(indx)] = nc.read_recvar(name, irecin)

                                    ncout.write_recvar(name, irec+irecin, data, j0, iY)
                                    del data

                        if bj == 0 and progress and not verbose:
                            sys.stderr.write('\n')

                    for nc in ncs:
                        nc.close()

                    if progress and not verbose:
                        if bj == 0 and nty > 1:
                            sys.stderr.write('Writing {0} more slices: '.format(nty-1))
                        else:
                            sys.stderr.write('.')

                if nty > 1 and progress and not verbose:
                    sys.stderr.write('\n')

                if havetime:
                    irec += nrec

        else:
            ncs = [nc]
            for fname in files0[1:]:
                try:
                    nc = netcdf_file(fname, 'r', **readopts)
                except IOError as e:
                    if e.errno == errno.EMFILE:
                        sys.exit('ERROR: Too many open files.  Try again with --many or increase the limit on open files.')
                    raise
                ncs.append(nc)

            # assemble non-record variable data
            if progress and not verbose: sys.stderr.write('Writing non-record variables\n')
            indstrings = {}
            for name,pos,isrec in ncout.begins:
                if not isrec:
                    if verbose: print name
                    prop = varprops[name]
                    vardims = prop['dimensions']
                    var = vars[name]
                    indx = len(vardims)*[slice(None)]
                    iX = prop['iX']
                    iY = prop['iY']
                    data = np.empty(var.shape, var.data.dtype.newbyteorder('>'))
                    for nc in ncs:
                        tn = nc.tile_number - 1
                        if iX is not None: indx[iX] = Xslice[tn]
                        if iY is not None: indx[iY] = Yslice[tn]
                        data[tuple(indx)] = nc.read_var(name)

                    ncout.write_var(name, data)
                    del data
                else:  # isrec
                    if not havetime:
                        raise ValueError('record variables, but no time')

            # assemble record variable data
            if havetime:
                irec = 0
                for fnames in filess:
                    if irec:
                        ncs = []
                        for fname in fnames:
                            nc = netcdf_file(fname, 'r', **readopts)
                            ncs.append(nc)

                    nrec = nc.numrecs
                    if progress and not verbose: sys.stderr.write('Writing {0} records: '.format(nrec))
                    for irecin in range(nrec):
                        if progress and not verbose: sys.stderr.write('.')
                        for name,pos,isrec in ncout.begins:
                            if isrec:
                                if verbose: print irec+irecin, name
                                prop = varprops[name]
                                vardims = prop['dimensions'][1:]
                                var = vars[name]
                                indx = len(vardims)*[slice(None)]
                                iX = prop['iX']
                                iY = prop['iY']
                                data = np.empty(var.data.shape[1:], var.data.dtype.newbyteorder('>'))
                                for nc in ncs:
                                    tn = nc.tile_number - 1
                                    if iX is not None: indx[iX-1] = Xslice[tn]
                                    if iY is not None: indx[iY-1] = Yslice[tn]
                                    data[tuple(indx)] = nc.read_recvar(name, irecin)

                                ncout.write_recvar(name, irec+irecin, data)
                                del data

                    irec += nrec

                    for nc in ncs:
                        nc.close()

                    if progress and not verbose: sys.stderr.write('\n')

    finally:
        ncout.close()