def dsres2meld(df, mfp, verbose=False, compression=True, single=True):
"""
This function reads in a file in 'dsres' format and then writes it
back out in meld format. Note there is a dependency in this code
on numpy and dymat.
"""
import numpy
from dymat import DyMatFile
# Read dsres file
mf = DyMatFile(df)
# Open a meld file to write to
meld = MeldWriter(mfp, compression=compression, single=single)
# Initialize a couple of internal data structures
tables = {}
signal_map = {}
alias_map = {}
# This is the key to use for "description" fields
DESC = "desc"
# We loop over the blocks in the dsres file and each block
# will end up being a table.
for block in mf.blocks():
# Extract the abscissa data for this block
if verbose:
print "Block: "+str(block)
(abscissa, aname, adesc) = mf.abscissa(block)
# Determine all columns in the dsres file and associate
# the signals and aliases with these columns
columns = {}
if verbose:
print "Abscissa: "+str(aname)+" '"+adesc+"'"
signals = []
aliases = []
# Loop over all variables in this block and either make them
# signals or aliases (if some other variable has already
# claimed that column)
for name in mf.names(block):
col = mf._vars[name][2]
if col in columns:
if verbose:
print " Alias "+name+" (of: "+columns[col]+")"
aliases.append((name, mf._vars[name][0],
columns[col], mf._vars[name][3]))
else:
if verbose:
print " Signal "+name+" ("+str(col)+")"
columns[col] = name
signals.append(name)
if verbose:
print "Number of columns: "+str(len(columns.keys()))
print "Number of signals: "+str(len(signals))
print "Number of aliases: "+str(len(aliases))
signal_map[block] = signals
alias_map[block] = aliases
# Generate table for this block (and store it away for later)
tables[block] = meld.add_table("T"+str(block))
# Add abscissa
tables[block].add_signal(aname, metadata={DESC: adesc})
for signal in signals:
tables[block].add_signal(signal, metadata={DESC: mf.description(signal)})
# Add aliases (and their metadata)
for alias in aliases:
transform = None
if alias[3]<0.0:
tables[block].add_alias(alias=alias[0], of=alias[2],
transform="aff(-1,0)",
metadata={DESC:mf.description(alias[0])})
else:
tables[block].add_alias(alias=alias[0], of=alias[2],
metadata={DESC:mf.description(alias[0])})
# Finalize structure
meld.finalize()
# Now loop again, this time with the intention to write data
for block in mf.blocks():
# Write abscissa for this block
(abscissa, aname, adesc) = mf.abscissa(block)
abscissa = list(abscissa.astype(numpy.float))
tables[block].write(aname, list(abscissa))
signals = signal_map[block]
print "Block: "+str(block)
print " Writing signals: "+str(signals)
# Then write signals (no need to write aliases)
for signal in signals:
vec = list(mf.data(signal).astype(numpy.float))
tables[block].write(signal, vec)
# Close the MeldWriter
meld.close()
