"""
gracePlot.py -- A high-level Python interface to the Grace plotting package
The intended purpose of gracePlot is to allow easy programmatic and interactive
command line plotting with convenience functions for the most common commands.
The Grace UI (or the grace_np module) can be used if more advanced
functionality needs to be accessed.
The data model in Grace, (mirrored in gracePlot) goes like this: Each grace
session is like virtual sheet of paper called a Plot. Each Plot can have
multiple Graphs, which are sets of axes (use gracePlot.multi() to get multiple
axes in gracePlot). Each Graph has multiple data Sets. Data Sets are added to
graphs with the plot and histoPlot functions in gracePlot.
The main python functions are plot() and histoPlot(). See their docstrings
for usage information. They can be called with any mix of Numpy arrays,
lists, tuples, or other sequences. In general, data is considered to be
stored in columns, so a matrix with three vectors x1, x2 and x3 would be:
[ [ x1[0], x2[0], x3[0] ],
[ x1[1], x2[1], x3[1] ],
[ x1[2], x2[2], x3[2] ],
[ x1[3], x2[3], x3[3] ] ]
Here's a simple example of a gracePlot session:
>>> import gracePlot
>>> p = gracePlot.gracePlot() # A grace session begins
>>> # Sequence arguments to plot() are X, Y, dy
>>> p.plot( [1,2,3,4,5], [10, 4, 2, 4, 10], [0.1, 0.4, 0.2, 0.4, 0.1],
... symbols=1 ) # A plot with errorbars
If you're doing a lot of histograms then you should get Konrad Hinsen's
Scientific Python package:
http://starship.python.net/crew/hinsen/scientific.html
histoPlot() knows how to automatically plot Histogram instances from the
Scientific.Statistics.Histogram module, so histogramming ends up being pretty
simple:
>>> from Scientific.Statistics.Histogram import Histogram
>>> joe = Histogram( some_data, 40 ) # 40 = number of bins
>>> p.histoPlot( joe ) # A histogram plot with correct axis limits
An important thing to realize about gracePlot is that it only has a one-way
communications channel with the Grace session. This means that if you make
changes to your plot interactively (such as changing the number/layout of
graphs) then gracePlot will have NO KNOWLEDGE of the changes. This should not
often be an issue, since the only state that gracePlot saves is the number and
layout of graphs, the number of Sets that each graph has, and the hold state
for each graph.
"""
# --UPDATES--
# 03/02/09: ported to Numpy by Mike McKerns (mmckerns@caltech.edu)
# 2016-06-23: adapted to newer python2&3 library grace_tread
# by Georges Khaznadar <georgesk@debian.org>
from __future__ import print_function
import sys, os.path
sys.path.append(os.path.dirname(os.path.abspath(__file__)))
if sys.version_info.major==3:
# old range was removed and xrange was renamed to range
# so activate the old name which is still used
xrange=range
__version__ = "0.5.3"
__author__ = "Nathaniel Gray <n8gray@caltech.edu>"
__date__ = "September 16, 2001"
from grace_thread import GraceProcess
import numpy as N
try:
from Scientific.Statistics.Histogram import Histogram
haveHisto = 1
except ImportError:
haveHisto = 0
class gracePlot:
def __init__(self, bufsize=-1):
self.grace = GraceProcess(bufsize=bufsize)
self.g = [ graceGraph(self.grace, 0) ]
self.curr_graph = self.g[0]
self.rows = 1
self.cols = 1
self.focus(0,0)
def _send(self, cmd):
#print (cmd)
self.grace.command(cmd)
def _flush(self):
#print ('flush()')
self.grace.flush()
def __del__(self):
"""Destroy the pipe but leave the grace window open for interaction.
This is the best action for the destructor so that unexpected crashes
don't needlessly destroy plots."""
self.grace = None
def exit(self):
"""Nuke the grace session. (more final than gracePlot.__del__())"""
self.grace.exit()
def redraw(self):
"""Refresh the plot"""
#print ('redraw')
self.grace('redraw')
def multi(self, rows, cols, offset=0.1, hgap=0.1, vgap=0.15):
"""Create a grid of graphs with the given number of <rows> and <cols>
"""
self._send( 'ARRANGE( %s, %s, %s, %s, %s )' % ( rows, cols, offset,
hgap, vgap ) )
self.rows = rows
self.cols = cols
if rows*cols > len(self.g):
nPlots = len(self.g)
for i in range( nPlots, (rows*cols - nPlots)+1 ):
self.g.append( graceGraph(self.grace, i) )
# Should we trim the last graphs if we now have *fewer* than before?
# I say yes.
elif rows*cols < len(self.g):
del self.g[rows*cols:]
self._flush()
self.redraw()
def save(self, filename, format='agr'):
"""Save the current plot
Default format is Grace '.agr' file, but other possible formats
are: x11, postscript, eps, pdf, mif, svg, pnm, jpeg, png, metafile
Note: Not all drivers are created equal. See the Grace documentation
for caveats that apply to some of these formats."""
devs = {'agr':'.agr', 'eps':'.eps', 'jpeg':'.jpeg', 'metafile':'',
'mif':'', 'pdf':'.pdf', 'png':'.png', 'pnm':'',
'postscript':'.ps', 'svg':'', 'x11':''}
try:
ext = devs[format.lower()]
except KeyError:
print ('Unknown format. Known formats are\n%s' % devs.keys())
return
if filename[-len(ext):] != ext:
filename = filename + ext
if ext == '.agr':
self._send('saveall "%s"' % filename)
else:
self._send('hardcopy device "%s"' % format.lower() )
self._send('print to "%s"' % filename)
self._send('print')
self._flush()
def focus( self, row, col ):
"""Set the currently active graph"""
self.curr_graph = self.g[row*self.cols + col]
self._send('focus g%s' % self.curr_graph.gID)
self._send('with g%s' % self.curr_graph.gID)
self._flush()
self.redraw()
for i in ['plot', 'histoPlot', 'title', 'subtitle', 'xlabel', 'ylabel',
'kill', 'clear', 'legend', 'hold', 'xlimit', 'ylimit',
'redraw']:
setattr( self, i, getattr(self.curr_graph, i) )
return self.curr_graph
def resize( self, xdim, ydim, rescale=1 ):
"""Change the page dimensions (in pp). If rescale==1, then also
rescale the current plot accordingly. Don't ask me what a pp is--I
don't know."""
if rescale:
self._send('page resize %s %s' % (xdim, ydim))
else:
self._send('page size %s %s' % (xdim, ydim))
def __getitem__( self, item ):
"""Access a specific graph. Can use either p[num] or p[row, col]."""
if type(item) == type(1):
return self.g[item]
elif type(item) == type( () ) and len(item) <= 2:
if item[0] >= self.rows or item[1] >= self.cols:
raise IndexError('graph index out of range')
return self.g[item[0]*self.cols + item[1]]
else:
raise TypeError('graph index must be integer or two integers')
class graceGraph:
def __init__(self, grace, gID):
self._hold = 0 # Set _hold=1 to add datasets to a graph
self.grace = grace
self.nSets = 0
self.gID = gID
def _send(self, cmd):
#print (cmd)
#raise NameError, "duh"
self.grace.command(cmd)
def _flush(self):
#print ('flush()')
self.grace.flush()
def _send_2(self, var, X, Y):
send = self.grace.command
for i in xrange(len(X)):
send( 'g%s.s%s point %s, %s' % (self.gID, var, X[i], Y[i]) )
if i%50 == 0:
self._flush()
self._flush()
def _send_3(self, var, X, Y, Z):
self._send_2(var, X, Y)
send = self.grace.command
for i in range(len(Z)):
send( 'g%s.s%s.y1[%s] = %s' %
(self.gID, var, i, Z[i]) )
if i%50 == 0:
self._flush()
self._flush()
def hold(self, onoff=None):
"""Turn on/off overplotting for this graph.
Call as hold() to toggle, hold(1) to turn on, or hold(0) to turn off.
Returns the previous hold setting.
"""
lastVal = self._hold
if onoff is None:
self._hold = not self._hold
return lastVal
if onoff not in [0, 1]:
raise RuntimeError ("Valid arguments to hold() are 0 or 1.")
self._hold = onoff
return lastVal
def title(self, titlestr):
"""Change the title of the plot"""
self._send('with g%s' % self.gID)
self._send('title "' + str(titlestr) + '"')
self.redraw()
def subtitle(self, titlestr):
"""Change the subtitle of the plot"""
self._send('with g%s' % self.gID)
self._send('subtitle "' + str(titlestr) + '"')
self.redraw()
def redraw(self):
"""Refresh the plot"""
self.grace('redraw')
def xlabel(self, label):
"""Change the x-axis label"""
self._send('with g%s' % self.gID)
self._send('xaxis label "' + str(label) + '"')
self.redraw()
def ylabel(self, label):
"""Change the y-axis label"""
self._send('with g%s' % self.gID)
self._send('yaxis label "' + str(label) + '"')
self.redraw()
def xlimit(self, lower=None, upper=None):
"""Set the lower and/or upper bounds of the x-axis."""
self._limHelper( 'x', lower, upper)
def ylimit(self, lower=None, upper=None):
"""Set the lower and/or upper bounds of the y-axis."""
self._limHelper( 'y', lower, upper)
def _limHelper(self, ax, lower, upper):
send = self._send
if lower is not None:
send('with g%s; world %smin %s' % (self.gID, ax, lower))
if upper is not None:
send('with g%s; world %smax %s' % (self.gID, ax, upper))
self.redraw()
def kill(self):
"""Kill the plot"""
self._send('kill g%s' % self.gID)
self._send('g%s on' % self.gID)
self.redraw()
self.nSets = 0
self._hold = 0
def clear(self):
"""Erase all lines from the plot and set hold to 0"""
for i in range(self.nSets):
self._send('kill g%s.s%s' % (self.gID, i))
self.redraw()
self.nSets=0
self._hold=0
def legend(self, labels):
"""Set the legend labels for the plot
Takes a list of strings, one string per dataset on the graph.
Note: <ctrl>-L allows you to reposition legends in Grace using
the mouse.
"""
if len(labels) != self.nSets:
raise RuntimeError( 'Wrong number of legends (%s) for number' \
' of lines in plot (%s).' % (len(labels), self.nSets))
for i in range(len(labels)):
self._send( ('g%s.s%s legend "' % (self.gID, i)) + labels[i] + '"' )
self._send('with g%s; legend on' % self.gID)
self.redraw()
def histoPlot(self, y, x_min=0, x_max=None, dy=None, edges=0,
fillcolor=2, edgecolor=1, labeled=0):
"""Plot a histogram
y contains a vector of bin counts
By default bin counts are plotted against bin numbers unless
x_min and/or x_max are specified
if edges == 0: # This is the default
x_min and x_max specify the lower and upper edges of the first
and last bins, respectively
else:
x_min and x_max specify the centers of the first and last bins
If dy is specified symmetric errorbars are plotted.
fillcolor and edgecolor are color numbers (0-15)
If labeled is set to 1 then labels are placed at each bin to show
the bin count
Note that this function can create *two* datasets in grace if you
specify error bars."""
if haveHisto and isinstance(y, Histogram):
self.histoPlot( y.array[:,1], x_min=y.min, x_max=y.max, edges=1,
dy=dy, fillcolor=fillcolor, edgecolor=edgecolor,
labeled=labeled )
return
# this is going to be ugly
y = N.array(y)
if x_max is None:
x_max = len(y)-1 + x_min
edges = 0
if x_max <= x_min:
raise RuntimeError ("x_max must be > x_min")
if dy is not None:
if len(dy) != len(y):
raise RuntimeError ('len(dy) != len(y)')
dy = N.array(dy)
if not self._hold: self.clear()
if edges:
# x_min and x_max are the outside edges of the first/last bins
binwidth = (x_max-x_min)/float(len(y))
edge_x = N.arange(len(y)+1 , dtype='d')*binwidth + x_min
cent_x = (edge_x + 0.5*binwidth)[0:-1]
else:
# x_min and x_max are the centers of the first/last bins
binwidth = (x_max-x_min)/float(len(y)-1)
cent_x = N.arange(len(y), dtype='d')*binwidth + x_min
edge_x = cent_x - 0.5*binwidth
edge_x = N.resize(edge_x, (len(cent_x)+1,))
edge_x[-1] = edge_x[-2] + binwidth
edge_y = y.copy() #N.zeros(len(y)+1)
edge_y = N.resize(edge_y, (len(y)+1,))
edge_y[-1] = 0
# Draw the edges:
me = 'g%s.s%s ' % (self.gID, self.nSets)
self._send( me + 'type xy' )
self._send( me + 'dropline on' )
self._send( me + 'line type 3' ) # step to right
self._send( me + 'line color ' + str(edgecolor) )
if fillcolor is not None:
self._send( me + 'fill type 2' ) #Solid
self._send( me + 'fill color ' + str(fillcolor) )
if labeled:
self._send( me + 'avalue on' )
self._flush()
self._send_2( self.nSets, edge_x, edge_y )
self.nSets = self.nSets + 1
# Draw the errorbars (if given)
if dy is not None:
me = 'g%s.s%s ' % (self.gID, self.nSets)
self._send( me + 'type xydy' )
self._send( me + 'line linestyle 0' ) #No connecting lines
self._send( me + 'errorbar color ' + str(edgecolor) )
self._flush()
self._send_3( self.nSets, cent_x, y, dy )
self.nSets = self.nSets + 1
#self._errPlot( cent_x, y, dy )
self._send('with g%s' % self.gID)
self._send('world ymin 0.0') # Make sure the y-axis starts at 0
self._send('autoscale')
self._send('redraw')
self._flush()
def _errPlot(self, X, Y, dy=None, symbols=None, styles=None, pType = 'xydy' ):
"""Line plot with error bars -- for internal use only
Do not use this! Use plot() with dy=something instead."""
if dy is None:
dy = Y
Y = X
X = N.arange(X.shape[0])
# Guarantee rank-2 matrices
if len(X.shape) == 1:
X.shape = (X.shape[0], 1)
if len(Y.shape) == 1:
Y.shape = (Y.shape[0], 1)
if len(dy.shape) == 1:
dy.shape = (dy.shape[0], 1)
if not ( Y.shape == dy.shape and
X.shape[0] == Y.shape[0] and
( X.shape[1] == Y.shape[1] or X.shape[1] == 1 ) ):
raise RuntimeError('X, Y, and dy have mismatched shapes')
if not self._hold: self.clear()
for i in xrange(self.nSets, Y.shape[1] + self.nSets):
me = 'g%s.s%s ' % (self.gID, i)
self._send( me + 'on')
self._send( me + 'type ' + pType)
mycolor = (i%15)+1
self._send( '%s line color %s' % (me, mycolor) )
self._send( '%s errorbar color %s' % (me, mycolor) )
if symbols is not None:
self._send( me + 'symbol %s' % ((i%10) + 1) ) # From 1 to 10
self._send( '%s symbol color %s' % (me, mycolor) )
if styles is not None:
self._send( me + 'line linestyle %s' %((i%8) + 1) ) # 1 to 8
self._flush()
if X.shape[1] == 1:
for i in range(Y.shape[1]):
self._send_3( i+self.nSets, X[:,0], Y[:,i], dy[:,i] )
# Send an upper and lower line too so that autoscaling works
self._send_2( i+self.nSets+Y.shape[1], X[:,0], Y[:,i]+dy[:,i] )
self._send_2( i+self.nSets+2*Y.shape[1], X[:,0],
Y[:,i]-dy[:,i] )
else:
for i in range(Y.shape[1]):
self._send_3( i+self.nSets, X[:,i], Y[:,i], dy[:,i] )
self._send_2( i+self.nSets+Y.shape[1], X[:,i], Y[:,i]+dy[:,i] )
self._send_2( i+self.nSets+2*Y.shape[1], X[:,i],
Y[:,i]-dy[:,i] )
self._send('with g%s' % self.gID)
self._send('autoscale')
#self._send('redraw')
self.nSets = self.nSets + Y.shape[1]
# Kill off the extra lines above/below
for i in range(self.nSets, self.nSets+2*Y.shape[1]):
self._send( 'KILL G%s.S%s' % (self.gID, i) )
self._send('redraw')
self._flush()
def plot(self, X, Y=None, dy=None, symbols=None, styles=None):
"""2-D line plot, with or without error bars
The arguments should be Numpy arrays of equal length.
X, Y, and dy can be rank-1 or rank-2 arrays (vectors or matrices).
In rank-2 arrays each column is treated as a dataset.
X can be rank-1 even if Y and DY are rank-2, so long as
len(X) == len( Y[:,0] )
If dy is not None then it must be the same shape as Y, and symmetric
error bars will be plotted with total height 2*dy.
Setting symbols=1 will give each dataset a unique symbol.
Setting styles=1 will give each dataset a unique linestyle
"""
X = N.array(X)
# if there's no Y, then just use X
if Y is None:
Y = X
X = N.arange(X.shape[0])
else:
Y = N.array(Y)
if dy is not None:
dy = N.array(dy)
self._errPlot(X, Y, dy, symbols=symbols, styles=styles)
return
# Guarantee rank-2 matrices
if len(X.shape) == 1:
X.shape = (X.shape[0], 1)
if len(Y.shape) == 1:
Y.shape = (Y.shape[0], 1)
if X.shape[0] != Y.shape[0] or ( # Different number of points per line
X.shape[1] != X.shape[1] and # Different number of lines
X.shape[1] != 1): # But if X is just 1 line it's ok.
raise RuntimeError('X and Y have mismatched shapes')
############# Grace commands start here ###########
if not self._hold: self.clear()
pType = 'xy' # At some point this might become an option
for i in range(self.nSets, Y.shape[1] + self.nSets):
me = 'g%s.s%s ' % (self.gID, i)
self._send( me + 'on')
self._send( me + 'type ' + pType)
self._send( '%s line color %s' % (me, (i%15)+1) )
if symbols is not None:
self._send( me + 'symbol %s' % ((i%15) + 1) ) # From 1 to 15
self._send( '%s symbol color %s' % (me, (i%15)+1) )
if styles is not None:
self._send( me + 'line linestyle %s' %((i%8) + 1) ) # 1 to 8
self._flush()
if X.shape[1] == 1:
for i in range(Y.shape[1]):
self._send_2( i+self.nSets, X[:,0], Y[:,i] )
else:
for i in range(Y.shape[1]):
self._send_2( i+self.nSets, X[:,i], Y[:,i] )
self._send('with g%s' % self.gID)
self._send('autoscale')
self._send('redraw')
self._flush()
self.nSets = self.nSets + Y.shape[1]
def _test():
from time import sleep
p = gracePlot(bufsize=1)
joe = N.arange(5,50)
p.plot(joe, joe**2, symbols=1)
p.title('Parabola')
sleep(2)
p.multi(2,2)
p.focus(1,1)
p.plot(joe, joe, styles=1)
p.hold(1)
p.plot(joe, N.log(joe), styles=1)
p.legend(['Linear', 'Logarithmic'])
p.xlabel('Abscissa')
p.ylabel('Ordinate')
sleep(2)
p.focus(1,0)
p.histoPlot(N.sin(joe*3.14/49.0), 5./49.*3.14, 3.14)
sleep(2)
p.exit()
if __name__=="__main__":
_test()