PvPython and PvBatch: Difference between revisions

From KitwarePublic
Jump to navigationJump to search
Line 118: Line 118:




===Generic Python Script for PvBatch on Linux:wq===
===Generic Python Script for PvBatch on Linux===
Here is an example running pvbatch, without having to make a trace.
Here is an example running pvbatch, without having to make a trace.


Line 127: Line 127:
**Change the output file from greenSphereScreenshot.png to redShpereScreenshotScreenshot.png
**Change the output file from greenSphereScreenshot.png to redShpereScreenshotScreenshot.png


*Make a file runner.py
*Make a file runner.sh
*Edit as follows:
*Edit as follows:
'''#!/usr/bin/python3'''
'''#!/usr/bin/bash'''


'''/pathToPvPython/pvbatch greenSphere.py'''
'''/pathToPvPython/pvbatch greenSphere.py'''
Line 135: Line 135:
'''/pathToPvPython/pvbatch redSphere.py'''
'''/pathToPvPython/pvbatch redSphere.py'''
*Run as follows:
*Run as follows:
'''python3 runner.py'''
'''source runner.sh'''


==Sandia National Labs specific section==
==Sandia National Labs specific section==

Revision as of 03:20, 12 January 2021

Introduction

ParaView can run without opening the ParaView GUI, and is controlled through Python. There are two Python interfaces - PvPython and PvBatch.

PvPython

PvPython is the Python intreface to ParaView. You can think of PvPython as ParaView with a Python interface. As we did with the Python Shell, you can manually type in commands to pvpython. The first thing you will want to do is import paraview simple, as follows:

from paraview.simple import *

PvPython can also read Python command files. Type pvpython --help for arguments. Running Python files looks like this:

fullPathToPvPython/pvpython fullPathToPythonCommandFile/commandFile.py
d:\alan\paraview\pvpython d:\alan\scripts\disk_out_ref-A.py
  • You will notice that pvpython will run the script and then exit. The output of the script is a screenshot or or other data product.
  • Anywhere that needs editing in the scripts above will be marked by the string editMeHere.
  • You will need to hard code in the paths to your data, and paths for output products.
  • The first time you run a script with pvpython, the output will be a postage stamp sized window. You can change this by finding and uncommenting the line renderView1.ViewSize.


Try making and running a script of your own. Alternatively, here is an example. Cut and paste the following into a file named greenSphere.py:

: File starts next line

#!/usr/bin/env pvpython

from paraview.simple import *

# Lets create a sphere

sphere=Sphere()

Show()

Render()

# get active view

renderView1 = GetActiveViewOrCreate('RenderView')

renderView1.ViewSize = [1670, 1091]

# get display properties

sphere1Display = GetDisplayProperties(sphere, view=renderView1)

# change solid color

sphere1Display.AmbientColor = [0.0, 1.0, 0.0]

sphere1Display.DiffuseColor = [0.0, 1.0, 0.0]

# save screenshot

SaveScreenshot('greenSphereScreenshot.png', renderView1, ImageResolution=[1670, 1091])

: File stops last line

Run this as follows: /pathToPvPythonpvpython greenSphere.py

PvBatch

PvBatch is like PvPython, with two exceptions. PvBatch only accepts commands from input scripts, and PvBatch will run in parallel if it was built using MPI. Input is exactly like PvPython.


Generic user specific section

This section describes how to use pvbatch when in the ParaView training class on Windows PCs.

  • If you are training on Linux, pvbatch will exist in the bin directory.
  • If you are training on OS X, open a terminal window, and cd to /Applications/YourParaViewVersion/Contents/bin. pvbatch will be located here.
  • If you are training on Windows, pvbatch does not exist. But, for a single process, such as this training, pvpython will substitute.

Lets create a Python trace.

Since we are on Windows, we will create a python trace, and use pvpython to process it.

  • Read exodus data, screenshot, movie.
    • Run ParaView.
    • Tools/ Start Trace
    • Read can.ex. Turn all of the variables on. Apply.
    • +Y
    • Go forward one timestep
    • Color by EQPS
    • Save Screenshot
    • Save Animation
    • Stop trace
    • Save this script on your desktop
    • Edit the file, change the following:
      • Correct the path to the input data and output screenshots or animations (not necessary, since you made the trace)
      • Uncomment the line that says renderView*.ViewSize. Change this to something reasonable (maybe 1920x1080)

Let's now batch run this Python trace.

  • Delete the Screenshot and Animation you made above. We want to recreate these
  • Open a CMD window. (On the Start button, type cmd, then click on Command Prompt.)
  • cd to the ParaView bin directory
    • cd c:\Program Files (x86)\ParaView 5.4.1\bin
  • Use pvpython to process our trace. Notice that pvpython understands forward slashes.
    • pvpython.exe c:/Users/myUserName/Desktop/trainingExampleScriptA.py
  • Open the Screenshot and Animation that you just made.


Let's edit the trace to accept arguments

  • Edit the python trace.
  • Right above the ExodusIIReader, enter the following code:
    • datasetIn = sys.argv[1]
    • directoryOut = sys.argv[2]
    • imageFilesOut = sys.argv[3]
    • print "datasetIn = "+datasetIn
    • print "directoryOut = "+directoryOut
    • print "imageFilesOut = "+imageFilesOut
  • Edit the canex2 = ExodusIIReader line as follows:
    • canex2 = ExodusIIReader(FileName=[datasetIn])
  • Edit the SaveScreenshot(...) line as follows:
    • SaveScreenshot(directoryOut+imageFilesOut+'.png', renderView1, ImageResolution=[1425, 1324])
  • Edit the SaveAnimation(...) line as follows:
    • SaveAnimation(directoryOut+imageFilesOut+'.avi', renderView1, ImageResolution=[1424, 1324], FrameWindow=[0, 43])
  • Now, run in a command window as follows:
    • pvpython.exe "c:/Users/myUserName/Desktop/trainingExampleScriptA.py" "C:/Users/myUsername/Desktop" "coolVizA"


Generic Python Script for PvBatch on Linux

Here is an example running pvbatch, without having to make a trace.

  • Make a file greenSphere.py, as described in the pvpython section.
  • Copy greenSphere.py to be redSphere.py.
  • Edit redSphere.py
    • Change AmbientColor and DiffuseColor to be [1.0, 0.0, 0.0]
    • Change the output file from greenSphereScreenshot.png to redShpereScreenshotScreenshot.png
  • Make a file runner.sh
  • Edit as follows:

#!/usr/bin/bash

/pathToPvPython/pvbatch greenSphere.py

/pathToPvPython/pvbatch redSphere.py

  • Run as follows:

source runner.sh

Sandia National Labs specific section

This section is specific to the clusters and environment at Sandia National Laboratories.



PvBatch on the clusters

ssh into one of the clusters. PvBatch can be run on the login nodes, and magically will acquire compute nodes and run your batch visualization in parallel. You will find test scripts at /projects/viz/training/paraview. These scripts are run as follows:


/projects/viz/paraview/bin/pvbatch_chama_mesa
This is version 5.4.0 of pvbatch.
Incorrect number of argument supplied. Expecting 4 but have 0
Usage: /projects/viz/paraview/bin/pvbatch_chama_mesa <Nodes> <Minutes> <HERT estimate> batchFileFullPath
An example is
/projects/viz/paraview/bin/pvbatch_chama_mesa 1 10 FY123456 /projects/viz/training/paraview/whipple-A.py


PvBatchOnNode on the clusters

pvbatch can be run on the same nodes as your simuation. Ask ParaView help for more information on how to use this feature.

Example scripts

Here are four examples. We are going to create scripts using the trace recorder, then run these scripts using pvbatch.

  • Read exodus data, screenshot, movie.
    • Run ParaView.
    • Start Trace
    • Read g1s1.e.16.[0-15]
    • Go to last timestep
    • Go back one timestep
    • Color by EQPS
    • Screenshot
    • Animation
    • Stop trace
    • Save this script on your cluster.
    • Edit the file, change the following:
      • Correct the path to the input data and output screenshots or animations
      • Uncomment the line that says renderView*.ViewSize. Change this to something reasonable (maybe 1920x1080)
  • Read exodus data, Clip, Slice, screenshot, movie.
    • Run ParaView.
    • Start Trace
    • Read g1s1.e.16.[0-15]
    • Go to last timestep
    • Go back one timestep
    • Color by EQPS
    • Clip
    • Slice/ Y Normal
    • Screenshot
    • Animation
    • Stop trace
    • Save this script on your cluster.
    • Edit the file, change the following:
      • Correct the path to the input data and output screenshots or animations
      • Uncomment the line that says renderView*.ViewSize. Change this to something reasonable (maybe 1920x1080)
      • Add the following line above ColorBy(....,('EQPS'))
        • g1s110fpse16Display.SetScalarBarVisibility(renderView1, False)
  • Read exodus data, 2d plots, screenshot, movie
    • Run ParaView.
    • Start Trace
    • Read g1s1.e.16.[0-15]
    • Select point if possible
    • Plot selection (or plot over line, if necessary)
    • Plot EQPS
    • Screenshot
    • Animation
    • Stop trace
    • Edit the file, change the following:
      • Correct the path to the input data and output screenshots or animations
      • Uncomment the line that says renderView*.ViewSize. Change this to something reasonable (maybe 1920x1080)
  • Read cth data, extractCTHPart, screenshot, movie.
    • Run ParaView.
    • Start Trace
    • Read cth-med/spcta.[0-31]
    • ExtractCTHParts - 1
    • ExtractCTHParts - 2
    • Stop trace

Where do you go next?

  • Python calculator and programmable fitler.

Acknowledgements

Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA-0003525.