Using Python to Access OSI Soft's PI Data

If you are not using PI historian (or don't know what it is), move on. This post is not for you.

First of all, for this to work, you will need pisdk installed. I think you already have it if you could access PI data using excel. You will then call it via COM service. Code snippet below:

import time
from datetime import datetime, timedelta
from win32com.client import Dispatch
tdMax = timedelta(minutes=15)
pisdk = Dispatch('PISDK.PISDK')
myServer = pisdk.Servers('PI_ServerName')
con =Dispatch('PISDKDlg.Connections')
con.Login(myServer,'username','password',1,0)
piTimeStart = Dispatch('PITimeServer.PITimeFormat')
piTimeEnd = Dispatch('PITimeServer.PITimeFormat')
piTimeStart.InputString = '08-25-2009 00:00:00'
piTimeEnd.InputString = '08-25-2009 23:59:59'
sampleAsynchStatus = Dispatch('PISDKCommon.PIAsynchStatus')
samplePoint = myServer.PIPoints['tagname']
sampleValues = samplePoint.Data.Summaries2(piTimeStart,piTimeEnd,'1h',5,0,sampleAsynchStatus)
t0 = datetime.now()
while True:
    try:
        valsum = sampleValues("Average").Value # retrieve the value
        break  # it will get out of infinite loop when there is no error
    except:  # it failed because server needs more time
        td = datetime.now() - t0
        if td > tdMax:
            print "It is taking so long to get PI data ! Exiting now ...."
            exit(1)
        time.sleep(3)
i = 1
while i < valsum.Count+1:
    print valsum(i).Value, valsum(i).ValueAttributes('Percentgood').Value
    i += 1
...
 61.4011819388 100.0
61.3148685875 100.0
61.4948477524 100.0
61.2000007629 100.0
61.2000007629 100.0
61.2000007629 100.0
63.8442935978 100.0
64.2453963685 100.0
66.205138361 100.0
71.4747024728 100.0
63.8786754349 100.0
64.3719732268 100.0
64.6296242787 100.0
61.8953031561 100.0
57.0552419993 100.0
57.2999992371 100.0
58.2445955483 100.0
57.5679091366 100.0
59.6997264523 100.0
60.9096459478 100.0
59.7756373596 100.0
55.1742569927 100.0
59.7151307987 100.0

For my officemates. That is just too much work, right? Actually, if you will write the application in VB, your code will look just like that. For this reason, I wrote a  convenience class to hide all those warts and simplify everything as follows:

import sys
sys.path.append(r'M:\EMS\pyhomebrew')

from osipi import osipi
# with time stamp
b = osipi('username','password') # username, then, password, and optionally, PI server
for x, y in b.getAverage('pi-tag here','04-01-2010 00:00','04-01-2010 08:00','1h',True):
    print x, y    # x is time, y is value
# without time stamp

b = osipi('username','password') # username, then, password, and optionally, PI server
for x in b.getAverage('pi-tag here','04-01-2010 00:00','04-01-2010 08:00','1h'):
    print x
# to get the compressed values
for x, y in b.getCompressed('pi-tag here','04-01-2010 00:00','04-01-2010 00:01',True):
    print x, y    # x is time, y is value

Why not use excel? You use excel when it is appropriate but there are instances when it is not. For one, excel has limitations on the number of rows. If you go back too far in time, excel could choke. I was told that the latest version of excel has virtually no limit on the number of rows - not true! I dumped the OAG database via rio and it could not read the whole thing!