This script will extract the PS-B and PS-E tables from DOE .sim building files.

extract_doe.py

import glob
import re
import csv

'''
 This script will extract the PS-B and PS-E tables from
 DOE .sim building files.

 This script was tested on DOE 2.1e .sim files.

 To run, please ensure that you have the files in a folder called:
   building_files
 And this script is located in the same folder as the building_files folder
 but not in that folder itself!

 It will iterate over all .sim files in the building_files folder
 Upon completion, it will add the files:
 * psb.csv
 * psb_total.csv
 * pse.csv
 * pse_elec.csv
 * pse_fuel.csv

 These are comma-separated variable files with the delimiter ','

 Please run this using Python 2.7, which can be obtained here:
 http://www.python.org/download/releases/2.7/

 This script is released without any support! Good luck!

 Alex Storer, Holly Samuelson
 December 2012
 Harvard University
'''

allfiles = glob.glob('building_files/*.sim')

# Write the PS-B Table
print "Extracting PS-B Data..."
fw = open('psb.csv','w')
dw = csv.DictWriter(fw,["filename","month","electricity","gas"])
dw.writeheader()
# Write the PS-B Table, with only totals
fwt = open('psb_total.csv','w')
dwt = csv.DictWriter(fwt,["filename","month","electricity","gas"])
for fname in allfiles:
    f = open(fname)
    print "PS-B>",fname
    d = dict()
    numMatches = 0
    for line in f:
        res = re.search('\(UNITS/MO\) \s* (\d+\.\d) (\s* (\d+\.\d))?',line)
        if res is not None:            
            numMatches = numMatches + 1
            d["month"] = numMatches
            d["electricity"] = res.group(1)
            d["gas"] = res.group(3)
            d["filename"] = fname
            dw.writerow(d)
        res = re.search('\(UNITS/YR\) \s* (\d+\.\d) (\s* (\d+\.\d))?',line)
        if res is not None:     
            numMatches = numMatches + 1
            d["month"] = "total"
            d["electricity"] = res.group(1)
            d["gas"] = res.group(3)
            d["filename"] = fname
            dw.writerow(d)
            dwt.writerow(d)
    f.close()
fw.close()
fwt.close()

# A general function to write PS-E tables with Electricity or Fuel
def doPSE(writeElec):
    if writeElec:
        fw = open('pse_elec.csv','w')
        print "Extracting PS-E Tables (Electricity only)>"
    else:
        fw = open('pse_fuel.csv','w')
        print "Extracting PS-E Tables (Fuel only)>"
    dw = csv.DictWriter(fw,["filename","month","misc-equipment","area-lights","space-cooling","pumps-misc","space-heating","domhot-water","vent-fans"])
    dw.writeheader()
    for fname in allfiles:
        if writeElec:
            print "PS-E (elec)>",fname
        else:
            print "PS-E (fuel)>",fname
        all_me = None
        all_al = None
        all_sc = None
        all_pm = None
        all_sh = None
        all_dw = None
        all_vf = None

        f = open(fname)
        inPSE = False
        useFuel = False
        useElec = False
        for line in f:
            res = re.search('REPORT- PS-E',line)
            if res is not None:
                inPSE = True
            res = re.search('REPORT- PS-F',line)
            if res is not None:
                inPSE = False
            res = re.search('FUEL END-USES IN MBTU',line)
            if res is not None:
                useFuel = True
                useElec = False
            res = re.search('ELECTRICAL END-USES IN KWH',line)
            if res is not None:
                useFuel = False
                useElec = True

            # do we write out for electricity or fuel?
            if writeElec:
                doWrite = useElec
            else:
                doWrite = useFuel
            res = re.search('MISC EQUIPMT',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_me = re.findall('\d+\.',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_vf = re.findall('\d+\.',line)            
            res = re.search('AREA LIGHTS',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_al = re.findall('\d+\.',line)
            res = re.search('SPACE COOL',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_sc = re.findall('\d+\.',line)
            res = re.search('PUMPS & MISC',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_pm = re.findall('\d+\.',line)
            res = re.search('SPACE HEAT',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_sh = re.findall('\d+\.',line)
            res = re.search('DOMHOT WATER',line)
            if (res is not None) and (inPSE) and (doWrite):
                all_dw = re.findall('\d+\.',line)
        m = 0
        for m in range(0,13):
            d = dict()
            if all_me is not None:
                d["misc-equipment"] = all_me[m]
            if all_al is not None:
                d["area-lights"] = all_al[m]
            if all_sc is not None:
                d["space-cooling"] = all_sc[m]
            if all_pm is not None:
                d["pumps-misc"] = all_pm[m]
            if all_sh is not None:
                d["space-heating"] = all_sh[m]        
            if all_dw is not None:
                d["domhot-water"] = all_dw[m]        
            if all_vf is not None:
                d["vent-fans"] = all_vf[m]    
            # Python counts from 0, so write months incremented by 1
            m = m+1
            if m<=12:
                d["month"] = m
            else:
                d["month"] = "total"
            d["filename"] = fname
            dw.writerow(d)
        f.close()
    fw.close()


# now generate PS-E tables for electricity and fuel separately
doPSE(writeElec = True)
doPSE(writeElec = False)

P.S. The pse.csv file only contains monthly MISC EQUIP.