def process(inputdir,outfilename,brtime1,brtime2):
# Empty dictionary to store the tweet counts for each user
userlist={}
# Process all the files in the input directory
filelist=tweetproc.jsonindir(inputdir)
for file in filelist:
# Returns a dictionary of user ids with the counts of each behavior response for that user
userbrcounts=tweetproc.tweetcodingbresponse(os.path.join(inputdir,file),brtime1,brtime2)
# Loop over all of the users in the userbrs dictionary returned by the bresponse method
# If the user is already in userlist, then add the behavior response counts, otherwise set the counts
for user in userbrcounts:
if user in userlist: # If user is already in the user list
for br in userbrcounts[user]: # Loop over all the behavior responses
userlist[user][br]+=userbrcounts[user][br] # Add each to the existing count
else:
userlist[user]=userbrcounts[user] # Just copy the behavior count dictionary over
print "Processing",outfilename
#pprint.pprint(userlist)
# Save the results
with io.open(outfilename,'w',encoding="utf-8",errors='ignore') as outfile:
# Loop over all users in the userlist and save the values to the CSV file
for user, brcounts in sorted(userlist.items()):
userstr=user
for br in brcounts:
userstr=userstr+","+str(brcounts[br])
userstr+="\n"
outfile.write(unicode(userstr))
brsummary={'handwash':0,
'handsanitize':0,
'cough':0,
'avoidgathering':0,
'avoidschool':0,
'total':0
}
# Loop over all users in the userlist and summarize the number of behavior responses
for user, brcounts in sorted(userlist.items()):
for br in brcounts:
print "user",user,"br",br,"brcounts[br]",brcounts[br],"brsum[br]",brsummary[br]
brsummary[br]+=brcounts[br]
sum=len(userlist)
# For convenience calculate the ratio of users for each category
# Write out the percentage of users with the count
for br in brsummary:
if sum>0:
percent=float(brsummary[br]*100)/float(sum)
else:
percent=0.0
outfile.write(unicode(str(br)+","+str(brsummary[br])+","+str(round(percent,2))+"\n"))
# Then write out the sum
outfile.write(unicode("sum,"+str(sum)+"\n"))
def process(inputdir,outfilename,tictime1,tictime2):
# Empty dictionary to store the tweet counts for each user
userlist={}
# Process all the files in the input directory
filelist=tweetproc.jsonindir(inputdir)
for file in filelist:
usercounts=tweetproc.tic(os.path.join(inputdir,file),tictime1,tictime2)
# Loop over all of the users in the usercounts dictionary returned by the tic method
# If the user is already in userlist, then add the count, otherwise set the count
for user in usercounts:
if user in userlist:
userlist[user]+=usercounts[user]
else:
userlist[user]=usercounts[user]
print "Processing",outfilename
with io.open(outfilename,'w',encoding="utf-8",errors='ignore') as outfile:
# Loop over all users in the userlist and save the values to the CSV file
for user, value in sorted(userlist.items()):
userstr=user+","+str(value)+"\n"
outfile.write(unicode(userstr))
# Loop over all users in the userlist and classify the user based on the number of tweets
# The count goes from 1-5.
counts=[0,0,0,0,0,0]
for user, value in sorted(userlist.items()):
if value >= 5:
counts[5]+=1
else:
counts[value]+=1
sum=len(userlist)
# For convenience calculate the ratio of users in each category
# Write out the percentage of users with 1, 2, 3, 4, 5+ tweets
for i in xrange(0,len(counts)):
if sum>0:
percent=float(counts[i]*100)/float(sum)
else:
percent=0.0
outfile.write(unicode(str(i)+","+str(counts[i])+","+str(round(percent,2))+"\n"))
# Then write out the sum
outfile.write(unicode("sum,"+str(sum)+"\n"))
#!/usr/bin/python
import tweetproc
import os
import io
inputdir="data/geoebola-sites-Kent-25000"
inputdir="data/geoebola"
outfilename="csv/out.geoebola.csv"
# Start with an empty string
outputstring=""
filelist=tweetproc.jsonindir(inputdir)
for file in filelist:
outputstring=outputstring+tweetproc.geocsv(os.path.join(inputdir,file))
with io.open(outfilename,'w',encoding="utf-8",errors='ignore') as outfile:
outfile.write(outputstring)
def process(inputdir,outfilename,tctime1,tctime2):
# Empty dictionary to store the tweet counts for each user
userlist={}
# Process all the files in the input directory
filelist=tweetproc.jsonindir(inputdir)
for file in filelist:
# Returns a dictionary of user ids with the counts of each behavior response for that user
usertccounts=tweetproc.tweetcoding(os.path.join(inputdir,file),tctime1,tctime2)
print usertccounts
# Loop over all of the users in the usertcs dictionary returned by the tcesponse method
# If the user is already in userlist, then add the behavior response counts, otherwise set the counts
for user in usertccounts:
if user in userlist: # If user is already in the user list
for tc in usertccounts[user]: # Loop over all the behavior responses
userlist[user][tc]+=usertccounts[user][tc] # Add each to the existing count
else:
userlist[user]=usertccounts[user] # Just copy the behavior count dictionary over
print "Processing",outfilename
#pprint.pprint(userlist)
# Save the results
with io.open(outfilename,'w',encoding="utf-8",errors='ignore') as outfile:
# Loop over all users in the userlist and save the values to the CSV file
for user, tccounts in sorted(userlist.items()):
userstr=user
for tc in tccounts:
userstr=userstr+","+str(tccounts[tc])
userstr+="\n"
outfile.write(unicode(userstr))
tcsummary={'concern':0,
'experience':0,
'opinion':0,
'sarcasm':0,
'relief':0,
'downplay':0,
'frustration':0,
'total':0
}
# Loop over all users in the userlist and summarize the number of behavior responses
for user, tccounts in sorted(userlist.items()):
for tc in tccounts:
print "user",user,"tc",tc,"tccounts[tc]",tccounts[tc],"tcsum[tc]",tcsummary[tc]
tcsummary[tc]+=tccounts[tc]
sum=len(userlist)
# For convenience calculate the ratio of users for each category
# Write out the percentage of users with the count
for tc in tcsummary:
if sum>0:
percent=float(tcsummary[tc]*100)/float(sum)
else:
percent=0.0
outfile.write(unicode(str(tc)+","+str(tcsummary[tc])+","+str(round(percent,2))+"\n"))
# Then write out the sum
outfile.write(unicode("sum,"+str(sum)+"\n"))