def fileFreq(text,dic): #get the word frequency for given texts using extracted dictionary
tf_file=defaultdict(int)
tf=wordFreq(text,stopword=stop,l=1)
for w in tf:
if w in dic:
tf_file[w]=tf_file[w]+1
return tf_file
def BOWFeature(json_files, dicPath, feature_path=''):
dic=loadDict(dicPath);label=0
dic_list=list(dic)
writer=open(feature_path,'w')
worg=open(feature_path+'_org','w')
for cat in os.listdir(base_path):
if not os.path .isdir(base_path+'/'+cat):
continue
for f in os.listdir(base_path+'/'+cat+'/jsons'):
if not os.path.isfile(base_path+'/'+cat+'/jsons/'+f): continue
jtexts = loadJson(base_path+'/'+cat+'/jsons/'+f)
jtexts_lower=dict((k.lower(), v) for k, v in jtexts.iteritems())
if not 'caption' in jtexts_lower: continue
tf=wordFreq(jtexts_lower['caption'].lower(), stopword=stop,l=1)
fn=os.path.splitext(f)[0]
worg.write(fn+str(label)+'\t'+'\t'+'\t'.join(w.encode('utf-8')+'\t'+str(tf[w]) for w in tf if w in dic_list)+'\n')
writer.write(fn+'\t'+str(label)+'\t'+'\t'.join(str(tf[w]) for w in dic)+'\n')
label=label+1
writer.close()
worg.close()
print 'BAG OF TEXTUAL FEATURE GENERATED'
def dicExt(path_list):#read all files under the list to generate dictionary
tf=defaultdict(lambda:defaultdict(int))
frequency = defaultdict(lambda:defaultdict(int))
label=0
count=0
clsize=[]
for cat in os.listdir(base_path):
if not os.path .isdir(base_path+'/'+cat):
continue
count=0
for f in os.listdir(base_path+'/'+cat+'/jsons'):
if not os.path.isfile(base_path+'/'+cat+'/jsons/'+f): continue
jtexts = loadJson(base_path+'/'+cat+'/jsons/'+f)
jtexts_lower=dict((k.lower(), v) for k, v in jtexts.iteritems())
if not 'caption' in jtexts_lower: continue
tf_file=wordFreq(jtexts_lower['caption'].lower(),stopword=stop,l=1)
for w in tf_file:
frequency[label][w]=frequency[label][w]+1
count=count+1
clsize.append(count)
label=label+1
return frequency, clsize
