if num <= 40:
a = (t[1], t[0])
new_list.append(a)
new_list.sort()
return new_list
def ls_convert(mylist):
new_list = []
for num, t in enumerate(mylist, 1):
if num <= 40:
a = str("%s:%s" % (t[0], t[1]))
new_list.append(a)
return new_list
print("------------------------")
for fn in ['Obama', 'Romney']:
if fn == 'Obama':
pairs = ls_wc(d_Obama)
else:
pairs = ls_wc(d_Romney)
high_count = 20
low_count = 2
body = ''
for word, cnt in pairs:
body = body + " " + make_HTML_word(word, cnt, high_count, low_count)
box = make_HTML_box(body) # creates HTML in a box
print_HTML_file(box, fn) # writes HTML to file name 'testFile.html'
clean_txt_inp_split.remove(another_word) # menghapus kata yang telah dibersihkan sebelumnya jika kata tersebut merupakan stop word
sorted_word_list = sorted(clean_txt_inp_split, reverse = True) # men-sortir kata-kata yang telah dibersihkan dari stop words dari yang terbesar
counter_word = Counter(sorted_word_list).most_common(50) # menghitung jumlah kemunculan kata-kata yang telah diurutkan, dan mengambil top 50
#----mencetak output----#
for i in range((len(counter_word))):
print("{:3d} : {:<14s}".format(counter_word[i][1], counter_word[i][0]), end='') # mencetak dengan format value 3 spasi dan key 14 spasi
if (i+1)%4==0:
print() # pindah baris jika sudah membuat 4 kolom
print('\n')
holder = input("Please type Enter to continue...")
file_inp.close() # menutup file_inp
file_inp_stop.close() # menutup file_inp_stop
#----membuat file HTML----#
high_count= counter_word[0][1] # nilai terbesar untuk kemunculan
low_count = counter_word[-1][1] # nilai terkecil untuk kemunculan
counter_word.sort() # mengurutkan berdasarkan abjad
body = '' # membuat string kosong untuk menyimpan body di loop
for word,cnt in counter_word:
body = body + " " + make_HTML_word(word,cnt,high_count,low_count) #menentukan body untuk box berdasarkan word cloud
box = make_HTML_box(body) #membuat box di HTML berdasarkan body
file_title = "A Word Cloud of " + file_name
print_HTML_file(box, file_title) # membuat file HTML
def ls_convert(mylist):
new_list = []
for num,t in enumerate(mylist,1):
if num <=40:
a = str("%s:%s" % (t[0],t[1]))
new_list.append(a)
return new_list
print("------------------------")
for fn in ['Obama', 'Romney']:
if fn == 'Obama':
pairs = ls_wc(d_Obama)
else:
pairs = ls_wc(d_Romney)
high_count=20
low_count=2
body=''
for word,cnt in pairs:
body = body + " " + make_HTML_word(word,cnt,high_count,low_count)
box = make_HTML_box(body) # creates HTML in a box
print_HTML_file(box,fn) # writes HTML to file name 'testFile.html'
print("{:3d}:{:15s}".format(word[i][0], str(word[i][1])), end="")
samping += 1
if samping == 4:
print("")
samping = 0
print("\n")
input("Please type Enter to continue ")
#=====================================
boi = ''
#karena yang masuk hanya 50 teratas, maka delete saja selain yang 50 pertama di list
#kalau kurang dari 50, for ini ga jalan kok
for i in range(50, len(word)):
del word[50]
#mensort sesuatu dengan menggunakan key lambda
#lambda itu function gitu seperti ini yang dipake :
"""
def lambda (i) :
return i[1]
"""
word = sorted(word, key=lambda i: i[1])
#bagian ini untuk membuat html yang diminta
for i in range(panjang):
boi += " " + htmlFunctions.make_HTML_word(word[i][1], word[i][0], maxc,
minc)
box = htmlFunctions.make_HTML_box(boi)
htmlFunctions.print_HTML_file(box, 'a word cloud of ' + ayy)
for i in range(len(list_akhir)):
print("{:>3d} : {:<14s}".format(list_akhir[i][0], list_akhir[i][1]),
end='')
batasan += 1
if batasan > 3:
print()
batasan = 0
list_akhir[i] = [list_akhir[i][1], list_akhir[i][0]]
#Mengurutkan list berdasarkan alfabet
list_akhir.sort()
#Meminta input enter dari user untuk melanjutkan membuat file html
input('\n\nPlease type Enter to continue ...')
#Membuat file html
body = ''
for word, cnt in list_akhir:
body = body + " " + htmlFunctions.make_HTML_word(
word, cnt, high_count, low_count)
box = htmlFunctions.make_HTML_box(body)
htmlFunctions.print_HTML_file(box, nama_file + ' Word Cloud')
#Jika terjadi error salah input atau ada interupsi dari user
except (FileNotFoundError):
print("File doesn't exist, please try again!")
except (IOError):
print('Input or output error, please try again!')
except (KeyboardInterrupt):
print('Keyboard interruption, please try again!')
for i in range(0, 38, 4):
print(romneyList[i][0], ":", romneyList[i][1], " ",
romneyList[i + 1][0], ":", romneyList[i + 1][1], " ",
romneyList[i + 2][0], ":", romneyList[i + 2][1], " ",
romneyList[i + 3][0], ":", romneyList[i + 3][1])
high_count = 200
low_count = 20
body = ""
i = 0
for cnt, word in obamaList:
body = body + " " + make_HTML_word(word, cnt, high_count, low_count)
i += 1
if i == 40:
break
box = make_HTML_box(body)
print_HTML_file(box, 'Obama')
body = ""
i = 0
for cnt, word in romneyList:
body = body + " " + make_HTML_word(word, cnt, high_count, low_count)
i += 1
if i == 40:
break
box = make_HTML_box(body)
print_HTML_file(box, 'Romney')