def build_word_cloud(text,back_coloring_path=None,font_path=None,txt_freq=None,scale=5):
"""
text: 词云的内容
back_coloring_path: 使用背景图片
font_path: 使用的字体所在路径
txt_freq: 词云权重,覆盖词云的内容
scale: 图片清晰度
"""
# 设置词云属性
wc = WordCloud(
background_color="white", # 背景颜色
max_words=2000, # 词云显示的最大词数
max_font_size=100, # 字体最大值
random_state=42,
scale=scale,
width=1000, height=860, margin=2, # 设置图片默认的大小,但是如果使用背景图片的话,那么保存的图片大小将会按照其大小保存,margin为词语边缘距离
)
wc.font_path = font_path if font_path else 'SIMHEI.TTF'
if back_coloring_path:
back_coloring = imread(back_coloring_path)
wc.mask = back_coloring
if text:
wc.generate(text)
if txt_freq:
wc.generate_from_frequencies(txt_freq)
if back_coloring_path:
image_colors = ImageColorGenerator(back_coloring)
plt.figure()
plt.imshow(wc)
plt.axis("off")
plt.show()
# 绘制词云
wc.to_file('WordCloudDefautColors.png')
if back_coloring_path:
image_colors = ImageColorGenerator(back_coloring)
plt.imshow(wc.recolor(color_func=image_colors))
else:
plt.imshow(wc)
plt.axis("off")
# 绘制背景图片为颜色的图片
plt.figure()
if back_coloring_path:
plt.imshow(back_coloring, cmap=plt.cm.gray)
else:
plt.imshow(wc)
plt.axis("off")
plt.show()
# 保存图片
wc.to_file('WordCloudColorsByImg.png')
print("请不要关闭页面，稍后会自动结束！！")
#打开颜色文件
f = open('颜色选择.csv')
yanse = f.read()
f.close()
result = yanse.split('\n', 164)
countnumber = 0
for x in result:
print(countnumber, ',164')
countnumber = countnumber + 1
stopwords = stopword.split("\n")
wc = WordCloud(prefer_horizontal=prefer_horizontal,
scale=scale,
width=width,
height=height,
stopwords=stopwords,
font_path="font.ttf",
background_color="white",
repeat=True,
relative_scaling=.5,
max_words=max_words,
max_font_size=300,
min_font_size=10,
colormap=x)
if use_mask == 1:
mask = np.array(Image.open("mask.png"))
wc.mask = mask
wc.generate(new_txt)
name = x + ".png"
wc.to_file(name)
def generate_wordcloud(text_path,
mask_path=None,
width=400,
height=400,
lan='en',
font_path=None,
want_worlds=[],
stop_words=[],
path_to_save='.'):
"""
generate a word cloud of the mask picture you provide and a word cloud color by your picture
:param text_path: use to generate words
:param mask_path: picture you want to show
:param width: the width of the word cloud picture, if mask_path is not provided
:param height: the height of the word cloud picture, if mask_path is not provided
:param lan: the language of your text
:param font_path: if lan is 'cn', a chinese font must provide
:param want_worlds: the special word you don't want to separate
:param stop_words: the words you don't want to show up in your word cloud
:param path_to_save: the directory you want to save your word cloud (!! is directory not file)
:return: no return
"""
image_colors = None
wc = WordCloud(background_color='white',
max_words=1000,
max_font_size=400,
random_state=42)
# check path
if not os.path.isfile(text_path):
print('## text_path is invalid !!')
return
if lan == 'cn':
# if the lan is cn, then the path of chinese font path can't be null
if not os.path.isfile(font_path):
print('## chinese font_path cannot be null !!')
return
text = _generate_cn_words(text=open(text_path).read(),
want_words=want_worlds,
stop_words=stop_words)
wc.font_path = font_path
else:
text = open(text_path).read()
if not os.path.isfile(font_path):
wc.font_path = font_path
if mask_path is None and width > 0 and height > 0:
wc.height = height
wc.width = width
elif os.path.isfile(mask_path):
mask = np.array(Image.open(mask_path))
wc.mask = mask
image_colors = ImageColorGenerator(mask)
else:
print('## mask_path is invalid !!')
return
wc.generate(text=text)
plt.imshow(wc.recolor(color_func=image_colors))
plt.axis('off')
plt.show()
if os.path.isdir(path_to_save):
wc.to_file(os.path.join(path_to_save, 'words.png'))
else:
print('## path_to_save is invalid !!')
return
if mask_path is not None:
img1 = Image.open(os.path.join(path_to_save, 'words.png'))
img2 = Image.open(mask_path)
width = img1.size[0]
height = img1.size[1]
for i in range(0, width):
for j in range(0, height):
data1 = (img1.getpixel((i, j)))
data2 = (img2.getpixel((i, j)))
if (data1[0] <= 250 or data1[1] <= 250 or data1[2] <= 250):
img1.putpixel((i, j), (data2[0], data2[1], data2[2], 255))
# if (data1[0] == 255
# and data1[1] == 255
# and data1[2] == 255):
# img1.putpixel((i, j), (205, 205, 205, 255))
plt.imshow(img1)
plt.axis('off')
plt.show()
img1.save(os.path.join(path_to_save, 'wordcloud.png'))
