def create_singel_word_cloud(text, path, colors=None):
filename = f"{path}.png"
with checkTimes(msg=f"Saved {filename} "):
if not text:
error("none text!")
return
stopwords = set(STOPWORDS) | set(read_stopwords())
list(map(stopwords.add, ["看", "都", "不"]))
jiebares = jieba.cut(text)
if conf.use_frequencies:
jiebares = dict(
filter(
lambda item: bool(item[0].strip())
and item[0].strip() not in stopwords,
Counter(jiebares).most_common(100000),
)
)
else:
jiebares = " ".join(jiebares)
wc = WordCloud(
background_color="black" if conf.is_dark else "white",
font_path="HYQiHei-25J.ttf",
max_words=2000,
mask=colors,
stopwords=stopwords,
max_font_size=100,
random_state=45,
)
(wc.generate_from_frequencies if conf.use_frequencies else wc.generate)(
jiebares
)
wc.recolor(color_func=ImageColorGenerator(colors))
wc.to_file(filename)
def create_xlsx(datas, columns, filename="res.xlsx"):
with checkTimes(msg=f"Created {filename} "):
xlsx = pandas.DataFrame(datas)
xlsx.rename(columns={_: __ for _, __ in enumerate(columns)}, inplace=True)
writer = pandas.ExcelWriter(
filename, engine="xlsxwriter", options={"strings_to_urls": False}
)
xlsx.to_excel(writer, "data")
writer.save()
def CreateXLSX(datas, columns, filename='res.xlsx'):
with checkTimes():
xlsx = pandas.DataFrame(datas)
xlsx.rename(columns=columns, inplace=True)
writer = pandas.ExcelWriter(
filename, options={'strings_to_urls': False})
xlsx.to_excel(writer, "data")
writer.save()
success('Created {filename}')
def create_json(datas, filename="res.json"):
with checkTimes(msg=f"Saved {filename} "):
with open(filename, "w") as f:
f.write(json.dumps(datas, ensure_ascii=False, indent=4))
conf.max_time = 20000
if new:
conf.new = new
if cons:
conf.maxConnections = cons
if excel:
conf.need_excel = excel
if words:
conf.need_words = words
if use_frequencies:
conf.use_frequencies = use_frequencies
if words_background:
conf.words_background = words_background
if is_dark:
conf.is_dark = is_dark
if graph:
conf.need_graph = graph
Spider().run()
if __name__ == "__main__":
with checkTimes():
try:
main()
except KeyboardInterrupt:
pass
except Exception as e:
raise e
finally:
info("finished")
def CreateJson(datas, filename='res.json'):
with checkTimes():
with open(filename, 'w') as f:
f.write(json.dumps(datas, ensure_ascii=False, indent=4))
success(f'Saved {filename}')
