parent
545f28324b
commit
a513d35444
@ -1,42 +1,42 @@
|
|||||||
from util import *
|
from util import *
|
||||||
|
|
||||||
@timing_decorator
|
@timing_decorator
|
||||||
def main_qzgy() -> None:
|
def main_qzgy() -> None:
|
||||||
"""简单方法爬取青藏高原数据"""
|
"""简单方法爬取青藏高原数据"""
|
||||||
print("\n开始爬取数据1……")
|
print("\n开始爬取数据1……")
|
||||||
qzgy = Crawler()
|
qzgy = QZGYDate()
|
||||||
content = qzgy.crawler_1()
|
content = qzgy.crawler_qzgy()
|
||||||
word_freqs = qzgy.get_freqs(content)
|
word_freqs = qzgy.get_freqs_of_Cn(content)
|
||||||
qzgy.print_freqs(word_freqs)
|
qzgy.print_freqs(word_freqs)
|
||||||
print("数据1词频获取结束。")
|
print("数据1词频获取结束。")
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
@timing_decorator
|
@timing_decorator
|
||||||
def main_lzwz() -> None:
|
def main_lzwz() -> None:
|
||||||
"""简单方法爬取泸州问政获取词频"""
|
"""简单方法爬取泸州问政获取词频"""
|
||||||
print("\n开始爬取数据2……")
|
print("\n开始爬取数据2……")
|
||||||
lzwz = Crawler()
|
lzwz = LZWZDataCrawler()
|
||||||
content = lzwz.get_data_2()
|
content = lzwz.get_data_lzwz()
|
||||||
word_freqs = lzwz.get_freqs(content)
|
word_freqs = lzwz.get_freqs_of_Cn(content)
|
||||||
lzwz.print_freqs(word_freqs)
|
lzwz.print_freqs(word_freqs)
|
||||||
print("数据2词频获取结束。")
|
print("数据2词频获取结束。")
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
@timing_decorator
|
@timing_decorator
|
||||||
def main_gdb() -> None:
|
def main_gdb() -> None:
|
||||||
"""简单方法爬取古德堡计划数据"""
|
"""简单方法爬取古德堡计划数据"""
|
||||||
print("\n开始爬取数据3……")
|
print("\n开始爬取数据3……")
|
||||||
gdb = Crawler()
|
gdb = GDBDate()
|
||||||
content = gdb.get_data_3()
|
content = gdb.get_data_gdb()
|
||||||
word_freqs = gdb.get_freqs_of_En(content)
|
word_freqs = gdb.get_freqs_of_En(content)
|
||||||
gdb.print_freqs(word_freqs)
|
gdb.print_freqs(word_freqs)
|
||||||
print("数据3词频获取结束。")
|
print("数据3词频获取结束。")
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
main_qzgy()
|
main_qzgy()
|
||||||
main_lzwz()
|
main_lzwz()
|
||||||
main_gdb()
|
main_gdb()
|
||||||
|
Loading…
Reference in new issue