import requests
from bs4 import BeautifulSoup
class Crawler:
result = ''
result1 = []
paraphrase0 = []
paraphrase1 = []
doubleparaphrase = []
EEparaphrase0 = []
y0 = []
y1 = []
flag=0
def __init__(self, param):
try:
url = 'http://dict.cn/{}'.format(param)
print(url)
r = requests.get(url)
html = BeautifulSoup(r.text, "lxml")
self.result1=[]
self.result = html.select('body > div#content > div.main > div.word > div.word-cont > h1.keyword')[
0].text ##单词本身
print(self.result)
for bdo in html.select('body > div#content > div.main > div.word > div.phonetic > span > bdo'):
self.result1.append(bdo.get_text())##单词音标
##paraphrase0保存解释的词性，paraphrase1以列表方式保存每个词性对应的解释
self.paraphrase0 = []
for span in html.select('body > div#content > div.main > div.section.def > div.layout.detail > span '):
self.paraphrase0.append(span.get_text())
print(self.paraphrase0)
self.paraphrase1 = []
for li in html.select('body > div#content > div.main > div.section.def > div.layout.detail > ol '):
self.paraphrase1.append(li.get_text())
print(self.paraphrase1)
self.doubleparaphrase = []
# example1=[]
# for li in html.select('body > div#content > div.main > div.section.def > div.layout.dual > ol >li'):
# example1.append((li.select('strong')[0].get_text()))
# print(example1)
for li in html.select(
'body > div#content > div.main > div.section.def > div.layout.dual > ol >li'): ###双语释义
self.doubleparaphrase.append((li.get_text()))
print(self.doubleparaphrase)
self.EEparaphrase0 = []
for li in html.select('body > div#content > div.main > div.section.def > div.layout.en > ol >li'): ##英英释义
self.EEparaphrase0.append(li.get_text())
print(self.EEparaphrase0)
self.y0 = [] ##例句中该单词词性
for b in html.select('body > div#content > div.main > div.section.sent > div.layout.sort > div >b'):
self.y0.append(b.get_text())
print(self.y0)
self.y1 = [] ##例句本身
for ol in html.select('body > div#content > div.main > div.section.sent > div.layout.sort > ol'):
self.y1.append(ol.get_text().split("\n"))
print(self.y1)
self.flag=1
except:
self.flag=0
webcrawler.py
from flask import Flask, render_template, request
import webcrawler
app = Flask(__name__)
@app.route('/',methods=['POST','GET'])
def hello_world():
return render_template('index0.html')
@app.route('/select', methods=['POST'])
def select():
x=webcrawler.Crawler(request.form["words"])
if x.flag==1:
return render_template('index1.html',result=x.result,paraphrase0=x.paraphrase0,paraphrase1=x.paraphrase1,CE=x.doubleparaphrase,EE=x.EEparaphrase0,y0=x.y0,y1=x.y1,result1=x.result1)
else:
return "查询失败"
if __name__ == '__main__':
app.run(debug=True)webpage.py
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>英语字典</title>
</head>
<body id="body">
<style>
.div1 {
height: 180px;
width: 380px;
border: 1px solid #8A8989;
margin: 0 auto;
position: absolute;
top: 0;
left: 0;
right: 0;
bottom: 0;
}
.input {
display: block;
width: 350px;
height: 40px;
margin: 10px auto;
}
.button {
background: #2066C5;
color: white;
font-size: 18px;
font-weight: bold;
height: 50px;
border-radius: 4px;
}
</style>
<div class="div1">
{# action:接受的路由 #}
<form action="/select" method="post">
<input type="text" class="input" name="words" placeholder="请输入单词">
<input type="submit" value="查询" class="input button">
</form>
</div>
</body>
</html>index0.html
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>{{result}}</title>
</head>
<style>
html,body{
height:100%;
}
body{
}
p {color: red;}
.button {
background: #2066C5;
color: white;
}
</style>
<body id="body">
<p style="font-size: 30px"><strong>音标</strong></p>
<table id="cparaphrase">
<tr>
<td style="font-size: 20px">英式发音</td ><td style="font-size: 20px">美式发音</td>
</tr>
<tr>
{% for i in result1 %}
<td>{{i}}</td>
{% endfor %}
</tr>
</table>
<p style="font-size: 30px"><strong>中文释义</strong></p>
<table id="cparaphrase0">
<tr>
<td>
<table id="cparaphrase1">
{% for i in paraphrase0 %}
<tr>
<th>{{i}}</th>
</tr>
{% endfor %}
</table>
</td>
<td>
<table id="cparaphrase2">
{% for i in paraphrase1 %}
<tr>
<td>{{i}}</td>
</tr>
{% endfor %}
</table>
</td>
</tr>
</table>
<p style="font-size: 30px"><strong>中英释义</strong></p>
<table id="ceparaphrase">
{% for i in CE %}
<tr>
<th>{{i}}</th>
</tr>
{% endfor %}
</table>
</table>
<p style="font-size: 30px"><strong>英英释义</strong></p>
<table id="eeparaphrase">
{% for i in EE %}
<tr>
<td>{{i}}</td>
</tr>
{% endfor %}
</table>
<p style="font-size: 30px">
<strong>例句</strong>
</p>
<table id="exqmple0">
<tr>
<td>
<table id="example1">
{% for i in y0 %}
<tr>
<th>{{i}}</th>
</tr>
{% endfor %}
</table>
</td>
<td>
<table id="example2">
{% for i in y1 %}
{%for j in i%}
<tr>
<td>{{j}}</td>
</tr>
{% endfor %}
{% endfor %}
</table>
</td>
</tr>
</table>
</body>
</html>index1.html
版权声明：本文为atri_ater原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。