[置顶] 【matlab 爬虫】用matlab做网络爬虫入门系列1

clc,clear
%%
% 设置不用科学计数法显示数据
format short g
% %
% 读取源代码
sourcefile=urlread('file:///D:/Program%20Files/MATLAB/R2013a/gui3/sheet.html');
% 正则表达式获取第一行数据存为data1
expr1='<td .*?>(.*?)</td>';

[datafile1, data_tokens1] = regexp(sourcefile, expr1, 'match', 'tokens');
data1 = zeros(size(data_tokens1));
for idx1 = 1:length(data_tokens1)
data1(idx1) = str2double(data_tokens1{idx1}{1});
end

%正则表达式获取第二行到最后一行存为data2
expr2='<td>(.*?)</td>';
[datafile2, data_tokens2] = regexp(sourcefile, expr2, 'match', 'tokens');
data2 = zeros(size(data_tokens2));
for idx2 = 16:length(data_tokens2)
data2(idx2) = str2double(data_tokens2{idx2}{1});
end
data2=data2(1,16:end);
% 合并data1和data2
data3=[data1 data2];
%%得到每个变量的数据
number=data3(1,1:15:end)';
Month=data3(1,2:15:end)';
Day=data3(1,3:15:end)';
Time=data3(1,4:15:end)';
p=data3(1,5:15:end)';
a=data3(1,6:15:end)';
v=data3(1,7:15:end)';
T=data3(1,8:15:end)';
Cp=data3(1,9:15:end)';
s=data3(1,10:15:end)';
t1=data3(1,11:15:end)';
S=data3(1,12:15:end)';
H=data3(1,13:15:end)';
P=data3(1,14:15:end)';
n=data3(1,15:15:end)';
% 转换成矩阵
data4=[number Month Day Time p a v T Cp s t1 S H P n]