#下載股票上市日期
#download the listingdate of one security
library(RCurl)
SH <- readLines("SH.txt")#獲取證券代碼列表
listing.date <- vector(length = length(SH))
url.date1 <- "http://quotes.money.163.com/trade/lsjysj_"
url.date2 <- ".html#01b07"
for (i in 1:length(SH))
{
  #解析網(wǎng)頁(yè)，得到listingdate
  cat(i,'\n')
  url.date <- paste(url.date1, SH[i], url.date2, sep="")
  xx <- getURL(url.date)
  posi <- regexpr("上市日",xx)
  listing.date[i] <- substring(xx,posi[1]-13,posi[1]-4)
}
listing.date.tab <- data.frame(code=SH,listingdate=listing.date,stringsAsFactors=FALSE)
#輸出
write.table(listing.date.tab,file="xxx.txt",sep="\t",quote=FALSE,row.name=FALSE)

#下載股票數(shù)據(jù)
library(RCurl)
#http://quotes.money.163.com/service/chddata.html?code=0600030&start=20030106&end=20140920&fields=TCLOSE;HIGH;LOW;TOPEN;LCLOSE;CHG;PCHG;TURNOVER;VOTURNOVER;VATURNOVER;TCAP;MCAP
url1    <- "http://quotes.money.163.com/service/chddata.html?code="
market  <- "1" # 1:深市，0:滬市
code    <- "000003"
url2    <- "&start="
start   <- "19900101"
url3    <- "&end="
end     <- "20140920"
url4    <- "&fields=TCLOSE;HIGH;LOW;TOPEN;LCLOSE;CHG;PCHG;TURNOVER;VOTURNOVER;VATURNOVER;TCAP;MCAP"

#文件的存放路徑
file.path   <- "F:/download/SS/"

#股票代碼+發(fā)行日期，格式：CODE制表符yyyymmdd
security <- readLines("SS.txt")
code <- vector(length = length(security))
listingdate <- vector(length = length(security))
security.tab <- data.frame(code, listingdate, stringsAsFactors=FALSE)

for (i in 1 : length(security))
{
  security.tab[i,] = strsplit(security[i],"\t")
}

for (i in 1 : dim(security.tab)[1])
{  
  code <- security.tab$code[i]
  start <- security.tab$listingdate[i]
  cat(i,"\t---",code,"\n")
  url <- paste(url1,market,code,url2,start,url3,end,url4,sep="")
  file <- paste(file.path,code,".csv",sep="")
  download.file(url,destfile=file, method="auto",quiet=T)
}

#include <vector>
#include <fstream>
#include <string>

int main(void)
{
    //處理股票數(shù)據(jù)
    const int N = 1000;
    ifstream fin;

    vector<string> file_list,ff;
    char  x[N];
    string ss;
    //獲取要處理的文件列表
    fin.open("fff.txt");
    while (!fin.eof())
    {
        fin.getline(x, N, '\n');
        ss.assign(x);
        if (ss.size() > 0)
        {
            file_list.push_back(ss);
        }
    }
    fin.close();
    //cout<<file_list.size()<<endl;
    
    ofstream fout, fout0,fout1,fout2,fout3,fout4,fout5;
    string obj_file("x.csv"),path1("F:/download/obj/"),path2("F:/download/outh/");

    for (int i = 0; i < file_list.size(); ++i)
    {
        cout<<i+1<<"\t\t"<<file_list[i]<<endl;
        int f_num = i / 500;
        obj_file[0] = '0'+f_num;
        fout.open(path2+obj_file,ofstream::out | ofstream::app);
        fin.open(path1+file_list[i]);
        bool first = true;
        while (!fin.eof())
        {
            fin.getline(x, N, '\n');
            ss.assign(x);
            if (ss.size() > 0 && !first)
            {
                //1到1004為滬市
                if (i+1<=1004)
                {
                    ss="SH,"+ss;
                }
                else
                {
                    ss="SS,"+ss;
                }
                ss.erase(7,1);
                ss.erase(9,1);
                ss.erase(12,1);
                fout<<ss;
            }
            first = false;
        }
        fin.close();
        fout.close();
    }
    
}

rm(list=ls())
library(RODBC)
channel <- odbcConnect(dsn="***",uid="***",pwd="***")

sql1 <- "SELECT dates, close, pre_close FROM idx
where
dates in
(
SELECT dates FROM idx where code='000001'
INTERSECT
SELECT dates FROM security where code='600000'
and
close > 0
and
dates >= to_date('20100101', 'yyyymmdd')
)
and
code = '000001'
order by dates asc"

sql2 <- "SELECT dates, close, pre_close FROM security
where
dates in
(
SELECT dates FROM idx where code='000001'
INTERSECT
SELECT dates FROM security where code='600000'
and
close > 0
and
dates >= to_date('20100101', 'yyyymmdd')
)
and
code = '600000'
order by dates asc"

i000001 <- sqlQuery(channel, sql1)
head(i000001)
s600000 <- sqlQuery(channel, sql2)
head(s600000)

t <- s600000$DATES
s <- log(s600000$CLOSE) - log(s600000$PRE_CLOSE)
i <- log(i000001$CLOSE) - log(i000001$PRE_CLOSE)

plot(i, s, pch = 20, xlab="000001", ylab="600000")
capm.lm <- lm(s~i)
abline(coef = capm.lm$coe, co="red", lwd=2)