从和讯读取财务数据

getHexunFinaceData.r

# getHexunFinaceData.r 
# 从和讯读取财务数据 
# version: 0.93 
# 调用语法：
#    getHexunFinaceData(stockid, start, end, type)
# 参数含义：
#   stockid  股票代码
#   start    起始年份
#   end      结束年份
#   type     财务报告类别，默认为"b"，表示资产负债表；"i"表示利润表；"c"表示现金流量表
# 例子： test <- getHexunFinaceData(600028,2001,2002,"b")

getHexunFinaceData <- function(stockid, 
                                start = 1991, 
                                end = as.numeric(substr(Sys.time(), 1, 4)), 
                                type = "b") {
  if (!is.vector(stockid)) {
    stop("参数stockid必须是向量形式！")
  }
  if (start > end) {
    stop("起始或结束年份输入有误！")
  }
  if (!type %in% c("b", "c", "i")) {
    stop("错误的财务报告类型！请检查参数type...")
  }
  require(stringr)
  stockid <- as.character(stockid)
  vnameOld <- vector()
  value.mat <- vector()
  total <- length(stockid) * (end - start + 1) * 4
  # create progress bar
  pb <- txtProgressBar(min = 0, max = total, style = 3)
  i <- 1
  for (stock in stockid) {
    while (nchar(stock) < 6) {
      stock <- paste("0", stock, sep = "")
    }
    if (nchar(stock) > 6) {
      warning(paste("invalid stock code: ", stock, sep = ""))
      next
    }
    for (year in start:end) {
      accountdates <- paste(year, c(".03.15", ".06.30", ".09.30", ".12.31"), sep = "")
      for (term in accountdates) {
        if (type == "b") {
          address <- "http://stockdata.stock.hexun.com/2008/zcfz.aspx?stockid="
        } else if (type == "i") {
          address <- "http://stockdata.stock.hexun.com/2008/lr.aspx?stockid="
        } else if (type == "c") {
          address <- "http://stockdata.stock.hexun.com/2008/xjll.aspx?stockid="
        }
        url <- paste(address, stock, "&accountdate=", term, sep = "")
        txt <- readLines(url)
        if (sessionInfo()$R.version$os == "linux-gnu") {
          txt <- iconv(txt, from = "gb2312", to = "UTF-8")
        }
        linenum <- grep("<span id=\"ControlEx1_lbl\">", txt)
        line <- txt[linenum]
        if (length(line) == 0 | line == "\t\t\t\t\t<span id=\"ControlEx1_lbl\"></span>") {
          setTxtProgressBar(pb, i)
          i <- i + 1
          next
        } else {
          vname.start <- gregexpr("<td class='dotborder' width='45%'><div class='tishi'><strong>", line)[[1]] + 61
          vname.end <- gregexpr("</strong></div></td><td>", line)[[1]] - 1
          vname <- str_sub(line, vname.start, vname.end)
          vname <- c("股票代码", vname)
          if (is.null(vnameOld) | all(vname == vnameOld)) {
            value.start <- gregexpr("<td><div class='tishi'>", line)[[1]] + 23
            value.end <- gregexpr("</div></td><tr>", line)[[1]] - 1
            value <- str_sub(line, value.start, value.end)
            value <- gsub(",", "", value)
            value <- c(stock, value)
            value.mat <- rbind(value.mat, value)
            vnameOld <- vname
            setTxtProgressBar(pb, i)
            i <- i + 1
          } else {
            msg <- paste("看来股票", stock, "的数据格式在", year, "年发生了变化！", sep = "")
            stop(msg)
          }
        }
      }
    }
  }
  close(pb)
  if (is.null(value.mat)) {
    return(NULL)
  } else {
    rownames(value.mat) <- NULL
    value.df <- as.data.frame(value.mat, stringsAsFactors = FALSE)
    value.df[value.df == "--"] <- NA
    value.df[, 3:(ncol(value.df) - 1)] <- apply(value.df[, 3:(ncol(value.df) - 1)], 2, as.numeric)
    colnames(value.df) <- vname
    return(value.df)
  }
} 

# 如果没有stringr包，就先安装该包
install.packages("stringr")

# 先加载程序 比如getHexunFinaceData.r放在 D盘下面的Rcode目录下
source("D:/Rcode/getHexunFinaceData.r")

# 函数调用
test1 <- getHexunFinaceData(600028,2001,2002) # 一只股票的资产负债表
test2 <- getHexunFinaceData(c(600028,600022),2011,2012) # 两只股票负债表

test3 <- getHexunFinaceData(c(600028,600022),2011,2012, "i") # 两只股票的利润表

# 如果要把读取的数据写到文件中
write.csv(test1,"D:/datafile.csv",row.names = FALSE)

2012年1月27日 增加了进度条

探测当前是哪种操作系统 还几种方式

if(.Platform$OS.type == "unix") {
   ...
}

或者

Sys.info()['sysname']

有时候和讯数据不够，对应的日期没有数据，希望对此做容错处理
是否用XML包里面的readHTMLtable更快？