0

我目前正在尝试使用包 doParallel 来并行化我的 RSelenium 网络爬虫(在 Docker 上运行)。我找到了这篇文章(Speed up web scraping using multiplie Rselenium browsers),并在此处复制@hdharrison 提供的答案:

library(RSelenium)
library(rvest)
library(magrittr)
library(foreach)
library(doParallel)

# using  docker run -d -p 4445:4444 selenium/standalone-chrome:3.5.3
# in windows
URLsPar <- c("https://stackoverflow.com/", "https://github.com/", 
             "http://www.bbc.com/", "http://www.google.com", 
             "https://www.r-project.org/", "https://cran.r-project.org",
             "https://twitter.com/", "https://www.facebook.com/")

appHTML <- c()

(cl <- (detectCores() - 1) %>%  makeCluster) %>% registerDoParallel
# open a remoteDriver for each node on the cluster
clusterEvalQ(cl, {
  library(RSelenium)
  remDr <- remoteDriver(remoteServerAddr = "192.168.99.100", port = 4445L, 
                        browserName = "chrome")
  remDr$open()
})
ws <- foreach(x = 1:length(URLsPar), 
              .packages = c("rvest", "magrittr", "RSelenium"))  %dopar%  {
                print(URLsPar[x])
                remDr$navigate(URLsPar[x])
                remDr$getTitle()[[1]]
              }
> ws
[[1]]
[1] "Stack Overflow - Where Developers Learn, Share, & Build Careers"

[[2]]
[1] "The world's leading software development platform · GitHub"

[[3]]
[1] "BBC - Homepage"

[[4]]
[1] "Google"

[[5]]
[1] "R: The R Project for Statistical Computing"

[[6]]
[1] "The Comprehensive R Archive Network"

[[7]]
[1] "Twitter. It's what's happening."

[[8]]
[1] "Facebook - Log In or Sign Up"     


# close browser on each node
clusterEvalQ(cl, {
  remDr$close()
})

stopImplicitCluster()

这似乎是我正在寻找的解决方案,但是当我运行它时,我遇到了以下错误消息:

checkForRemoteErrors(lapply(cl, recvResult)) 中的错误:

3个节点产生错误;第一个错误:httr 调用中的未定义错误。httr 输出:无法连接到 192.168.99.100 端口 4445:连接被拒绝

这是“docker ps”输出:

CONTAINER ID        IMAGE                              COMMAND                  CREATED             STATUS              PORTS                    NAMES
f2d62f6b293b        selenium/standalone-chrome:3.5.3   "/opt/bin/entry_poin…&quot;   36 minutes ago      Up 35 minutes       0.0.0.0:4445->4444/tcp   recursing_austin

我知道我必须为每个内核打开一个新的浏览器,但我认为这就是问题所在:当我减少内核时,产生的错误就会减少。

如果我可以提供更多详细信息,请告诉我!提前致谢!

4

1 回答 1

1

与此同时,我能够弄清楚如何解决我的错误。如果其他人面临同样的问题,我会在这里发表评论。我无法解释它背后的逻辑,但是当我替换时我的代码按预期运行

remDr <- remoteDriver(remoteServerAddr = "192.168.99.100", port = 4445L, 
                       browserName = "chrome")

经过

remDr <- remoteDriver(port = 4445L) 

并使用 Firefox 浏览器而不是 Chrome。

于 2019-02-11T09:37:13.770 回答