r - 在 R 中,将杂乱的数据抓取并组织到数据框中
问题描述
我们正在尝试收集有关大学篮球教练的一般信息。这是我试图抓取的两个示例页面:
- https://gobonnies.sbu.edu/sports/m-baskbl/coaches/index
- https://uofoathletics.com/sports/wbkb/coaches/index
我们的理想输出是:
data.frame(
name = c('Mark Schmidt', 'Sean Neal', 'Matt Pappano', 'Steve Curran', 'Tray Woodall', NA, 'Dominique Broadus'),
title = c("Head Men's Basketball Coach", "Assistant Men's Basketball Coach", "Director Of Basketball Operations", "Associate Head Coach, Men's Basketball", "Assistant Men's Basketball Coach", "Head Women's Basketball Coach", "Assistant Women's Basketball Coach"),
email = c(NA, 'sneal@sbu.edu', 'mpappano@sbu.edu', 'scurran@sbu.edu', 'twoodall@sbu.edu', NA, 'dbroadus@ozarks.edu'),
phone = c('716-375-2207', '716-375-2257', '716-375-2218', '716-375-2258', '716-375-2259', '479-979-1325', '479-979-1325'),
stringsAsFactors = FALSE
)
name title email phone
1 Mark Schmidt Head Men's Basketball Coach <NA> 716-375-2207
2 Sean Neal Assistant Men's Basketball Coach sneal@sbu.edu 716-375-2257
3 Matt Pappano Director Of Basketball Operations mpappano@sbu.edu 716-375-2218
4 Steve Curran Associate Head Coach, Men's Basketball scurran@sbu.edu 716-375-2258
5 Tray Woodall Assistant Men's Basketball Coach twoodall@sbu.edu 716-375-2259
6 <NA> Head Women's Basketball Coach <NA> 479-979-1325
7 Dominique Broadus Assistant Women's Basketball Coach dbroadus@ozarks.edu 479-979-1325
这导致我们的问题有几个原因:
- 在这两个页面上,数据都没有保存在表格中,而是
divs
为每个人单独保存。 - 有一些缺失的数据。有 2 封电子邮件丢失和一个姓名丢失。
到目前为止,这是我们得到的:
# go to pages, grab person bios
page1 <- 'https://gobonnies.sbu.edu/sports/m-baskbl/coaches/index' %>% read_html()
page1_bios <- page1 %>% html_nodes('div.coach-bios .coach-bio .info')
page2 <- 'https://uofoathletics.com/sports/wbkb/coaches/index' %>% read_html()
page2_bios <- page2 %>% html_nodes('div.coach-bios .coach-bio .info')
# turn bios into 1-column dataframes (not really what we need)
page1_list <- lapply(page1_bios, function(x) paste(x %>% html_children() %>% html_text(), collapse = " "))
page1_bios_df <- unlist(page1_list) %>% as.data.frame()
page2_list <- lapply(page2_bios, function(x) paste(x %>% html_children() %>% html_text(), collapse = " "))
page2_bios_df <- unlist(page2_list) %>% as.data.frame()
我们并不是那么接近,事实上我们也不确定这是否可能做到。我认为即使列名错误,我们也需要首先将数据放入数据框中,然后检查列的内容(例如,查找电子邮件的@符号,查找电话号码的#s,查找“教练”一词标题等)尝试正确命名它们。
解决方案
实现所需结果的一种选择可能如下所示。基本上,我的方法使用特定的 CSS 选择器逐个提取所需的信息:
library(rvest)
library(magrittr)
# go to pages, grab person bios
page1 <- 'https://gobonnies.sbu.edu/sports/m-baskbl/coaches/index' %>% read_html()
page1_bios <- page1 %>% html_nodes('div.coach-bios .coach-bio .info')
page2 <- 'https://uofoathletics.com/sports/wbkb/coaches/index' %>% read_html()
page2_bios <- page2 %>% html_nodes('div.coach-bios .coach-bio .info')
get_bios <- function(x) {
data.frame(
name = x %>% html_node("span.name") %>% html_text(),
title = x %>% html_node("p:nth-of-type(2)") %>% html_text(),
email = x %>% html_node("p.email a") %>% html_attr("href"),
phone = x %>% html_node("p:last-of-type") %>% html_text()
)
}
# turn bios into 1-column dataframes (not really what we need)
page1_list <- lapply(page1_bios, get_bios)
page2_list <- lapply(page2_bios, get_bios)
bios_df <- do.call("rbind", c(page1_list, page2_list))
bios_df$email <- gsub("^mailto:(.*)$", "\\1", bios_df$email)
bios_df$phone <- gsub("^Phone:\\s(.*)$", "\\1", bios_df$phone)
bios_df
#> name title email
#> 1 Mark Schmidt Head Men's Basketball Coach <NA>
#> 2 Steve Curran Associate Head Coach, Men's Basketball scurran@sbu.edu
#> 3 Sean Neal Assistant Men's Basketball Coach sneal@sbu.edu
#> 4 Tray Woodall Assistant Men's Basketball Coach twoodall@sbu.edu
#> 5 Matt Pappano Director Of Basketball Operations mpappano@sbu.edu
#> 6 Head Women's Basketball Coach <NA>
#> 7 Dominique Broadus Assistant Women's Basketball Coach dbroadus@ozarks.edu
#> phone
#> 1 716-375-2207
#> 2 716-375-2258
#> 3 716-375-2257
#> 4 716-375-2259
#> 5 716-375-2218
#> 6 479-979-1325
#> 7 479-979-1325
推荐阅读
- c++ - 我收到此代码的分段错误?
- typescript - Typescript、Babel 7、Karma、Jasmine、代码覆盖率配置
- ios - 选中时 Tabman 图像色调颜色
- python - 使用python为每行的唯一代码创建新列
- algorithm - 在 goroutine 中设置广播和多接收
- ajax - 如何使用ajax从数据库中获取最后一条记录
- python - Django在用户模型中显示多对多关系
- c++ - 用数据填充 QVector 的最有效方法是什么?
- vue.js - Facebook Oauth2 无法手动打开同意对话框 Vuejs
- android - 如何在 Android 中创建水平和垂直滚动的列表视图?