Question

我正在尝试从USNO API中提取JSON月球周期数据。问题是我得到的数据中包含两个JSON数据数组。我看不到指定从天文台返回的内容的方法，因此我认为我需要在R中清理它。这是我的代码：

library(sqldf);
library(jsonlite);

curr_date <- Sys.Date();
Q_date <- format.Date(curr_date, "%m/%d/%Y");
moon_call <- paste0("http://api.usno.navy.mil/moon/phase?date=",Q_date,"&nump=4");

moon_json <- fromJSON(moon_call, simplifyDataFrame =  TRUE);

moon_phases <- do.call("rbind.fill", lapply(moon_json$phasedata, as.data.frame));

我得到的数据如下：

"","error","apiversion","year","month","day","numphases","datechanged","phasedata.phase","phasedata.date","phasedata.time"
"1",FALSE,"2.1.0",2018,8,29,4,FALSE,"Last Quarter","2018 Sep 03","02:37"
"2",FALSE,"2.1.0",2018,8,29,4,FALSE,"New Moon","2018 Sep 09","18:01"
"3",FALSE,"2.1.0",2018,8,29,4,FALSE,"First Quarter","2018 Sep 16","23:15"
"4",FALSE,"2.1.0",2018,8,29,4,FALSE,"Full Moon","2018 Sep 25","02:52"

将其转换为数据框时，我得到以下信息：

"","X[[i]]"
"1","Last Quarter"
"2","New Moon"
"3","First Quarter"
"4","Full Moon"
"5","2018 Sep 03"
"6","2018 Sep 09"
"7","2018 Sep 16"
"8","2018 Sep 25"
"9","02:37"
"10","18:01"
"11","23:15"
"12","02:52"

但是我想要的是一个选中了phasedata.phase/.date/.time列的数据框：

"","phase","date","time"
"1","Last Quarter","2018 Sep 03","02:37"
"2","New Moon","2018 Sep 09","18:01"
"3","First Quarter","2018 Sep 16","23:15"
"4","Full Moon","2018 Sep 25","02:52"

Answer 1

R允许您直接从数据帧moon_json中extract the three columns，就像您想要的那样：

moon_phases <- moon_json[, c('phasedata.phase', 'phasedata.date', 'phasedata.time')]

（do.call("rbind.fill", lapply(..., as.data.frame))不需要任何东西，这只是切片和级联的一种低效且受折磨的方式。）

然后，您要重命名df列，以放置phasedata.前缀：

names(moon_phases) <- c('phase', 'date', 'time')

或：names(moon_phases) <- gsub('^phasedata\.', '', names(moon_phases))

此外，您通常不希望像1,2,3...这样的数据框上使用显式行名moon_json，所以只需要row.names(moon_json) <- NULL或data.frame(..., row.names=NULL)或as.data.frame(..., row.names=NULL)

（{jsonlite（或其他R json软件包之一）应该具有执行此清理和自动重命名的选项，我不知道，我使用的不是很多，请检查并选择可以减轻刮花痛苦的包装。）

将JSON数据拉入R Dataframe

1 个答案: