regex - R 正则表达式提取文本不立即遵循模式

Question

我有一个数据框，其中有一列包含半结构化文本。我正在寻找某些信息，但它们有一个我不知道如何利用的模式。

head(events5)
     flight_history_id         date_time_recorded              event
862          280604281 2012-11-13 17:55:58.673-08 Arrival Estimation
5514         280316415 2012-11-12 17:25:08.814-08    Time Adjustment
5527         280315758 2012-11-12 16:10:06.968-08    Time Adjustment
5539         280316495 2012-11-12 15:38:10.099-08    Time Adjustment
5545         280303866 2012-11-13 06:51:45.831-08    Time Adjustment
5558         280303866 2012-11-12 05:50:23.547-08    Time Adjustment
                                                                               data_updated
862                                                     EGA- Based on Distance and Airspeed
5514                        EGD- New=11/13/12 20:20, DGATE- New=A1, EGA- New=11/13/12 21:00
5527         EGD- New=11/13/12 19:05, DGATE- New=A7, EGA- New=11/13/12 20:50, AGATE- New=C5
5539                        EGD- New=11/13/12 06:35, DGATE- New=A1, EGA- New=11/13/12 07:15
5545 AGD- Old=11/13/12 07:40 New=11/13/12 07:38, EGA- Old=11/13/12 10:25 New=11/13/12 10:18
5558         EGD- New=11/13/12 07:45, DGATE- New=11, EGA- New=11/13/12 10:25, AGATE- New=A1

这是输出：

structure(list(flight_history_id = c(280604281L, 280316415L, 
280315758L, 280316495L, 280303866L, 280303866L, 280271465L, 280271465L, 
280271465L, 280271465L), date_time_recorded = structure(c(10L, 
4L, 3L, 2L, 7L, 1L, 9L, 8L, 6L, 5L), .Label = c("2012-11-12 05:50:23.547-08", 
"2012-11-12 15:38:10.099-08", "2012-11-12 16:10:06.968-08", "2012-11-12 17:25:08.814-08", 
"2012-11-12 17:58:14.268-08", "2012-11-13 05:16:41.01-08", "2012-11-13 06:51:45.831-08", 
"2012-11-13 07:38:19.593-08", "2012-11-13 07:54:22.588-08", "2012-11-13 17:55:58.673-08"
), class = "factor"), event = structure(c(1L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L), .Label = c("Arrival Estimation", "Time Adjustment"
), class = "factor"), data_updated = structure(c(3L, 8L, 7L, 
4L, 2L, 5L, 1L, 10L, 9L, 6L), .Label = c("AGD- New=11/13/12 10:22, EGA- Old=11/13/12 11:20 New=11/13/12 11:09", 
"AGD- Old=11/13/12 07:40 New=11/13/12 07:38, EGA- Old=11/13/12 10:25 New=11/13/12 10:18", 
"EGA- Based on Distance and Airspeed", "EGD- New=11/13/12 06:35, DGATE- New=A1, EGA- New=11/13/12 07:15", 
"EGD- New=11/13/12 07:45, DGATE- New=11, EGA- New=11/13/12 10:25, AGATE- New=A1", 
"EGD- New=11/13/12 08:55, EGA- New=11/13/12 09:45, AGATE- New=A1", 
"EGD- New=11/13/12 19:05, DGATE- New=A7, EGA- New=11/13/12 20:50, AGATE- New=C5", 
"EGD- New=11/13/12 20:20, DGATE- New=A1, EGA- New=11/13/12 21:00", 
"EGD- Old=11/13/12 08:55 New=11/13/12 10:05, EGA- Old=11/13/12 09:45 New=11/13/12 10:55", 
"EGD- Old=11/13/12 10:05 New=11/13/12 10:30, EGA- Old=11/13/12 10:55 New=11/13/12 11:20"
), class = "factor")), .Names = c("flight_history_id", "date_time_recorded", 
"event", "data_updated"), row.names = c(862L, 5514L, 5527L, 5539L, 
5545L, 5558L, 5564L, 5566L, 5570L, 5572L), class = "data.frame")

我正在尝试从 column'New='开始提取部分的字段部分。问题是有时两者之间有一个“Old=”字段。我查看了使用 stringr::str_split 并将列转换为列表，但仍然有同样的问题。我认为这是一个正则表达式/gsub 问题，但不知道具体如何实现。'EGA''data_updated'

有什么建议么？

编辑：我希望最终产品是具有 EGA/New 值的向量/列。所以是这样的：

NA
11/13/12 21:00
11/13/12 20:50
11/13/12 07:15
11/13/12 10:18
11/13/12 10:25

Answer 1

那个怎么样？

EGA-\s*(?:Old=\d{2}\/\d{2}\/\d{2}\s+\d{2}:\d{2}\s*)?New=(\d{2}\/\d{2}\/\d{2}\s+\d{2}:\d{2})

Answer 2

EGA这是一个替代方法，与@Pharaoh 的正则表达式相比，在-tag之后使用无限量的文本

/EGA-(?:(?!New=|").)+(New=[:\d\/ ]+)/g

如果"否定前瞻中的 太笼统，您可以将其更改为EGD.

演示+说明： http ://regex101.com/r/qY4uA0

如果不需要，请删除标志和分隔符。

Answer 3

与@BlueMagister 的概念类似，但没有那么多嵌套：

gsub(".*EGA-.*New=([0-9/]{8} [0-9:]{5}).*|.*EGA.*", "\\1", events5$data_updated)
# [1] ""               "11/13/12 21:00" "11/13/12 20:50" "11/13/12 07:15"
# [5] "11/13/12 10:18" "11/13/12 10:25" "11/13/12 11:09" "11/13/12 11:20"
# [9] "11/13/12 10:55" "11/13/12 09:45"

基本上，第一部分gsub说寻找“EGA-”，它不一定出现在字符串的开头，并在遇到“New =”时开始捕获。捕获模式([0-9/]{8} [0-9:]{5})表示您首先需要一个带有数字和斜线的 8 个字符长的字符串，然后是一个空格，然后是一个带有数字和冒号的 5 个字符长的字符串。然后这个|字符用来表示在它之前或之后寻找 EGA。

要向您添加新信息，您data.frame可以使用以下内容：

within(events5, {
  EGA.New <- gsub(".*EGA-.*New=([0-9/]{8} [0-9:]{5}).*|.*EGA-.*", 
                  "\\1", events5$data_updated)
})

Answer 4

这是一个从头到尾的解决方案：

##formats of the date-times
date.format <- "\\d{2}/\\d{2}/\\d{2} \\d{2}:\\d{2}"
##puts this together into a regex
regex.string <- paste0("EGA- (?:Old=",date.format," )?New=(",date.format,")")

##if pattern matches, get the part of "new" we are looking for
##otherwise NA
ifelse(grepl(regex.string,events5$data_updated),
  gsub(paste0("^.*",regex.string,".*$"),"\\1",events5$data_updated),
  NA_character_)