R正则表达式中的负先行查找

Question

我有一个文本位置：

locality <- "NEAR HAENA BEACH PARK, N 22 13 W 159 34 500 meters from coast"

我只想从这个字符串中提取坐标，不包括"500“，因为它不是坐标的一部分；它指的是离海岸的距离。

我已经编写了这个更通用的regex命令来提取具有类似模式的坐标，在最后我有一个负的前视，这样实际上是距离的数字就不包括在内。

这是可行的：

> capture <- gregexpr("([0-9]*\\.?[0-9]+)?(\\$O)?(\\s)?[0-9]*\\.?[0-9]+(\\')?(\\$O)?\\s([0-9]*\\.?[0-9]+(\\')?)(?!\\sMI\\b|KM\\b|M\\b|MILES|KILOMETERS|METERS)", locality, ignore.case = TRUE, perl = TRUE)
> regmatches(locality, capture)
[[1]]
[1] " 22 13"  " 159 34"

但是我遗漏了北边和西边的参考文献。如果我希望拾取这些N和W字母，这将不再正常工作：

> capture <- gregexpr("(N(\\s|\\b)|S(\\s|\\b)|E(\\s|\\b)|W(\\s|\\b))([0-9]*\\.?[0-9]+)?(\\$O)?(\\s)?[0-9]*\\.?[0-9]+(\\')?(\\$O)?\\s([0-9]*\\.?[0-9]+(\\')?)(?!\\sMI\\b|KM\\b|M\\b|MILES|KILOMETERS|METERS)", locality, ignore.case = TRUE, perl = TRUE)
> regmatches(locality, capture)
[[1]]
[1] "N 22 13"      "W 159 34 500"

换句话说，通过在正则表达式的开头只添加(N(\\s|\\b)|S(\\s|\\b)|E(\\s|\\b)|W(\\s|\\b))，前视就不再起作用。我认为先行只适用于紧接在它之前的圆括号中的片段。

为了扩展这一点，我从评论中吸收了一个建议，并包括了这个位置的许多变体，我希望这个正则表达式能够处理这些建议。

locality <- c(
            "NEAR HAENA BEACH PARK, N 22 13 W 159 34 500 meters from coast",
            "NEAR HAENA BEACH PARK, N 22 13 45 W 159 34 23 500 meters from coast",
            "NEAR HAENA BEACH PARK, N 22 13 12.32 W 159 34  500.4 meters from coast",
            "NEAR HAENA BEACH PARK, E 22 13 S 159 34 500 meters from coast",
            "NEAR HAENA BEACH PARK, N 22 13' W 159 34' 500 meters from coast",
            "NEAR HAENA BEACH PARK, N 22 13' W 159 34' 500 km from coast"
            "NEAR HAENA BEACH PARK, N 22 13' W 159 34' 500 distance from coast"
            )


regex <- "[NSEW]\\b([0-9]*\\.?[0-9]+)?(\\$O)?(\\s)?[0-9]*\\.?[0-9]+(\\')?(\\$O)?\\s([0-9]*\\.?[0-9]+(\\')?)(?!\\sMI\\b|KM\\b|M\\b|MILES|KILOMETERS|METERS)"

> capture <- gregexpr(regex, locality[1], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[1], capture)
[[1]]
[1] "N 22 13"  "W 159 34"

> 
> capture <- gregexpr(regex, locality[2], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[2], capture)
[[1]]
[1] "N 22 13"  "W 159 34"

> 
> capture <- gregexpr(regex, locality[3], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[3], capture)
[[1]]
[1] "N 22 13"  "W 159 34"

> 
> capture <- gregexpr(regex, locality[4], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[4], capture)
[[1]]
[1] "E 22 13"  "S 159 34"

> 
> capture <- gregexpr(regex, locality[5], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[5], capture)
[[1]]
[1] "N 22 13'"  "W 159 34'"

> 
> capture <- gregexpr(regex, locality[6], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[6], capture)
[[1]]
[1] "N 22 13'"  "W 159 34'"

> 
> capture <- gregexpr(regex, locality[7], ignore.case = TRUE, perl = TRUE)
> regmatches(locality[7], capture)
[[1]]
[1] "N 22 13'"  "W 159 34'"

看起来有几个方面不起作用。在第二个位置，秒数没有被接收到。此外，先行不应该影响最后一个位置，但它确实影响了(但这可能与秒数的问题相同)。

Answer 1

也许你可以试试这个

location_N <- which(strsplit(locality, "\\s")[[1]]=="N")
stringr::word(locality,location_N,location_N+5)