如何从非结构化文本中提取某些项目？

Question

I have an extremely unstructured data frame (df) in R, which includes a text column.

An example of the df$text looks like this

John Smith 3.8 GPA johnsmith@gmail.com, https://link.com

I am trying to extract the GPA out of the field and save to a new column called df$GPA but am unable to get it to work.

I have tried:

df$gpa <- sub('[0-9].[0-9] GPA',"\\1", df$text)

But that returns the whole block of text.

I am also trying to extract the url but am unsure how to do that as well.Does anybody have any suggestions?

Answer 1

Here's a solution using positive lookahead in (?=GPA) and str_extract from the package stringr :

df$GPA <- str_extract(df$text, "\\d+\\.\\d+\\s(?=GPA)")

A sub solution with backreference would be this:

df$GPA <- sub(".*(\\d+\\.\\d+).*", "\\1", df$text)

Result:

df
                                                      text GPA
1 John Smith 3.8 GPA johnsmith@gmail.com, https://link.com 3.8

Data:

df <- data.frame(text = "John Smith 3.8 GPA johnsmith@gmail.com, https://link.com")

Answer 2

We can use a regex lookaround to extract the numeric part

library(stringr)
df$GPA <- str_extract(df$text, "[0-9.]+(?=\\s*GPA)")
df$GPA
#[1] "3.8"

---

Or in base R with regmatches/regexpr

regmatches(df$text, regexpr("[0-9.]+(?=\\s*GPA)", df$text, perl = TRUE))

data

df <- data.frame(text = 'John Smith 3.8 GPA johnsmith@gmail.com, https://link.com', stringsAsFactors = FALSE)