Hashtag Extract function in R Programming

Question

I am trying to create an hashtag extraction function in R. This function will extract a hashtags from a post, if there are any, else will give a blank. My function is like

hashtag_extract= function(text){
              match = str_extract_all(text,"#\\S+")
              if (match) { 
                 return match
                 }else{
               return ''}}
String="#letsdoit #Tonewbeginnign world is on a new#route

But my function is not working, showing me tons of errors.like 1st error is

Error: unexpected symbol in:
      "  if (match) { 
     return match"

so I want to apply it as

hashatag_extract(string)

and answer should come like

#letsdoit  ##Tonewbeginnign   #route

And eventually I will use sapply to apply this function on whole column, that's why the If part is important. Please ignore my indentation for R, since its not important for R, but every suggestion will be helpful

What exactly is the question (relevant here at SO)? Does the given function not work, or not in all cases, or is it missing "features" ? please add to the question. Thanks. — Dilettant, Aug 05 '16 at 11:48
Please read `?return`: R calls it like a function `return(match)`, with parens. — r2evans, Aug 05 '16 at 11:58
@r2evans- I removed the return from my function, now it is showing me error - "Error in if (match) { : argument is not interpretable as logical" — Manu Sharma, Aug 05 '16 at 12:33
The output from `str_extract_all` is not logical. It doesn't make sense to use it as `if(match)`. Change it to `if (length(match) > 0)` and your example seems to work. — cory, Aug 05 '16 at 14:24

score 11 · Answer 1 · answered Aug 05 '16 at 12:21

Hashtag regexes aren't that simple
I'm not sure you understand the commonly accepted "rules" for hashtags
I do not believe str_extract_all() is returning what you think it is
Just use stringi which stringr functions are built on top of
Folks rly need to stop analyzing tweets

This should handle most, if not all, cases:

get_tags <- function(x) {
  # via http://stackoverflow.com/a/5768660/1457051
  twitter_hashtag_regex <- "(^|[^&\\p{L}\\p{M}\\p{Nd}_\u200c\u200d\ua67e\u05be\u05f3\u05f4\u309b\u309c\u30a0\u30fb\u3003\u0f0b\u0f0c\u00b7])(#|\uFF03)(?!\uFE0F|\u20E3)([\\p{L}\\p{M}\\p{Nd}_\u200c\u200d\ua67e\u05be\u05f3\u05f4\u309b\u309c\u30a0\u30fb\u3003\u0f0b\u0f0c\u00b7]*[\\p{L}\\p{M}][\\p{L}\\p{M}\\p{Nd}_\u200c\u200d\ua67e\u05be\u05f3\u05f4\u309b\u309c\u30a0\u30fb\u3003\u0f0b\u0f0c\u00b7]*)"
  stringi::stri_match_all_regex(x, hashtag_regex) %>% 
    purrr::map(~.[,4]) %>% 
    purrr::flatten_chr()

}

tests <- c("#teste_teste      //underscore accepted",
           "#teste-teste      //Hyphen not accepted",
           "#leof_gfg.sdfsd   //dot not accepted",
           "#f34234@45#6fgh6  // @ not accepted",
           "#leo#leo2#asd     //followed hastag without space ",
           "#6663             // only number accepted",
           "_#asd_            // hashtag can't start or finish with underscore",
           "-#sdfsdf-         // hashtag can't start or finish with hyphen",
           ".#sdfsdf.         // hashtag can't start or finish with dot",
           "#leo_leo__leo__leo____leo // decline followed underline")


get_tags(tests)
##  [1] "teste_teste"              "teste"                   
##  [3] "leof_gfg"                 "f34234"                  
##  [5] "leo"                      NA                        
##  [7] NA                         "sdfsdf"                  
##  [9] "sdfsdf"                   "leo_leo__leo__leo____leo"

your_string <- "#letsdoit #Tonewbeginnign world is on a new#route"

get_tags(your_string)
## [1] "letsdoit"       "Tonewbeginnign"

You'll need to tweak the function if you need each set of hashtags to be grouped with each input vector but you really didn't provide much detail on what you're really trying to accomplish.

Thank you so much for this. Though, str_extract_all is working really nice for me, Still, pls help me with my function — Manu Sharma, Aug 05 '16 at 12:28
you should have everything you need with this answer. i'll gladly delete it if it doesn't meet your requirements. — hrbrmstr, Aug 05 '16 at 12:42
nice workflow of `map` and `flatten_chr`. I thought the `.x` instead of the placeholder was mandatory such as `map(~.x[,4])`. Good to know — aurelien, Aug 05 '16 at 13:54

score 1 · Answer 2 · answered Aug 06 '16 at 07:43

@manu sharma I would say you need not apply if else inside. Let the non-matching rows take values as 'NA'. And after applying the function you change it to blank. Hope my code helps you:

   aaa <- readLines("C:\\MY_FOLDER\\NOI\\file2sample.txt")
 ttt <- function(x){

  r <- sapply(x, function(x) { matches <- str_match(x,"#\\w+\\s+")})
  r


  }

 y <-ttt(aaa)
 y[is.na(y)]=''

score 0 · Answer 3 · answered Aug 08 '16 at 09:31

0

Thanks everyone for all the help, I got it worked somehow, thought it is almost similar as Shalini's answer 1.replacing all NAs on message

message[is.na(message)]='abc'

2.function for extracting the Hashtags

hashtag_extrac= function(text){
match = str_extract_all(text,"#\\S+")
if (match!= "") { 
match
} else {
'' }}

applying function on whole column

hashtags= sapply(message, hashtag_extrac)

answered Aug 08 '16 at 09:31

Manu Sharma

1,593
4
25
48

Why is that if statement there? It doesn't do anything... If it's not blank, then do nothing. If it is blank, make it blank. I'm baffled why you don't use the much higher quality answer above. – cory Aug 08 '16 at 12:29
Thank you so much! But I will request, keep calm, even in scripts, we have our own cases and uses, which some of the times we can't explain in a que and certainly they are better answers – Manu Sharma Aug 08 '16 at 13:45
So you accept @Shalini's answer - I understand it that way, or did I misread it? – Dilettant Aug 09 '16 at 05:32

Hashtag Extract function in R Programming

3 Answers3

applying function on whole column

Linked

Related