Using gsub function with multiple criteria in R

Question

Folllow up on question Searching for unique values in dataframe and creating a table with them

Here is how my data looks like

    UUID    Source
1   Jane    http//mywebsite.com44bb00?utm_source=ADW&utm_medium=banner&utm_campaign=Monk&gclid1234
2   Mike    http//mywebsite.com44bb00?utm_source=Google&utm_medium=cpc&utm_campaign=DOG&gclid1234
3   John    http//mywebsite.com44bb00?utm_source=Yahoo&utm_medium=banner&utm_campaign=DOG&gclid1234
4   Sarah   http//mywebsite.com44bb00?utm_source=Facebookdw&utm_medium=cpc&utm_campaign=CAT&gclid1234
5   Michael http//mywebsite.com44bb00?utm_source=Twitter&utm_medium=GDNr&utm_campaign=CAT&gclid1234
6   Bob     http//mywebsite.com44bb00?utm_source=ADW&utm_medium=GDN&utm_campaign=DOG&gclid1234
7   Mark    http//mywebsite.com44bb00?utm_source=Twitter&utm_medium=banner&utm_campaign=MONK&gclid1234
8   Anna    http//mywebsite.com44bb00?utm_source=Facebook&utm_medium=banner&utm_campaign=MONK&gclid1234

And here is the desired output of what I am trying to achieve

    NAME    UTM_SOURCE  UTM_MEDIUM  UTM_CAMPAIGN
1   Jane    ADW             banner     Monk
2   Mike    Google          cpc        DOG
3   John    Yahoo           banner     DOG
4   Sarah   Faceboo         cpc        CAT
5   Michael Twitter         GDN        CAT
6   Bob     ADW             GDN        DOG
7   Mark    Twitter         banner     MONK
8   Anna    Facebook        banner     MONK

So in other words what I want is to obtain a specific piece of information based on a criteria. Example: search in the dataframe for the value "utmsource=" and when found, copy whatever information is found between the "=" and "&" signs. In the case of User no1 (Jame) if you look in the original file, her Source URL contains the value "utm_source=ADW". In the output file, the "ADW" bit is extracted and imputed in a new column named "utm_source". Same principle for all other users and other dimmensions (utm_medium & utm_campaign)

I understand that the function gsub can help me. Here is what I have tried so far:

> file1 <- read.csv("C:/Users/Dumitru Ostaciu/Desktop/Users.csv")
> file1 <- transform(file1, Source = as.character(Source))
> file2 <- gsub(".*\\?utm_source=", "", file1$Source)

And this is the result I got

  UUID  SOURCE
    1   ADW&utm_medium=banner&utm_campaign=Monk&gclid1234
    2   Google&utm_medium=cpc&utm_campaign=DOG&gclid1234
    3   Yahoo&utm_medium=banner&utm_campaign=DOG&gclid1234
    4   Facebookdw&utm_medium=cpc&utm_campaign=CAT&gclid1234
    5   Twitter&utm_medium=GDNr&utm_campaign=CAT&gclid1234
    6   ADW&utm_medium=GDN&utm_campaign=DOG&gclid1234
    7   Twitter&utm_medium=banner&utm_campaign=MONK&gclid1234
    8   Facebook&utm_medium=banner&utm_campaign=MONK&gclid1234

I have 2 questions about this:

1) In the output that I got, the function copied everything that followed the value "utm_source-" . How do I add another dimension to make the formula copy only what is between "=" and "&"

2) How do i keep the values that were initially in the first column (UUID) , Jane, Mike, John, etc?

You'd be better off using a tool specifically designed to work with urls, but the urls in your example are pretty mangled. — hadley, Nov 20 '12 at 14:15

Andrie · Accepted Answer · 2012-11-20T12:27:43.073

You need to do two things:

Use gsub to strip the website name from your Source
Use strsplit to separate the remaining string at each occurrence of ?

Read in the data:

x <- read.table(text="
UUID    Source
1   Jane    http//mywebsite.com44bb00?utm_source=ADW&utm_medium=banner&utm_campaign=Monk&gclid1234
2   Mike    http//mywebsite.com44bb00?utm_source=Google&utm_medium=cpc&utm_campaign=DOG&gclid1234
3   John    http//mywebsite.com44bb00?utm_source=Yahoo&utm_medium=banner&utm_campaign=DOG&gclid1234
4   Sarah   http//mywebsite.com44bb00?utm_source=Facebookdw&utm_medium=cpc&utm_campaign=CAT&gclid1234
5   Michael http//mywebsite.com44bb00?utm_source=Twitter&utm_medium=GDNr&utm_campaign=CAT&gclid1234
6   Bob     http//mywebsite.com44bb00?utm_source=ADW&utm_medium=GDN&utm_campaign=DOG&gclid1234
7   Mark    http//mywebsite.com44bb00?utm_source=Twitter&utm_medium=banner&utm_campaign=MONK&gclid1234
8   Anna    http//mywebsite.com44bb00?utm_source=Facebook&utm_medium=banner&utm_campaign=MONK&gclid1234", header=TRUE, stringsAsFactors=FALSE)

Use strsplit to separate the Source string at each ?:

z <- matrix(
  unlist(strsplit(gsub(".*\\?", "", x$Source), "\\&")), 
  ncol=4, byrow=TRUE)
cbind(x$UUID, gsub(".*=", "", z))

     [,1]      [,2]         [,3]     [,4]   [,5]       
[1,] "Jane"    "ADW"        "banner" "Monk" "gclid1234"
[2,] "Mike"    "Google"     "cpc"    "DOG"  "gclid1234"
[3,] "John"    "Yahoo"      "banner" "DOG"  "gclid1234"
[4,] "Sarah"   "Facebookdw" "cpc"    "CAT"  "gclid1234"
[5,] "Michael" "Twitter"    "GDNr"   "CAT"  "gclid1234"
[6,] "Bob"     "ADW"        "GDN"    "DOG"  "gclid1234"
[7,] "Mark"    "Twitter"    "banner" "MONK" "gclid1234"
[8,] "Anna"    "Facebook"   "banner" "MONK" "gclid1234"

And then convert to a data frame and add names:

z <- matrix(
  unlist(strsplit(gsub(".*\\?", "", x$Source), "\\&")), 
  ncol=4, byrow=TRUE)
z <- cbind(x$UUID, gsub(".*=", "", z))
z <- as.data.frame(z[, -5])
names(z) <- c("UUID", "UTM_SOURCE", "UTM_MEDIUM", "UTM_CAMPAIGN")
z

     UUID UTM_SOURCE UTM_MEDIUM UTM_CAMPAIGN
1    Jane        ADW     banner         Monk
2    Mike     Google        cpc          DOG
3    John      Yahoo     banner          DOG
4   Sarah Facebookdw        cpc          CAT
5 Michael    Twitter       GDNr          CAT
6     Bob        ADW        GDN          DOG
7    Mark    Twitter     banner         MONK
8    Anna   Facebook     banner         MONK

Up until the cbind it works fine, but when I do that, the result I get, instead of names is a series of numnbers. Does this has to do with transforming column UUID from the initial file to"as.character" as well? — David Rogers, Nov 20 '12 at 12:30

David Rogers · Answer 2 · 2012-11-20T12:54:45.207

Here is how I did it

> file1 <- read.csv("C:/Users/Dumitru Ostaciu/Desktop/Users.csv")
> file1 <- transform(file1, Source = as.character(Source))
> z <- matrix(
     unlist(strsplit(gsub(".*\\?", "", file1$Source), "\\&")), 
     ncol=4, byrow=TRUE)
> file2 <- cbind(file1$UUID, gsub(".*=", "", z))

This is the result that I got

    V1  V2          V3      V4      V5
1   3   ADW         banner  Monk    gclid1234
2   7   Google      cpc     DOG     gclid1234
3   4   Yahoo       banner  DOG     gclid1234
4   8   Facebookdw  cpc     CAT     gclid1234
5   6   Twitter     GDNr    CAT     gclid1234
6   2   ADW         GDN     DOG     gclid1234
7   5   Twitter     banner  MONK    gclid1234
8   1   Facebook    banner  MONK    gclid1234

I need to point out that my real data will have 500.000 rows and in the first column there will be a unique designation.

How do I fix this to have the names show up in V1? What was my mistake?

Using gsub function with multiple criteria in R

2 Answers2

Linked