Parse Claritin json string

badJSON.R

library(jsonlite)

url <- "http://www.claritin.com/webservice/allergyforecast.php?zip=33956"

df <- fromJSON(url)

url <- "http://www.claritin.com/webservice/allergyforecast.php?zip=33956"
res <- readLines(url)
res <- gsub('^"|"%', "", gsub("\\\\", "", res))
str(jsonlite::fromJSON(res))

'data.frame': 1 obs. of  3 variables:
 $ pollenForecast :'data.frame':  1 obs. of  6 variables:
  ..$ zip      : chr "33956"
  ..$ city     : chr "SAINT JAMES CITY"
  ..$ state    : chr " FL"
  ..$ forecast :List of 1
  .. ..$ : num  3.5 3.4 5.4 4.5
  ..$ pp       : chr "Ragweed, Grass and Chenopods."
  ..$ timestamp: chr "8/11/2015 12:00:00 AM"
 $ weatherForecast:'data.frame':  1 obs. of  5 variables:
  ..$ date    : chr "2015-08-11T07:00:00-0400"
  ..$ city    : chr "SAINT JAMES CITY"
  ..$ state   : chr " FL"
  ..$ zip     : chr "33956"
  ..$ forecast:List of 1
  .. ..$ :'data.frame': 6 obs. of  9 variables:
  .. .. ..$ lowF       : chr  "79" "79" "77" "77" ...
  .. .. ..$ highF      : chr  "94" "89" "88" "88" ...
  .. .. ..$ iconNight  : chr  "29" "29" "29" "33" ...
  .. .. ..$ skyDay     : int  0 30 30 38 38 38
  .. .. ..$ skyNight   : int  29 29 29 33 47 47
  .. .. ..$ phraseNight: chr  "Partly Cloudy" "Partly Cloudy" "Partly Cloudy" "Mostly Clear" ...
  .. .. ..$ date       : chr  "2015-08-11T07:00:00-0400" "2015-08-12T07:00:00-0400" "2015-08-13T07:00:00-0400" "2015-08-14T07:00:00-0400" ...
  .. .. ..$ iconDay    : chr  NA "30" "30" "38" ...
  .. .. ..$ phraseDay  : chr  NA "Partly Cloudy" "Partly Cloudy" "AM Thunderstorms" ...
 $ result         : logi TRUE

Code:

library(httr)
library(jsonlite)

url <- "http://www.claritin.com/webservice/allergyforecast.php?zip=33956"

foo <- GET(url)
c <- content(foo, as="text")
c <- substring(c, 4)
json <- fromJSON(fromJSON(c))
str(json)

Output:

'data.frame':   1 obs. of  3 variables:
 $ pollenForecast :'data.frame':        1 obs. of  6 variables:
  ..$ zip      : chr "33956"
  ..$ city     : chr "SAINT JAMES CITY"
  ..$ state    : chr " FL"
  ..$ forecast :List of 1
  .. ..$ : num  3.5 3.4 5.4 4.5
  ..$ pp       : chr "Ragweed, Grass and Chenopods."
  ..$ timestamp: chr "8/11/2015 12:00:00 AM"
 $ weatherForecast:'data.frame':        1 obs. of  5 variables:
  ..$ date    : chr "2015-08-11T07:00:00-0400"
  ..$ city    : chr "SAINT JAMES CITY"
  ..$ state   : chr " FL"
  ..$ zip     : chr "33956"
  ..$ forecast:List of 1
  .. ..$ :'data.frame': 6 obs. of  9 variables:
  .. .. ..$ lowF       : chr  "79" "79" "77" "77" ...
  .. .. ..$ highF      : chr  "93" "89" "88" "88" ...
  .. .. ..$ iconNight  : chr  "29" "27" "29" "31" ...
  .. .. ..$ skyDay     : int  0 30 30 38 38 38
  .. .. ..$ skyNight   : int  29 27 29 31 47 47
  .. .. ..$ phraseNight: chr  "Partly Cloudy" "Mostly Cloudy" "Partly Cloudy" "Clear" ...
  .. .. ..$ date       : chr  "2015-08-11T07:00:00-0400" "2015-08-12T07:00:00-0400" "2015-08-13T07:00:00-0400" "2015-08-14T07:00:00-0400" ...
  .. .. ..$ iconDay    : chr  NA "30" "30" "38" ...
  .. .. ..$ phraseDay  : chr  NA "Partly Cloudy" "Partly Cloudy" "AM Thunderstorms" ...
 $ result         : logi TRUE

those extraneous 4 bytes show up alot in httr-json & xml responses. i need to poke at the httr/curl code more to see if i can narrow that down.

In case anyone else comes late to this thread. Jeroen Ooms explains that this is a UTF-8 byte order mark (BOM). The development version of jsonlite on github removes this mark automatically, supplies a warning, and then continues to parse the json. Will be version 0.9.17 once it's on CRAN.