Error parsing JSON file with the jsonlite package

ANOTHER UPDATE

You can use the ndjson package to process this ndjson/streaming JSON data. It’s faster than jsonlite::stream_in() and always produces a completely “flat” data frame:

system.time(bitly01 <- ndjson::stream_in("usagov_bitly_data2013-05-17-1368832207.gz"))
##    user  system elapsed 
##   0.146   0.004   0.154

system.time(bitly02 <- jsonlite::stream_in(file("usagov_bitly_data2013-05-17-1368832207.gz"), verbose=FALSE, pagesize=10000))
##    user  system elapsed 
##   0.419   0.008   0.427

If we examine the resultant data frame2, you’ll see ndjson expands ll into ll.0 and ll.1 where you get a list column in jsonlite that you have to deal with later.

ndjson:

dplyr::glimpse(bitly01)
## Observations: 3,959
## Variables: 19
## $ a           <chr> "Mozilla/5.0 (Linux; U; Android 4.1.2; en-us; HTC_PN071 Build/JZO54K) AppleWebKit/534.30 ...
## $ al          <chr> "en-US", "en-us", "en-US,en;q=0.5", "en-US", "en", "en-US", "en-US,en;q=0.5", "en-us", "e...
## $ c           <chr> "US", NA, "US", "US", NA, "US", "US", NA, "AU", NA, "US", "US", "US", "US", "US", "US", "...
## $ cy          <chr> "Anaheim", NA, "Fort Huachuca", "Houston", NA, "Mishawaka", "Hammond", NA, "Sydney", NA, ...
## $ g           <chr> "15r91", "ifIpBW", "10DaxOu", "TysVFU", "10IGW7m", "13GrCeP", "YmtpnZ", "13oM0hV", "15r91...
## $ gr          <chr> "CA", NA, "AZ", "TX", NA, "IN", "WI", NA, "02", NA, "OH", "MD", "KY", "OR", "IL", "TX", "...
## $ h           <chr> "10OBm3W", "ifIpBW", "10DaxOt", "TChsoQ", "10IGW7l", "13GrCeP", "YmtpnZ", "15PUeH0", "10O...
## $ hc          <dbl> 1365701422, 1302189369, 1368814585, 1354719206, 1368738258, 1368130510, 1363711958, 13687...
## $ hh          <chr> "j.mp", "1.usa.gov", "1.usa.gov", "1.usa.gov", "1.usa.gov", "1.usa.gov", "1.usa.gov", "go...
## $ l           <chr> "pontifier", "bitly", "jaxstrong", "o_5004fs3lvd", "peacecorps", "bitly", "bitly", "nasat...
## $ ll.0        <dbl> 33.8161, NA, 31.5273, 29.7633, NA, 41.6123, 45.0070, NA, -33.8615, NA, 39.5151, 39.1317, ...
## $ ll.1        <dbl> -117.9794, NA, -110.3607, -95.3633, NA, -86.1381, -92.4591, NA, 151.2055, NA, -84.3983, -...
## $ nk          <dbl> 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ r           <chr> "direct", "http://www.usa.gov/", "http://www.facebook.com/l.php?u=http%3A%2F%2F1.usa.gov%...
## $ t           <dbl> 1368832205, 1368832207, 1368832209, 1368832209, 1368832208, 1368832209, 1368832210, 13688...
## $ tz          <chr> "America/Los_Angeles", "", "America/Phoenix", "America/Chicago", "", "America/Indianapoli...
## $ u           <chr> "http://www.nsa.gov/", "http://answers.usa.gov/system/selfservice.controller?CONFIGURATIO...
## $ _heartbeat_ <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, N...
## $ kw          <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, N...

jsonlite:

dplyr::glimpse(bitly02)
## Observations: 3,959
## Variables: 18
## $ a           <chr> "Mozilla/5.0 (Linux; U; Android 4.1.2; en-us; HTC_PN071 Build/JZO54K) AppleWebKit/534.30 ...
## $ c           <chr> "US", NA, "US", "US", NA, "US", "US", NA, "AU", NA, "US", "US", "US", "US", "US", "US", "...
## $ nk          <int> 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
## $ tz          <chr> "America/Los_Angeles", "", "America/Phoenix", "America/Chicago", "", "America/Indianapoli...
## $ gr          <chr> "CA", NA, "AZ", "TX", NA, "IN", "WI", NA, "02", NA, "OH", "MD", "KY", "OR", "IL", "TX", "...
## $ g           <chr> "15r91", "ifIpBW", "10DaxOu", "TysVFU", "10IGW7m", "13GrCeP", "YmtpnZ", "13oM0hV", "15r91...
## $ h           <chr> "10OBm3W", "ifIpBW", "10DaxOt", "TChsoQ", "10IGW7l", "13GrCeP", "YmtpnZ", "15PUeH0", "10O...
## $ l           <chr> "pontifier", "bitly", "jaxstrong", "o_5004fs3lvd", "peacecorps", "bitly", "bitly", "nasat...
## ## $ al          <chr> "en-US", "en-us", "en-US,en;q=0.5", "en-US", "en", "en-US", "en-US,en;q=0.5", "en-us", "e...
## $ hh          <chr> "j.mp", "1.usa.gov", "1.usa.gov", "1.usa.gov", "1.usa.gov", "1.usa.gov", "1.usa.gov", "go...
## $ r           <chr> "direct", "http://www.usa.gov/", "http://www.facebook.com/l.php?u=http%3A%2F%2F1.usa.gov%...
## $ u           <chr> "http://www.nsa.gov/", "http://answers.usa.gov/system/selfservice.controller?CONFIGURATIO...
## $ t           <int> 1368832205, 1368832207, 1368832209, 1368832209, 1368832208, 1368832209, 1368832210, 13688...
## $ hc          <int> 1365701422, 1302189369, 1368814585, 1354719206, 1368738258, 1368130510, 1363711958, 13687...
## $ cy          <chr> "Anaheim", NA, "Fort Huachuca", "Houston", NA, "Mishawaka", "Hammond", NA, "Sydney", NA, ...
## $ ll          <list> [<33.8161, -117.9794>, NULL, <31.5273, -110.3607>, <29.7633, -95.3633>, NULL, <41.6123, ...
## $ _heartbeat_ <int> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, N...
## $ kw          <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, N...

UPDATE

The latest version of the jsonlite package supports streaming JSON (which is what this actually is). You can now read it with one line like so:

json_file <- stream_in(file("usagov_bitly_data2013-05-17-1368832207"))

See also Jeroen’s answer below for stream-parsing it directly over http.


OLD ANSWER

It turns out this is a “pseudo-JSON” file. I come across these in many naive API systems I work in. Each line is valid JSON, but the individual objects aren’t in a JSON array. You need to use readLines and then build your own, valid JSON array from it and pass that into fromJSON:

library(jsonlite)

# read in individual JSON lines
json_file <- "usagov_bitly_data2013-05-17-1368832207"

# turn it into a proper array by separating each object with a "," and
# wrapping that up in an array with "[]"'s.

dat <- fromJSON(sprintf("[%s]", paste(readLines(json_file), collapse=",")))

dim(dat)

## [1] 3959   18

str(dat)

## 'data.frame': 3959 obs. of  18 variables:
##  $ a          : chr  "Mozilla/5.0 (Linux; U; Android 4.1.2; en-us; HTC_PN071 Build/JZO54K) AppleWebKit/534.30 (KHTML, like Gecko) Version/4.0 Mobile "| __truncated__ "Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.1; .NET CLR 1.1.4322; .NET CLR 2.0.50727; .NET CLR 3.0.04506.30; .NET CLR 3.0.4"| __truncated__ "Mozilla/5.0 (Windows NT 6.1; rv:21.0) Gecko/20100101 Firefox/21.0" "Mozilla/5.0 (Linux; U; Android 4.1.2; en-us; SGH-T889 Build/JZO54K) AppleWebKit/534.30 (KHTML, like Gecko) Version/4.0 Mobile S"| __truncated__ ...
##  $ c          : chr  "US" NA "US" "US" ...
##  $ nk         : int  0 0 1 1 0 0 1 0 0 0 ...
##  $ tz         : chr  "America/Los_Angeles" "" "America/Phoenix" "America/Chicago" ...
##  $ gr         : chr  "CA" NA "AZ" "TX" ...
##  $ g          : chr  "15r91" "ifIpBW" "10DaxOu" "TysVFU" ...
##  $ h          : chr  "10OBm3W" "ifIpBW" "10DaxOt" "TChsoQ" ...
##  $ l          : chr  "pontifier" "bitly" "jaxstrong" "o_5004fs3lvd" ...
##  $ al         : chr  "en-US" "en-us" "en-US,en;q=0.5" "en-US" ...
##  $ hh         : chr  "j.mp" "1.usa.gov" "1.usa.gov" "1.usa.gov" ...
## ... (goes on for a while, many columns)

I combined the readLines in with the paste/sprintf call since the object.size of the resultant (temporary) object is 2,025,656 bytes (~2MB) and didn’t feel like doing an rm on a separate temporary variable.

Leave a Comment