I have a data.table where one of the columns contains JSON. I am trying to extract the content so that each variable is a column.
library(jsonlite)
library(data.table)
df<-data.table(a=c('{"tag_id":"34","response_id":2}',
'{"tag_id":"4","response_id":1,"other":4}',
'{"tag_id":"34"}'),stringsAsFactors=F)
The desired result, that does not refer to the "other" variable:
tag_id response_id
1 "34" 2
2 "4" 1
3 "34" NA
I have tried several versions of:
parseLog <- function(x){
if (is.na(x))
e=c(tag_id=NA,response_id=NA)
else{
j=fromJSON(x)
e=c(tag_id=as.integer(j$tag_id),response_id=j$response_id)
}
e
}
that seems to work well to retrieve a list of vectors (or lists if c is replaced by list) but when I try to convert the list to data.table something doesn´t work as expected.
parsed<-lapply(df$a,parseLog)
rparsed<-do.call(rbind.data.frame,parsed)
colnames(rparsed)<-c("tag_id","response_id")
Because of the missing value in the third row. How can I solve it in a R-ish clean way? How can I make that my parse method returns an NA for the missing value. Alternative, Is there a parameter "fill" like there is for rbind that can be used in rbind.data.frame or analogous method?
The dataset I am using has 11M rows so performance is important.
Additionally, there is an equivalent method to rbind.data.frame to obtain a data.table. How would that be used? When I check the documentation it refers me to rbindlist but it complains the parameter is not used and if call directly(without do.call it complains about the type of parsed):
rparsed<-do.call(rbindlist,fill=T,parsed)
EDIT: The case I need to cover is more general, in a set of 11M records all the possible circumstances happen:
df<-data.table(a=c('{"tag_id":"34","response_id":2}',
'{"trash":"34","useless":2}',
'{"tag_id":"4","response_id":1,"other":4}',
NA,
'{"response_id":"34"}',
'{"tag_id":"34"}'),stringsAsFactors=F)
and the output should only contain tag_id and response_id columns.
There might be a simpler way but this seems to be working:
library(data.table)
library(jsonlite)
df[, json := sapply(a, fromJSON)][, rbindlist(lapply(json, data.frame), fill=TRUE)]
#or if you need all the columns :
#df[, json := sapply(a, fromJSON)][,
# c('tag_id', 'response_id') := rbindlist(lapply(json, data.frame), fill=TRUE)]
Output:
> df[, json := sapply(a, fromJSON)][, rbindlist(lapply(json, data.frame), fill=TRUE)]
tag_id response_id
1: 34 2
2: 4 1
3: 34 NA
EDIT:
This solution comes after the edit of the question with additional requests.
There are lots of ways to do this but I find the simplest one is at the creation of the data.frame like this:
df[, json := sapply(a, fromJSON)][,
rbindlist(lapply(json, function(x) data.frame(x)[-3]), fill=TRUE)]
# tag_id response_id
#1: 34 2
#2: 4 1
#3: 34 NA
Related
I have a list of more than 100,000 json files from which I want to get a data.table with only a few variables. Unfortunately the files are complex. The content of each json file looks like:
Sample 1
$id
[1] "10.1"
$title
$title$value
[1] "Why this item"
$itemsource
$itemsource$id
[1] "AA"
$date
[1] "1992-01-01"
$itemType
[1] "art"
$creators
list()
Sample 2
$id
[1] "10.2"
$title
$title$value
[1] "We need this item"
$itemsource
$itemsource$id
[1] "AY"
$date
[1] "1999-01-01"
$itemType
[1] "art"
$creators
type name firstname surname affiliationIds
1 Person Frank W. Cornell. Frank W. Cornell. a1
2 Person David A. Chen. David A. Chen. a1
$affiliations
id name
1 a1 Foreign Affairs Desk, New York Times
What I need from this set of files is a table with creator names, item ids and dates. For the two sample files above:
id date name firstname lastname creatortype
"10.1" "1992-01-01" NA NA NA NA
"10.2" "1999-01-01" Frank W. Cornell. Frank W. Cornell. Person
"10.2" "1999-01-01" David A. Chen. David A. Chen. Person
What I have done so far:
library(parallel)
library(data.table)
library(jsonlite)
library(dplyr)
filelist = list.files(pattern="*.json",recursive=TRUE,include.dirs =TRUE)
parsed = mclapply(filelist, function(x) fromJSON(x),mc.cores=24)
data = rbindlist(mclapply(1:length(parsed), function(x) {
a = data.table(item = parsed[[x]]$id, date = list(list(parsed[[x]]$date)), name = list(list(parsed[[x]]$name)), creatortype = list(list(parsed[[x]]$creatortype))) #ignoring the firstname/lastname fields here for convenience
b = data.table(id = a$item, date = unlist(a$date), name=unlist(a$name), creatortype=unlist(a$creatortype))
return(b)
},mc.cores=24))
However, on the last step, I get this error:
"Error in rbindlist(mclapply(1:length(parsed), function(x){:
Item 1 of list is not a data.frame, data.table or list"
Thanks in advance for your suggestions.
Related questions include:
Extract data from list of lists [R]
R convert json to list to data.table
I want to convert JSON file into data.table in r
How can read files from directory using R?
Convert R data table column from JSON to data table
from the error message, i suppose this basically means that one of the results from mclapply() is empty, by empty I mean either NULL or data.table with 0 row, or simply encounters an error within the parallel processing.
what you could do is:
add more checks inside the mclapply() like try-error or check the class of b and nrow of b, whether b is empty or not
when you use rbindlist, add argument fill = T
hope this solves ur problem.
I have recently started using R and have a task regarding parsing json in R to get a non-json format. For this, i am using the "fromJSON()" function. I have tried to parse json as a text file. It runs successfully when i do it with just a single row entry. But when I try it with multiple row entries, i get the following error:
fromJSON("D:/Eclairs/Printing/test3.txt")
Error in feed_push_parser(readBin(con, raw(), n), reset = TRUE) :
lexical error: invalid char in json text.
[{'CategoryType':'dining','City':
(right here) ------^
> fromJSON("D:/Eclairs/Printing/test3.txt")
Error in feed_push_parser(readBin(con, raw(), n), reset = TRUE) :
parse error: trailing garbage
"mumbai","Location":"all"}] [{"JourneyType":"Return","Origi
(right here) ------^
> fromJSON("D:/Eclairs/Printing/test3.txt")
Error in feed_push_parser(readBin(con, raw(), n), reset = TRUE) :
parse error: after array element, I expect ',' or ']'
:"mumbai","Location":"all"} {"JourneyType":"Return","Origin
(right here) ------^
The above errors are due to three different formats in which i tried to parse the json text, but the result was the same, only the location suggested by changed.
Please help me to identify the cause of this error or if there is a more efficient way o performing the task.
The original file that i have is an excel sheet with multiple columns and one of those columns consists of json text. The way i tried right now is by extracting just the json column and converting it to a tab separated text and then parsing it as:
fromJSON("D:/Eclairs/Printing/test3.txt")
Please also suggest if this can be done more efficiently. I need to map all the columns in the excel to the non-json text as well.
Example:
[{"CategoryType":"dining","City":"mumbai","Location":"all"}]
[{"CategoryType":"reserve-a-table","City":"pune","Location":"Kothrud,West Pune"}]
[{"Destination":"Mumbai","CheckInDate":"14-Oct-2016","CheckOutDate":"15-Oct-2016","Rooms":"1","NoOfPax":"3","NoOfAdult":"3","NoOfChildren":"0"}]
Consider reading in the text line by line with readLines(), iteratively saving the JSON dataframes to a growing list:
library(jsonlite)
con <- file("C:/Path/To/Jsons.txt", open="r")
jsonlist <- list()
while (length(line <- readLines(con, n=1, warn = FALSE)) > 0) {
jsonlist <- append(jsonlist, list(fromJSON(line)))
}
close(con)
jsonlist
# [[1]]
# CategoryType City Location
# 1 dining mumbai all
# [[2]]
# CategoryType City Location
# 1 reserve-a-table pune Kothrud,West Pune
# [[3]]
# Destination CheckInDate CheckOutDate Rooms NoOfPax NoOfAdult NoOfChildren
# 1 Mumbai 14-Oct-2016 15-Oct-2016 1 3 3 0
I wanted to import a .txt file in R but the format is really special and it's looks like a json format but I don't know how to import it. There is an example of my data:
{"datetime":"2015-07-08 09:10:00","subject":"MMM","sscore":"-0.2280","smean":"0.2593","svscore":"-0.2795","sdispersion":"0.375","svolume":"8","sbuzz":"0.6026","lastclose":"155.430000000","companyname":"3M Company"},{"datetime":"2015-07-07 09:10:00","subject":"MMM","sscore":"0.2977","smean":"0.2713","svscore":"-0.7436","sdispersion":"0.400","svolume":"5","sbuzz":"0.4895","lastclose":"155.080000000","companyname":"3M Company"},{"datetime":"2015-07-06 09:10:00","subject":"MMM","sscore":"-1.0057","smean":"0.2579","svscore":"-1.3796","sdispersion":"1.000","svolume":"1","sbuzz":"0.4531","lastclose":"155.380000000","companyname":"3M Company"}
To deal with this is used this code:
test1 <- read.csv("C:/Users/test1.txt", header=FALSE)
## Import as 5 observations (5th is all empty) of 1700 variables
#(in fact 40 observations of 11 variables). In fact when I imported the
#.txt file, it's having one line (5th obs) empty, and 4 lines of data and
#placed next to each other 4 lines of data of 11 variables.
# Get the different lines
part1=test1[1:10]
part2=test1[11:20]
part3=test1[21:30]
part4=test1[31:40]
...
## Remove the empty line (there were an empty line after each)
part1=part1[-5,]
part2=part2[-5,]
part3=part3[-5,]
...
## Rename the columns
names(part1)=c("Date Time","Subject","Sscore","Smean","Svscore","Sdispersion","Svolume","Sbuzz","Last close","Company name")
names(part2)=c("Date Time","Subject","Sscore","Smean","Svscore","Sdispersion","Svolume","Sbuzz","Last close","Company name")
names(part3)=c("Date Time","Subject","Sscore","Smean","Svscore","Sdispersion","Svolume","Sbuzz","Last close","Company name")
...
## Assemble data to have one dataset
data=rbind(part1,part2,part3,part4,part5,part6,part7,part8,part9,part10)
## Formate Date Time
times <- as.POSIXct(data$`Date Time`, format='{datetime:%Y-%m-%d %H:%M:%S')
data$`Date Time` <- times
## Keep only the Date
data$Date <- as.Date(times)
## Formate data - Remove text
data$Subject <- gsub("subject:", "", data$Subject)
data$Sscore <- gsub("sscore:", "", data$Sscore)
...
So My code is working to reinstate the data but it's maybe very difficult and more long I know there is better ways to do it, so if you could help me with that I would be very grateful.
There are many packages that read JSON, e.g. rjson, jsonlite, RJSONIO (they will turn in up a google search) - just pick one and give it a go.
e.g.
library(jsonlite)
json.text <- '{"datetime":"2015-07-08 09:10:00","subject":"MMM","sscore":"-0.2280","smean":"0.2593","svscore":"-0.2795","sdispersion":"0.375","svolume":"8","sbuzz":"0.6026","lastclose":"155.430000000","companyname":"3M Company"},{"datetime":"2015-07-07 09:10:00","subject":"MMM","sscore":"0.2977","smean":"0.2713","svscore":"-0.7436","sdispersion":"0.400","svolume":"5","sbuzz":"0.4895","lastclose":"155.080000000","companyname":"3M Company"},{"datetime":"2015-07-06 09:10:00","subject":"MMM","sscore":"-1.0057","smean":"0.2579","svscore":"-1.3796","sdispersion":"1.000","svolume":"1","sbuzz":"0.4531","lastclose":"155.380000000","companyname":"3M Company"}'
x <- fromJSON(paste0('[', json.text, ']'))
datetime subject sscore smean svscore sdispersion svolume sbuzz lastclose companyname
1 2015-07-08 09:10:00 MMM -0.2280 0.2593 -0.2795 0.375 8 0.6026 155.430000000 3M Company
2 2015-07-07 09:10:00 MMM 0.2977 0.2713 -0.7436 0.400 5 0.4895 155.080000000 3M Company
3 2015-07-06 09:10:00 MMM -1.0057 0.2579 -1.3796 1.000 1 0.4531 155.380000000 3M Company
I paste the '[' and ']' around your JSON because you have multiple JSON elements (the rows in the dataframe above) and for this to be well-formed JSON it needs to be an array, i.e. [ {...}, {...}, {...} ] rather than {...}, {...}, {...}.
I'm currently doing Cox Proportional Hazards Modeling using Rpy2 - I imagine my question will cover other functions and the results from calling them as well though.
After I run the function, I have a variable which contains the results from the function, in the form of a vector. I have tried explicitly converting this to a DataFrame (resultsDataFrame = DataFrame(resultVector)). There are no errors returned when doing this. However, when I do resultsDataFrame.to_csvfile(filename) I get the following error:
Traceback (most recent call last):
File "<pyshell#171>", line 1, in <module>
modelFrame.to_csvfile('/Users/fortylashes/Documents/Matthews_Research/Cox_PH/ResultOutput_Exp1.csv')
File "/Library/Python/2.7/site-packages/rpy2/robjects/vectors.py", line 1031, in to_csvfile
'col.names': col_names, 'qmethod': qmethod, 'append': append})
RRuntimeError: Error in as.data.frame.default(x[[i]], optional = TRUE, stringsAsFactors = stringsAsFactors) :
cannot coerce class ""coxph"" to a data.frame
Furthermore, when I simply do:
for result in resultVector:
print (result)
I get an extremely long list of results- including information on each entry in the dataset used in the model, for each variable (so 9,000 records x 9 variables = 81,000 unneeded results). The results I really need are at the bottom of this vector and look like this:
coef exp(coef) se(coef) z p
age_age6574 -0.057775 0.944 0.05469 -1.056 2.9e-01
age_age75plus -0.020795 0.979 0.04891 -0.425 6.7e-01
sex_female -0.005304 0.995 0.03961 -0.134 8.9e-01
stage_late -0.261609 0.770 0.04527 -5.779 7.5e-09
access -0.000494 1.000 0.00069 -0.715 4.7e-01
Likelihood ratio test=36.6 on 5 df, p=7.31e-07 n= 9752, number of events= 2601
*NOTE: There were several more variables for which data was reported in the initial results (the 9,000 x 9 that I was talking about) but weren't actually used in the model.
I was wondering if there was a way to explicitly get this data, put it in one long ordered row, and then output it to a csv file?
::::UPDATE::::
When I call theModel.names I get a list of the various measures which can be called by numerical index:
[1] "coefficients" "var" "loglik"
[4] "score" "iter" "linear.predictors"
[7] "residuals" "means" "concordance"
[10] "method" "n" "nevent"
[13] "terms" "assign" "wald.test"
[16] "y" "formula" "call"
From this I can get the coefficients, which can then be exponentiated. I have not found, however, the p-value, the z score or the likelihood test ratio, which I will need.
I'm making word frequency tables with R and the preferred output format would be a JSON file. sth like
{
"word" : "dog",
"frequency" : 12
}
Is there any way to save the table directly into this format? I've been using the write.csv() function and convert the output into JSON but this is very complicated and time consuming.
set.seed(1)
( tbl <- table(round(runif(100, 1, 5))) )
## 1 2 3 4 5
## 9 24 30 23 14
library(rjson)
sink("json.txt")
cat(toJSON(tbl))
sink()
file.show("json.txt")
## {"1":9,"2":24,"3":30,"4":23,"5":14}
or even better:
set.seed(1)
( tab <- table(letters[round(runif(100, 1, 26))]) )
a b c d e f g h i j k l m n o p q r s t u v w x y z
1 2 4 3 2 5 4 3 5 3 9 4 7 2 2 2 5 5 5 6 5 3 7 3 2 1
sink("lets.txt")
cat(toJSON(tab))
sink()
file.show("lets.txt")
## {"a":1,"b":2,"c":4,"d":3,"e":2,"f":5,"g":4,"h":3,"i":5,"j":3,"k":9,"l":4,"m":7,"n":2,"o":2,"p":2,"q":5,"r":5,"s":5,"t":6,"u":5,"v":3,"w":7,"x":3,"y":2,"z":1}
Then validate it with http://www.jsonlint.com/ to get pretty formatting. If you have multidimensional table, you'll have to work it out a bit...
EDIT:
Oh, now I see, you want the dataset characteristics sink-ed to a JSON file. No problem, just give us a sample data, and I'll work on a code a bit. Practically, you need to carry out the data into desirable format, hence convert it to JSON. list should suffice. Give me a sec, I'll update my answer.
EDIT #2:
Well, time is relative... it's a common knowledge... Here you go:
( dtf <- structure(list(word = structure(1:3, .Label = c("cat", "dog",
"mouse"), class = "factor"), frequency = c(12, 32, 18)), .Names = c("word",
"frequency"), row.names = c(NA, -3L), class = "data.frame") )
## word frequency
## 1 cat 12
## 2 dog 32
## 3 mouse 18
If dtf is a simple data frame, yes, data.frame, if it's not, coerce it! Long story short, you can do:
toJSON(as.data.frame(t(dtf)))
## [1] "{\"V1\":{\"word\":\"cat\",\"frequency\":\"12\"},\"V2\":{\"word\":\"dog\",\"frequency\":\"32\"},\"V3\":{\"word\":\"mouse\",\"frequency\":\"18\"}}"
I though I'll need some melt with this one, but simple t did the trick. Now, you only need to deal with column names after transposing the data.frame. t coerces data.frames to matrix, so you need to convert it back to data.frame. I used as.data.frame, but you can also use toJSON(data.frame(t(dtf))) - you'll get X instead of V as a variable name. Alternatively, you can use regexp to clean the JSON file (if needed), but it's a lousy practice, try to work it out by preparing the data.frame.
I hope this helped a bit...
These days I would typically use the jsonlite package.
library("jsonlite")
toJSON(mydatatable, pretty = TRUE)
This turns the data table into a JSON array of key/value pair objects directly.
RJSONIO is a package "that allows conversion to and from data in Javascript object notation (JSON) format". You can use it to export your object as a JSON file.
library(RJSONIO)
writeLines(toJSON(anobject), "afile.JSON")