Related
I have a json object served from an api as follows:
{
"workouts": [
{
"id": 92527291,
"starts": "2021-06-28T15:42:44.000Z",
"minutes": 30,
"name": "Indoor Cycling",
"created_at": "2021-06-28T16:12:57.000Z",
"updated_at": "2021-06-28T16:12:57.000Z",
"plan_id": null,
"workout_token": "ELEMNT BOLT A1B3:59",
"workout_type_id": 12,
"workout_summary": {
"id": 87540207,
"heart_rate_avg": "152.0",
"calories_accum": "332.0",
"created_at": "2021-06-28T16:12:58.000Z",
"updated_at": "2021-06-28T16:12:58.000Z",
"power_avg": "185.0",
"distance_accum": "17520.21",
"cadence_avg": "87.0",
"ascent_accum": "0.0",
"duration_active_accum": "1801.0",
"duration_paused_accum": "0.0",
"duration_total_accum": "1801.0",
"power_bike_np_last": "186.0",
"power_bike_tss_last": "27.6",
"speed_avg": "9.73",
"work_accum": "332109.0",
"file": {
"url": "https://cdn.wahooligan.com/wahoo-cloud/production/uploads/workout_file/file/FPoJBPZo17BvTmSomq5Y_Q/2021-06-28-154244-ELEMNT_BOLT_A1B3-59-0.fit"
}
}
}
],
"total": 55,
"page": 1,
"per_page": 1,
"order": "descending",
"sort": "starts"
}
I want to get the data into a dataframe. However, lots of the columns seem to have a dtype of object. I assume that this is because some of the numeric values in the json are double quoted. What is the best and most efficient way to avoid this (the json potentially has many workouts elements)?
Is it to fix the returned json? Or to iterate through the dataframe columns and convert the objects to floats?
Thank you
Martyn
IIUC, you can try:
df = pd.json_normalize(json_data, meta=[
'total', 'page', 'per_page', 'order', 'sort'], record_path='workouts').convert_dtypes()
Try using pandas.to_numeric. Here are the docs.
https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.to_numeric.html
I have been studying official documentation of JMESPath and a few other resources. However I was not successful with the following task:
my data structure is a json from vimeo api (video list):
data array contains lots of objects, each object is the uploaded file that has many attributes and various options.
"data": [
{
"uri": "/videos/00001",
"name": "Video will be added.mp4",
"description": null,
"type": "video",
"link": "https://vimeo.com/00001",
"duration": 9,
"files":[
{
"quality": "hd",
"type": "video/mp4",
"width": 1440,
"height": 1440,
"link": "https://player.vimeo.com/external/4443333.sd.mp4",
"created_time": "2020-09-01T19:10:01+00:00",
"fps": 30,
"size": 10807854,
"md5": "643d9f18e0a63e0630da4ad85eecc7cb",
"public_name": "UHD 1440p",
"size_short": "10.31MB"
},
{
"quality": "sd",
"type": "video/mp4",
"width": 540,
"height": 540,
"link": "https://player.vimeo.com/external/44444444.sd.mp4",
"created_time": "2020-09-01T19:10:01+00:00",
"fps": 30,
"size": 1345793,
"md5": "cb568939bb7b276eb468d9474c1f63f6",
"public_name": "SD 540p",
"size_short": "1.28MB"
},
... other data
]
},
... other uploaded files
]
Filter I need to apply is that duration needs to be less than 10 and width of file needs to be 540 and the result needs to contain a link (url) from files
I have managed to get only one of structure-levels working:
data[].files[?width == '540'].link
I need to extract this kind of list
[
{
"uri": "/videos/111111",
"link": "https://player.vimeo.com/external/4123112312.sd.mp4"
},
{
"uri": "/videos/22222",
"link": "https://player.vimeo.com/external/1231231231.sd.mp4"
},
...other data
]
Since the duration is in your data array, you will have to add this filter at that level.
You will also have to use what is described under the section filtering and selecting nested data because you only care of one specific type of file under the files array, so, you can use the same type of query structure | [0] in order to pull only the first element of the filtered files array.
So on your reduced exemple, the query:
data[?duration < `10`].{ uri: uri, link: files[?width == `540`].link | [0] }
Would yield the expected:
[
{
"uri": "/videos/00001",
"link": "https://player.vimeo.com/external/44444444.sd.mp4"
}
]
I'm trying to use some results exported in JSON of a script called "Mixed Content Scan" (it's a script in order to search on a website if there is some mixed HTTP/HTTPS content and if all your pages are ok in HTTPS).
I'm a beginner with JSON, I read and watched a lot of tutorials in order to understand how to structure JSON data but I'm stumbling on something.
Here is a sample of my data (first 3 lines) :
{"message":"Scanning https://mywebsite.com/","context":[],"level":250,"level_name":"NOTICE","channel":"MCS","datetime":{"date":"2018-10-05 23:48:50.268196","timezone_type":3,"timezone":"America/New_York"},"extra":[]}
{"message":"00000 - https://mywebsite.com/","context":[],"level":400,"level_name":"ERROR","channel":"MCS","datetime":{"date":"2018-10-05 23:48:50.760948","timezone_type":3,"timezone":"America/New_York"},"extra":[]}
{"message":"http://mywebsite.com/wp-content/uploads/2015/03/image.jpg","context":[],"level":300,"level_name":"WARNING","channel":"MCS","datetime":{"date":"2018-10-05 23:48:50.761082","timezone_type":3,"timezone":"America/New_York"},"extra":[]}
I know I need to wrap my data around some {} or [] (tried both), but I think I'm missing something, for example, every JSON data validator websites are telling me that I have an error between 2 lines when I add a "," when I try to have multiple results into it.
How can I upgrade this raw data in order for a JSON validator to validate it?
Thanks!
How's this
[{
"message": "Scanning https://mywebsite.com/",
"context": [],
"level": 250,
"level_name": "NOTICE",
"channel": "MCS",
"datetime": {
"date": "2018-10-05 23:48:50.268196",
"timezone_type": 3,
"timezone": "America/New_York"
},
"extra": []
}, {
"message": "00000 - https://mywebsite.com/",
"context": [],
"level": 400,
"level_name": "ERROR",
"channel": "MCS",
"datetime": {
"date": "2018-10-05 23:48:50.760948",
"timezone_type": 3,
"timezone": "America/New_York"
},
"extra": []
}, {
"message": "http://mywebsite.com/wp-content/uploads/2015/03/image.jpg",
"context": [],
"level": 300,
"level_name": "WARNING",
"channel": "MCS",
"datetime": {
"date": "2018-10-05 23:48:50.761082",
"timezone_type": 3,
"timezone": "America/New_York"
},
"extra": []
}]
Entries in an array need to be separated by commas.
I'd like to perform a JOIN query on incoming JSON events and JSON reference data in Azure Stream Analytics.
The JSON events look like the following:
{"devicename":"12345","value":25}
The JSON Reference data (simplified) looks like the following:
{
"Id": "configuration",
"Sites": [{
"Id": "b83939b5-6502-4140-b807-205162ac4939",
"Name": "site1",
"Location": {
"Latitude": 5.000,
"Longitude": 5.000
},
"Lines": [{
"Id": "e707a451-948a-498e-80de-d61bc448a5ef",
"Name": "line1",
"Sections": [{
"Id": "d17d762c-4291-4912-9dcf-72113c9f0b4b",
"Name": "section1",
"Sensors": [{
"Id": "S1_PL1_LS1_M1_device_001",
"Name": "sensor1",
"ExpectedAVG": 55,
"ExpectedMIN": 35,
"ExpectedMAX": 75,
"TypeOfSensor": "Humidity"
}, {
"Id": "S1_PL1_LS1_M1_device_002",
"Name": "sensor2",
"ExpectedAVG": 105,
"ExpectedMIN": 90,
"ExpectedMAX": 118,
"TypeOfSensor": "Temperature"
}]
}]
}]
}]
}
Now I'd like to perform a JOIN on the incoming device data and the reference data.
SELECT
i.devicename as pk,
i.value as rk,
system.timestamp as time,
r.Sites.Lines.Sections.Sensors.Temperature as Temperature,
r.Sites.Lines.Sections.Sensors.Humidity as Humidity
INTO
output
FROM
IoTHubStream i
INNER JOIN
Reference r ON r.Sites.Lines.Sections.Sensors.Id = i.devicename
Now how can this be done? The problem is that the reference data can change and I would like to handle this dynamically.
I'm thinking about using a WHERE statement but I don't know how this could look like, some advice/help is really welcome.
Thanks in advance!
You can access the array elements using GetArrayElements described in this page https://msdn.microsoft.com/en-us/library/azure/mt298451.aspx
If your data changes make it impossible to use GetArrayElements, please share details about how it changes.
EDIT 2014-05-01: I tried fromJSON first (as suggested below), but that only parsed the first line. I found out that there were commas missing between the brackets of each JSON line so I changed that in TextEdit and saved the file. I also added [ at the beginning of the file and ] at the end and then it worked with JSON. Now the next step: from a list (with embedded lists) to a dataframe (or csv).
I get a data package from edX every now and then on the courses we are evaluating. Some of these are just plain .csv files which are quite easy to handle, others are more difficult for me (not having a CS or programming background).
I have 2 files I want to open and parse into csv files for analysis in R. I have tried many many json2csv tools out there, but to no avail. I also tried the simple methods described here to turn json into csv.
The data is confidential, so I cannot share the entire data set, but will share the first two lines of the file, maybe that helps. The problem is that nowhere I find anything about .mongo files, which to me seems quite strange, do they even exist? Or is this just a JSON file that may be corrupted (which could explain the errors)?
Any suggestions are welcome.
The first 2 lines in one of the .mongo files:
{
"_id": {
"$oid": "52d1e62c350e7a3156000009"
},
"votes": {
"up": [
],
"down": [
],
"up_count": 0,
"down_count": 0,
"count": 0,
"point": 0
},
"visible": true,
"abuse_flaggers": [
],
"historical_abuse_flaggers": [
],
"parent_ids": [
],
"at_position_list": [
],
"body": "the delft university accredited course with the scholarship (fundamentals of water treatment) is supposed to start in about a month's time. But have the scholarship list been published? Any tentative date??",
"course_id": "DelftX/CTB3365x/2013_Fall",
"_type": "Comment",
"endorsed": false,
"anonymous": false,
"anonymous_to_peers": false,
"author_id": "269835",
"comment_thread_id": {
"$oid": "52cd40c5ab40cf347e00008d"
},
"author_username": "tachak59",
"sk": "52d1e62c350e7a3156000009",
"updated_at": {
"$date": 1389487660636
},
"created_at": {
"$date": 1389487660636
}
}{
"_id": {
"$oid": "52d0a66bcb3eee318d000012"
},
"votes": {
"up": [
],
"down": [
],
"up_count": 0,
"down_count": 0,
"count": 0,
"point": 0
},
"visible": true,
"abuse_flaggers": [
],
"historical_abuse_flaggers": [
],
"parent_ids": [
{
"$oid": "52c63278100c07c0d1000028"
}
],
"at_position_list": [
],
"body": "I got it. Thank you!",
"course_id": "DelftX/CTB3365x/2013_Fall",
"_type": "Comment",
"endorsed": false,
"anonymous": false,
"anonymous_to_peers": false,
"parent_id": {
"$oid": "52c63278100c07c0d1000028"
},
"author_id": "2655027",
"comment_thread_id": {
"$oid": "52c4f303b03c4aba51000013"
},
"author_username": "dmoronta",
"sk": "52c63278100c07c0d1000028-52d0a66bcb3eee318d000012",
"updated_at": {
"$date": 1389405803386
},
"created_at": {
"$date": 1389405803386
}
}{
"_id": {
"$oid": "52ceea0cada002b72c000059"
},
"votes": {
"up": [
],
"down": [
],
"up_count": 0,
"down_count": 0,
"count": 0,
"point": 0
},
"visible": true,
"abuse_flaggers": [
],
"historical_abuse_flaggers": [
],
"parent_ids": [
{
"$oid": "5287e8d5906c42f5aa000013"
}
],
"at_position_list": [
],
"body": "if u please send by mail \n",
"course_id": "DelftX/CTB3365x/2013_Fall",
"_type": "Comment",
"endorsed": false,
"anonymous": false,
"anonymous_to_peers": false,
"parent_id": {
"$oid": "5287e8d5906c42f5aa000013"
},
"author_id": "2276302",
"comment_thread_id": {
"$oid": "528674d784179607d0000011"
},
"author_username": "totah1993",
"sk": "5287e8d5906c42f5aa000013-52ceea0cada002b72c000059",
"updated_at": {
"$date": 1389292044203
},
"created_at": {
"$date": 1389292044203
}
}
R doesn't have "native" support for these files but there is a JSON parser with the rjson package. So I might load my .mongo file with:
myfile <- "path/to/myfile.mongo"
myJSON <- readLines(myfile)
myNiceData <- fromJSON(myJSON)
Since RJson converts into a data structure that fits the object being read, you'll have to do some additional snooping but once you have an R data type you shouldn't have any trouble working with it from there.
Another package to consider when parsing JSON data is jsonlite. It will make data frames for you so you can write them to a csv format with write.table or some other applicable method for writing objects.
NOTE: if it is easier to connect to the MongoDB and get the data from a request, then RMongo may be a good bet. The R-Bloggers also made a post about using RMongo that has a nice little walkthrough.
I used RJSON as suggested by #theWanderer and with the help of a colleague wrote the following code to parse the data into columns, choosing the specific columns that are needed, and checking each of the instances if they return the right variables.
Entire workflow:
Checked some of the data in jsonlint - corrected the errors → },{ instead of }{ between each line and [ and ] at the beginning and end of the file
Made a smaller file to play with, containing about 11 JSON lines
Used the code below to parse the datafile - however, checking the different listItems first if they are not lists themselves (that gives problems) // as you will see, I also removed things like \n because that gave errors and added an empty value for parent_id if there is none in the data (otherwise it would mix up the data)
The code to import the .mongo file into R and then parse it into CSV:
library(rjson)
###### set working directory to write out the data file
setwd("/your/favourite/dir/json to csv/")
#never ever convert strings to factors
options(stringsAsFactors = FALSE)
#import the .mongo file to R
temp.data = fromJSON(file="temp.mongo", method="C", unexpected.escape="error")
file.remove("temp.csv") ## removes the old datafile if there is one
## (so the data is not appended to the file,
## but a new file is created)
listItem = temp.data[[1]] ## prepare the listItem the first time
for (listItem in temp.data){
parent_id = ""
if (length(listItem$parent_id)>0){
parent_id = listItem$parent_id
}
write.table(t(c(
listItem$votes$up_count, listItem$visible, parent_id,
gsub("\n", "", listItem$body), listItem$course_id, unlist(listItem["_type"]),
listItem$endorsed, listItem$anonymous, listItem$author_id,
unlist(listItem$comment_thread_id), listItem$author_username,
as.POSIXct(unlist(listItem$created_at)/1000, origin="1970-01-01"))), # end t(), c()
file="temp.csv", sep="\t", append=TRUE, row.names=FALSE, col.names=FALSE)
}