dasel
Dasel (short for data-selector) allows you to query and modify data structures using selector strings.
Comparable to jq / yq, but supports JSON, YAML, TOML, XML and CSV with zero runtime dependencies.
One tool to rule them all
Say good bye to learning new tools just to work with a different data format.
Dasel uses a standard selector syntax no matter the data format. This means that once you learn how to use dasel you immediately have the ability to query/modify any of the supported data types without any additional tools or effort.
Issue vs Discussion
I have enabled discussions on this repository.
I am aware there may be some confusion when deciding where you should communicate when reporting issues, asking questions or raising feature requests so this section aims to help us align on that.
Please raise an issue if:
- You find a bug.
- You have a feature request and can clearly describe your request.
Please open a discussion if:
- You have a question.
- You're not sure how to achieve something with dasel.
- You have an idea but don't quite know how you would like it to work.
- You have achieved something cool with dasel and want to show it off.
- Anything else!
Features
- Query/select data from structured data files.
- Update data in structured data files.
- Create data files.
- Supports multiple data formats/types.
- Convert between data formats/types.
- Uses a standard query/selector syntax across all data formats.
- Zero runtime dependencies.
- Available on Linux, Mac and Windows.
- Available to import and use in your own projects.
- Run via Docker.
- Faster than jq/yq.
- Self update.
Table of contents
- Dasel
- One tool to rule them all
- Issue vs discussion
- Features
- Table of contents
- Playground
- Installation
- Notes
- Usage
- Supported file types
- Selectors
- Examples
- Benchmarks
Playground
You can test out dasel commands using the playground.
Source code for the playground can be found at github.com/TomWright/daselplayground.
Installation
You can import dasel as a package and use it in your applications, or you can use a pre-built binary to modify files from the command line.
Docker
Run dasel in docker using the image ghcr.io/tomwright/dasel.
echo '{"name": "Tom"}' | docker run -i --rm ghcr.io/tomwright/dasel:latest -p json '.name'
"Tom"
Docker images are pushed to the github container repository: tomwright/dasel.
If you want to use a specific version of dasel simply change latest to the desired version.
latest- The latest released version.dev- The latest build frommasterbranch.v*.*.*- The build from the given release.
Homebrew
Dasel is part of homebrew-core making it available without any additional taps.
brew install dasel
ASDF (Cross Platform)
Using asdf-vm and the asdf-dasel plugin
asdf plugin add dasel https://github.com/asdf-community/asdf-dasel.git
asdf list all dasel
asdf install dasel <version>
asdf global dasel <version>
Manual
You can download a compiled executable from the latest release.
Note: Don't forget to put the binary somewhere in your PATH.
Linux
This one liner should work for you - be sure to change the targeted release executable if needed. It currently targets dasel_linux_amd64.
curl -s https://api.github.com/repos/tomwright/dasel/releases/latest | grep browser_download_url | grep linux_amd64 | cut -d '"' -f 4 | wget -qi - && mv dasel_linux_amd64 dasel && chmod +x dasel
mv ./dasel /usr/local/bin/dasel
MacOS
You may have to brew install wget in order for this to work.
curl -s https://api.github.com/repos/tomwright/dasel/releases/latest | grep browser_download_url | grep darwin_amd64 | cut -d '"' -f 4 | wget -qi - && mv dasel_darwin_amd64 dasel && chmod +x dasel
mv ./dasel /usr/local/bin/dasel
Windows 10
The most convenient installation method is via scoop command-line installer. Issue the following commands in your terminal and dasel will be available:
scoop bucket add extras
scoop install dasel
You can then use
scoop update dasel
to update later on.
Self Update
If you have dasel installed you can easily upgrade to the latest release using:
dasel update
If you have a development version of dasel you should run the following:
dasel update --dev
Note: This is a dasel feature and will cause the current executable to be replaced. This may cause side effects when used in conjuction with a package manager.
Development Version
You can go get the main package to build and install dasel for you.
go get github.com/tomwright/dasel/cmd/dasel@master
You may need to prefix the command with GO111MODULE=on in order for this to work.
Note that doing this will set the version to development
Using dasel as a go package
As with any other go package, just use go get.
go get github.com/tomwright/dasel
Once imported you do something like the following:
package main
import (
"encoding/json"
"fmt"
"github.com/tomwright/dasel"
)
func main() {
var data interface{}
_ = json.Unmarshal([]byte(`[{"name": "Tom"}, {"name": "Jim"}]`), &data)
rootNode := dasel.New(data)
result, _ := rootNode.Query(".[0].name")
printNodeValue(result) // Tom
results, _ := rootNode.QueryMultiple(".[*].name")
printNodeValue(results...) // Tom\nJim
_ = rootNode.Put(".[0].name", "Frank")
printNodeValue(rootNode) // [map[name:Frank] map[name:Jim]]
_ = rootNode.PutMultiple(".[*].name", "Joe")
printNodeValue(rootNode) // [map[name:Joe] map[name:Joe]]
outputBytes, _ := json.Marshal(rootNode.InterfaceValue())
fmt.Println(string(outputBytes)) // [{"name":"Joe"},{"name":"Joe"}]
}
func printNodeValue(nodes ...*dasel.Node) {
for _, n := range nodes {
fmt.Println(n.InterfaceValue())
}
}
From then on the rest of the docs should be enough.
Just know that when using the command-line tool the -m,--multiple flag tells dasel to use QueryMultiple/PutMultiple instead of Query/Put.
If the information provided here isn't good enough please open a discussion.
Notes
Preserved formatting and ordering
The formatting of files can be changed while being processed. Dasel itself doesn't make these changes, rather the act of marshaling the results.
In short, the output files may have properties in a different order but the actual contents will be as expected.
Memory usage
Dasel's method of querying data requires that the entire input document is stored in memory.
You should keep this in mind as the maximum filesize it can process will be limited by your system's available resources (specifically RAM).
Converting between formats
Dasel allows you to specify different input/output formats using the -r,--read and -w,--write flags.
E.g.
echo '{"name": "Tom"}{"name": "Jim"}' | dasel -r json -w yaml .
name: Tom
---
name: Jim
This works well in general but you may run into issues when converting between data formats that don't typically play well together.
If you have any questions or concerns around this please raise a discussion.
Usage
dasel -h
An important note is that if no sub-command is given, dasel will default to select.
Select
dasel select -f <file> -p <parser> -r <read_parser> -w <write_parser> -m <selector>
Arguments
-f, --file
Specify the file to query. This is required unless you are piping in data.
If piping in data you can optionally pass -f stdin/-f -.
-r, --read
Specify the parser to use when reading the input data.
This is required if you are piping in data, otherwise dasel will use the given file extension to guess which parser to use.
See supported parsers.
-w, --write
Specify the parser to use when writing the output data.
If not provided dasel will attempt to use the --out and --read flags to determine which parser to use.
See supported parsers.
-p, --parser
Shorthand for -r <value> -w <value>
-m, --multiple
Tells dasel to select multiple items.
This causes the dynamic selector to return all matching results rather than the first, and enables the any index selector.
All matches will be output on a new line.
E.g.
echo '[{"name": "Tom"}, {"name": "Jim"}]' | dasel -p json -m '.[*].name'
"Tom"
"Jim"
-s, --selector, <selector>
Specify the selector to use. See Selectors for more information.
If no selector flag is given, dasel assumes the first argument given is the selector.
This is required.
--plain
By default, dasel formats the output using the specified parser.
If this flag is used no formatting occurs and the results output as a string.
-n, --null
This flag tells dasel to output null instead of ValueNotFound errors.
With the flag:
$ echo '[1]' | dasel -p json -n '.[1]'
null
Without the flag:
$ echo '[1]' | dasel -p json '.[1]'
Error: could not query node: could not find value: no value found for selector: .[1]: [1]
Usage:
dasel select -f <file> -p <json,yaml> -s <selector> [flags]
Flags:
-f, --file string The file to query.
-h, --help help for select
-m, --multiple Select multiple results.
-n, --null Output null instead of value not found errors.
-p, --parser string Shorthand for -r FORMAT -w FORMAT.
--plain Alias of -w plain
-r, --read string The parser to use when reading.
-s, --selector string The selector to use when querying the data structure.
-w, --write string The parser to use when writing.
Error: could not write custom output: could not query node: could not find value: no value found for selector: .[1]: [1]
exit status 1
-c, --compact
This tells dasel to output compact data where possible. E.g. not pretty printing JSON.
Example
Select the image within a kubernetes deployment manifest file:
dasel select -f deployment.yaml "spec.template.spec.containers.(name=auth).image"
"tomwright/auth:v1.0.0"
Piping data into the select:
cat deployment.yaml | dasel select -p yaml "spec.template.spec.containers.(name=auth).image"
"tomwright/auth:v1.0.0"
Put
dasel put <type> -f <file> -o <out> -p <parser> -m <selector> <value>
echo "name: Tom" | ./dasel put string -p yaml "name" Jim
name: Jim
Arguments
type
The type of value you want to put.
Available arguments:
stringintboolobject- see Put Object
-f, --file
Specify the file to query. This is required unless you are piping in data.
If piping in data you can optionally pass -f stdin/-f -.
-o, --out
Specify the output file. If present, results will be written to the given file. If not present, results will be written to the input file (or stdout if none given).
To force output to be written to stdout, pass -o stdout/-o -.
-r, --read
Specify the parser to use when reading the input data.
This is required if you are piping in data, otherwise dasel will use the given file extension to guess which parser to use.
See supported parsers.
-w, --write
Specify the parser to use when writing the output data.
If not provided dasel will attempt to use the --out and --read flags to determine which parser to use.
See supported parsers.
-p, --parser
Shorthand for -r <value> -w <value>
-m, --multiple
Tells dasel to put multiple items.
This causes the dynamic selector to return all matching results rather than the first, and enables the any index selector.
E.g.
echo '[{"name": "Tom"}, {"name": "Jim"}]' | dasel put string -p json -m '.[*].name' Frank
[
{
"name": "Frank"
},
{
"name": "Frank"
}
]
-s, --selector, <selector>
Specify the selector to use. See Selectors for more information.
If no selector flag is given, dasel assumes the first argument given is the selector.
This is required.
-c, --compact
This tells dasel to output compact data where possible. E.g. not pretty printing JSON.
value
The value to write.
Dasel will parse this value as a string, int, or bool from this value depending on the given type.
This is required.
Creating properties
When putting data dasel will create items if they don't exist.
You can create an entire record from scratch by piping in an empty record, and then piping dasel commands together.
echo '' | dasel put string -p yaml -s '.propa' A | dasel put string -p yaml -s '.propb' B
propa: A
propb: B
This can be used to change multiple values or to create an entire document.
Put Object
Putting objects works slightly differently to a standard put, but the same principles apply.
dasel put object -f <file> -o <out> -p <parser> -m -t <type> <selector> <values>
If you want to create an empty object just omit the type flag and the values.
Arguments
-t, --type
The type of value you want to put.
You must repeat this argument for each value provided.
Available arguments:
stringintbool
-f, --file
Specify the file to query. This is required unless you are piping in data.
If piping in data you can optionally pass -f stdin/-f -.
-o, --out
Specify the output file. If present, results will be written to the given file. If not present, results will be written to the input file (or stdout if none given).
To force output to be written to stdout, pass -o stdout/-o -.
-r, --read
Specify the parser to use when reading the input data.
This is required if you are piping in data, otherwise dasel will use the given file extension to guess which parser to use.
See supported parsers.
-w, --write
Specify the parser to use when writing the output data.
If not provided dasel will attempt to use the --out and --read flags to determine which parser to use.
See supported parsers.
-p, --parser
Shorthand for -r <value> -w <value>
-m, --multiple
Tells dasel to put multiple items.
This causes the dynamic selector to return all matching results rather than the first, and enables the any index selector.
E.g.
echo '[{"name": "Tom"}, {"name": "Jim"}]' | dasel put object -p json -m -t string '.[*]' 'name=Frank'
[
{
"name": "Frank"
},
{
"name": "Frank"
}
]
-s, --selector, <selector>
Specify the selector to use. See Selectors for more information.
If no selector flag is given, dasel assumes the first argument given is the selector.
This is required.
-c, --compact
This tells dasel to output compact data where possible. E.g. not pretty printing JSON.
values
A space-separated list of key=value pairs.
Dasel will parse each value as a string, int, or bool depending on the related type.
Example
echo "" | dasel put object -p yaml -t string -t int "my.favourites" colour=red number=3
Results in the following:
my:
favourites:
colour: red
number: 3
Put Document
You are able to put entire documents using put document.
dasel put document -f <file> -o <out> -p <parser> -m <selector> -d <document-parser> <document>
Arguments
-f, --file
Specify the file to query. This is required unless you are piping in data.
If piping in data you can optionally pass -f stdin/-f -.
-o, --out
Specify the output file. If present, results will be written to the given file. If not present, results will be written to the input file (or stdout if none given).
To force output to be written to stdout, pass -o stdout/-o -.
-d, --document-parser, <document-parser>
Specify the parser to use when reading the document value.
If no value is provided, the read parser is used.
See supported parsers.
-r, --read
Specify the parser to use when reading the input data.
This is required if you are piping in data, otherwise dasel will use the given file extension to guess which parser to use.
See supported parsers.
-w, --write
Specify the parser to use when writing the output data.
If not provided dasel will attempt to use the --out and --read flags to determine which parser to use.
See supported parsers.
-p, --parser
Shorthand for -r <value> -w <value>
-m, --multiple
Tells dasel to put multiple items.
This causes the dynamic selector to return all matching results rather than the first, and enables the any index selector.
-s, --selector, <selector>
Specify the selector to use. See Selectors for more information.
If no selector flag is given, dasel assumes the first argument given is the selector.
This is required.
-c, --compact
This tells dasel to output compact data where possible. E.g. not pretty printing JSON.
document, <document>
The document you want to put, as a string.
This is required.
Example
echo '{"people":[]}' | dasel put document -p json -d yaml '.people.[]' 'name: Tom
colours:
- red
- green
- blue'
Results in the following:
{
"people": [
{
"colours": [
"red",
"green",
"blue"
],
"name": "Tom"
}
]
}
Supported file types
Dasel attempts to find the correct parser for the given file type, but if that fails you can choose which parser to use with the -p or --parser flag.
JSON
-p json
Using golang.org/pkg/encoding/json.
Multi-document files
Multi-document files are decoded into an array, with [0] being the first document, [1] being the second and so on.
Once decoded, you can access them using any of the standard selectors provided by Dasel.
TOML
-p toml
Using github.com/pelletier/go-toml.
YAML
-p yaml
Using gopkg.in/yaml.v2.
Multi-document files
Multi-document files are decoded into an array, with [0] being the first document, [1] being the second and so on.
Once decoded, you can access them using any of the standard selectors provided by Dasel.
XML
-p xml
Using github.com/clbanning/mxj.
XML Documents
XML documents within dasel are stored as a map of values.
This is just how dasel stores data and is required for the general functionality to work. An example of a simple documents representation is as follows:
<Person active="true">
<Name main="yes">Tom</Name>
<Age>27</Age>
</Person>
map[
Person:map[
-active:true
Age:27
Name:map[
#text:Tom
-main:true
]
]
]
In general this won't affect you, but on the odd occasion in specific instances it could lead to unexpected output.
If you are struggling with this please open a discussion for support. This will also help me know when the docs aren't sufficient.
Debugging
You can run select commands with the --plain flag to see the raw data that is stored within dasel. This can help you figure out the exact properties you may need to target when it isn't immediately obvious.
Arrays/Lists
Due to the way that XML is decoded, dasel can only detect something as a list if there are at least 2 items.
If you try to use list selectors (dynamic, index, append) when there are less than 2 items in the list you will get an error.
There are no plans to introduce a workaround for this but if there is enough demand it may be worked on in the future.
CSV
-p csv
Using golang.org/pkg/encoding/csv.
Plain
-p plain
This outputs the data using fmt.Sprint(x), displaying whatever underlying value is present as a string.
Adding data
New columns will be detected and added to the end of the CSV output.
Column deletion is not supported.
Selectors
Selectors define a path through a set of data.
Selectors are made up of different parts separated by a dot ., each part being used to identify the next node in the chain.
The following YAML data structure will be used as a reference in the following examples.
name: Tom
preferences:
favouriteColour: red
colours:
- red
- green
- blue
colourCodes:
- name: red
rgb: ff0000
- name: green
rgb: 00ff00
- name: blue
rgb: 0000ff
You can escape values in selectors using a backslash \. The main use for this is to allow you to target fields that contain a dot or space in their name.
Property
Property selectors are used to reference a single property of an object.
Just use the property name as a string.
dasel select -f ./tests/assets/example.yaml -s "name"
Tom
name==Tom
Keys and Indexes
You can use the property selector with a value of - to return a list of all the keys/indexes in the current node.
echo '{"a":{"c": [1, 2, 3]},"b":{}}' | dasel -p json -m '.a.c.-'
"0"
"1"
"2"
This must be used in conjunction with -m,--multiple and cannot be used in put commands.
Child Elements
Just separate the child element from the parent element using a .:
dasel select -f ./tests/assets/example.yaml -s "preferences.favouriteColour"
red
preferences.favouriteColour==red
Index
When you have a list, you can use square brackets to access a specific item in the list by its index.
dasel select -f ./tests/assets/example.yaml -s "colours.[1]"
green
colours.[0]==redcolours.[1]==greencolours.[2]==blue
Next Available Index
The next available index selector is used when adding to a list of items. It allows you to append to a list.
colours.[]
Any Index
The any index selector is used to select all items of a list or map.
colours.[*]
This must be used in conjunction with -m,--multiple.
Dynamic
Dynamic selectors are used with lists/maps when you don't know the index/property of the item, but instead want to find the index based on some other criteria.
Dasel currently supports key/query=value checks but I aim to support more check types in the future.
Look ups are defined in brackets. You can use multiple dynamic selectors within the same part to perform multiple checks.
dasel select -f ./tests/assets/example.yaml -s "colourCodes.(name=red).rgb"
ff0000
dasel select -f ./tests/assets/example.yaml -s "colourCodes.(name=blue)(rgb=0000ff)"
map[name:blue rgb:0000ff]
colourCodes.(name=red).rgb==ff0000colourCodes.(name=green).rgb==00ff00colourCodes.(name=blue).rgb==0000ffcolourCodes.(name=blue)(rgb=0000ff).rgb==0000ff
If you want to refer to the value of a non-object value in a list, just define the key as value or ., meaning the current value. This may look something like (value=2).
Using queries in dynamic selectors
When performing a check dasel creates a new root node at the current position and then selects data using the given key as the query.
This allows you to perform complex queries such as...
echo `{
"users": [
{
"name": {
"first": "Tom",
"last": "Wright"
},
"addresses": [
{
"primary": true,
"number": 123
},
{
"primary": false,
"number": 456
}
]
}
]
}` | dasel -p json '.users.(.addresses.(.primary=true).number=123).name.first'
"Tom"
The above query in plain English may read as...
Give me the first name of the user who's primary address is at number 123
The resolution of that query looks something like this:
.users.(.addresses.(.primary=true).number=123).name.first
.users.(.addresses.[0].number=123).name.first
.users.[0].name.first
Search
Search selectors recursively search all the data below the current node and returns all the results - this means they can only be used in multi select/put commands.
The syntax is as follows:
.(?:key=value)
If key is:
.orvalue- dasel checks if the current nodes value isvalue.-orkeyValue- dasel checks if the current nodes key/name/index value isvalue.- Else dasel uses the
keyas a selector itself and compares the result againstvalue.
Search Example
{
"users": [
{
"primary": true,
"name": {
"first": "Tom",
"last": "Wright"
}
},
{
"primary": false,
"extra": {
"name": {
"first": "Joe",
"last": "Blogs"
}
},
"name": {
"first": "Jim",
"last": "Wright"
}
}
]
}
Search for all objects with a key of name and output the first name of each:
dasel -p json -m '.(?:-=name).first'
"Tom"
"Joe"
"Jim"
Search for all objects with a last name of Wright and output the first name of each:
dasel -p json -m '.(?:name.last=Wright).name.first'
"Tom"
"Jim"
Examples
General
Filter JSON API results
The following line will return the download URL for the latest macOS dasel release:
curl https://api.github.com/repos/tomwright/dasel/releases/latest | dasel -p json --plain '.assets.(name=dasel_darwin_amd64).browser_download_url'
jq to dasel
The follow examples show a set of jq commands and the equivalent in dasel.
Select a single value
echo '{"name": "Tom"}' | jq '.name'
"Tom"
echo '{"name": "Tom"}' | dasel -p json '.name'
"Tom"
Select a nested value
echo '{"user": {"name": "Tom", "age": 27}}' | jq '.user.age'
27
echo '{"user": {"name": "Tom", "age": 27}}' | dasel -p json '.user.age'
27
Select an array index
echo '[1, 2, 3]' | jq '.[1]'
2
echo '[1, 2, 3]' | dasel -p json '.[1]'
2
Append to an array of strings
echo '["a", "b", "c"]' | jq '. += ["d"]'
[
"a",
"b",
"c",
"d"
]
echo '["a", "b", "c"]' | dasel put string -p json -s '.[]' d
[
"a",
"b",
"c",
"d"
]
Update a string value
echo '["a", "b", "c"]' | jq '.[1] = "d"'
[
"a",
"d",
"c"
]
echo '["a", "b", "c"]' | dasel put string -p json -s '.[1]' d
[
"a",
"d",
"c"
]
Update an int value
echo '[1, 2, 3]' | jq '.[1] = 5'
[
1,
5,
3
]
echo '[1, 2, 3]' | dasel put int -p json -s '.[1]' 5
[
1,
5,
3
]
Overwrite an object
echo '{"user": {"name": "Tom", "age": 27}}' | jq '.user = {"name": "Frank", "age": 25}'
{
"user": {
"name": "Frank",
"age": 25
}
}
echo '{"user": {"name": "Tom", "age": 27}}' | dasel put object -p json -s '.user' -t string -t int name=Frank age=25
{
"user": {
"age": 25,
"name": "Frank"
}
}
Append to an array of objects
echo '{"users": [{"name": "Tom"}]}' | jq '.users += [{"name": "Frank"}]'
{
"users": [
{
"name": "Tom"
},
{
"name": "Frank"
}
]
}
echo '{"users": [{"name": "Tom"}]}' | dasel put object -p json -s '.users.[]' -t string name=Frank
{
"users": [
{
"name": "Tom"
},
{
"name": "Frank"
}
]
}
yq to dasel
The follow examples show a set of yq commands and the equivalent in dasel.
Select a single value
echo 'name: Tom' | yq '.name'
"Tom"
echo 'name: Tom' | dasel -p yaml '.name'
Tom
Select a nested value
echo 'user:
name: Tom
age: 27' | yq '.user.age'
27
echo 'user:
name: Tom
age: 27' | dasel -p yaml '.user.age'
27
Select an array index
echo '- 1
- 2
- 3' | yq '.[1]'
2
echo '- 1
- 2
- 3' | dasel -p yaml '.[1]'
2
Append to an array of strings
echo '- a
- b
- c' | yq --yaml-output '. += ["d"]'
- a
- b
- c
- d
echo '- a
- b
- c' | dasel put string -p yaml -s '.[]' d
- a
- b
- c
- d
Update a string value
echo '- a
- b
- c' | yq --yaml-output '.[1] = "d"'
- a
- d
- c
echo '- a
- b
- c' | dasel put string -p yaml -s '.[1]' d
- a
- d
- c
Update an int value
echo '- 1
- 2
- 3' | yq --yaml-output '.[1] = 5'
- 1
- 5
- 3
echo '- 1
- 2
- 3' | dasel put int -p yaml -s '.[1]' 5
- 1
- 5
- 3
Overwrite an object
echo 'user:
name: Tom
age: 27' | yq --yaml-output '.user = {"name": "Frank", "age": 25}'
user:
name: Frank
age: 25
echo 'user:
name: Tom
age: 27' | dasel put object -p yaml -s '.user' -t string -t int name=Frank age=25
user:
age: 25
name: Frank
Append to an array of objects
echo 'users:
- name: Tom' | yq --yaml-output '.users += [{"name": "Frank"}]'
users:
- name: Tom
- name: Frank
echo 'users:
- name: Tom' | dasel put object -p yaml -s '.users.[]' -t string name=Frank
users:
- name: Tom
- name: Frank
Kubernetes
The following should work on a kubernetes deployment manifest. While kubernetes isn't for everyone, it does give some good example use-cases.
Select the image for a container named auth
dasel select -f deployment.yaml -s "spec.template.spec.containers.(name=auth).image"
tomwright/x:v2.0.0
Change the image for a container named auth
dasel put string -f deployment.yaml -s "spec.template.spec.containers.(name=auth).image" "tomwright/x:v2.0.0"
Update replicas to 3
dasel put int -f deployment.yaml -s "spec.replicas" 3
Add a new env var
dasel put object -f deployment.yaml -s "spec.template.spec.containers.(name=auth).env.[]" -t string -t string name=MY_NEW_ENV_VAR value=MY_NEW_VALUE
Update an existing env var
dasel put string -f deployment.yaml -s "spec.template.spec.containers.(name=auth).env.(name=MY_NEW_ENV_VAR).value" NEW_VALUE
XML Examples
XML has some slight differences (such as attributes) that should be documented.
Query attributes
Decoded attributes are set as properties on the related object with a prefix of -.
echo '<data>
<users primary="true">
<name>Tom</name>
</users>
<users primary="false">
<name>Frank</name>
</users>
</data>' | dasel -p xml '.data.users[0].-primary'
true
Filtering on attributes
We can also filter on attributes since they are defined against the related object.
echo '<data>
<users primary="true">
<name>Tom</name>
</users>
<users primary="false">
<name>Frank</name>
</users>
</data>' | dasel -p xml '.data.users.(-primary=true).name'
Tom
Benchmarks
In my tests dasel has been up to 3x faster than jq and 15x faster than yq.
See the benchmark directory.
![`dasel put -t json -s 'array.[]' -v](https://avatars.githubusercontent.com/u/14150873?v=4)
