Tips, Tricks, Troubleshooting
Yaml files can be surprisingly lenient in what can be parsed as a yaml file. A reasonable way of validation a yaml file is to ensure the top level is a map or array (although it is valid yaml to have scalars at the top level, but often this is not what you want). This can be done by:
yq --exit-status 'tag == "!!map" or tag== "!!seq"' file.txt > /dev/null
Feel free to use multiple lines in your expression to improve readability.
Use
with if you need to make several updates to the same path.yq --inplace '
with(.a.deeply.nested;
. = "newValue" | . style="single") |
with(.b.another.nested;
. = "cool" | . style="folded")
' my_file.yaml
Given a yaml file like
coolActions:
- create
- edit
- delete
You can create a bash array named
actions by:> readarray actions < <(yq '.coolActions[]' sample.yaml)
> echo "${actions[1]}"
edit
For a given yaml file like:
identities:
- arn: "arn:aws:iam::ARN1"
group: "system:masters"
user: "user1"
- arn: "arn:aws:iam::ARN2"
group: "system:masters"
user: "user2"
You can loop over the results in a bash loop like:
# load array into a bash array
# output each entry as a single line json
readarray identityMappings < <(yq -o=j -I=0 '.identities[]' test.yml )
for identityMapping in "${identityMappings[@]}"; do
# identity mapping is a single json snippet representing a single entry
roleArn=$(echo "$identityMapping" | yq '.arn' -)
echo "roleArn: $roleArn"
done
The
strenv operator is a great way to handle special characters in strings:VAL='.a |[email protected] == "string2"' yq '.a = strenv(VAL)' example.yaml
yq doesn't have a way of updating multiple files in a single command (yet?) - but you can use your shell's built in tools like find:find *.yaml -exec yq '. += "cow"' -i {} \;
This will run the
'. += "cow"' expression against every matching file, and update it in place (-i).There are a couple of tricks to getting the right string representation, take a look at string operators for more details:
PS > yq -n '.test = ""something""'
test: something
PS >
To merge all given yaml files into one, use the
reduce operator with the * (multiply) operator. Note the use of ea or eval-all to load all files into memory so that they can be merged.yq ea '. as $item ireduce ({}; . * $item )' file1.yml file2.yml ...
To see the original source file and line number of your merged result, you can pre-process the files and add that information in as line comments, then perform the merge.
yq ea '(.. lineComment |= filename + ":" + line) | select(fi==0) * select(fi==1)' data1.yaml data2.yaml
To create a new
yaml file simply:yq -n '.someNew="content"' > newfile.yml
The best way to run a diff is to use
yq to normalise the yaml files and then just use diff. Here is a simple example of using pretty print -P to normalise the styling and running diff:diff <(yq -P 'sort_keys(..)' file1.yaml) <(yq -P 'sort_keys(..)' file2.yaml)
This way you can use the full power of
diff and normalise the yaml files as you like.You may also want to remove all comments using
... comments=""Like
diff and other bash commands, you can use <(exp) to pipe in multiple streams of data into yq. instance:yq '.apple' <(curl -s https://somewhere/data1.yaml) <(cat file.yml)
The most important thing to remember to do is to have brackets around the LHS expression - otherwise what
yq will do is first filter by the selection, and then, separately, update the filtered result and return that subset.yq '(.foo.bar[] | select(name == "fred) | .apple) = "cool"'
In order to combine multiple yaml files into a single file (with
--- separators) you can just:yq '.' somewhere/*.yaml
yq 'with(.a.deeply ; .nested = "newValue" | .other= "newThing")' sample.yml
The first argument expression sets the root context, and the second expression runs against that root context.
yq has not yet added if expressions - however you should be able to use with and select to achieve the same outcome. Lets use an example:- animal: cat
- animal: dog
- animal: frog
Now, if you were using good ol' jq - you may have a script with
ifs like so:jq ' .[] |=
if (.animal == "cat") then
.noise = "meow" |
.whiskers = true
elif (.animal == "dog") then
.noise = "woof" |
.happy = true
else
.noise = "??"
end
' < file.yaml
Using
yq - you can get the same result by:yq '.[] |= (
with(select(.animal == "cat");
.noise = "meow" |
.whiskers = true
) |
with(select(.animal == "dog");
.noise = "woof" |
.happy = true
) |
with(select(.noise == null);
.noise = "???"
)
)' < file.yml
Note that the logic isn't quite the same, as there is no concept of 'else'. So you may need to put additional logic in the expressions, as this has for the 'else' logic.
The merge functionality from yaml v1.1 (e.g.
<<:has actually been removed in the 1.2 spec. Thankfully, yq underlying yaml parser still supports that tag - and it's extra nice in that it explicitly puts the !!merge tag on key of the map entry. This tag tells other yaml parsers that this entry is a merge entry, as opposed to a regular string key that happens to have a value of <<:. This is backwards compatible with the 1.1 spec of yaml, it's simply an explicit way of specifying the type (for instance, you can use a !!str tag to enforce a particular value to be a string.Although this does affect the readability of the yaml to humans, it still works and processes fine with various yaml processors.
Last modified 2mo ago