I've a YAML file representing the releases of a project:
releases:
- name: v0.1.0
changes:
added:
- Second feature (#4)
fixed:
- Second fix (#5)
- name: v0.0.1
changes:
fixed:
- First fix (#3)
security:
- name: v0.0.1
changes:
added:
- Bootstrap the project (#1)
- First feature (#2)
I'm searching the correct recipe to get the following result:
v0.1.0
Second feature (#4)
Second fix (#5)
v0.0.1
First fix (#3)
v0.0.1
Bootstrap the project (#1)
First feature (#2)
For now, the closest I've gotten is:
v0.1.0
v0.0.1
v0.0.1
Second feature (#4)
Bootstrap the project (#1)
First feature (#2)
Second fix (#5)
First fix (#3)
With the following recipe:
.releases[] | (
.name,
.changes.added[],
.changes.fixed[]
)
From what I understand, with my current recipe, all the items in releases will be provided to the filter .name, then all the items in releases will be provided to .changes.added[], etc.
Is there a way to provide one item to the three filters, then another one, etc.?
Thanks in advance!
Edit (October 1st, 2023): I'm using mikefarah flavor of yq.
Your expected result is the default behavior with jq, and thus also with kislyuk/yq. With mikefarah/yq, however, it's different, but you can induce this behavior by storing the iteration value in a variable:
.releases[] as $r | (
$r.name,
$r.changes.added[],
$r.changes.fixed[]
)
v0.1.0
Second feature (#4)
Second fix (#5)
v0.0.1
First fix (#3)
v0.0.1
Bootstrap the project (#1)
First feature (#2)
Note: Depending on your actual use-case (especially wrt "interfering" values), you also might want to consider employing a recursive descent .., which traverses the tree in document order. Reduce this stream by using a reasonable discriminator, e.g. for this sample, it suffices filtering for strings to get your desired output:
.releases | .. | select(tag == "!!str")