Use dynamic configuration model in bundles (#1098)
## Changes
This is a fundamental change to how we load and process bundle
configuration. We now depend on the configuration being represented as a
`dyn.Value`. This representation is functionally equivalent to Go's
`any` (it is variadic) and allows us to capture metadata associated with
a value, such as where it was defined (e.g. file, line, and column). It
also allows us to represent Go's zero values properly (e.g. empty
string, integer equal to 0, or boolean false).
Using this representation allows us to let the configuration model
deviate from the typed structure we have been relying on so far
(`config.Root`). We need to deviate from these types when using
variables for fields that are not a string themselves. For example,
using `${var.num_workers}` for an integer `workers` field was impossible
until now (though not implemented in this change).
The loader for a `dyn.Value` includes functionality to capture any and
all type mismatches between the user-defined configuration and the
expected types. These mismatches can be surfaced as validation errors in
future PRs.
Given that many mutators expect the typed struct to be the source of
truth, this change converts between the dynamic representation and the
typed representation on mutator entry and exit. Existing mutators can
continue to modify the typed representation and these modifications are
reflected in the dynamic representation (see `MarkMutatorEntry` and
`MarkMutatorExit` in `bundle/config/root.go`).
Required changes included in this change:
* The existing interpolation package is removed in favor of
`libs/dyn/dynvar`.
* Functionality to merge job clusters, job tasks, and pipeline clusters
are now all broken out into their own mutators.
To be implemented later:
* Allow variable references for non-string types.
* Surface diagnostics about the configuration provided by the user in
the validation output.
* Some mutators use a resource's configuration file path to resolve
related relative paths. These depend on `bundle/config/paths.Path` being
set and populated through `ConfigureConfigFilePath`. Instead, they
should interact with the dynamically typed configuration directly. Doing
this also unlocks being able to differentiate different base paths used
within a job (e.g. a task override with a relative path defined in a
directory other than the base job).
## Tests
* Existing unit tests pass (some have been modified to accommodate)
* Integration tests pass
2024-02-16 19:41:58 +00:00
|
|
|
package mutator
|
|
|
|
|
|
|
|
import (
|
|
|
|
"context"
|
|
|
|
|
|
|
|
"github.com/databricks/cli/bundle"
|
|
|
|
"github.com/databricks/cli/libs/dyn"
|
|
|
|
"github.com/databricks/cli/libs/dyn/convert"
|
|
|
|
"github.com/databricks/cli/libs/dyn/dynvar"
|
2024-02-19 10:44:51 +00:00
|
|
|
"github.com/databricks/cli/libs/log"
|
Use dynamic configuration model in bundles (#1098)
## Changes
This is a fundamental change to how we load and process bundle
configuration. We now depend on the configuration being represented as a
`dyn.Value`. This representation is functionally equivalent to Go's
`any` (it is variadic) and allows us to capture metadata associated with
a value, such as where it was defined (e.g. file, line, and column). It
also allows us to represent Go's zero values properly (e.g. empty
string, integer equal to 0, or boolean false).
Using this representation allows us to let the configuration model
deviate from the typed structure we have been relying on so far
(`config.Root`). We need to deviate from these types when using
variables for fields that are not a string themselves. For example,
using `${var.num_workers}` for an integer `workers` field was impossible
until now (though not implemented in this change).
The loader for a `dyn.Value` includes functionality to capture any and
all type mismatches between the user-defined configuration and the
expected types. These mismatches can be surfaced as validation errors in
future PRs.
Given that many mutators expect the typed struct to be the source of
truth, this change converts between the dynamic representation and the
typed representation on mutator entry and exit. Existing mutators can
continue to modify the typed representation and these modifications are
reflected in the dynamic representation (see `MarkMutatorEntry` and
`MarkMutatorExit` in `bundle/config/root.go`).
Required changes included in this change:
* The existing interpolation package is removed in favor of
`libs/dyn/dynvar`.
* Functionality to merge job clusters, job tasks, and pipeline clusters
are now all broken out into their own mutators.
To be implemented later:
* Allow variable references for non-string types.
* Surface diagnostics about the configuration provided by the user in
the validation output.
* Some mutators use a resource's configuration file path to resolve
related relative paths. These depend on `bundle/config/paths.Path` being
set and populated through `ConfigureConfigFilePath`. Instead, they
should interact with the dynamically typed configuration directly. Doing
this also unlocks being able to differentiate different base paths used
within a job (e.g. a task override with a relative path defined in a
directory other than the base job).
## Tests
* Existing unit tests pass (some have been modified to accommodate)
* Integration tests pass
2024-02-16 19:41:58 +00:00
|
|
|
)
|
|
|
|
|
|
|
|
type resolveVariableReferences struct {
|
|
|
|
prefixes []string
|
|
|
|
}
|
|
|
|
|
|
|
|
func ResolveVariableReferences(prefixes ...string) bundle.Mutator {
|
|
|
|
return &resolveVariableReferences{prefixes: prefixes}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (*resolveVariableReferences) Name() string {
|
|
|
|
return "ResolveVariableReferences"
|
|
|
|
}
|
|
|
|
|
|
|
|
func (m *resolveVariableReferences) Validate(ctx context.Context, b *bundle.Bundle) error {
|
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
func (m *resolveVariableReferences) Apply(ctx context.Context, b *bundle.Bundle) error {
|
|
|
|
prefixes := make([]dyn.Path, len(m.prefixes))
|
|
|
|
for i, prefix := range m.prefixes {
|
|
|
|
prefixes[i] = dyn.MustPathFromString(prefix)
|
|
|
|
}
|
|
|
|
|
|
|
|
// The path ${var.foo} is a shorthand for ${variables.foo.value}.
|
|
|
|
// We rewrite it here to make the resolution logic simpler.
|
|
|
|
varPath := dyn.NewPath(dyn.Key("var"))
|
|
|
|
|
|
|
|
return b.Config.Mutate(func(root dyn.Value) (dyn.Value, error) {
|
|
|
|
// Synthesize a copy of the root that has all fields that are present in the type
|
|
|
|
// but not set in the dynamic value set to their corresponding empty value.
|
|
|
|
// This enables users to interpolate variable references to fields that haven't
|
|
|
|
// been explicitly set in the dynamic value.
|
|
|
|
//
|
|
|
|
// For example: ${bundle.git.origin_url} should resolve to an empty string
|
|
|
|
// if a bundle isn't located in a Git repository (yet).
|
|
|
|
//
|
|
|
|
// This is consistent with the behavior prior to using the dynamic value system.
|
|
|
|
//
|
|
|
|
// We can ignore the diagnostics return valuebecause we know that the dynamic value
|
|
|
|
// has already been normalized when it was first loaded from the configuration file.
|
|
|
|
//
|
|
|
|
normalized, _ := convert.Normalize(b.Config, root, convert.IncludeMissingFields)
|
|
|
|
lookup := func(path dyn.Path) (dyn.Value, error) {
|
|
|
|
// Future opportunity: if we lookup this path in both the given root
|
|
|
|
// and the synthesized root, we know if it was explicitly set or implied to be empty.
|
|
|
|
// Then we can emit a warning if it was not explicitly set.
|
|
|
|
return dyn.GetByPath(normalized, path)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Resolve variable references in all values.
|
2024-02-19 10:44:51 +00:00
|
|
|
root, err := dynvar.Resolve(root, func(path dyn.Path) (dyn.Value, error) {
|
Use dynamic configuration model in bundles (#1098)
## Changes
This is a fundamental change to how we load and process bundle
configuration. We now depend on the configuration being represented as a
`dyn.Value`. This representation is functionally equivalent to Go's
`any` (it is variadic) and allows us to capture metadata associated with
a value, such as where it was defined (e.g. file, line, and column). It
also allows us to represent Go's zero values properly (e.g. empty
string, integer equal to 0, or boolean false).
Using this representation allows us to let the configuration model
deviate from the typed structure we have been relying on so far
(`config.Root`). We need to deviate from these types when using
variables for fields that are not a string themselves. For example,
using `${var.num_workers}` for an integer `workers` field was impossible
until now (though not implemented in this change).
The loader for a `dyn.Value` includes functionality to capture any and
all type mismatches between the user-defined configuration and the
expected types. These mismatches can be surfaced as validation errors in
future PRs.
Given that many mutators expect the typed struct to be the source of
truth, this change converts between the dynamic representation and the
typed representation on mutator entry and exit. Existing mutators can
continue to modify the typed representation and these modifications are
reflected in the dynamic representation (see `MarkMutatorEntry` and
`MarkMutatorExit` in `bundle/config/root.go`).
Required changes included in this change:
* The existing interpolation package is removed in favor of
`libs/dyn/dynvar`.
* Functionality to merge job clusters, job tasks, and pipeline clusters
are now all broken out into their own mutators.
To be implemented later:
* Allow variable references for non-string types.
* Surface diagnostics about the configuration provided by the user in
the validation output.
* Some mutators use a resource's configuration file path to resolve
related relative paths. These depend on `bundle/config/paths.Path` being
set and populated through `ConfigureConfigFilePath`. Instead, they
should interact with the dynamically typed configuration directly. Doing
this also unlocks being able to differentiate different base paths used
within a job (e.g. a task override with a relative path defined in a
directory other than the base job).
## Tests
* Existing unit tests pass (some have been modified to accommodate)
* Integration tests pass
2024-02-16 19:41:58 +00:00
|
|
|
// Rewrite the shorthand path ${var.foo} into ${variables.foo.value}.
|
|
|
|
if path.HasPrefix(varPath) && len(path) == 2 {
|
|
|
|
path = dyn.NewPath(
|
|
|
|
dyn.Key("variables"),
|
|
|
|
path[1],
|
|
|
|
dyn.Key("value"),
|
|
|
|
)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Perform resolution only if the path starts with one of the specified prefixes.
|
|
|
|
for _, prefix := range prefixes {
|
|
|
|
if path.HasPrefix(prefix) {
|
|
|
|
return lookup(path)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return dyn.InvalidValue, dynvar.ErrSkipResolution
|
|
|
|
})
|
2024-02-19 10:44:51 +00:00
|
|
|
if err != nil {
|
|
|
|
return dyn.InvalidValue, err
|
|
|
|
}
|
|
|
|
|
|
|
|
// Normalize the result because variable resolution may have been applied to non-string fields.
|
|
|
|
// For example, a variable reference may have been resolved to a integer.
|
|
|
|
root, diags := convert.Normalize(b.Config, root)
|
|
|
|
for _, diag := range diags {
|
|
|
|
// This occurs when a variable's resolved value is incompatible with the field's type.
|
|
|
|
// Log a warning until we have a better way to surface these diagnostics to the user.
|
|
|
|
log.Warnf(ctx, "normalization diagnostic: %s", diag.Summary)
|
|
|
|
}
|
|
|
|
return root, nil
|
Use dynamic configuration model in bundles (#1098)
## Changes
This is a fundamental change to how we load and process bundle
configuration. We now depend on the configuration being represented as a
`dyn.Value`. This representation is functionally equivalent to Go's
`any` (it is variadic) and allows us to capture metadata associated with
a value, such as where it was defined (e.g. file, line, and column). It
also allows us to represent Go's zero values properly (e.g. empty
string, integer equal to 0, or boolean false).
Using this representation allows us to let the configuration model
deviate from the typed structure we have been relying on so far
(`config.Root`). We need to deviate from these types when using
variables for fields that are not a string themselves. For example,
using `${var.num_workers}` for an integer `workers` field was impossible
until now (though not implemented in this change).
The loader for a `dyn.Value` includes functionality to capture any and
all type mismatches between the user-defined configuration and the
expected types. These mismatches can be surfaced as validation errors in
future PRs.
Given that many mutators expect the typed struct to be the source of
truth, this change converts between the dynamic representation and the
typed representation on mutator entry and exit. Existing mutators can
continue to modify the typed representation and these modifications are
reflected in the dynamic representation (see `MarkMutatorEntry` and
`MarkMutatorExit` in `bundle/config/root.go`).
Required changes included in this change:
* The existing interpolation package is removed in favor of
`libs/dyn/dynvar`.
* Functionality to merge job clusters, job tasks, and pipeline clusters
are now all broken out into their own mutators.
To be implemented later:
* Allow variable references for non-string types.
* Surface diagnostics about the configuration provided by the user in
the validation output.
* Some mutators use a resource's configuration file path to resolve
related relative paths. These depend on `bundle/config/paths.Path` being
set and populated through `ConfigureConfigFilePath`. Instead, they
should interact with the dynamically typed configuration directly. Doing
this also unlocks being able to differentiate different base paths used
within a job (e.g. a task override with a relative path defined in a
directory other than the base job).
## Tests
* Existing unit tests pass (some have been modified to accommodate)
* Integration tests pass
2024-02-16 19:41:58 +00:00
|
|
|
})
|
|
|
|
}
|