logctx
Package logctx provides a way to decorate structured log entries with metadata
added to a context.Context.
Currently, this works with Zap only. But it would be trivial to support other structured logging libraries.
WithMeta creates a new context which contains a hash table of arbitrary metadata strings which can later be easily added to a structured log entry.
For example, if you want to decorate a call tree with some data:
func DoBusinessLogic(ctx context.Context, userID string) error {
ctx = logctx.WithMeta(ctx, logctx.Meta{"user_id": userID})
GetResource(ctx, ...)
}
You can wrap contexts with this helper as much as you want:
func DoBusinessLogic(ctx context.Context, userID string) error {
ctx = logctx.WithMeta(ctx, logctx.Meta{"user_id": userID})
GetResource(ctx, ...)
}
func GetResource(ctx context.Context, ...) {
ctx = logctx.WithMeta(ctx, logctx.Meta{"something_else": xyz})
CallAnotherThing(ctx, ...)
}
Then, when you need to log it out, use logctx.Zap.
Zap will wrap your Zap log fields with any available metadata from the given
context. Any context returned from calls to WithMeta will work in this
function and provide a “context” field to the log entry. If the given context
was not decorated with WithMeta then this function does nothing and just
passes your fields unmodified.
It’s best used directly in a zap log call, with the spread operator:
func (s *service) DoBusinessLogic(ctx context.Context, userID string) error {
s.l.Info("i am doing the thing", logctx.Zap(
zap.String("event_specific", "information"),
)...)
}
In this example, assuming a function higher up in the call chain used the
WithMeta to add a user_id, the log entry for this will be:
{
"level": "info",
"msg": "i am doing the thing",
"context": {
"user_id": "the_user_id"
}
}
