[brianparks]

From your page: "I tried jq, which seems very good. However I was not quite able to grasp its syntax, so I gave up searching and I created my own."

I use jq on an almost daily basis, so I can sympathize with your plight. jq's syntax is documented quite well (if rather succinctly), but one non-intuitive thing is that it is possible to end up with non-JSON output if you use jq's array ([]) operator rather than the map() filter. However, I still prefer its syntax as one pipeline similar to UNIX one-liners rather than several command-line args, probably purely out of similarity to how I think about the data. I think about its transformations, rather than as independent selection and projection steps. This may be personal preference, but I'm curious as to your thoughts on syntactical choices.