fzf is an interactive tool for fuzzy-finding and will take input from stdin. For example, `ls | fzf` will allow you to search through a list of files in your working directory. It also has an optional preview pane that can be used to show the output of a command.
In this case, the empty string is provided as input because we're not interested in selecting options. You can then type your jq filter in fzf's query box and the command will run against filename.json in the preview pane on every kepress (the {q} is your currently typed query).
I sometimes use jiq for this purpose, but I don't like that it tries to evaluate the query as you type, pressing enter exits the interactive session, and that you can't see the source and output at the same time. Jid has similar issues.
Jqp seems to solve all of these problems, plus it has syntax highlighting. I'm going to use this tool all the time.
I've played with jid in jiq and IIRC they were very slow with large inputs since they re-parse the JSON blob every time the query changes. I wonder if this solves that problem--I'll bookmark it and find out next time I need to wrangle some JSON!
wow, thanks so much for the kind words! Didn't think people would see this, let alone for it to show up on Hacker News haha. Hope you get a lot of use out of it!
I often write some pretty complex jq queries. Once a query starts getting complex, I like to spread it over multiple lines for better readability. It would be cool if Jqp would let me insert newlines by pressing Shift+Enter.
Yeah for sure! Initially I had tried to do something like this, hit a roadblock and just went with the 1 line input in order to get a v1 out. Definitely something I'd consider as I continue to iterate on it.
I had fun writing a tree-sitter grammar[1] for jq, mostly based on the gojq parser. I realized it was probably in pretty good shape when a snippet that I thought was valid had a syntax error, which tree-sitter was happy to confirm. jq is a lot of fun to write and think with, and the stdlib is good but quirky. It would really benefit from some more libraries, so it's great to see tools to facilitate writing in it!
What would be really awesome (although likely means ripping apart jq) would be some intermediate query caching. I often browse through large-ish (~1-2GB) JSON files and do something similar to the example animation in that I repeatedly filter and look for something. It would be awesome if it were smart enough to reasonable cache "filter A" when later reasoning about "filter A and filter B".
jq is a tool that can query (search/extract/format) JSON (JavaScript readable data files).
I figured from the README that TUI probably stands for Text User Interface as it seems to support keystrokes in a way that an editor like Vi or Emacs work in a terminal (yeah I hadn't of it either, but I guess it is a retro acronym from GUI which is more well known)
echo '' | fzf --preview 'jq {q} < filename.json'
This approach can be generalized to other tools, to some extent: https://github.com/pawelduda/fzf-live-repl