The query language Unquery is a fast, powerful and flexible language for querying structured documents. It was originally developed as the query language for XCiteDB, a fast and reliable noSQL database for structured documents, supporting XML, JSON, and other features such as temporal versioning and branches.
This repository contains unq, a stand-alone command-line tool for querying, transforming and analyzing JSON files using Unquery. It reads either a single or multiple JSON files and an Unquery query, and produce a single JSON output based on the query.
To build the tool, you'll need cmake version 3.6.2 or higher. You can install it with the package manager on most Linux distros, or follow the directions here.
To build
cd unq
cmake .
makeThe executable would be in bin/Release/unq.
To run unq, you'll need to provide the query, either as a file (with the option -f) or as a command-line option (with the option -c), and a list of files to query. For example, if you want to collect the value of the field firstName from all the json files in the current directory, the query you need is:
["firstName"]
You can store the above query in a file, say query.unq, and then run:
unq -f query.unq *.json
Or alternatively, you can write:
unq -c '["firstName"]' *.json
It also allows to query CSV files by converting them on-the-fly to JSON and then processing the query. Example for a command line for processing CSV:
unq -f query.unq -csv -delim ";" users.csv
There already over a dozen different quey languages for json. Why do we need yet another one?
While there are a lot of query languages for json, Unquery is different. And, we believe, better. The approach we used for designing Unquery is based on two principles:
- Any Unquery query is a well-formed json.
- The structure of the query matches the structure of the result.
The advantages of Unquery are:
- Readability: because te structure of the query matches the structure of the result, it is easy to unserstand what the query does.
- Succint: complex queries can be writen using relatively shot queries.
- Simple: there are only few constructs the user need to learn to write complex queries.
- Powerfull: we do not compromise on the expressive power of the language.
- Recursive structure: json is recursive, and allows unlimited nesting depth of objects and arrays. Similarly, Unquery is capable of querying nested arrays and objects with ease.
- Guarantee to produce well-formed json.
- Fully embrace json: query json files with queries written in json, and results in json.
- Fast. The implementation is written in C++, and use RapidJSON, the fastest c++ json library, to parse json files.
- The same query language is used in XCiteDB, a powerfull database for XML and JSON.
What are the disadvantages of Unquery, compared to other query languages?
Unquery does have some disadvantages as well:
- The queries are written in json, and may come with some syntactic overhead. Mostly, lots of double quotes, which are required for strings.
- It can only produce well-formed json. So if you need non-json results, it wouldn't be a good match.
- The implmentation is in C++ (and not Javascript). So it cannot run on the client's side inside the browser.
It looks like unq is still being developped. Is it ready to be used?
The implementation is fully working. All the features shown in the tutorial (and many more) are already implemented. It is still a beta version, and may still have bugs. The more you use it, the better it would become.
Where is the reference manual?
The full language manual is at here: Language Reference
Also, check out the tutorial for more examples.
Having a command line utility is cool. But I want to use it in a Python/NodeJS/C++ project. Do you plan to have language bindings for those languages?
Yes.
Can I use it? Do I need to pay?
unq is an open source project, disctributed under the terms Apache 2.0. So it is completely free.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent