Metadata-Version: 2.1
Name: liquer-framework
Version: 0.2.6
Summary: LiQuer - Query in (URL) link
Home-page: https://github.com/orest-d/liquer
Author: Orest Dubay
Author-email: orest3.dubay@gmail.com
License: UNKNOWN
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Description-Content-Type: text/markdown
Requires-Dist: flask

# LiQuer (Link Query) 

LiQuer is a minimalistic python framework on top of flask, which simplifies creation of web services and web interfaces,
particularly in connection to data science: working with data frames, charts and reports.
The core of LiQuer is a simplistic query language, that can represent a chain of transformations (*commands*)
applied to an arbitrary data object (e.g. a dataframes, text, json, images). LiQuer query is safe to use inside URLs
and is interpreted and executed by a LiQuer server that is available as a flask blueprint.
LiQuer is very modular and designed to be extremely easy to extend (typically just by simple decorators and use of conventions).
Results of LiQuer queries can be cached for added performance.

Please visit [LiQuer website](https://orest-d.github.io/liquer/) for more info.

# Install

```
pip install liquer-framework
```

# TODO

- [x] improve command catalogue (in progress)
- [ ] support for dataframe iterator (maybe)
- [x] namespaces
- [x] lift the requirement for a state to be cloneable - at least for some data types
- [x] make indicator of *volatile* state, that can never be cached (volatile=immutable?)
- [x] specify metadata in command decorator
- [x] cache categories/levels dependent on metadata
- [x] basic keras support
- [x] basic scikit support
- [x] composite state (djson)
- [x] meta extension
- [x] command/query GUI
- [x] simple GUI
- [ ] remote liquer service as cache
- [ ] automatic dependent query recording
- [ ] implied volatile (query result dependent on a volatile query should be volatile)
- [ ] attributes to prevent cloning and caching
- [ ] prevent crashing on non-cloneable types - automatic volatile ?
- [ ] keras history support
- [ ] numpy support
- [ ] fix volatility logic - now just the state does not get cloned 
- [ ] parquet support


