A lightweight Python-3.6+ lib for solving & executing graphs of functions
Project description
It’s a DAG all the way down!
Lightweight computation graphs for Python
Graphtik is a library to design, plot & execute graphs of functions (a.k.a pipeline) that consume and populate (possibly nested) data, based on whether values for those data (a.k.a dependencies) exist.
The API posits a fair compromise between features and complexity, without precluding any.
It can be used as is to build machine learning pipelines for data science projects.
It should be extendable to act as the core for a custom ETL engine, a workflow-processor for interdependent tasks & files like GNU Make, or a spreadsheet calculation engine.
Features
Can assemble existing functions without modifications into pipelines.
dependency resolution can bypass calculation cycles based on data given and asked.
Support functions with optional <optionals> input args and/or varargs <varargish>.
Support functions with partial outputs; keep working even if certain endured operations fail.
Support alias of function provides to avoid the need for trivial conveyor operations.
Default conveyor operation to easily pass (possibly nested) dependencies around.
Merge <operation merging> or nest <operation nesting> sub-pipelines.
Hierarchical dependencies <subdoc> may access data values deep in solution with json pointer path expressions.
Hierarchical dependencies annotated as implicit imply which subdoc dependency the function reads or writes in the parent-doc.
Denote and schedule sideffects on dependency values, to update them repeatedly, avoiding cycles (e.g. to add columns into pandas.DataFrames).
Deterministic pre-decided execution plan (excepting partial-outputs or endured operations).
Early eviction of intermediate results from solution, to optimize memory footprint.
Solution tracks all intermediate overwritten <overwrite> values for the same dependency.
Parallel execution (but underdeveloped).
Elaborate plotting with configurable plot themes.
Integration with Sphinx sites with the new graphtik directive.
Authored with debugging in mind.
Anti-features
It’s not an orchestrator for long-running tasks, nor a calendar scheduler - Apache Airflow and Luigi may help for that.
It’s not really a parallelizing optimizer, neither a map-reduce framework - look additionally at Dask, IpyParallel, Celery, Hive, Pig, Spark, Hadoop, etc.
Quick start
Here’s how to install:
pip install graphtik
OR with various “extras” dependencies, such as, for plotting:
pip install graphtik[plot]
- . Tip::
Supported extras:
- plot
for plotting with Graphviz,
- matplot
for plotting in maplotlib windows
- sphinx
for embedding plots in sphinx-generated sites,
- test
for running pytests,
- dill
may help for pickling parallel tasks - see marshalling term and set_marshal_tasks() configuration.
- all
all of the above, plus development libraries, eg black formatter.
- dev
like all
Let’s build a graphtik computation graph that produces x3 outputs out of 2 inputs a and b:
a x b
a - a x b
|a - a x b| ^ 3
>>> from graphtik import compose, operation >>> from operator import mul, sub
>>> @operation(name="abs qubed", ... needs=["a_minus_ab"], ... provides=["abs_a_minus_ab_cubed"]) ... def abs_qubed(a): ... return abs(a) ** 3
Compose the abspow function along the mul & sub built-ins into a computation graph:
>>> graphop = compose("graphop",
... operation(needs=["a", "b"], provides=["ab"])(mul),
... operation(needs=["a", "ab"], provides=["a_minus_ab"])(sub),
... abs_qubed,
... )
>>> graphop
Pipeline('graphop', needs=['a', 'b', 'ab', 'a_minus_ab'],
provides=['ab', 'a_minus_ab', 'abs_a_minus_ab_cubed'],
x3 ops: mul, sub, abs qubed)
Run the graph and request all of the outputs:
>>> graphop(a=2, b=5)
{'a': 2, 'b': 5, 'ab': 10, 'a_minus_ab': -8, 'abs_a_minus_ab_cubed': 512}
… or request a subset of outputs:
>>> solution = graphop.compute({'a': 2, 'b': 5}, outputs=["a_minus_ab"])
>>> solution
{'a_minus_ab': -8}
… and plot the results (if in jupyter, no need to create the file):
>>> solution.plot('executed_3ops.svg') # doctest: +SKIP
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file graphtik-10.1.0.tar.gz.
File metadata
- Download URL: graphtik-10.1.0.tar.gz
- Upload date:
- Size: 167.8 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.5.0.1 requests/2.24.0 setuptools/50.0.0 requests-toolbelt/0.9.1 tqdm/4.48.2 CPython/3.8.0
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | 7f560fc30f8f7bef6cf24956a4304c1a652827297b4062b05d333e58717e7e00 |
|
| MD5 | 92581ad6f4e6c6a24af7afb52296fe31 |
|
| BLAKE2b-256 | 1db4a506e1282b27921d0c0c450309ec87bab8746023aa71877a88d3d7f88db8 |
File details
Details for the file graphtik-10.1.0-py2.py3-none-any.whl.
File metadata
- Download URL: graphtik-10.1.0-py2.py3-none-any.whl
- Upload date:
- Size: 114.9 kB
- Tags: Python 2, Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.2.0 pkginfo/1.5.0.1 requests/2.24.0 setuptools/50.0.0 requests-toolbelt/0.9.1 tqdm/4.48.2 CPython/3.8.0
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | f68021a6bf9124782f1c71cbc5ae2a2aa2b6be4a2c47d2281e0b5cd5b30820df |
|
| MD5 | 121f7b8fd640cadd6e1638ab6187901f |
|
| BLAKE2b-256 | 938c6b14a3049a4f34830777ae3d101859956ab82fa6faa595526775fbf34e44 |
