By GitHub convention, the CONTRIBUTING file specifies the guidelines for contribution, which allows more potential collabrators to contribute to EasyGraph.

Example outline of the CONTRIBUTING file:

• Coding style (e.g. PEP8) and the formatter to use (yapf, black, autopep8)
• Issue labeling rules
• Ways to contribute

Also consider adding a Code of Conduct file if this repo becomes popular. :)

I think this is a common problem for several graph metrics and could be easily supported.

Take node degree as an example. Suppose we have a graph with 100k nodes and 500k edges. I just want to know the degree values of 100 nodes within that graph.

One direct way is to just compute the degree values of these 100 nodes. Existing implementations seem to require computing all the nodes and then selecting the values corresponding to the desired nodes.

• I have searched the issues of this repo and believe that this is not a duplicate.

Issue

I am wondering if you can add the calculation complexity (something like O(N^2)) in the documentation.
It would provide programmers with more information about how to pick an appropriate function.

The comments in the .cpp and .h files are shown as garbled text on machines using UTF-8 charset.

@Unparalleled-Calvin Can you fix them? I'd love to be able to read your comments. :)

• I have searched the issues of this repo and believe that this is not a duplicate.
• I have searched the documentation and believe that my question is not covered.

Feature Request

1. Hi all, I am a little curious about the support for creating graph objects with node attributes and edge attributes.
2. Does EasyGraph support other graph types such as dynamic graphs and some high-order networks, such as hypergraphs?

Thanks a lot!

SHS를 사용하는 방법

We're working on the legacy codebase and the coding styles of different part of this package varies wildly.

To improve code quality, let's choose a formatter for the code and add a pre-push hook to ensure that all codes pushed are properly formatted.

Popular formatters includes:

• https://github.com/psf/black (Opinionated, non-configurable)
• https://github.com/google/yapf (My personal fav)
• https://github.com/hhatto/autopep8

Error was encountered when running eg.constraint and eg.hierarchy on an GraphC object representing the karate club dataset.

Info

OS: macOS Monterey 12.4, Intel

python: CPython 3.9.13

Python 3.9.13 (main, May 24 2022, 21:28:31) 
[Clang 13.1.6 (clang-1316.0.21.2)] on darwin

python-easygraph: 0.2a38

To reproduce the error, run

## Setup
git clone https://github.com/tddschn/easygraph-test.git
cd easygraph-test

# install pinned dependencies
# CPython >=3.9,<3.10 is required
poetry install # install poetry first: https://python-poetry.org/
# activate venv
poetry shell

## Run tests

poetry run pytest

Error when running the above tests

AttributeError: 'GraphC' object has no attribute 'is_multigraph'

Feature Request

Please support python 3.11.

• I am on the latest EasyGraph version.
• I have searched the issues of this repo and believe that this is not a duplicate.

• OS version and name:
• EasyGraph version:

Issue

To ensure the changes made don't break existing features, it's a good practice to set up CI run on that is triggered every pull requrests and commit to the master branch, for automated testing across all supported python versions and operating systems. Commits to master will be blocked until all the tests have passed.

See this guide for Building and testing Python on GitHub Actions.

Questions:

• How much test coverage are we looking at here?
• What features requires more attention and tests?
• Testing framework to use? (pytest, unittest)
• How do we test and keep track of the performance of the functions? Do we compare the perfs with that of popular libraries?
• How do we test ML related features like graph embedding in a deterministic way so that the test case runs produce the same results?

OS: macOS Monterey 12.3
CPU: Intel

$ which python3
/usr/local/Caskroom/miniconda/base/envs/lab/bin/python3

$ python3 --version
Python 3.9.10

$ conda install -c fudanmsn Python-EasyGraph
...
# installed easygraph from
# + python-easygraph              0.2a35  py39_0                  fudanmsn/osx-64           3MB
...

$ python3 -c 'import easygraph'
^CTraceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/easygraph/__init__.py", line 8, in <module>
    import easygraph.functions
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/easygraph/functions/__init__.py", line 5, in <module>
    from easygraph.functions.drawing import *
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/easygraph/functions/drawing/__init__.py", line 3, in <module>
    from .plot import *
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/easygraph/functions/drawing/plot.py", line 4, in <module>
    import statsmodels.api as sm
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/statsmodels/api.py", line 105, in <module>
    from .graphics import api as graphics
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/statsmodels/graphics/api.py", line 1, in <module>
    from . import tsaplots as tsa
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/statsmodels/graphics/tsaplots.py", line 11, in <module>
    from statsmodels.tsa.stattools import acf, pacf
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/statsmodels/tsa/stattools.py", line 19, in <module>
    from scipy.signal import correlate
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/scipy/signal/__init__.py", line 329, in <module>
    from ._spectral_py import *
  File "/usr/local/Caskroom/miniconda/base/envs/lab/lib/python3.9/site-packages/scipy/signal/_spectral_py.py", line 8, in <module>
    from ._spectral import _lombscargle
KeyboardInterrupt

What is a test entry point?
You've got tons of test cases, and you want to run a simple pytest all_tests.py to test them all.
Here the all_tests.py is the test entry point of all tests.
In the all_tests.py, you may include all the tests for classes, readwritte etc.

Rationale:

• Ergonomic for local dev and testing
• Portable to CI

parameter nodes and weight are not used. please fix the problem.

Having installed Easy-Graph with the dependency Gensim>=4.1.2, the execution failed outputting that the Gensim Word2Vec class could not recognize the skipgram argument "size". Digging around, I found that the learn_embeddings() function (link below) inserts the keyword 'size' in skip_gram_params, although the Gensim Word2Vec class requires a 'vector_size' argument. The parameter must be renamed to solve the issue.

Who should do it?

The person controlling EasyGraph package on PyPI and the creator of GitHub org easy-graph, who's probably @ICYPOLE.

Steps to take:

Go to https://pypi.org/manage/account/#api-tokens and create a new API token. If you have the project on PyPI already, limit the token scope to just that project. You can call it something like GitHub Actions CI/CD — project-org/project-repo in order for it to be easily distinguishable in the token list. Don’t close the page just yet — you won’t see that token again.

In a separate browser tab or window, go to the Settings tab of your target repository and then click on Secrets in the left sidebar.

Create a new secret called PYPI_API_TOKEN and copy-paste the token from the first step.

• I have searched the issues of this repo and believe that this is not a duplicate.
• I have searched the documentation and believe that my question is not covered.

Version

0.2a39, Installed through pypi and Linux Script in repo. Tested using Ubuntu 20.04 TLS

Issue

For HITS AND MAX-D structural holes, it took forever to return results of a large graph with 80K relationships.
Is there any way to speed up the performance? This is because only single core is used as processing at a time.

Source data: https://github.com/aritan7/pgraph/raw/main/node10k.csv

import easygraph as eg
from easygraph.functions.community.modularity_max_detection import *
G = eg.Graph()
G.add_edges_from_file(file='node10k.csv')
c = eg.greedy_modularity_communities(G)
maxd = eg.get_structural_holes_MaxD(G,k=50, C=c)

Update:

1. Using get_structural_holes_HIS() would cause a memory leak.

• I have searched the issues of this repo and believe that this is not a duplicate.

Issue

when I try to run:
model1 = eg.functions.graph_embedding.sdne.SDNE(G,hidden_size=[256, 128]) # The order of model LINE. 'first'，'second' or 'all'.
model1.train(batch_size=3000, epochs=40, verbose=2)
y = model1.get_embeddings() # Returns the graph embedding results.

I get the error like:
ValueError: Failed to convert a NumPy array to a Tensor (Unsupported numpy type: NPY_INT).

I'm wondering what's wrong?

I think it'd be beneficial to adopt a git workflow,
e.g. constantly pushing to dev branch and only periodically merging dev into master.

master is always stable and won't break for the end users.

Commits on master can also be set up to trigger an automatic release to PyPI & conda, via CI pipelines.

This work is solid and wonderful! I like it.
Hope there are some functions designed for tree. For example, given a tree, extract some sub-trees according their name or id, or flatten the tree as a linear vector to compress it. Besides, is it possible to design an efficient recursive tree for BPE (Byte Pair Encoding)? This can greatly improves the processing efficiency of tree-based or NLP-based tasks.
The above is just my personal suggestions for further improvement.
Wish this project get better and better!

• I have searched the issues of this repo and believe that this is not a duplicate.
• I have searched the documentation and believe that my question is not covered.

Feature Request

The calculation of Effective Size(a feature of structural hole theory) only supports the undirected graphs. However, I think this metric is also useful for directed graphs. The "networkx" package has supported this by "networkx.effective_size".
I propose this issue for requesting the calculation support of structural hole theory metrics for directed graphs. Thanks!

See https://github.com/easy-graph/Easy-Graph/runs/6224787310?check_suite_focus=true

• I have searched the issues of this repo and believe that this is not a duplicate.
• I have searched the documentation and believe that my question is not covered.

Issue

I just used the function "eg.functions.not_sorted.pagerank" to calculate the pageranks of a graph, with about 50,000 nodes.
It runs a long time, and I check the system status using the command "htop". It shows that the calculation only uses one core of CPU, and wastes all other CPU cores. So I propose the issue, hope for parallel running characteristics and better usage of more CPU cores.

See https://github.com/easy-graph/Easy-Graph/blob/f66e200c2319941dbde8391a24a119cc10208795/easygraph/functions/graph_generator/classic.py#L22.

n of type "int" is not iterable.

Might results in

TypeError: cannot unpack non-iterable int object

Better to have two examples. One could be related to the structural hole theory, and the other one could be related to other graph properties. I feel that some users have no idea about the structural hole theory, and thus better to have another example for them.

I believe that the computation of bridges (for undirected graphs) is incorrect. I have the following example, where easygraph fails to correctly compute the bridges:

Consider as undirected graph G a clique of size 4, i.e. with 4 nodes [0, 1, 2, 3] and edges [(0, 1), (0, 2), (0, 3), (1, 2), (1, 3), (2, 3)].
Then, we remove edges [(0, 1), (0, 2)] by repeatedly calling G.remove_edge(...).
After removing these edges, node 0 will have a single edge (0, 3) connecting it to the remainder of the graph. Hence, edge (0, 3) must be a bridge.
However, computing the bridges with easygraph.bridges(G) returns an emtpy set. This is an error.

Here is a minimal test case:

import easygraph as eg
G = eg.Graph()
G.add_edges([ (0, 3), (1, 2), (1, 3), (2, 3), ])
bridges = eg.bridges(G)
assert (0, 3) in bridges

when I run:
y,x44 = eg.functions.graph_embedding.node2vec.node2vec(G,
I get the following error:
AttributeError: 'function' object has no attribute 'node2vec'

easygraph/functions/path

Issue

the annotation of Dijkstra function need to be updated:
the source node could be any type that is hashable, if the graph is created by add_edges_from_file(), the type of source_node() should be string

On the reference pages in the documentation I don't get any details shown, only headings are visible, e.g.

https://easy-graph.github.io/docs/reference/easygraph.classes.html

only has the following content:

easygraph.classes package
Submodules
easygraph.classes.base module
easygraph.classes.directed_graph module
easygraph.classes.directed_multigraph module
easygraph.classes.graph module
easygraph.classes.graphviews module
easygraph.classes.hypergraph module
easygraph.classes.multigraph module
easygraph.classes.operation module
Module contents