Rchitecture

This package is my report for the architectural recovery assignment on the Software Architecture course at ITU in Spring 2021. The purpose is to recover meaningful views from a software project. I went with the default example, the Zeeguu-Api for analysis.

Approach and dependencies

After some initial struggles with parsing the python AST trees in R I settled on the following :

  • kavon/yaml-ast and maligree/python-ast-explorer for parsing python AST into data structures.
  • 📦reticulate for calling those python scripts from R
  • 📦data.tree for parsing yaml (and directories) into data.tree objects that could then be converted into igraph objects
  • 📦igraph behind the scenes and for implementations of different graph algorithms
  • 📦tidygraph for a tidy interface to manipulate the graphs
  • 📦ggraph for plotting graphs

Furthermore, I am using 📦tidyverse for most of the data manipulation and additional plotting. The documentation website is built using 📦pkgdown.

Reproduction

You can install this package from github. R dependencies should be installed automatically for reproducing what worked out. For dependencies for things that didn’t work out and python, some debugging might be required. Please ask :)

remotes::install_github("benjaminschwetz/Rchitecture")

Example: Zeeguu Api

For the assignment, I made some summary and a dependency view. They are in vignette(zeeguu_api, package = "Rchitecture)

What didn’t work

  • ❌ Using the structure of the ast tree. Not sure if it is my lack of python knowledge or lack of tree wrangling skills but I didn’t find a good way to use the tree for anything, except splitting the script code into tokens. I killed a lot of time with this.
  • ❌ Simplifying the graph using edgebundling from shochastics/edgebundle. I think my graph was too messy. for the algorithm. see edgebundling_failed(). Depends on datashader (python) package.

What I didn’t have time for in the end

  • ⏱️ Making a graph of external dependencies. I started parsing out the requirements.txt but didn’t have time to generate a separate graph viz for it. See generate_dependency_graph()
  • ⏱ Refactor code. A lot of stuff could be reused,/ is copy pasted across funtions right now.
  • ⏱ Parameterize the notebook: give it a github path and autogenerate the viz.