Dependency visualizer is meant for visualizing dependencies of golang based projects and perform analysis on them. The visualizer captures modules in a repo and their dependencies as an adjacency list. Essentially the module becomes the source vertex and each dependency becomes the destination vertex to which the module points to. Thus we form a directed graph. By nature if your project is compiling the graph is also acyclic, since golang compiler breaks on having cyclical dependencies, thus for a successfully compiled project we form a DAG.
Once we get the dependency graph we can use it to perform analysis leveraging various graph algorithms. Few of the analysis currently performed by dependency visualizer include:
- Just by default, rendering a graph tells us how the components are connected.
- On clicking a particular vertex, we traverse the graph with that vertex as source and thus get the dependency graph of the said vertex.
- We can filter the graph to only show modules that point to a dependency with specific keywords, this can be useful in filtering only the dependencies which are grouped together by a common prefix. Currently adding more features is in progress.
- perform topological sorting on the graph: topological sorting can be done on directed acyclic graphs to get an order in which vertices can be traversed such that when a particular vertex is visited, all vertices that point to the said vertex have already been visited.
- Get a minimum cut of the graph: this gives the minimum set of vertices required to break the graph into two disjoint sets, this can be used to find out critical dependencies which become the choke points of the repository as well.
This application leverages network module provided by vis.js to render a graph. The dependencies of a module, can be found using parser package provided by go. Recursively walk a directory we can form an adjacency list with each module as the source and dependencies as destination.
Requires docker to be installed on your machine.
From the root of the repo run docker build -t <my_image_name> -f Docker/dockerfile ., replacing <my_image_name> with your preferred image name.
First identify the module you want to analyze(should usually have 1 or more golang projects), then mount it as a volume, map the port 5000 on container to port 5000 on your machine, provide the image id.Please follow the below steps for a detailed guide:
docker image ls | grep <your image name>
docker run -t -d -v <folder to analyse>:/testfolder/ -p 5000:5000 --name=volume_analyser <image name>
Example:
docker run -t -d -v C:\Users\gupta\OneDrive\Documents\test:/testfolder/ -p 5000:5000 --name=graph_analyser f0f67cc624db
Once above steps are done, application can be interacted with on localhost:5000
