cvelazquezr / saner-2022 Goto Github PK

The folders in this project contain the code, datasets and results for the paper "Uncovering Library Features from API Usage on Stack Overflow".

The description of each folder in the repository is described below:

Folder feature_lists_libraries

This folder contains the 50 initial libraries considered to perform the study. The files in the folder list a number of features for each library. However, many of the features in the linked web pages, do not have accompanying code.

Folder tutorial_features

This folder contains the 7 sets of tutorial features we were able to extract from the tutorial web sites. A copy of the folder is used for the evaluation.

Folder github_usages

This folder contains the API usages extracted from GitHub usages. A copy of the folder is used for the evaluation.

Project island_parser

This project contains the grammar and code for our Island Parser in order to extract information from code snippets. We based our parser in the Java 8 Language Specification here. This project requires a version of Java installed in the system, preferably Java 8.

Project feature_me

The project extracts the code and textual information from Stack Overflow (SO) posts, apply the Island Parser and generate vector combinations for each SO answer where is a library usage.

Project github_usages_extractor

The project clones GitHub repositories using our seven selected libraries and extracts from them the API usages using the Island parser.

Project clusters_exploration

This projects serves as the tool for clustering exploration where the clusters for each library can be visualised. There are two ways of executing this tool:

1. A manual ways which requires the installation of all packages.
2. An automatic way (preferred if you only want to see the tool in action) in the form of a Docker image and container.

Both ways of deploying the tool are explained next. The automatic way is explained first since it requires less instructions.

Automatic deployment with Docker

You need to have Docker installed in your system to execute this part. Please follow the installation instructions for Docker here.

After the Docker installation (or if you have it already installed), make sure that it is running. Then in the root folder of this repository execute the following to build the Docker image:

docker build -t features-libraries .

Note: Please include the dot (e.g., ".") at the end of the command. It is not a typographic error :)

It should take some minutes while all libraries are downloaded, configured, and the system gets ready for the deployment, i.e., you could grab some liquid (:coffee:, :tea:) in the meantime.

A FINISHED message should be displayed after the process is done. To check whether the image was built simply type:

docker images

A similar output as the one below should be produced:

To deploy the application simply run the built image specifying a port like the 8080:

docker run -p 8080:3838 features-libraries

When the application starts listening for requests as the image below, open your browser in localhost and the selected port.

If you selected 8080 click here: http://localhost:8080. Otherwise, replace the port (after the second colon) with the one you selected.

Manual deployment

If instead, you prefer to do a manual setup, follow the next instructions.

You need to install R and Python as programming languages into your system.

Then, you could run the application in the RStudio IDE. Download it here and follow the installation instructions.

Please install the following libraries for R. You can do so in RStudio or in the R console:

• dynamicTreeCut
• stringr
• cluster
• clValid
• dplyr
• shiny
• shinycssloaders
• treemap
• highcharter
• RColorBrewer
• ggplot2
• parallel
• lvplot
• knitr
• rmarkdown
• reticulate

For Python, please install the following packages:

• numpy
• pandas
• scikit-learn

After the installation of the packages/libraries, configure the reticulate R library by adding the following line to the file ~/.Renviron:

RETICULATE_PYTHON="full/path/to/your/python/executable"

Note: If the file is not created, please do so.

Please, add the full path to your installed Python executable. Most of the times this path is located in /usr/bin/python3 or /usr/local/bin/python3 for Unix-based systems.

Finally, you can run the application in RStudio by opening one of the files related to the Shiny configuration (i.e., global.R, server.R or ui.R) and pressing the top green play button called Run App.

Folder uncovered_features

The folder contains the generated features from the clusters_exploration tool above. They are presented in the form of condensed HTML files accessible from any web browser.

Folder images

Finally, this folder contain some snapshots of the cluster_exploration tool.