tue-sevol/assignment-1-2020

A template repository for the first assignment of Software Evolution at Eindhoven University of Technology 2019-2020.

Introduction

This repository serves as an example starting point for the first assignment of Software Evolution 2020-2021. Inside is a DockerFile and example Python script that reads input from a directory /input, and that writes output to /output/links.csv containing the identified trace links.

The source code you hand in should be a docker project, with in the root directory a Dockerfile. If you are unfamilar with Docker please take some time to read up on its concepts, and following a getting started guide.

This readme describes the requirements your solution should satisfy, including the format of the input and output files, and how this template can be used, as well as what docker commands we will use while grading your submission.

Please note that this repository is a starting point for a Python 3 project. However, you are free to use any programming language and environment, as long as the project you submit is a Docker project that runs on Docker for Linux containers, and which can be run using the commands specified below.

Requirements

Your program should read two .csv files, low.csv, and high.csv from the folder /input, match the requirements in high.csv with low.csv as described in the assignment, and then output a links.csv in the /output directory. To determine which match type should be used, the first command line argument provided to your program is an integer specifying which match type should be used.

• 0: No filtering.
• 1: Similarity of at least .25.
• 2: Similarity of at least .67 of the most similar low level requirement.
• 3: Your own custom technique.

Each .csv file used should use , as a delimiter between cells, \n as a delimiter between lines, and " as a quote character.

The files in the input directory have two columns, id, and text. Where id is the identifier of the requirement, and text is the contents of the requirement.

links.csv, generated by your program, should have two columns. id, which contains the id of the high level requirement, and links which is a comma separated list of all low level requirement ids that link to that particular high level requirement.

Note: if for a particular high level requirement you find no links, then there should be an entry in links.csv for that high level requirement with an empty links list.

Finally, if a file input/links.csv is present, you should compare this to the links generated by your tool (output/links.csv), and use the ground truth to compute the precision, recall, and F-measure.

Using this DockerFile

To run the project the following docker command will be used:

docker build -t 2imp25-assignment-1 ./

This builds a docker image according to the DockerFile, and tags the built image as 2imp25-assignment-1.

For example to run the docker image using the matcher that requires a similarity of at least .25 (match type: 1) we execute the following command.

Windows (Powershell): docker run --rm -v "$pwd\dataset-1:/input" -v "$pwd\output:/output" 2imp25-assignment-1 1

Linux: docker run --rm -v "$PWD/dataset-1:/input" -v "$PWD/output:/output" 2imp25-assignment-1 1

The command runs the previously built container. The -v commands mounts two directories (Volumes) on the host file system in the container. One $PWD/dataset-1:/input maps the host directory ./dataset-1 to the container folder ./input, and the second one maps the host directory ./output to container directory /output. 2imp25-assignment-1 is the name of the tag that was given during the build step, and the final argument 1 is the match type that should be used by the trace link finder.

By providing a different host directory for the input mount a different dataset can be provided to the program. e.g. this repository provides two datasets, dataset-1 and dataset-2. However, when grading we will be using a third, previously unseen dataset to evaluate how your algorithm performs.

Therefore, please take care such that we can run your submitted source code using the two described docker commands, and extract a links.csv file. If your DockerFile fails to build, no output file is written to /output/links.csv, or the /output/links.csv file is not formatted correctly, points will be subtracted.

In summary, your submission should conform to the following requirements:

• Your submission should contain a single Dockerfile. The folder you submit should not be larger than 100MBs.
• We should be able to build your submission using docker build -t assignment-1 ./. This step should not take longer than 3 minutes, if it takes longer we will terminate the build.
• We should be able to run your submission using the following command docker run --rm -v "$PWD/dataset-1:/input" -v "$PWD/output:/output" 2imp25-assignment-1 TYPE, where Type is one of {0, 1, 2, 3}. Note: We will use this exact command. You are free to add optional parameters when testing or debugging your submission, however, these parameters should be optional. If we cannot run your submission using the given command points will be subtracted.
• Running your tool on the two provided datasets should not take longer than 2 minutes, if it takes longer your submission will be terminated.