This is the tutorial page originally designed for ASE 2023. It might not be updated on every release, so please open an issue if something does not work. Original slides of the ASE 2023 tutorial: Google slides

Table of contents:

• Task 0: Setup
• Task 1: Getting started
• Task 2: Branch coverage
• Task 3: Dynamic call graph
• Task 4: Analysis hook hierarchy
• Task 5: Modify execution

To isolate the packages that you install during this tutorial either use Docker, or use a virtual environment.

1. Install Docker (if it is your first time using Docker, Docker Desktop is recommended)
2. Build the image: docker build -t dynapyt_tutorial .
3. Run bash in the container: docker run -it dynapyt_tutorial /bin/bash

Note: Changing the code in your host machine does not change the code inside the container. So you either need to share the directory using -v option when running docker (-v .:/dynapyt/), or repeat steps 2 and 3 each time you modify or implement an analysis.

1. Install virtual environment:
   • With pipx: pipx install virtualenv
   • With pip: python -m pip install --user virtualenv
2. Create a directory for the virtual environment: mkdir ~/.dynapyt_virtualenv
3. Create the virtual environment: virtualenv dynapyt_virtualenv ~/.dynapyt_virtualenv
4. Activate the virtual environment: source ~/.dynapyt_virtualenv/bin/activate

• Install the latest release:

  pip install dynapyt

  Or
• Install from source:

  git clone https://github.com/sola-st/DynaPyt.git
  cd DynaPyt
  pip install .

python -m dynapyt.run_instrumentation \
 --analysis dynapyt.analyses.TraceAll.TraceAll \
 --directory task1

• Checkout the instrumented code in task1/main.py and compare it to the original code in task1/main.py.orig.
• Checkout the metadata file task1/main-dynapyt.json.

python -m dynapyt.run_analysis \
 --analysis dynapyt.analyses.TraceAll.TraceAll \
 --entry task1/main.py

• Checkout the TraceAll analysis.

Generate a log of branches taken and their truth values.

1. Create a file under tutorial-analyses/src/tutorial_analyses called BranchCoverageAnalysis.py.
2. Implement the analysis class as a subclass of dynapyt.analyses.BaseAnalysis.
3. Build the analysis: pip install ./tutorial-analyses

python -m dynapyt.run_instrumentation \
 --analysis tutorial_analyses.BranchCoverageAnalysis.BranchCoverageAnalysis \
 --directory task2

python -m dynapyt.run_analysis \
 --analysis tutorial_analyses.BranchCoverageAnalysis.BranchCoverageAnalysis \
 --entry task2/main.py

Generate the graph of dynamic function calls.

1. Create a file under tutorial-analyses/src/tutorial_analyses called CallGraphAnalysis.py.
2. Implement the analysis class as a subclass of dynapyt.analyses.BaseAnalysis.
3. Build the analysis: pip install ./tutorial-analyses

python -m dynapyt.run_instrumentation \
 --analysis tutorial_analyses.CallGraphAnalysis.CallGraphAnalysis \
 --directory task3

python -m dynapyt.run_analysis \
 --analysis tutorial_analyses.CallGraphAnalysis.CallGraphAnalysis \
 --entry task3/main.py

String concatenation with += is slower than "".join(). Find cases that a string gets appended in a loop.

1. Create a file under tutorial-analyses/src/tutorial_analyses called SlowStringConcatAnalysis.py.
2. Implement the analysis class as a subclass of dynapyt.analyses.BaseAnalysis.
3. Build the analysis: pip install ./tutorial-analyses

python -m dynapyt.run_instrumentation \
 --analysis tutorial_analyses.SlowStringConcatAnalysis.SlowStringConcatAnalysis \
 --directory task4

python -m dynapyt.run_analysis \
 --analysis tutorial_analyses.SlowStringConcatAnalysis.SlowStringConcatAnalysis \
 --entry task4/main.py

Force the execution to always take the opposite branch.

1. Create a file under tutorial-analyses/src/tutorial_analyses called OppositeBranchAnalysis.py.
2. Implement the analysis class as a subclass of dynapyt.analyses.BaseAnalysis.
3. Build the analysis: pip install ./tutorial-analyses

python -m dynapyt.run_instrumentation \
 --analysis tutorial_analyses.OppositeBranchAnalysis.OppositeBranchAnalysis \
 --directory task5

python -m dynapyt.run_analysis \
 --analysis tutorial_analyses.OppositeBranchAnalysis.OppositeBranchAnalysis \
 --entry task5/main.py