Awesome

Emerge

Emerge (or emerge-viz) is a code analysis tool to gather insights about source code structure, metrics, dependencies and complexity of software projects. You can use it to scan the source code of a project, calculate metric results and statistics, map the source code to graph structures (e.g. a dependency graph or a filesystem graph), export the results in other file formats and even create an interactive web application for further exploration. Emerge currently has scanning support for the following languages: C, C++, Groovy, Java, JavaScript, TypeScript, Kotlin, ObjC, Ruby, Swift, Python. The structure, coloring and clustering is calculated and based on the idea of combining a force-directed graph simulation and Louvain modularity. emerge is mainly written in Python 3 and is tested on macOS, linux and modern web browsers (i.e. latest Safari, Chrome, Firefox, Edge).

Goals of this project

emerge (/ɪˈməːdʒ/)

• to appear by coming out of something or out from behind something
• to become known, especially as a result of examining something or asking questions about it

The main goal of this project is to create a free/ open source tool, that can easily be used by anyone with interest in software development, architecture, metrics and visualization to gather more insights about those topics. It should facilitate/ support getting a better understanding of a given software project by using an exploratory approach.

The following features are currently supported by emerge

• File scan support for the following languages: C, C++, Groovy, Java, JavaScript, TypeScript, Kotlin, ObjC, Ruby, Swift, Python
• Basic entity scan/extraction (e.g. classes) for the following languages: Groovy, Java, Kotlin, Swift
• Basic implementation of the following software metrics: SLOC, Number of Methods, Fan-In/Fan-Out, Modularity (Louvain)
• Infer meaning by feature/semantic keyword extraction based on term frequency-inverse document frequency
• Logging support with configurable log levels
• Configuration support based on YAML syntax to configure multiple/specific analyses
• Create a language/project configuration directly from an included configuration template
• Export of scan results/ metrics/ statistics for the following formats/ outputs
  • File scan
    • Dependency graph
  • Entity scan
    • Dependency graph
    • Inheritance graph
    • Complete graph (composition of dependency and inheritance graph)
  • A Filesystem graph that shows the project filesystem hierarchy as a graph
  • GraphML
  • Graphviz DOT format
  • JavaScript format suited for a D3 force graph simulation
  • Interactive HTML/ web application for interactive, exploratory analysis and data visualization of your project based on graph structures
    • HTML app is based on Bootstrap
    • Force-directed graph simulation by D3
    • The node colors are based on Louvain modularity with a bit of post-processing to make the graph coloring more deterministic and stable
    • Fast full-screen UI rendering on HTML canvas
    • Visualization of files, entities and given metrics
    • Dark mode support
    • Reactive visual live search of files/ entities (e.g. classes)
    • Selection and highlighting of individual nodes
    • Concave hull visualization of single clusters
    • Display of cluster metrics to facilitate comparability
    • Interactivity given by translation, zooming, dragging and hovering over nodes
  • Tabular console output
  • Tabular file output
  • JSON file output

How to install and use emerge as a user

Basically there are two ways to install emerge. If you're familiar with pip (a virtual environment by using pyenv, virtualenv and virtualenvwrapper is recommended, but not needed) you can simply install the latest version of emerge with the following few steps.

0️⃣ ~ (Optional) Setup a virtual environment with pyenv

The recommended way would be to use a virtual env, you can do this by using the following example:

pyenv install 3.10.0
pyenv virtualenv 3.10.0 venv-3.10.0
pyenv activate venv-3.10.0

1️⃣ ~ Install emerge with pip

You can simply install emerge by using pip.

1️⃣.1️⃣ ~ (ubuntu) install prerequisites

On Ubuntu 20.04+ pleaase make sure that the packages graphviz and graphviz-dev are installed, i.e.

apt-get install graphviz graphviz-dev 

1️⃣.2️⃣ ~ Install with pip

pip install emerge-viz

and then simply execute it like this

(emerge) user@host ~ % emerge
usage: emerge [-h] [-c YAMLCONFIG] [-v] [-d] [-e] [-a LANGUAGE]

🔎 Welcome to emerge x.y.z (yyyy-mm-dd hh:mm:ss)

options:
  -h, --help            show this help message and exit
  -c YAMLCONFIG, --config YAMLCONFIG
                        set yaml config file
  -v, --verbose         set logging level to INFO
  -d, --debug           set logging level to DEBUG
  -e, --error           set logging level to ERROR
  -a LANGUAGE, --add-config LANGUAGE
                        add a new config from a template, where LANGUAGE is one of [JAVA, SWIFT, C, CPP, GROOVY, JAVASCRIPT,
                        TYPESCRIPT, KOTLIN, OBJC, RUBY, PY]

2️⃣ ~ Create and adjust project configuration

You can create a simple project config adhoc from the command line and then simply adjust the necessary source/export paths

(emerge) user@host tmp % pwd
/Users/user1/tmp
(emerge) user@host tmp % emerge -a java
✅ created config file from template: /Users/user1/tmp/java-template.yaml

and then simply adjust the necessary paths (analyses/source_directory and export/directory):

(emerge) user@host tmp % cat java-template.yaml 
---
project_name: java_project_example
loglevel: info
analyses:
- analysis_name: full java check
  source_directory: /Users/user1/emerge/project/source
  only_permit_languages:
  - java
  only_permit_file_extensions:
  - .java
  file_scan:
  - number_of_methods
  - source_lines_of_code
  - dependency_graph
  - fan_in_out
  - louvain_modularity
  - tfidf
  entity_scan:
  - dependency_graph
  - source_lines_of_code
  - number_of_methods
  - fan_in_out
  - louvain_modularity
  - tfidf
  export:
  - directory: /Users/user1/emerge/project/export
  - graphml
  - dot
  - json
  - tabular_file
  - tabular_console_overall
  - d3
(emerge) user@host tmp %

3️⃣ ~ Start a scan

After this you can simply start a scan by

(emerge) user@host tmp % emerge -c java-template.yaml
2021-12-04 21:18:15   analysis I 👉 starting to analyze java_project_example
2021-12-04 21:18:15   analysis I ⏩ performing analysis 1/1: full java check
2021-12-04 21:18:15   analysis I 👉 starting to create filesystem graph in full java check
2021-12-04 21:18:15   analysis I ⏩ starting scan at directory: ...
...
...
...
2021-12-04 21:18:27   analysis I ✅ all your generated/exported data can be found here: /Users/user1/tmp/java
2021-12-04 21:18:27   analysis I ✅ copy the following path to your browser and start your web app: 👉 file:///Users/user1/tmp/java/html/emerge.html
2021-12-04 21:18:27   analysis I ✅ total runtime of analysis: 00:00:10 + 154 ms

4️⃣ ~ Start your web app

Now just copy the above mentioned file:// path to any modern web browser and interactively expore your configured codebase 😉

How to install and use emerge from source (e.g. for development)

You can clone this repository and install it by following this instruction:

1️⃣ ~ Clone this repository

git clone https://github.com/glato/emerge.git

2️⃣.1️⃣ ~ (macOS) Install the graphviz package first

brew install graphviz

2️⃣.2️⃣ ~ (macOS) Create a virtual environment

Check of you have the latest Python 3 installed on your macOS. I recommend installing/using Python 3 from Homebrew. Create a Python 3 virtual environment (optionally within the project structure)

cd emerge
pip3 install virtualenv
virtualenv -p python3 venv

2️⃣ ~ (ubuntu) Create a virtual environment

Install required packages and create a Python 3 virtual environment (optionally within the project structure)

apt-get install python3-venv python3-dev graphviz graphviz-dev
cd emerge
python3 -m venv venv

3️⃣ ~ Before using/working with the tool, activate the virtual environment

source venv/bin/activate

4️⃣ ~ (macOS) Install all dependencies

Install all required dependencies for the project with pip

pip install -r requirements.txt

4️⃣ ~ (ubuntu) Install all dependencies

Install the wheel package, after that install all required dependencies for the project with pip

pip install wheel
pip install -r requirements.txt

5️⃣ ~ Running unit tests from the command line

Execute the following from the cloned project root:

python -m unittest discover -v -s ./emerge -p "test_*.py"

otherwise execute the script run_tests.py:

python run_tests.py

If you got in any trouble executing the tests, check this woraround.

6️⃣ ~ Running emerge as a standalone tool

(emerge) user@host emerge % python emerge.py 
usage: emerge.py [-h] [-c YAMLCONFIG] [-v] [-d] [-e] [-a LANGUAGE]

🔎 Welcome to emerge x.y.z (yyyy-mm-dd hh:mm:ss)

options:
  -h, --help            show this help message and exit
  -c YAMLCONFIG, --config YAMLCONFIG
                        set yaml config file
  -v, --verbose         set logging level to INFO
  -d, --debug           set logging level to DEBUG
  -e, --error           set logging level to ERROR
  -a LANGUAGE, --add-config LANGUAGE
                        add a new config from a template, where LANGUAGE is one of [JAVA, SWIFT, C, CPP, GROOVY, JAVASCRIPT,
                        TYPESCRIPT, KOTLIN, OBJC, RUBY, PY]

7️⃣ ~ You're ready to go

Let's quickly try to run emerge on its own codebase

python emerge.py -c configs/emerge.yaml

This should produce a similar output:

...   analysis I 👉 starting to analyze emerge
...   analysis I ⏩ performing analysis 1/1: self-check
...   analysis I 👉 starting to create filesystem graph in self-check
...   analysis I ⏩ starting scan at directory: .
...   ...
...   analysis I 👉 the following statistics were collected in self-check
+-------------------------------------+-------------------+
|                      statistic name | value             |
+-------------------------------------+-------------------+
|                    scanning_runtime | 00:00:00 + 61 ms  |
|                       scanned_files | 32                |
|                       skipped_files | 176               |
|                        parsing_hits | 313               |
|                      parsing_misses | 141               |
|              extracted_file_results | 32                |
|       file_results_creation_runtime | 00:00:00 + 538 ms |
|    number-of-methods-metric-runtime | 00:00:00 + 4 ms   |
| source-lines-of-code-metric-runtime | 00:00:00 + 11 ms  |
|   louvain-modularity-metric-runtime | 00:00:00 + 161 ms |
|           fan-in-out-metric-runtime | 00:00:00 + 4 ms   |
|                       total_runtime | 00:00:00 + 786 ms |
+-------------------------------------+-------------------+
...   analysis I 👉 the following overall metrics were collected in self-check
+----------------------------------------------+----------------------------+
|                                  metric name | value                      |
+----------------------------------------------+----------------------------+
|                avg-number-of-methods-in-file | 13.0                       |
|                             avg-sloc-in-file | 151.41                     |
|                          total-sloc-in-files | 4845                       |
|         louvain-communities-dependency-graph | 3                          |
|          louvain-modularity-dependency-graph | 0.21                       |
| louvain-biggest-communities-dependency-graph | 0.49, 0.46, 0.05, 0.0, 0.0 |
|                  avg-fan-in-dependency-graph | 5.55                       |
|                 avg-fan-out-dependency-graph | 5.55                       |
|                  max-fan-in-dependency-graph | 29                         |
|             max-fan-in-name-dependency-graph | typing                     |
|                 max-fan-out-dependency-graph | 19                         |
|            max-fan-out-name-dependency-graph | emerge/appear.py           |
+----------------------------------------------+----------------------------+
...   analysis I ✅ all your generated/exported data can be found here: /Users/user1/tmp/python
...   analysis I ✅ copy the following path to your browser and start your web app: 👉 file:///Users/user1/tmp/python/html/emerge.html
...   analysis I ✅ total runtime of analysis: 00:00:00 + 786 ms

8️⃣ ~ Start your web app

Now just copy the above mentioned file:// path to any modern web browser and interactively expore the emerge codebase 😉

8️⃣.1️⃣ ~ Currently emerge offers the following keyboard shortcuts in the interactive web app

• ℹ️ Hovering over a node and pressing ⬆️ + s to select and highlight or deselect a specific node
• ℹ️ Resetting the currently active node selection by pressing ⬆️ + r
• ℹ️ Fading all unselected nodes to have a more highlighted visualization of you currenty selected nodes by pressing ⬆️ + f

And now let's make this more interesting ...

Further configuration (using emerge on other projects)

If you wand to use emerge on other projects, you can simple copy or customize one of the existing configuration templates from the emerge/configs directory.

9️⃣ ~ Scan a real project

For a quick run, it should be enough to adjust source_directory, directory in export.

---
project_name: c-example-project
loglevel: info
analyses:
- analysis_name: check_c_files
  source_directory: /Users/user1/emerge/project/source/github/linux-5.8.5/crypto
  only_permit_languages:
  - c
  only_permit_file_extensions:
  - .c
  - .h
  ignore_dependencies_containing:
  - string.h
  file_scan:
  - number_of_methods
  - source_lines_of_code
  - dependency_graph
  - louvain_modularity
  - fan_in_out
  - tfidf
  export:
  - directory: /Users/user1/emerge/project/export
  - graphml
  - dot
  - json
  - tabular_file
  - tabular_console_overall
  - d3

1️⃣0️⃣ ~ Run emerge with a specific yaml configuration

After customizing a present config (e.g. config/c-template.yaml) or creating your own, just run emerge again with this new config

python emerge.py -c configs/c-template.yaml

After the scan, your scan output (including your interactive web app) can be found at the directory that you created and set in the config parameter export -> directory, as seen in the logs above.

A full YAML configuration that contains both file and entity scan has the following format:

---
project_name: java_project_example
loglevel: info
analyses:
- analysis_name: check_java_files_and_classes
  source_directory: /Users/user1/emerge/project/source
  only_permit_languages:
  - java
  only_permit_file_extensions:
  - .java
  ignore_dependencies_containing:
  - java.util
  file_scan:
  - number_of_methods
  - source_lines_of_code
  - dependency_graph
  - fan_in_out
  - louvain_modularity
  - tfidf
  entity_scan:
  - dependency_graph
  - source_lines_of_code
  - number_of_methods
  - fan_in_out
  - louvain_modularity
  - tfidf
  export:
  - directory: /Users/user1/emerge/project/export
  - graphml
  - dot
  - json
  - tabular_file
  - tabular_console_overall
  - d3

The yaml configuration is basically defined at the following levels:

project level

analysis level

file_scan metrics

entity_scan metrics

export configuration

Supported scan types and file extensions

Emerge supports the following file extensions and scan types per language, whereas a file_scan simply calculates metrics and maps nodes within graph structures to scanned files and an entity_scan tries to extract more fine-grained entities from files e.g. classes or structs.

Interpretation of graphs

The interpretation of such graphs can often be very subjective and project dependent. The following examples should help to recognize certain patterns through indicators and hints.

⭐️ Modularity

The magic of uncovering modularity lies in applying a community detection algorithm e.g. Louvain optimization to a force-directed graph, so that both distances and coloring influence the result. The following example includes several indicators for a modular codebase.

1. In the first example on the left you can spot multiple coherent colored clusters which show a low coupling by a certain distance (= generated by the force-directed graph).

2. In the second example on the right the same graph is rendered with activated cluster hulls. On this example the hulls show minimal to no overlapping. Such hints can be indicators for a good software architecture e.g. in terms of modularity, abstraction and well defined interfaces.

⭐️ Codebases with different characteristics

1. In the following example on the left you can see structures with increased modularity e.g. cluster 1 and 2.
2. At the same time you can see another cluster 3 that shows increased overlappings with other clusters. In addition to that an active SLOC metric even shows some huge entities (e.g. classes) which could indicate code smells like god classes. Such code smells like increased coupling and god classes may indicate increased maintenace efforts and error-proneness.

⭐️ Big Ball of Mud

"A BIG BALL OF MUD is haphazardly structured, sprawling, sloppy, duct-tape and bailing wire, spaghetti code jungle" (B. Foote, J. Yoder, 1997). This kind of graph often represents a less optimal architecture. To verify this kind of spaghetti code jungle, one can simply enable hull rendering for all clusters to finally determine: there is only one big cluster after all.

⭐️ Abstract irrelevant dependencies

Sometimes it can help to better understand the complexity of a software architecture if irrelevant dependencies are ignored.

1. Besides being shocked to see the Big Ball of Mud with irrelevant dependencies like java.lang, java.util or any third party dependencies that does not directly belong to a project ...
2. ... you can remove irrelevant dependencies configuratively with the key ignore_dependencies_containing. With a comparatively activated fan-out metric, one recognizes more scattering, some distant hub nodes and clearer clusters. All of these are possible clues to the real (= often more understandable) architecture underneath.

Further development

• Parsing of further entity types for more languages is planned for further development. Contributions are very welcome ❤️
• Everyone is invited to contribute to this project, whether the contribution is related with development, testing, bug reporting or any other support. I would appreciate any help. See Contributing and Credits for further details.