Calculate code metrics in various languages
This tool tries to calculate the following metrics for many, many programming languages
- Comment to Code percentage
- Cyclomatic complexity according to McCabe
- Difficulty according to Halstead
- Effort according to Halstead
- Fan-Out
- Lines of code
- Maintainability index
- Metric according to pylint
- Metric according to TIOBE
- Number of delivered bugs according to Halstead
- Time required to program according to Halstead
- Volume according to Halstead
This tool was heavily inspired by metrics
For the following programming language we do offer the complete feature set.
- Bash/Shell
- C
- CoffeeScript
- C++
- C#
- Dart
- Go
- Groovy
- Haskell
- Java
- JavaScript
- Julia
- Kotlin
- Lisp
- Lua
- Objective-C
- Perl
- PHP
- Python
- Ruby
- Rust
- TCL
- TypeScript
- Zig
Other languages work too, but might have a more limited support of what metrics can be extracted.
simply run
pip3 install multimetric- git clone this repository
- cd to <clone folder>
- run
pip3 install .
usage: multimetric [-h] [--warn_compiler WARN_COMPILER] [--warn_duplication WARN_DUPLICATION] [--warn_functional WARN_FUNCTIONAL] [--warn_standard WARN_STANDARD]
[--warn_security WARN_SECURITY] [--coverage COVERAGE] [--dump] [--verbose] [--jobs JOBS] [--bugpredict {old,new}] [--maintindex {sei,classic,microsoft}]
files [files ...]
Calculate code metrics in various languages
positional arguments:
files Files to parse
options:
-h, --help show this help message and exit
--warn_compiler WARN_COMPILER
File(s) holding information about compiler warnings
--warn_duplication WARN_DUPLICATION
File(s) holding information about code duplications
--warn_functional WARN_FUNCTIONAL
File(s) holding information about static code analysis findings
--warn_standard WARN_STANDARD
File(s) holding information about language standard violations
--warn_security WARN_SECURITY
File(s) File(s) holding information about found security issue
--coverage COVERAGE File(s) with compiler warningsFile(s) holding information about testing coverage
--dump Just dump the token tree
--verbose Verbose logging output
--jobs JOBS Run x jobs in parallel
--bugpredict {old,new}
Method how to calculate the bug prediction
--maintindex {sei,classic,microsoft}
Method how to calculate the maintainability index
Currently you could import files of the following types for --warn_* or --coverage
Following information can be read
<file> = full path to file
<severity> = severity [error, warning, info]
<content> = optional string
Note: you could also add a single line, then <content>
has to be a number reflecting to total number of findings
File formats
csv: CSV file of following line format
<file>,<severity>,[<content>]
json: JSON file
<file>: {
["content": <content>,]
"severity": <severity>
}
By default tool guesses the content type by the filename, if that doesn't work for you please see below
Output will be written to stdout as json.
filescontains a list of each file passed by CLIoverallcontains the calculated values for all passed filesstatscontains the statistically calculated values over all files passed see Statistical additions
| item | description | range | recommendation |
|---|---|---|---|
| comment_ratio | Comment to Code percentage | 0..100 | > 30.0 |
| cyclomatic_complexity | Cyclomatic complexity according to McCabe | 0..(inf) | < 10 |
| fanout_external | Number imports from out of tree modules | 0..(inf) | |
| fanout_internal | Number imports from same source tree modules | 0..(inf) | |
| halstead_bugprop | Number of delivered bugs according to Halstead | 0..(inf) | < 0.05 |
| halstead_difficulty | Difficulty according to Halstead | 0..(inf) | |
| halstead_effort | Effort according to Halstead | 0..(inf) | |
| halstead_timerequired | Time required to program according to Halstead | 0..(inf) | |
| halstead_volume | Volume according to Halstead | 0..(inf) | |
| lang | list of identified programming languages | list | |
| loc | Lines of code | 1..(inf) | |
| maintainability_index | Maintainability index | 0..100 | > 80.0 |
| operands_sum | Number of used operands | 1..(inf) | |
| operands_uniq | Number of unique used operands | 1..(inf) | |
| operators_sum | Number of used operators | 1..(inf) | |
| operators_uniq | Number of unique used operators | 1..(inf) | |
| pylint | General quality score according to pylint | 0..100 | > 80.0 |
| tiobe_compiler | Compiler warnings score according to TIOBE | 0..100 | > 90.0 |
| tiobe_complexity | Complexity according to TIOBE | 0..100 | > 80.0 |
| tiobe_coverage | Coverage according to TIOBE | 0..100 | > 80.0 |
| tiobe_duplication | Code duplications score according to TIOBE | 0..100 | > 80.0 |
| tiobe_fanout | Fan-Out score according to TIOBE | 0..100 | > 80.0 |
| tiobe_functional | Functional defect score according to TIOBE | 0..100 | > 90.0 |
| tiobe_security | Security score according to TIOBE | 0..100 | > 90.0 |
| tiobe_standard | Language standard score according to TIOBE | 0..100 | > 80.0 |
| tiobe | General quality score according to TIOBE | 0..100 | > 80.0 |
The item stats contains in addition to the above mentioned the following items, which by themselves contain all the items mentioned at Item structure
max= the maximum value of all items of the metricmean= statistical mean over all items of the metricmedian= statistical median over all items of the metricmin= the minimum value of all items of the metricsd= standard deviation over all items of the metric (needs more than one file to be passed by CLI)
Feel free to create issues or pull requests
flowchart TD
A[I want to contribute!] --> C(Code your changes)
C -->|Flake8 successful| E[Open PR]
C -->|pytest successful| E[Open PR]
C -->|Contribution guideline accepted| E[Open PR]
C -->|README updated| E[Open PR]
flowchart TD
A[I want to contribute a new language config!] --> A1(Find permissively licensed test code)
A1 --> B(Pygments contains a lexer for the language)
B --> C(Create fanout settings)
C --> D[Code your changes]
D -->|Flake8 successful| E[Open PR]
D -->|pytest successful| E[Open PR]
D -->|Contribution guideline accepted| E[Open PR]
D -->|README updated| E[Open PR]