4 System Views

Overall Architecture

flowchart LR  
    MantraCfg@{ shape: doc, label: "mantra.json5"}

    ReqJson@{ shape: docs, label: "Optional: Requirements
    [JSON RequirementSchema]"}

    AnnotationJson@{ shape: docs, label: "Optional: Annotations
    [JSON AnnotationSchema]"}

    TestRunJson@{ shape: docs, label: "Optional: Test Runs
    [JSON TestRunSchema]"}

    ReviewJson@{ shape: docs, label: "Optional: ReviewJson
    [JSON ReviewSchema]"}

    Mantra["mantra
    CLI Tool"]

    db[("mantra.db
    SQLite")]

    MantraCfg-->|"Configuration"|Mantra
    ReqJson-->|mantra collect|Mantra
    AnnotationJson-->|mantra collect|Mantra
    TestRunJson-->|mantra collect|Mantra
    ReviewJson-->|mantra collect|Mantra

    Mantra<-->|"Collected Data"|db

    Report@{ shape: docs, label: "Reports
    [HTML, JSON]"}

    Mantra-->|"mantra report"|Report

Loading

Main mantra Commands

• mantra collect

  Used to collect traceability information. The command uses the configuration in mantra.toml to know what files/folders to search for information.

  The result is a SQLite database that contains all collected traceability information.

• mantra report

  Used to generate a (custom) traceability report. Supported output formats are currently only HTML and JSON.

  The Tera templating engine is used to generate HTML reports. It is possible to provide custom templates that may access the data from the JSON report.

Trace Collection

Trace collection is integrated into mantra for all text file types. This includes code files for most programming languages.

By default, mantra uses regular expressions to detect requirement traces. This enables mantra to support tracing for various languages out of the box.

To get advanced trace information, such as associated code elements and spans, mantra uses tree-sitter to create a generic AST per file. This AST may then be traversed to detect requirements traces.

The associated span information that is collected with the tree-sitter approach enables mantra to combine code coverage data from tests with the collected traces to calculate the requirement coverage.

Currently, the AST approach is only implemented for Rust code.