MathML EPUB Validator

This application scans EPUB files and validates MathML content based on customizable rule sets. It generates structured JSON reports and human‑readable HTML summaries.

Features

• Reads EPUB files from a predefined folder
• Extracts and validates MathML from HTML/XHTML/XML entries
• Supports optional rule filtering using rules.txt
• Supports running only the rules tagged with a given library, language and/or MathML version
• Outputs JSON and HTML validation reports

Processing Flow

1. Optionally load rule definitions from rules.txt.
2. Enumerate all EPUB files recursively in the input directory.
3. Open each EPUB as a ZIP archive.
4. Parse HTML/XHTML/XML files.
5. Run MathML validation rules.
6. Produce JSON and HTML summary reports.

Setup

Instructions on how to download and run the validator can be found here: How to download and run the validator.

Usage

Place EPUB files inside the folder:

C:/mathml_validator

Run the program. Reports will be saved to:

C:/mathml_validator/report/json
C:/mathml_validator/report/html

Running only some rules

Each rule may optionally be tagged with a Library, Language and/or MathML_version in Tests.cs:

new TestRule {
    Id = "math-wrong-number-of-children-3",
    Library = "tibi",
    Language = "no",
    MathML_version = "MathML3",
    ...
}

To run only the matching rules, set the filter in settings.txt next to the EPUB files (C:/mathml_validator/settings.txt). A commented-out template is written automatically on the first run:

library        = tibi
language       = no
mathml_version = MathML3

Notes:

• Without a settings file, or with all values empty, every active rule runs, exactly as before.
• Any combination of the three keys can be used; each one is optional.
• The values can also be set directly in code (filterLibrary / filterLanguage / filterMathMLVersion in Program.cs); settings.txt overrides them when it supplies a non-empty value.
• Matching is case-insensitive, and a rule may list several values: MathML_version = "MathML3,MathML4".
• A rule with no Library/Language/MathML_version is treated as generic and runs for every filter. To make untagged rules be skipped when a filter is given, set IncludeUntaggedRules = false in RulesLoader.cs.
• The library/language/MathML version filter is applied after the rules.txt ID filter, so the two can be combined.