Architecture

This document describes the internal architecture of Simpesys, including core data structures and the build pipeline.

Simpesys processes Markdown documents through a multi-stage pipeline. The system maintains a document dictionary that maps document keys to processed document objects. During initialization, documents are traversed through breadth-first search starting from the root document.

Core Components

Simpesys

The Simpesys class is the main entry point. It manages document loading and processing, and provides access to the processed document dictionary.

class Simpesys {
  constructor({ config, hooks }: DeepPartial<Context>);
  init(options?: { syncMetadata?: boolean }): Promise<Simpesys>;
  getDocuments(): DocumentDict;
  getDocument(key: string): Document | undefined;
  getConfig(): Config;
}

The constructor accepts a partial configuration. Unspecified options use default values. The init() method must be called before accessing documents.

Context

The Context type combines configuration and hooks into a single object passed to internal functions.

type Context = {
  config: Config;
  hooks: Hooks;
};

For details on available options, see the Configuration document.

Document

The Document interface represents information about a built document.

DocumentDict

The DocumentDict type is a record that maps document keys to Document objects:

type DocumentDict = Record<string, Document>;

A document key is a unique identifier derived from the relative path from the root document without the .md extension. For example, the key for features/internal-links.md is features/internal-links.

Reference

The Reference interface represents a backlink relationship.

interface Reference {
  document: Document; // Document containing the link
  sentences: string[]; // Sentences mentioning the link
}

Build Pipeline

Initialization proceeds through the following stages:

Document Traversal

Starting from the root document, all documents are traversed using breadth-first search, with each document processed as follows:

1. Read the Markdown file from the filesystem.
2. If a manipulateMarkdown hook is defined, execute it to modify the Markdown.
3. Extract the document title from the first level 1 heading.
4. Load or create metadata timestamps.
5. Add the document to the dictionary.
6. Add subdocuments listed in the subdocuments section to the processing queue.

Link Resolution

After all documents are traversed, internal links are processed:

1. Resolve and label internal links in each document.
2. Match links against the document dictionary.
3. If there are unresolved links and an onInternalLinkUnresolved hook is defined, execute it to handle the error.
4. For each valid link, add a backlink reference to the target document.

HTML Conversion

Finally, each document is converted to HTML:

1. Add backlinks section.
2. Insert table of contents before the title.
3. Convert Markdown to HTML.
4. Replace internal link syntax with HTML anchor tags using the renderInternalLink hook.

Document Types

Documents are classified into two types:

The type affects how documents are displayed in the subdocuments section and can be customized in custom rendering logic. By default, documents mentioned under a level 3 Publications heading are classified as publication documents. The publication section title can be configured using the config.docs.publicationsSectionTitle option.

Backlinks

• Simpesys