Struct Builder 

pub struct Builder { /* private fields */ }

Configures and runs a Markdown → HTML build.

All directories are explicit — there are no ambient globals. templates_dir and output_dir must be set via Builder::templates and Builder::output before Builder::build is called.

Implementations

impl Builder

pub fn new(input_dir: impl Into<PathBuf>) -> Self

Starts a builder rooted at input_dir, the directory of .md source files.

pub fn templates(self, dir: impl Into<PathBuf>) -> Self

Sets the directory of Tera templates.

pub fn output(self, dir: impl Into<PathBuf>) -> Self

Sets the output directory that mirrors input_dir, one .html file per .md file.

pub fn default_template(self, name: impl Into<String>) -> Self

Sets the template used for pages that have no template: frontmatter key.

pub fn link_base(self, base: impl Into<String>) -> Self

Sets the base path used to rewrite [x](x.md)-style links to clean URLs.

pub fn data_json(self, name: impl Into<String>) -> Self

Opts into writing a data.json index of every non-draft page to name (relative to output_dir) on every Builder::build — for a search index, table of contents, or “recent items” list to consume.

Each entry has id, title, date, updated, version, url, summary, tags, and pinned — the frontmatter-sourced fields default to "" ([] for tags, false for pinned) when absent. A page with draft: true in its frontmatter is excluded from both this index and the HTML build output.

Off by default; explicit over implicit, like the rest of Builder’s optional features. Regenerated by both build() and crate::Watcher::tick (whenever a .md file was added, removed, or modified — a template-only change never alters index content, so it’s skipped then).

pub fn processor(self, processor: impl MarkdownProcessor + 'static) -> Self

Registers a MarkdownProcessor to run on every page’s markdown body before title/template resolution. Processors run in registration order.

pub fn analyzer(self, analyzer: impl MarkdownAnalyzer + 'static) -> Self

Registers a MarkdownAnalyzer to run on every page’s markdown body after processing, extracting metadata for the template. Analyzer results are merged into the Tera context under page.extensions.<name()>.

pub fn watch(&self) -> Result<Watcher<'_>, DocError>

Starts a watch session: an initial full Builder::build, then incremental rebuilds via crate::Watcher::tick whenever a .md or template file’s mtime changes. See Builder::build for the same required-configuration panics.

pub fn build(&self) -> Result<(), DocError>

Walks input_dir and (re-)renders each non-draft .md file whose output isn’t already up to date, writing the result under output_dir. If Builder::data_json is set, also (re-)writes the page index.

A page’s HTML is skipped when output_path already exists and is at least as new as both the .md file and every template file (the render cache — cache::is_up_to_date internally). This makes repeat build() calls incremental for free: no in-memory state, no cache to invalidate — the filesystem’s own mtimes decide.

Panics

Panics if .templates() or .output() were not called first — this is a programmer error (missing required configuration), not a runtime data failure.