Document library

This commit is contained in:
Joscha 2024-12-02 17:27:42 +01:00
parent cb2a6da25c
commit cfd83df713
4 changed files with 325 additions and 27 deletions

View file

@ -1,7 +1,11 @@
use std::collections::{btree_map::Entry, BTreeMap, HashMap};
/// <https://html.spec.whatwg.org/multipage/syntax.html#elements-2>
#[derive(Clone, Copy, PartialEq, Eq)]
/// The kind of an element.
///
/// Follows the [definitions from the HTML standard][spec].
///
/// [spec]: https://html.spec.whatwg.org/multipage/syntax.html#elements-2
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum ElementKind {
Void,
Template,
@ -11,27 +15,71 @@ pub enum ElementKind {
Normal,
}
#[derive(Clone)]
/// A single bit of [`Element`] content.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Content {
/// A raw string to be rendered without any checks.
///
/// Can also be constructed using [`Self::raw`].
///
/// # Warning
///
/// This is an escape hatch for including arbitrary text. Using it
/// incorrectly may result in security vulnerabilities in the rendered HTML.
Raw(String),
/// Plain text.
///
/// Can also be constructed using [`Self::text`].
Text(String),
/// An HTML comment (`<!-- ... -->`).
///
/// Can also be constructed using [`Self::comment`].
Comment(String),
/// A child [`Element`].
///
/// Can also be constructed using [`Self::element`].
Element(Element),
}
impl Content {
/// Construct [`Content::Raw`], a raw string to be rendered without any
/// checks.
///
/// # Warning
///
/// This is an escape hatch for including arbitrary text. Using it
/// incorrectly may result in security vulnerabilities in the rendered HTML.
pub fn raw(str: impl ToString) -> Self {
Self::Raw(str.to_string())
}
/// Construct [`Content::Text`], plain text.
pub fn text(str: impl ToString) -> Self {
Self::Text(str.to_string())
}
/// Construct [`Content::Comment`], an HTML comment (`<!-- ... -->`).
pub fn comment(str: impl ToString) -> Self {
Self::Comment(str.to_string())
}
/// Construct [`Content::Element`], a child [`Element`].
///
/// Instead of calling `Content::element(foo)`, you can also use
/// `foo.into()`.
pub fn element(e: impl Into<Element>) -> Self {
Self::Element(e.into())
}
/// Construct a doctype of the form `<!DOCTYPE html>`.
///
/// # Example
///
/// ```
/// use el::Content;
/// let doctype = Content::doctype();
/// assert_eq!(doctype, Content::raw("<!DOCTYPE html>"));
/// ```
pub fn doctype() -> Self {
Self::raw("<!DOCTYPE html>")
}
@ -55,15 +103,61 @@ impl From<Element> for Content {
}
}
#[derive(Clone)]
/// An HTML element.
///
/// SVG and MathML elements are also modelled using this type.
///
/// Errors (e.g. illegal characters or an element of [`ElementKind::Void`]
/// having children) are deferred until rendering and are not checked during
/// element construction. See also [`crate::Render`] and [`crate::Error`].
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Element {
/// The tag name of the element.
pub name: String,
/// What kind of element this is.
///
/// # Warning
///
/// The element kind affects the correctness of the rendered output.
/// Choosing an incorrect kind may result in security vulnerabilities in the
/// rendered HTML. See [`ElementKind`] for more details.
pub kind: ElementKind,
/// The attributes (e.g. `id` or `class`) of the element.
///
/// This map does not take into account case insensitivity of attributes.
/// Any attributes contained in the map will appear in the rendered output.
pub attributes: BTreeMap<String, String>,
/// The children of the element.
pub children: Vec<Content>,
}
impl Element {
/// Create a new element of a specific [`ElementKind`].
///
/// See also [`Self::normal`] to create elements of kind
/// [`ElementKind::Normal`].
///
/// # Warning
///
/// The element kind affects the correctness of the rendered output.
/// Choosing an incorrect kind may result in security vulnerabilities in the
/// rendered HTML. See [`ElementKind`] for more details.
///
/// # Example
///
/// ```
/// use el::{Element, ElementKind, html, svg};
///
/// let p = Element::new("p", ElementKind::Normal);
/// let link = Element::new("link", ElementKind::Void);
/// let script = Element::new("script", ElementKind::RawText);
/// let svg = Element::new("svg", ElementKind::Foreign);
///
/// assert_eq!(p, html::p(()));
/// assert_eq!(link, html::link(()));
/// assert_eq!(script, html::script(()));
/// assert_eq!(svg, svg::svg(()));
/// ```
pub fn new(name: impl ToString, kind: ElementKind) -> Self {
let mut name = name.to_string();
if kind != ElementKind::Foreign {
@ -78,26 +172,109 @@ impl Element {
}
}
/// Create a new element of the kind [`ElementKind::Normal`].
///
/// `Element::normal(foo)` is equivalent to calling `Element::new(foo,
/// ElementKind::Normal)`.
///
/// # Warning
///
/// The element kind affects the correctness of the rendered output.
/// Choosing an incorrect kind may result in security vulnerabilities in the
/// rendered HTML. See [`ElementKind`] for more details.
///
/// # Example
///
/// ```
/// use el::{Element, ElementKind};
/// let element = Element::normal("custom");
/// assert_eq!(element.kind, ElementKind::Normal);
/// ```
pub fn normal(name: impl ToString) -> Self {
Self::new(name, ElementKind::Normal)
}
/// Add components to the element in-place.
///
/// To add multiple components, either call this function repeatedly or use
/// a type like tuples, arrays, [`Vec`], [`Option`], [`Result`] to combine
/// multiple components. See [`ElementComponent`] for more info.
///
/// # Example
///
/// ```
/// use el::{Attr, html::*};
/// let mut element = p(());
///
/// // Adding single components
/// element.add("some text");
/// element.add(Attr::class("foo"));
///
/// // Adding multiple components
/// element.add((Attr::id("bar"), " ", em("and"), " some more text"));
/// ```
pub fn add(&mut self, c: impl ElementComponent) {
c.add_to_element(self);
}
/// A more builder-pattern-like version of [`Self::add`].
///
/// Instead of a mutable reference, this function takes ownership of the
/// element before returning it again. This can be more ergonomic in some
/// cases.
///
/// # Example
///
/// ```
/// use el::{Attr, html::*};
///
/// let element = p(())
/// // Adding single components
/// .with("some text")
/// .with(Attr::class("foo"))
/// // Adding multiple components
/// .with((Attr::id("bar"), " ", em("and"), " some more text"));
/// ```
pub fn with(mut self, c: impl ElementComponent) -> Self {
self.add(c);
self
}
}
/// A component can add itself to an [`Element`] by modifying it.
///
/// A component usually represents either a bit of content or an attribute for
/// the element it is being added to. Some components (e.g. tuples, arrays,
/// [`Vec`], [`Option`], [`Result`]) consist of further components. This creates
/// a flexible API for building [`Element`]s:
///
/// ```
/// use el::{Attr, Render, html::*};
/// let p = p((
/// Attr::id("foo"),
/// Attr::class("bar"),
/// Attr::class("baz"),
/// "Hello ", em("world"), "!",
/// ));
/// assert_eq!(
/// p.render_to_string().unwrap(),
/// r#"<p class="bar baz" id="foo">Hello <em>world</em>!</p>"#,
/// );
/// ```
pub trait ElementComponent {
/// Add a component to an element, consuming the component in the process.
fn add_to_element(self, element: &mut Element);
}
// Attributes
/// An element attribute, used during [`Element`] construction.
///
/// # Example
///
/// ```
/// use el::{Attr, html::*};
/// let p = p(Attr::class("foo"));
/// assert_eq!(p.attributes["class"], "foo");
/// ```
pub struct Attr {
name: String,
value: String,
@ -292,11 +469,11 @@ element_component_tuple!(C1, C2, C3, C4, C5, C6, C7);
element_component_tuple!(C1, C2, C3, C4, C5, C6, C7, C8);
element_component_tuple!(C1, C2, C3, C4, C5, C6, C7, C8, C9);
/// An HTML document.
/// A full HTML document including doctype.
///
/// A `Document(el)` is basically the same as `[Content::doctype(), el.into()]`
/// for the purposes of the [`crate::Render`] trait.
#[derive(Clone)]
/// for the purposes of the [`Render`][crate::Render] trait.
#[derive(Debug, Clone)]
pub struct Document(pub Element);
impl From<Element> for Document {

View file

@ -1,7 +1,71 @@
//! Create HTML by manipulating elements as structured data. Inspired by the
//! clojure library [hiccup][hiccup].
//! # el
//!
//! Write, modify, and safely render HTML elements as simple data structures.
//!
//! This library is inspired by [hiccup] and named after a small helper function
//! I once wrote in JS.
//!
//! [hiccup]: https://github.com/weavejester/hiccup
//!
//! ## Library overview
//!
//! The basic data structure is the [`Element`], which can be rendered to a
//! [`String`] using the [`Render`] trait. Custom elements can be constructed
//! using [`Element::normal`] or [`Element::new`]. Once constructed, elements
//! can be modified by accessing their fields or using the [`Element::add`] or
//! [`Element::with`] methods, though this is usually not necessary.
//!
//! Constructor functions for all (non-deprecated) HTML tags can be found in the
//! [`html`] module, SVG tags in [`svg`] and MathML tags in [`mathml`]. These
//! three modules are designed to be wildcard-included for more concise code,
//! either on a per-function or per-file basis.
//!
//! Element construction uses the [`ElementComponent`] trait, which represents
//! not only element contents but also attributes. Tuples, arrays, [`Vec`],
//! [`Option`], and [`Result`] can be used to combine components. The order of
//! content components is preserved. To set attributes, include [`Attr`] values
//! as components.
//!
//! If you want to render an entire web page, wrap an [`html::html`] element in
//! a [`Document`]. When rendered, documents include the `<!DOCTYPE html>`
//! annotation required by the standard.
//!
//! ## Usage example
//!
//! ```
//! use el::{Attr, Render, html::*};
//!
//! let page: String = html((
//! head((
//! meta(Attr::new("charset", "utf-8")),
//! meta((
//! Attr::new("name", "viewport"),
//! Attr::new("content", "width=device-width, initial-scale=1"),
//! )),
//! title("Example page"),
//! )),
//! body((
//! h1((Attr::id("heading"), "Example page")),
//! p(("This is an example for a ", em("simple"), " web page.")),
//! )),
//! ))
//! .render_to_string()
//! .unwrap();
//! ```
//!
//! ## Axum support
//!
//! The [axum] crate is supported via the optional `axum` feature flag. When it
//! is enabled, [`Document`] implements axum's `IntoResponse` trait and can be
//! returned directly from handlers. In order to prevent accidentally returning
//! incomplete HTML documents, [`Element`] does not implement `IntoResponse`.
//!
//! ```toml
//! [dependencies]
//! el = { version = "...", features = ["axum"] }
//! ```
//!
//! [axum]: https://crates.io/crates/axum
#[cfg(feature = "axum")]
mod axum;

View file

@ -6,33 +6,42 @@ use crate::{
Document,
};
/// The cause of an [`Error`].
#[derive(Debug)]
pub enum ErrorCause {
/// An error occurred while formatting a value.
Format(fmt::Error),
InvalidTagName(String),
InvalidAttrName(String),
/// A name is not a valid tag name.
InvalidTagName { name: String },
/// A name is not a valid attribute name.
InvalidAttrName { name: String },
/// A child is in a place where it is not allowed (e.g. it is the child of a
/// [`ElementKind::Void`] element).
InvalidChild,
InvalidRawText(String),
/// Text inside a [`ElementKind::RawText`] element contains forbidden
/// structures.
InvalidRawText { text: String },
}
/// An error that can occur during element rendering.
#[derive(Debug)]
pub struct Error {
reverse_path: Vec<String>,
reverse_path: Vec<(usize, Option<String>)>,
cause: ErrorCause,
}
impl Error {
pub fn new(cause: ErrorCause) -> Self {
pub(crate) fn new(cause: ErrorCause) -> Self {
Self {
reverse_path: vec![],
cause,
}
}
pub fn at(mut self, index: usize, child: &Content) -> Self {
pub(crate) fn at(mut self, index: usize, child: &Content) -> Self {
self.reverse_path.push(match child {
Content::Element(el) => format!("{index}[{}]", el.name),
_ => index.to_string(),
Content::Element(el) => (index, Some(el.name.clone())),
_ => (index, None),
});
self
}
@ -66,6 +75,11 @@ impl Error {
})
.collect::<String>()
}
/// The cause of the error.
pub fn cause(&self) -> &ErrorCause {
&self.cause
}
}
impl fmt::Display for Error {
@ -74,10 +88,10 @@ impl fmt::Display for Error {
match &self.cause {
ErrorCause::Format(error) => write!(f, "{error}")?,
ErrorCause::InvalidTagName(name) => write!(f, "Invalid tag name {name:?}")?,
ErrorCause::InvalidAttrName(name) => write!(f, "Invalid attribute name {name:?}")?,
ErrorCause::InvalidTagName { name } => write!(f, "Invalid tag name {name:?}")?,
ErrorCause::InvalidAttrName { name } => write!(f, "Invalid attribute name {name:?}")?,
ErrorCause::InvalidChild => write!(f, "Invalid child")?,
ErrorCause::InvalidRawText(text) => write!(f, "Invalid raw text {text:?}")?,
ErrorCause::InvalidRawText { text } => write!(f, "Invalid raw text {text:?}")?,
}
Ok(())
@ -92,11 +106,20 @@ impl From<fmt::Error> for Error {
}
}
/// A wrapper around [`std::result::Result`] with the error [`Error`].
pub type Result<T> = std::result::Result<T, Error>;
/// Render an [`Element`] or a [`Document`] to a [`fmt::Write`]; usually a
/// [`String`].
///
/// To implement this trait, only [`Self::render`] needs to be implemented.
pub trait Render {
/// Render to a writer.
fn render<W: fmt::Write>(&self, w: &mut W) -> Result<()>;
/// Render directly to a [`String`].
///
/// This method is implemented by default and uses [`Self::render`].
fn render_to_string(&self) -> Result<String> {
let mut result = String::new();
self.render(&mut result)?;
@ -137,11 +160,15 @@ impl Render for Element {
fn render<W: fmt::Write>(&self, w: &mut W) -> Result<()> {
// Checks
if !check::is_valid_tag_name(&self.name) {
return Err(Error::new(ErrorCause::InvalidTagName(self.name.clone())));
return Err(Error::new(ErrorCause::InvalidTagName {
name: self.name.clone(),
}));
}
for name in self.attributes.keys() {
if !check::is_valid_attribute_name(name) {
return Err(Error::new(ErrorCause::InvalidAttrName(name.clone())));
return Err(Error::new(ErrorCause::InvalidAttrName {
name: name.clone(),
}));
}
}
@ -174,9 +201,9 @@ impl Render for Element {
Content::Text(text) if check::is_valid_raw_text(&self.name, text) => {
write!(w, "{text}").map_err(|e| e.into())
}
Content::Text(text) => {
Err(Error::new(ErrorCause::InvalidRawText(text.clone())))
}
Content::Text(text) => Err(Error::new(ErrorCause::InvalidRawText {
text: text.clone(),
})),
_ => Err(Error::new(ErrorCause::InvalidChild)),
},
ElementKind::EscapableRawText => match child {