mdbook/crates/mdbook-html/src/html_handlebars/hbs_renderer.rs

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

707 lines
25 KiB
Rust
Raw Normal View History

use super::helpers;
use super::static_files::StaticFiles;
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
use crate::html::ChapterTree;
use crate::html::{build_trees, render_markdown, serialize};
use crate::theme::Theme;
use crate::utils::ToUrlPath;
use anyhow::{Context, Result, bail};
use handlebars::Handlebars;
use mdbook_core::book::{Book, BookItem, Chapter};
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
use mdbook_core::config::{BookConfig, Config, HtmlConfig};
use mdbook_core::utils;
use mdbook_renderer::{RenderContext, Renderer};
use serde_json::json;
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
use std::collections::{BTreeMap, HashMap};
use std::fs::{self, File};
use std::path::{Path, PathBuf};
use tracing::error;
use tracing::{debug, info, trace, warn};
/// The HTML renderer for mdBook.
#[derive(Default)]
#[non_exhaustive]
pub struct HtmlHandlebars;
impl HtmlHandlebars {
/// Returns a new instance of [`HtmlHandlebars`].
pub fn new() -> Self {
HtmlHandlebars
}
fn render_chapter(
&self,
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
chapter_tree: &ChapterTree<'_>,
prev_ch: Option<&Chapter>,
next_ch: Option<&Chapter>,
mut ctx: RenderChapterContext<'_>,
) -> Result<()> {
// FIXME: This should be made DRY-er and rely less on mutable state
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
let ch = chapter_tree.chapter;
2020-03-09 22:34:28 +01:00
let path = ch.path.as_ref().unwrap();
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
// "print.html" is used for the print page.
if path == Path::new("print.md") {
bail!("{} is reserved for internal use", path.display());
};
2020-03-09 22:34:28 +01:00
if let Some(ref edit_url_template) = ctx.html_config.edit_url_template {
let full_path = ctx.book_config.src.to_str().unwrap_or_default().to_owned()
+ "/"
+ ch.source_path
.clone()
.unwrap_or_default()
.to_str()
.unwrap_or_default();
let edit_url = edit_url_template.replace("{path}", &full_path);
ctx.data
.insert("git_repository_edit_url".to_owned(), json!(edit_url));
}
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
let mut content = String::new();
serialize(&chapter_tree.tree, &mut content);
2020-03-09 22:34:28 +01:00
let ctx_path = path
.to_str()
.with_context(|| "Could not convert path to str")?;
2020-03-09 22:34:28 +01:00
let filepath = Path::new(&ctx_path).with_extension("html");
2020-05-10 09:23:40 -07:00
let book_title = ctx
.data
.get("book_title")
.and_then(serde_json::Value::as_str)
.unwrap_or("");
2020-03-09 22:34:28 +01:00
let title = if let Some(title) = ctx.chapter_titles.get(path) {
title.clone()
} else if book_title.is_empty() {
ch.name.clone()
} else {
ch.name.clone() + " - " + book_title
2020-05-10 09:23:40 -07:00
};
2020-03-09 22:34:28 +01:00
ctx.data.insert("path".to_owned(), json!(path));
ctx.data.insert("content".to_owned(), json!(content));
ctx.data.insert("chapter_title".to_owned(), json!(ch.name));
ctx.data.insert("title".to_owned(), json!(title));
ctx.data.insert(
"path_to_root".to_owned(),
2023-05-13 09:44:11 -07:00
json!(utils::fs::path_to_root(path)),
2020-03-09 22:34:28 +01:00
);
if let Some(ref section) = ch.number {
ctx.data
.insert("section".to_owned(), json!(section.to_string()));
}
let redirects = collect_redirects_for_path(&filepath, &ctx.html_config.redirect)?;
if !redirects.is_empty() {
ctx.data.insert(
"fragment_map".to_owned(),
json!(serde_json::to_string(&redirects)?),
);
}
let mut nav = |name: &str, ch: Option<&Chapter>| {
let Some(ch) = ch else { return };
let path = ch
.path
.as_ref()
.unwrap()
.with_extension("html")
.to_url_path();
let obj = json!( {
"title": ch.name,
"link": path,
});
ctx.data.insert(name.to_string(), obj);
};
nav("previous", prev_ch);
nav("next", next_ch);
2020-03-09 22:34:28 +01:00
// Render the handlebars template with the data
debug!("Render template");
let rendered = ctx.handlebars.render("index", &ctx.data)?;
// Write to file
debug!("Creating {}", filepath.display());
utils::fs::write_file(&ctx.destination, &filepath, rendered.as_bytes())?;
if prev_ch.is_none() {
2020-03-09 22:34:28 +01:00
ctx.data.insert("path".to_owned(), json!("index.md"));
ctx.data.insert("path_to_root".to_owned(), json!(""));
ctx.data.insert("is_index".to_owned(), json!(true));
2020-03-09 22:34:28 +01:00
let rendered_index = ctx.handlebars.render("index", &ctx.data)?;
debug!("Creating index.html from {}", ctx_path);
utils::fs::write_file(&ctx.destination, "index.html", rendered_index.as_bytes())?;
}
Ok(())
}
fn render_404(
&self,
ctx: &RenderContext,
html_config: &HtmlConfig,
2021-05-31 20:27:52 -07:00
src_dir: &Path,
handlebars: &mut Handlebars<'_>,
data: &mut serde_json::Map<String, serde_json::Value>,
) -> Result<()> {
let destination = &ctx.destination;
let content_404 = if let Some(ref filename) = html_config.input_404 {
let path = src_dir.join(filename);
std::fs::read_to_string(&path)
.with_context(|| format!("unable to open 404 input file {path:?}"))?
} else {
// 404 input not explicitly configured try the default file 404.md
let default_404_location = src_dir.join("404.md");
if default_404_location.exists() {
std::fs::read_to_string(&default_404_location).with_context(|| {
format!("unable to open 404 input file {default_404_location:?}")
})?
} else {
"# Document not found (404)\n\nThis URL is invalid, sorry. Please use the \
navigation bar or search to continue."
.to_string()
}
};
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
let options = crate::html::HtmlRenderOptions::new(
Path::new("404.md"),
html_config,
ctx.config.rust.edition,
);
let html_content_404 = render_markdown(&content_404, &options);
let mut data_404 = data.clone();
let base_url = if let Some(site_url) = &html_config.site_url {
site_url
} else {
debug!(
"HTML 'site-url' parameter not set, defaulting to '/'. Please configure \
this to ensure the 404 page work correctly, especially if your site is hosted in a \
subdirectory on the HTTP server."
);
"/"
};
data_404.insert("base_url".to_owned(), json!(base_url));
// Set a dummy path to ensure other paths (e.g. in the TOC) are generated correctly
data_404.insert("path".to_owned(), json!("404.md"));
data_404.insert("content".to_owned(), json!(html_content_404));
2022-03-27 17:17:20 -07:00
let mut title = String::from("Page not found");
if let Some(book_title) = &ctx.config.book.title {
title.push_str(" - ");
title.push_str(book_title);
}
data_404.insert("title".to_owned(), json!(title));
let rendered = handlebars.render("index", &data_404)?;
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
let output_file = utils::fs::get_404_output_file(&html_config.input_404);
utils::fs::write_file(destination, output_file, rendered.as_bytes())?;
debug!("Creating 404.html ✓");
Ok(())
}
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
fn render_print_page(
2018-05-16 12:08:23 -05:00
&self,
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
ctx: &RenderContext,
handlebars: &Handlebars<'_>,
2018-05-16 12:08:23 -05:00
data: &mut serde_json::Map<String, serde_json::Value>,
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
chapter_trees: Vec<ChapterTree<'_>>,
) -> Result<String> {
let print_content = crate::html::render_print_page(chapter_trees);
if let Some(ref title) = ctx.config.book.title {
data.insert("title".to_owned(), json!(title));
} else {
// Make sure that the Print chapter does not display the title from
// the last rendered chapter by removing it from its context
data.remove("title");
}
data.insert("is_print".to_owned(), json!(true));
data.insert("path".to_owned(), json!("print.md"));
data.insert("content".to_owned(), json!(print_content));
2018-05-16 12:08:23 -05:00
data.insert(
"path_to_root".to_owned(),
json!(utils::fs::path_to_root(Path::new("print.md"))),
);
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
debug!("Render template");
let rendered = handlebars.render("index", &data)?;
Ok(rendered)
}
2020-01-24 11:01:44 +08:00
fn register_hbs_helpers(&self, handlebars: &mut Handlebars<'_>, html_config: &HtmlConfig) {
2018-05-16 12:08:23 -05:00
handlebars.register_helper(
"toc",
Box::new(helpers::toc::RenderToc {
no_section_label: html_config.no_section_label,
}),
);
handlebars.register_helper("fa", Box::new(helpers::fontawesome::fa_helper));
}
2017-06-15 18:17:16 +08:00
fn emit_redirects(
&self,
root: &Path,
handlebars: &Handlebars<'_>,
redirects: &HashMap<String, String>,
) -> Result<()> {
if redirects.is_empty() {
return Ok(());
}
debug!("Emitting redirects");
let redirects = combine_fragment_redirects(redirects);
for (original, (dest, fragment_map)) in redirects {
// Note: all paths are relative to the build directory, so the
// leading slash in an absolute path means nothing (and would mess
// up `root.join(original)`).
2020-12-31 15:37:34 -05:00
let original = original.trim_start_matches('/');
let filename = root.join(original);
if filename.exists() {
// This redirect is handled by the in-page fragment mapper.
continue;
}
if dest.is_empty() {
bail!(
"redirect entry for `{original}` only has source paths with `#` fragments\n\
There must be an entry without the `#` fragment to determine the default \
destination."
);
}
debug!("Redirecting \"{}\"\"{}\"", original, dest);
self.emit_redirect(handlebars, &filename, &dest, &fragment_map)?;
}
Ok(())
}
fn emit_redirect(
&self,
handlebars: &Handlebars<'_>,
original: &Path,
destination: &str,
fragment_map: &BTreeMap<String, String>,
) -> Result<()> {
if let Some(parent) = original.parent() {
std::fs::create_dir_all(parent)
.with_context(|| format!("Unable to ensure \"{}\" exists", parent.display()))?;
}
let js_map = serde_json::to_string(fragment_map)?;
let ctx = json!({
"fragment_map": js_map,
"url": destination,
});
let f = File::create(original)?;
handlebars
.render_to_write("redirect", &ctx, f)
.with_context(|| {
format!(
"Unable to create a redirect file at \"{}\"",
original.display()
)
})?;
Ok(())
}
}
impl Renderer for HtmlHandlebars {
fn name(&self) -> &str {
"html"
}
fn render(&self, ctx: &RenderContext) -> Result<()> {
let book_config = &ctx.config.book;
let html_config = ctx.config.html_config().unwrap_or_default();
let src_dir = ctx.root.join(&ctx.config.book.src);
let destination = &ctx.destination;
let book = &ctx.book;
let build_dir = ctx.root.join(&ctx.config.build.build_dir);
2019-07-21 04:32:28 +08:00
if destination.exists() {
utils::fs::remove_dir_content(destination)
.with_context(|| "Unable to remove stale HTML output")?;
2019-07-21 04:32:28 +08:00
}
trace!("render");
let mut handlebars = Handlebars::new();
let theme_dir = match html_config.theme {
Some(ref theme) => {
let dir = ctx.root.join(theme);
if !dir.is_dir() {
bail!("theme dir {} does not exist", dir.display());
}
dir
}
2018-05-14 14:52:29 -05:00
None => ctx.root.join("theme"),
};
let theme = Theme::new(theme_dir);
debug!("Register the index handlebars template");
handlebars.register_template_string("index", String::from_utf8(theme.index.clone())?)?;
debug!("Register the head handlebars template");
handlebars.register_partial("head", String::from_utf8(theme.head.clone())?)?;
debug!("Register the redirect handlebars template");
handlebars
.register_template_string("redirect", String::from_utf8(theme.redirect.clone())?)?;
debug!("Register the header handlebars template");
handlebars.register_partial("header", String::from_utf8(theme.header.clone())?)?;
debug!("Register the toc handlebars template");
handlebars.register_template_string("toc_js", String::from_utf8(theme.toc_js.clone())?)?;
handlebars
.register_template_string("toc_html", String::from_utf8(theme.toc_html.clone())?)?;
debug!("Register handlebars helpers");
self.register_hbs_helpers(&mut handlebars, &html_config);
let mut data = make_data(&ctx.root, book, &ctx.config, &html_config, &theme)?;
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
let chapter_trees = build_trees(book, &html_config, ctx.config.rust.edition);
2023-05-13 09:44:11 -07:00
fs::create_dir_all(destination)
.with_context(|| "Unexpected error when constructing destination path")?;
let mut static_files = StaticFiles::new(&theme, &html_config, &ctx.root)?;
// Render search index
#[cfg(feature = "search")]
{
let default = mdbook_core::config::Search::default();
let search = html_config.search.as_ref().unwrap_or(&default);
if search.enable {
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
super::search::create_files(&search, &mut static_files, &chapter_trees)?;
}
}
debug!("Render toc js");
{
let rendered_toc = handlebars.render("toc_js", &data)?;
static_files.add_builtin("toc.js", rendered_toc.as_bytes());
debug!("Creating toc.js ✓");
}
if html_config.hash_files {
static_files.hash_files()?;
}
debug!("Copy static files");
let resource_helper = static_files
.write_files(&destination)
.with_context(|| "Unable to copy across static files")?;
handlebars.register_helper("resource", Box::new(resource_helper));
debug!("Render toc html");
{
data.insert("is_toc_html".to_owned(), json!(true));
data.insert("path".to_owned(), json!("toc.html"));
let rendered_toc = handlebars.render("toc_html", &data)?;
utils::fs::write_file(destination, "toc.html", rendered_toc.as_bytes())?;
debug!("Creating toc.html ✓");
data.remove("path");
data.remove("is_toc_html");
}
utils::fs::write_file(
destination,
".nojekyll",
b"This file makes sure that Github Pages doesn't process mdBook's output.\n",
)?;
if let Some(cname) = &html_config.cname {
utils::fs::write_file(destination, "CNAME", format!("{cname}\n").as_bytes())?;
}
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
for (i, chapter_tree) in chapter_trees.iter().enumerate() {
let previous = (i != 0).then(|| chapter_trees[i - 1].chapter);
let next = (i != chapter_trees.len() - 1).then(|| chapter_trees[i + 1].chapter);
let ctx = RenderChapterContext {
handlebars: &handlebars,
destination: destination.to_path_buf(),
data: data.clone(),
book_config: book_config.clone(),
html_config: html_config.clone(),
chapter_titles: &ctx.chapter_titles,
};
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
self.render_chapter(chapter_tree, previous, next, ctx)?;
}
2020-05-13 14:45:35 +02:00
// Render 404 page
if html_config.input_404 != Some("".to_string()) {
self.render_404(ctx, &html_config, &src_dir, &mut handlebars, &mut data)?;
2020-05-13 14:45:35 +02:00
}
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
// Render the print version.
if html_config.print.enable {
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
let print_rendered =
self.render_print_page(ctx, &handlebars, &mut data, chapter_trees)?;
Add a new HTML rendering pipeline This rewrites the HTML rendering pipeline to use a tree data structure, and implements a custom HTML serializer. The intent is to make it easier to make changes and to manipulate the output. This should make some future changes much easier. This is a large change, but I'll try to briefly summarize what's changing: - All of the HTML rendering support has been moved out of mdbook-markdown into mdbook-html. For now, all of the API surface is private, though we may consider ways to safely expose it in the future. - Instead of using pulldown-cmark's html serializer, this takes the pulldown-cmark events and translates them into a tree data structure (using the ego-tree crate to define the tree). See `tree.rs`. - HTML in the markdown document is parsed using html5ever, and then lives inside the same tree data structure. See `tokenizer.rs`. - Transformations are then applied to the tree data structure. For example, adding header links or hiding code lines. - Serialization is a simple process of writing out the nodes to a string. See `serialize.rs`. - The search indexer works on the tree structure instead of re-rendering every chapter twice. See `html_handlebars/search.rs`. - The print page now takes a very different approach of taking the same tree structure built for rendering the chapters, and applies transformations to it. This avoid re-parsing everything again. See `print.rs`. - I changed the linking behavior so that links on the print page link to items on the print page instead of outside the print page. - There are a variety of small changes to how it serializes as can be seen in the changes to the tests. Some highlights: - Code blocks no longer have a second layer of `<pre>` tags wrapping it. - Fixed a minor issue where a rust code block with a specific edition was having the wrong classes when there was a default edition. - Drops the ammonia dependency, which significantly reduces the number of dependencies. It was only being used for a very minor task, and we can handle it much more easily now. - Drops `pretty_assertions`, they are no longer used (mostly being migrated to the testsuite). There's obviously a lot of risk trying to parse everything to such a low level, but I think the benefits are worth it. Also, the API isn't super ergonomic compared to say javascript (there are no selectors), but it works well enough so far. I have not run this through rigorous benchmarking, but it does have a very noticeable performance improvement, especially in a debug build. I expect in the future that we'll want to expose some kind of integration with extensions so they have access to this tree structure (or some kind of tree structure). Closes https://github.com/rust-lang/mdBook/issues/1736
2025-09-16 20:14:00 -07:00
utils::fs::write_file(destination, "print.html", print_rendered.as_bytes())?;
debug!("Creating print.html ✓");
}
self.emit_redirects(&ctx.destination, &handlebars, &html_config.redirect)
.context("Unable to emit redirects")?;
// Copy all remaining files, avoid a recursive copy from/to the book build dir
utils::fs::copy_files_except_ext(&src_dir, destination, true, Some(&build_dir), &["md"])?;
2025-06-12 21:33:50 +02:00
info!("HTML book written to `{}`", destination.display());
Ok(())
}
}
2018-05-16 12:08:23 -05:00
fn make_data(
root: &Path,
book: &Book,
config: &Config,
html_config: &HtmlConfig,
theme: &Theme,
2018-05-16 12:08:23 -05:00
) -> Result<serde_json::Map<String, serde_json::Value>> {
trace!("make_data");
let mut data = serde_json::Map::new();
data.insert(
"language".to_owned(),
json!(config.book.language.clone().unwrap_or_default()),
);
data.insert(
"text_direction".to_owned(),
json!(config.book.realized_text_direction()),
);
2018-05-16 12:08:23 -05:00
data.insert(
"book_title".to_owned(),
json!(config.book.title.clone().unwrap_or_default()),
);
data.insert(
"description".to_owned(),
json!(config.book.description.clone().unwrap_or_default()),
);
if theme.favicon_png.is_some() {
data.insert("favicon_png".to_owned(), json!("favicon.png"));
}
if theme.favicon_svg.is_some() {
data.insert("favicon_svg".to_owned(), json!("favicon.svg"));
}
if let Some(ref live_reload_endpoint) = html_config.live_reload_endpoint {
data.insert(
"live_reload_endpoint".to_owned(),
json!(live_reload_endpoint),
);
2016-04-02 04:46:05 +02:00
}
let default_theme = match html_config.default_theme {
Some(ref theme) => theme.to_lowercase(),
None => "light".to_string(),
};
data.insert("default_theme".to_owned(), json!(default_theme));
let preferred_dark_theme = match html_config.preferred_dark_theme {
Some(ref theme) => theme.to_lowercase(),
None => "navy".to_string(),
};
data.insert(
"preferred_dark_theme".to_owned(),
json!(preferred_dark_theme),
);
if html_config.mathjax_support {
data.insert("mathjax_support".to_owned(), json!(true));
}
// Add check to see if there is an additional style
if !html_config.additional_css.is_empty() {
let mut css = Vec::new();
for style in &html_config.additional_css {
match style.strip_prefix(root) {
2018-05-16 12:08:23 -05:00
Ok(p) => css.push(p.to_str().expect("Could not convert to str")),
Err(_) => css.push(style.to_str().expect("Could not convert to str")),
}
}
data.insert("additional_css".to_owned(), json!(css));
}
2017-06-06 20:34:57 +02:00
// Add check to see if there is an additional script
if !html_config.additional_js.is_empty() {
2017-06-06 20:34:57 +02:00
let mut js = Vec::new();
for script in &html_config.additional_js {
match script.strip_prefix(root) {
2017-06-06 20:34:57 +02:00
Ok(p) => js.push(p.to_str().expect("Could not convert to str")),
Err(_) => js.push(script.to_str().expect("Could not convert to str")),
2017-06-06 20:34:57 +02:00
}
}
data.insert("additional_js".to_owned(), json!(js));
}
if html_config.playground.editable && html_config.playground.copy_js {
data.insert("playground_js".to_owned(), json!(true));
if html_config.playground.line_numbers {
data.insert("playground_line_numbers".to_owned(), json!(true));
}
2017-06-29 00:35:20 -04:00
}
if html_config.playground.copyable {
data.insert("playground_copyable".to_owned(), json!(true));
}
2017-06-29 00:35:20 -04:00
2020-09-22 13:24:13 -07:00
data.insert("print_enable".to_owned(), json!(html_config.print.enable));
data.insert("fold_enable".to_owned(), json!(html_config.fold.enable));
data.insert("fold_level".to_owned(), json!(html_config.fold.level));
data.insert(
"sidebar_header_nav".to_owned(),
json!(html_config.sidebar_header_nav),
);
let search = html_config.search.clone();
if cfg!(feature = "search") {
2018-06-13 13:11:25 -05:00
let search = search.unwrap_or_default();
data.insert("search_enabled".to_owned(), json!(search.enable));
2018-07-23 12:45:01 -05:00
data.insert(
"search_js".to_owned(),
json!(search.enable && search.copy_js),
);
} else if search.is_some() {
warn!("mdBook compiled without search support, ignoring `output.html.search` table");
2018-05-16 12:08:23 -05:00
warn!(
"please reinstall with `cargo install mdbook --force --features search`to use the \
search feature"
)
}
2018-05-16 12:08:23 -05:00
if let Some(ref git_repository_url) = html_config.git_repository_url {
data.insert("git_repository_url".to_owned(), json!(git_repository_url));
}
let git_repository_icon = match html_config.git_repository_icon {
Some(ref git_repository_icon) => git_repository_icon,
None => "fab-github",
};
let git_repository_icon_class = match git_repository_icon.split('-').next() {
Some("fa") => "regular",
Some("fas") => "solid",
Some("fab") => "brands",
_ => "regular",
};
data.insert("git_repository_icon".to_owned(), json!(git_repository_icon));
data.insert(
"git_repository_icon_class".to_owned(),
json!(git_repository_icon_class),
);
let mut chapters = vec![];
for item in book.iter() {
// Create the data to inject in the template
let mut chapter = BTreeMap::new();
match *item {
2020-03-20 21:18:07 -05:00
BookItem::PartTitle(ref title) => {
chapter.insert("part".to_owned(), json!(title));
}
BookItem::Chapter(ref ch) => {
if let Some(ref section) = ch.number {
chapter.insert("section".to_owned(), json!(section.to_string()));
}
chapter.insert(
"has_sub_items".to_owned(),
json!((!ch.sub_items.is_empty()).to_string()),
);
chapter.insert("name".to_owned(), json!(ch.name));
2020-02-29 17:55:45 +01:00
if let Some(ref path) = ch.path {
let p = path
.to_str()
.with_context(|| "Could not convert path to str")?;
2020-02-29 17:55:45 +01:00
chapter.insert("path".to_owned(), json!(p));
}
}
BookItem::Separator => {
chapter.insert("spacer".to_owned(), json!("_spacer_"));
}
}
chapters.push(chapter);
}
data.insert("chapters".to_owned(), json!(chapters));
debug!("[*]: JSON constructed");
Ok(data)
}
struct RenderChapterContext<'a> {
2020-01-24 11:01:44 +08:00
handlebars: &'a Handlebars<'a>,
destination: PathBuf,
data: serde_json::Map<String, serde_json::Value>,
book_config: BookConfig,
html_config: HtmlConfig,
chapter_titles: &'a HashMap<PathBuf, String>,
}
/// Redirect mapping.
///
/// The key is the source path (like `foo/bar.html`). The value is a tuple
/// `(destination_path, fragment_map)`. The `destination_path` is the page to
/// redirect to. `fragment_map` is the map of fragments that override the
/// destination. For example, a fragment `#foo` could redirect to any other
/// page or site.
type CombinedRedirects = BTreeMap<String, (String, BTreeMap<String, String>)>;
fn combine_fragment_redirects(redirects: &HashMap<String, String>) -> CombinedRedirects {
let mut combined: CombinedRedirects = BTreeMap::new();
// This needs to extract the fragments to generate the fragment map.
for (original, new) in redirects {
if let Some((source_path, source_fragment)) = original.rsplit_once('#') {
let e = combined.entry(source_path.to_string()).or_default();
if let Some(old) = e.1.insert(format!("#{source_fragment}"), new.clone()) {
error!(
"internal error: found duplicate fragment redirect \
{old} for {source_path}#{source_fragment}"
);
}
} else {
let e = combined.entry(original.to_string()).or_default();
e.0 = new.clone();
}
}
combined
}
/// Collects fragment redirects for an existing page.
///
/// The returned map has keys like `#foo` and the value is the new destination
/// path or URL.
fn collect_redirects_for_path(
path: &Path,
redirects: &HashMap<String, String>,
) -> Result<BTreeMap<String, String>> {
let path = format!("/{}", path.to_url_path());
if redirects.contains_key(&path) {
bail!(
"redirect found for existing chapter at `{path}`\n\
Either delete the redirect or remove the chapter."
);
}
let key_prefix = format!("{path}#");
let map = redirects
.iter()
.filter_map(|(source, dest)| {
source
.strip_prefix(&key_prefix)
.map(|fragment| (format!("#{fragment}"), dest.to_string()))
})
.collect();
Ok(map)
}