zola/components/rendering/src/markdown.rs

475 lines
18 KiB
Rust
Raw Normal View History

2019-12-21 21:52:39 +00:00
use lazy_static::lazy_static;
2017-03-07 12:34:31 +00:00
use pulldown_cmark as cmark;
use regex::Regex;
use syntect::html::{start_highlighted_html_snippet, IncludeBackground};
2017-03-27 14:17:33 +00:00
2019-12-21 21:52:39 +00:00
use crate::context::RenderContext;
use crate::table_of_contents::{make_table_of_contents, Heading};
use config::highlighting::THEME_SET;
2019-01-23 18:20:02 +00:00
use errors::{Error, Result};
use front_matter::InsertAnchor;
2018-10-31 07:18:57 +00:00
use utils::site::resolve_internal_link;
use utils::slugs::slugify_anchors;
2019-12-23 08:21:51 +00:00
use utils::vec::InsertMany;
2018-05-06 20:58:39 +00:00
use self::cmark::{Event, LinkType, Options, Parser, Tag};
2018-05-06 20:58:39 +00:00
mod codeblock;
mod fence;
use self::codeblock::CodeBlock;
const CONTINUE_READING: &str = "<span id=\"continue-reading\"></span>";
const ANCHOR_LINK_TEMPLATE: &str = "anchor-link.html";
2018-08-24 21:37:39 +00:00
#[derive(Debug)]
pub struct Rendered {
pub body: String,
pub summary_len: Option<usize>,
pub toc: Vec<Heading>,
pub internal_links_with_anchors: Vec<(String, String)>,
pub external_links: Vec<String>,
}
// tracks a heading in a slice of pulldown-cmark events
2019-01-16 09:09:23 +00:00
#[derive(Debug)]
struct HeadingRef {
2019-01-18 14:46:18 +00:00
start_idx: usize,
end_idx: usize,
level: u32,
Allow manual specification of header IDs (#685) Justification for this feature is added in the docs. Precedent for the precise syntax: Hugo. Hugo puts this syntax behind a preference named headerIds, and automatic header ID generation behind a preference named autoHeaderIds, with both enabled by default. I have not implemented a switch to disable this. My suggestion for a workaround for the improbable case of desiring a literal “{#…}” at the end of a header is to replace `}` with `&#125;`. The algorithm I have used is not identical to [that which Hugo uses][0], because Hugo’s looks to work at the source level, whereas here we work at the pulldown-cmark event level, which is generally more sane, but potentially limiting for extremely esoteric IDs. Practical differences in implementation from Hugo (based purely on reading [blackfriday’s implementation][0], not actually trying it): - I believe Hugo would treat `# Foo {#*bar*}` as a heading with text “Foo” and ID `*bar*`, since it is working at the source level; whereas this code turns it into a heading with HTML `Foo {#<em>bar</em>}`, as it works at the pulldown-cmark event level and doesn’t go out of its way to make that work (I’m not familiar with pulldown-cmark, but I get the impression that you could make it work Hugo’s way on this point). The difference should be negligible: only *very* esoteric hashes would include magic Markdown characters. - Hugo will automatically generate an ID for `{#}`, whereas what I’ve coded here will yield a blank ID instead (which feels more correct to me—`None` versus `Some("")`, and all that). In practice the results should be identical. Fixes #433. [0]: https://github.com/russross/blackfriday/blob/a477dd1646916742841ed20379f941cfa6c5bb6f/block.go#L218-L234
2019-05-20 20:08:49 +00:00
id: Option<String>,
}
impl HeadingRef {
fn new(start: usize, level: u32) -> HeadingRef {
HeadingRef { start_idx: start, end_idx: 0, level, id: None }
}
}
2018-05-06 20:58:39 +00:00
// We might have cases where the slug is already present in our list of anchor
// for example an article could have several titles named Example
// We add a counter after the slug if the slug is already present, which
// means we will have example, example-1, example-2 etc
fn find_anchor(anchors: &[String], name: String, level: u8) -> String {
if level == 0 && !anchors.contains(&name) {
2018-11-19 14:04:22 +00:00
return name;
2018-05-06 20:58:39 +00:00
}
let new_anchor = format!("{}-{}", name, level + 1);
if !anchors.contains(&new_anchor) {
return new_anchor;
}
find_anchor(anchors, name, level + 1)
}
2017-03-07 12:34:31 +00:00
// Returns whether the given string starts with a schema.
//
// Although there exists [a list of registered URI schemes][uri-schemes], a link may use arbitrary,
// private schemes. This function checks if the given string starts with something that just looks
// like a scheme, i.e., a case-insensitive identifier followed by a colon.
//
// [uri-schemes]: https://www.iana.org/assignments/uri-schemes/uri-schemes.xhtml
fn starts_with_schema(s: &str) -> bool {
lazy_static! {
static ref PATTERN: Regex = Regex::new(r"^[0-9A-Za-z\-]+:").unwrap();
}
PATTERN.is_match(s)
}
// Colocated asset links refers to the files in the same directory,
// there it should be a filename only
2018-05-07 19:03:51 +00:00
fn is_colocated_asset_link(link: &str) -> bool {
2018-09-30 19:15:09 +00:00
!link.contains('/') // http://, ftp://, ../ etc
&& !starts_with_schema(link)
}
// Returns whether a link starts with an HTTP(s) scheme.
fn is_external_link(link: &str) -> bool {
link.starts_with("http:") || link.starts_with("https:")
2018-05-07 19:03:51 +00:00
}
2019-06-02 18:21:06 +00:00
fn fix_link(
link_type: LinkType,
link: &str,
context: &RenderContext,
internal_links_with_anchors: &mut Vec<(String, String)>,
2019-06-02 18:21:06 +00:00
external_links: &mut Vec<String>,
) -> Result<String> {
2019-03-15 20:24:06 +00:00
if link_type == LinkType::Email {
return Ok(link.to_string());
}
// A few situations here:
// - it could be a relative link (starting with `@/`)
// - it could be a link to a co-located asset
// - it could be a normal link
let result = if link.starts_with("@/") {
match resolve_internal_link(&link, &context.permalinks) {
Ok(resolved) => {
if resolved.anchor.is_some() {
internal_links_with_anchors
.push((resolved.md_path.unwrap(), resolved.anchor.unwrap()));
}
resolved.permalink
}
Err(_) => {
return Err(format!("Relative link {} not found.", link).into());
}
}
} else if is_colocated_asset_link(&link) {
format!("{}{}", context.current_page_permalink, link)
} else {
if is_external_link(link) {
external_links.push(link.to_owned());
}
link.to_string()
};
Ok(result)
}
/// get only text in a slice of events
fn get_text(parser_slice: &[Event]) -> String {
let mut title = String::new();
for event in parser_slice.iter() {
2019-05-29 18:33:34 +00:00
match event {
Event::Text(text) | Event::Code(text) => title += text,
2019-06-02 18:21:06 +00:00
_ => continue,
}
}
title
}
fn get_heading_refs(events: &[Event]) -> Vec<HeadingRef> {
let mut heading_refs = vec![];
for (i, event) in events.iter().enumerate() {
match event {
Event::Start(Tag::Heading(level)) => {
heading_refs.push(HeadingRef::new(i, *level));
}
Event::End(Tag::Heading(_)) => {
let msg = "Heading end before start?";
heading_refs.last_mut().expect(msg).end_idx = i;
}
_ => (),
}
}
heading_refs
}
pub fn markdown_to_html(content: &str, context: &RenderContext) -> Result<Rendered> {
lazy_static! {
static ref EMOJI_REPLACER: gh_emoji::Replacer = gh_emoji::Replacer::new();
}
2018-05-06 20:58:39 +00:00
// the rendered html
2018-05-07 19:03:51 +00:00
let mut html = String::with_capacity(content.len());
2017-03-27 14:17:33 +00:00
// Set while parsing
let mut error = None;
2018-05-06 20:58:39 +00:00
let mut highlighter: Option<CodeBlock> = None;
let mut inserted_anchors: Vec<String> = vec![];
let mut headings: Vec<Heading> = vec![];
let mut internal_links_with_anchors = Vec::new();
let mut external_links = Vec::new();
2017-06-16 04:00:48 +00:00
let mut opts = Options::empty();
let mut has_summary = false;
let mut in_html_block = false;
opts.insert(Options::ENABLE_TABLES);
opts.insert(Options::ENABLE_FOOTNOTES);
opts.insert(Options::ENABLE_STRIKETHROUGH);
2020-06-11 16:29:05 +00:00
opts.insert(Options::ENABLE_TASKLISTS);
2020-11-27 10:11:19 +00:00
if context.config.markdown.smart_punctuation {
opts.insert(Options::ENABLE_SMART_PUNCTUATION);
}
2017-03-27 14:17:33 +00:00
{
let mut events = Parser::new_ext(content, opts)
.map(|event| {
match event {
Event::Text(text) => {
// if we are in the middle of a highlighted code block
if let Some(ref mut code_block) = highlighter {
let html = code_block.highlight(&text);
Event::Html(html.into())
} else if context.config.markdown.render_emoji {
let processed_text = EMOJI_REPLACER.replace_all(&text);
Event::Text(processed_text.to_string().into())
} else {
// Business as usual
Event::Text(text)
}
2017-03-27 14:17:33 +00:00
}
2020-02-16 18:17:41 +00:00
Event::Start(Tag::CodeBlock(ref kind)) => {
2020-11-21 11:38:43 +00:00
let language = match kind {
cmark::CodeBlockKind::Fenced(fence_info) => {
let fence_info = fence::FenceSettings::new(fence_info);
2020-11-21 11:38:43 +00:00
fence_info.language
}
2020-11-21 11:38:43 +00:00
_ => None,
};
2020-11-21 11:38:43 +00:00
if !context.config.highlight_code() {
if let Some(lang) = language {
2021-01-02 08:29:28 +00:00
let html = format!(
r#"<pre><code class="language-{}" data-lang="{}">"#,
lang, lang
);
return Event::Html(html.into());
}
2019-03-15 20:24:06 +00:00
return Event::Html("<pre><code>".into());
}
2018-05-07 16:38:04 +00:00
let theme = &THEME_SET.themes[context.config.highlight_theme()];
2020-02-16 18:17:41 +00:00
match kind {
cmark::CodeBlockKind::Indented => (),
cmark::CodeBlockKind::Fenced(fence_info) => {
// This selects the background color the same way that
// start_coloured_html_snippet does
let color = theme
.settings
.background
.unwrap_or(::syntect::highlighting::Color::WHITE);
highlighter = Some(CodeBlock::new(
fence_info,
&context.config,
IncludeBackground::IfDifferent(color),
2021-02-02 19:49:57 +00:00
context
.tera_context
.get("page")
.or(context.tera_context.get("section"))
.map(|x| {
x.as_object()
.unwrap()
.get("relative_path")
.unwrap()
.as_str()
.unwrap()
}),
));
2020-02-16 18:17:41 +00:00
}
};
let snippet = start_highlighted_html_snippet(theme);
let mut html = snippet.0;
if let Some(lang) = language {
2021-01-02 08:29:28 +00:00
html.push_str(&format!(
r#"<code class="language-{}" data-lang="{}">"#,
lang, lang
));
} else {
html.push_str("<code>");
}
Event::Html(html.into())
2017-03-27 14:17:33 +00:00
}
Event::End(Tag::CodeBlock(_)) => {
if !context.config.highlight_code() {
2019-03-15 20:24:06 +00:00
return Event::Html("</code></pre>\n".into());
}
// reset highlight and close the code block
highlighter = None;
Event::Html("</code></pre>".into())
}
2019-03-15 20:24:06 +00:00
Event::Start(Tag::Image(link_type, src, title)) => {
if is_colocated_asset_link(&src) {
2019-03-15 20:24:06 +00:00
let link = format!("{}{}", context.current_page_permalink, &*src);
return Event::Start(Tag::Image(link_type, link.into(), title));
}
2018-05-07 19:03:51 +00:00
2019-03-15 20:24:06 +00:00
Event::Start(Tag::Image(link_type, src, title))
}
Event::Start(Tag::Link(link_type, link, title)) if link.is_empty() => {
error = Some(Error::msg("There is a link that is missing a URL"));
Event::Start(Tag::Link(link_type, "#".into(), title))
}
2019-03-15 20:24:06 +00:00
Event::Start(Tag::Link(link_type, link, title)) => {
let fixed_link = match fix_link(
link_type,
&link,
context,
&mut internal_links_with_anchors,
&mut external_links,
) {
Ok(fixed_link) => fixed_link,
Err(err) => {
error = Some(err);
return Event::Html("".into());
}
};
if is_external_link(&link)
&& context.config.markdown.has_external_link_tweaks()
{
let mut escaped = String::new();
// write_str can fail but here there are no reasons it should (afaik?)
2020-11-25 10:53:14 +00:00
cmark::escape::escape_href(&mut escaped, &link)
.expect("Could not write to buffer");
Event::Html(
context
.config
.markdown
.construct_external_link_tag(&escaped, &title)
.into(),
)
} else {
Event::Start(Tag::Link(link_type, fixed_link.into(), title))
}
}
Event::Html(ref markup) => {
if markup.contains("<!-- more -->") {
has_summary = true;
Event::Html(CONTINUE_READING.into())
2020-11-21 11:38:43 +00:00
} else if in_html_block && markup.contains("</pre>") {
in_html_block = false;
Event::Html(markup.replacen("</pre>", "", 1).into())
} else if markup.contains("pre data-shortcode") {
in_html_block = true;
let m = markup.replacen("<pre data-shortcode>", "", 1);
if m.contains("</pre>") {
in_html_block = false;
2020-11-21 11:38:43 +00:00
Event::Html(m.replacen("</pre>", "", 1).into())
} else {
2020-11-21 11:38:43 +00:00
Event::Html(m.into())
}
2020-11-21 11:38:43 +00:00
} else {
event
}
}
_ => event,
}
})
.collect::<Vec<_>>(); // We need to collect the events to make a second pass
let mut heading_refs = get_heading_refs(&events);
let mut anchors_to_insert = vec![];
// First heading pass: look for a manually-specified IDs, e.g. `# Heading text {#hash}`
Allow manual specification of header IDs (#685) Justification for this feature is added in the docs. Precedent for the precise syntax: Hugo. Hugo puts this syntax behind a preference named headerIds, and automatic header ID generation behind a preference named autoHeaderIds, with both enabled by default. I have not implemented a switch to disable this. My suggestion for a workaround for the improbable case of desiring a literal “{#…}” at the end of a header is to replace `}` with `&#125;`. The algorithm I have used is not identical to [that which Hugo uses][0], because Hugo’s looks to work at the source level, whereas here we work at the pulldown-cmark event level, which is generally more sane, but potentially limiting for extremely esoteric IDs. Practical differences in implementation from Hugo (based purely on reading [blackfriday’s implementation][0], not actually trying it): - I believe Hugo would treat `# Foo {#*bar*}` as a heading with text “Foo” and ID `*bar*`, since it is working at the source level; whereas this code turns it into a heading with HTML `Foo {#<em>bar</em>}`, as it works at the pulldown-cmark event level and doesn’t go out of its way to make that work (I’m not familiar with pulldown-cmark, but I get the impression that you could make it work Hugo’s way on this point). The difference should be negligible: only *very* esoteric hashes would include magic Markdown characters. - Hugo will automatically generate an ID for `{#}`, whereas what I’ve coded here will yield a blank ID instead (which feels more correct to me—`None` versus `Some("")`, and all that). In practice the results should be identical. Fixes #433. [0]: https://github.com/russross/blackfriday/blob/a477dd1646916742841ed20379f941cfa6c5bb6f/block.go#L218-L234
2019-05-20 20:08:49 +00:00
// (This is a separate first pass so that auto IDs can avoid collisions with manual IDs.)
for heading_ref in heading_refs.iter_mut() {
let end_idx = heading_ref.end_idx;
Allow manual specification of header IDs (#685) Justification for this feature is added in the docs. Precedent for the precise syntax: Hugo. Hugo puts this syntax behind a preference named headerIds, and automatic header ID generation behind a preference named autoHeaderIds, with both enabled by default. I have not implemented a switch to disable this. My suggestion for a workaround for the improbable case of desiring a literal “{#…}” at the end of a header is to replace `}` with `&#125;`. The algorithm I have used is not identical to [that which Hugo uses][0], because Hugo’s looks to work at the source level, whereas here we work at the pulldown-cmark event level, which is generally more sane, but potentially limiting for extremely esoteric IDs. Practical differences in implementation from Hugo (based purely on reading [blackfriday’s implementation][0], not actually trying it): - I believe Hugo would treat `# Foo {#*bar*}` as a heading with text “Foo” and ID `*bar*`, since it is working at the source level; whereas this code turns it into a heading with HTML `Foo {#<em>bar</em>}`, as it works at the pulldown-cmark event level and doesn’t go out of its way to make that work (I’m not familiar with pulldown-cmark, but I get the impression that you could make it work Hugo’s way on this point). The difference should be negligible: only *very* esoteric hashes would include magic Markdown characters. - Hugo will automatically generate an ID for `{#}`, whereas what I’ve coded here will yield a blank ID instead (which feels more correct to me—`None` versus `Some("")`, and all that). In practice the results should be identical. Fixes #433. [0]: https://github.com/russross/blackfriday/blob/a477dd1646916742841ed20379f941cfa6c5bb6f/block.go#L218-L234
2019-05-20 20:08:49 +00:00
if let Event::Text(ref mut text) = events[end_idx - 1] {
if text.as_bytes().last() == Some(&b'}') {
if let Some(mut i) = text.find("{#") {
let id = text[i + 2..text.len() - 1].to_owned();
inserted_anchors.push(id.clone());
while i > 0 && text.as_bytes()[i - 1] == b' ' {
i -= 1;
}
heading_ref.id = Some(id);
Allow manual specification of header IDs (#685) Justification for this feature is added in the docs. Precedent for the precise syntax: Hugo. Hugo puts this syntax behind a preference named headerIds, and automatic header ID generation behind a preference named autoHeaderIds, with both enabled by default. I have not implemented a switch to disable this. My suggestion for a workaround for the improbable case of desiring a literal “{#…}” at the end of a header is to replace `}` with `&#125;`. The algorithm I have used is not identical to [that which Hugo uses][0], because Hugo’s looks to work at the source level, whereas here we work at the pulldown-cmark event level, which is generally more sane, but potentially limiting for extremely esoteric IDs. Practical differences in implementation from Hugo (based purely on reading [blackfriday’s implementation][0], not actually trying it): - I believe Hugo would treat `# Foo {#*bar*}` as a heading with text “Foo” and ID `*bar*`, since it is working at the source level; whereas this code turns it into a heading with HTML `Foo {#<em>bar</em>}`, as it works at the pulldown-cmark event level and doesn’t go out of its way to make that work (I’m not familiar with pulldown-cmark, but I get the impression that you could make it work Hugo’s way on this point). The difference should be negligible: only *very* esoteric hashes would include magic Markdown characters. - Hugo will automatically generate an ID for `{#}`, whereas what I’ve coded here will yield a blank ID instead (which feels more correct to me—`None` versus `Some("")`, and all that). In practice the results should be identical. Fixes #433. [0]: https://github.com/russross/blackfriday/blob/a477dd1646916742841ed20379f941cfa6c5bb6f/block.go#L218-L234
2019-05-20 20:08:49 +00:00
*text = text[..i].to_owned().into();
}
}
}
}
// Second heading pass: auto-generate remaining IDs, and emit HTML
for heading_ref in heading_refs {
let start_idx = heading_ref.start_idx;
let end_idx = heading_ref.end_idx;
let title = get_text(&events[start_idx + 1..end_idx]);
2019-12-23 08:21:51 +00:00
let id = heading_ref.id.unwrap_or_else(|| {
find_anchor(
&inserted_anchors,
slugify_anchors(&title, context.config.slugify.anchors),
2019-12-23 08:21:51 +00:00
0,
)
});
inserted_anchors.push(id.clone());
// insert `id` to the tag
let html = format!("<h{lvl} id=\"{id}\">", lvl = heading_ref.level, id = id);
2019-03-15 20:24:06 +00:00
events[start_idx] = Event::Html(html.into());
// generate anchors and places to insert them
if context.insert_anchor != InsertAnchor::None {
let anchor_idx = match context.insert_anchor {
InsertAnchor::Left => start_idx + 1,
InsertAnchor::Right => end_idx,
InsertAnchor::None => 0, // Not important
};
let mut c = tera::Context::new();
c.insert("id", &id);
c.insert("level", &heading_ref.level);
2019-01-23 18:20:02 +00:00
let anchor_link = utils::templates::render_template(
&ANCHOR_LINK_TEMPLATE,
&context.tera,
c,
&None,
)
.map_err(|e| Error::chain("Failed to render anchor link template", e))?;
2019-03-15 20:24:06 +00:00
anchors_to_insert.push((anchor_idx, Event::Html(anchor_link.into())));
}
// record heading to make table of contents
let permalink = format!("{}#{}", context.current_page_permalink, id);
let h =
Heading { level: heading_ref.level, id, permalink, title, children: Vec::new() };
headings.push(h);
}
if context.insert_anchor != InsertAnchor::None {
events.insert_many(anchors_to_insert);
}
2017-03-07 12:34:31 +00:00
cmark::html::push_html(&mut html, events.into_iter());
2017-03-07 12:34:31 +00:00
}
if let Some(e) = error {
2019-08-24 20:23:08 +00:00
Err(e)
} else {
Ok(Rendered {
summary_len: if has_summary { html.find(CONTINUE_READING) } else { None },
body: html,
toc: make_table_of_contents(headings),
internal_links_with_anchors,
external_links,
})
2017-03-27 14:17:33 +00:00
}
2017-03-07 12:34:31 +00:00
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_starts_with_schema() {
// registered
assert!(starts_with_schema("https://example.com/"));
assert!(starts_with_schema("ftp://example.com/"));
assert!(starts_with_schema("mailto:user@example.com"));
assert!(starts_with_schema("xmpp:node@example.com"));
assert!(starts_with_schema("tel:18008675309"));
assert!(starts_with_schema("sms:18008675309"));
assert!(starts_with_schema("h323:user@example.com"));
// arbitrary
assert!(starts_with_schema("zola:post?content=hi"));
// case-insensitive
assert!(starts_with_schema("MailTo:user@example.com"));
assert!(starts_with_schema("MAILTO:user@example.com"));
}
#[test]
fn test_is_external_link() {
assert!(is_external_link("http://example.com/"));
assert!(is_external_link("https://example.com/"));
assert!(is_external_link("https://example.com/index.html#introduction"));
assert!(!is_external_link("mailto:user@example.com"));
assert!(!is_external_link("tel:18008675309"));
assert!(!is_external_link("#introduction"));
assert!(!is_external_link("http.jpg"))
}
}