new function to pull the links from a chunk of markdown

This commit is contained in:
QuietMisdreavus 2017-12-21 14:54:16 -06:00 committed by Manish Goregaokar
parent bc072ed0ca
commit 473fcfd49a

View file

@ -527,6 +527,7 @@ struct MyOpaque {
*const hoedown_buffer, *const hoedown_renderer_data,
libc::size_t),
toc_builder: Option<TocBuilder>,
links: Option<Vec<String>>,
}
extern {
@ -555,201 +556,203 @@ impl hoedown_buffer {
}
}
extern fn hoedown_block(ob: *mut hoedown_buffer, orig_text: *const hoedown_buffer,
lang: *const hoedown_buffer, data: *const hoedown_renderer_data,
line: libc::size_t) {
unsafe {
if orig_text.is_null() { return }
let opaque = (*data).opaque as *mut hoedown_html_renderer_state;
let my_opaque: &MyOpaque = &*((*opaque).opaque as *const MyOpaque);
let text = (*orig_text).as_bytes();
let origtext = str::from_utf8(text).unwrap();
let origtext = origtext.trim_left();
debug!("docblock: ==============\n{:?}\n=======", text);
let mut compile_fail = false;
let mut ignore = false;
let rendered = if lang.is_null() || origtext.is_empty() {
false
} else {
let rlang = (*lang).as_bytes();
let rlang = str::from_utf8(rlang).unwrap();
let parse_result = LangString::parse(rlang);
compile_fail = parse_result.compile_fail;
ignore = parse_result.ignore;
if !parse_result.rust {
(my_opaque.dfltblk)(ob, orig_text, lang,
opaque as *const hoedown_renderer_data,
line);
true
} else {
false
}
};
let lines = origtext.lines().filter_map(|l| map_line(l).for_html());
let text = lines.collect::<Vec<&str>>().join("\n");
if rendered { return }
PLAYGROUND.with(|play| {
// insert newline to clearly separate it from the
// previous block so we can shorten the html output
let mut s = String::from("\n");
let playground_button = play.borrow().as_ref().and_then(|&(ref krate, ref url)| {
if url.is_empty() {
return None;
}
let test = origtext.lines()
.map(|l| map_line(l).for_code())
.collect::<Vec<&str>>().join("\n");
let krate = krate.as_ref().map(|s| &**s);
let (test, _) = test::make_test(&test, krate, false,
&Default::default());
let channel = if test.contains("#![feature(") {
"&amp;version=nightly"
} else {
""
};
// These characters don't need to be escaped in a URI.
// FIXME: use a library function for percent encoding.
fn dont_escape(c: u8) -> bool {
(b'a' <= c && c <= b'z') ||
(b'A' <= c && c <= b'Z') ||
(b'0' <= c && c <= b'9') ||
c == b'-' || c == b'_' || c == b'.' ||
c == b'~' || c == b'!' || c == b'\'' ||
c == b'(' || c == b')' || c == b'*'
}
let mut test_escaped = String::new();
for b in test.bytes() {
if dont_escape(b) {
test_escaped.push(char::from(b));
} else {
write!(test_escaped, "%{:02X}", b).unwrap();
}
}
Some(format!(
r#"<a class="test-arrow" target="_blank" href="{}?code={}{}">Run</a>"#,
url, test_escaped, channel
))
});
let tooltip = if ignore {
Some(("This example is not tested", "ignore"))
} else if compile_fail {
Some(("This example deliberately fails to compile", "compile_fail"))
} else {
None
};
s.push_str(&highlight::render_with_highlighting(
&text,
Some(&format!("rust-example-rendered{}",
if ignore { " ignore" }
else if compile_fail { " compile_fail" }
else { "" })),
None,
playground_button.as_ref().map(String::as_str),
tooltip));
hoedown_buffer_put(ob, s.as_ptr(), s.len());
})
}
}
extern fn hoedown_header(ob: *mut hoedown_buffer, text: *const hoedown_buffer,
level: libc::c_int, data: *const hoedown_renderer_data,
_: libc::size_t) {
// hoedown does this, we may as well too
unsafe { hoedown_buffer_put(ob, "\n".as_ptr(), 1); }
// Extract the text provided
let s = if text.is_null() {
"".to_owned()
} else {
let s = unsafe { (*text).as_bytes() };
str::from_utf8(&s).unwrap().to_owned()
};
// Discard '<em>', '<code>' tags and some escaped characters,
// transform the contents of the header into a hyphenated string
// without non-alphanumeric characters other than '-' and '_'.
//
// This is a terrible hack working around how hoedown gives us rendered
// html for text rather than the raw text.
let mut id = s.clone();
let repl_sub = vec!["<em>", "</em>", "<code>", "</code>",
"<strong>", "</strong>",
"&lt;", "&gt;", "&amp;", "&#39;", "&quot;"];
for sub in repl_sub {
id = id.replace(sub, "");
}
let id = id.chars().filter_map(|c| {
if c.is_alphanumeric() || c == '-' || c == '_' {
if c.is_ascii() {
Some(c.to_ascii_lowercase())
} else {
Some(c)
}
} else if c.is_whitespace() && c.is_ascii() {
Some('-')
} else {
None
}
}).collect::<String>();
let opaque = unsafe { (*data).opaque as *mut hoedown_html_renderer_state };
let opaque = unsafe { &mut *((*opaque).opaque as *mut MyOpaque) };
let id = derive_id(id);
let sec = opaque.toc_builder.as_mut().map_or("".to_owned(), |builder| {
format!("{} ", builder.push(level as u32, s.clone(), id.clone()))
});
// Render the HTML
let text = format!("<h{lvl} id='{id}' class='section-header'>\
<a href='#{id}'>{sec}{}</a></h{lvl}>",
s, lvl = level, id = id, sec = sec);
unsafe { hoedown_buffer_put(ob, text.as_ptr(), text.len()); }
}
extern fn hoedown_codespan(
ob: *mut hoedown_buffer,
text: *const hoedown_buffer,
_: *const hoedown_renderer_data,
_: libc::size_t
) -> libc::c_int {
let content = if text.is_null() {
"".to_owned()
} else {
let bytes = unsafe { (*text).as_bytes() };
let s = str::from_utf8(bytes).unwrap();
collapse_whitespace(s)
};
let content = format!("<code>{}</code>", Escape(&content));
unsafe {
hoedown_buffer_put(ob, content.as_ptr(), content.len());
}
// Return anything except 0, which would mean "also print the code span verbatim".
1
}
pub fn render(w: &mut fmt::Formatter,
s: &str,
print_toc: bool,
html_flags: libc::c_uint) -> fmt::Result {
extern fn block(ob: *mut hoedown_buffer, orig_text: *const hoedown_buffer,
lang: *const hoedown_buffer, data: *const hoedown_renderer_data,
line: libc::size_t) {
unsafe {
if orig_text.is_null() { return }
let opaque = (*data).opaque as *mut hoedown_html_renderer_state;
let my_opaque: &MyOpaque = &*((*opaque).opaque as *const MyOpaque);
let text = (*orig_text).as_bytes();
let origtext = str::from_utf8(text).unwrap();
let origtext = origtext.trim_left();
debug!("docblock: ==============\n{:?}\n=======", text);
let mut compile_fail = false;
let mut ignore = false;
let rendered = if lang.is_null() || origtext.is_empty() {
false
} else {
let rlang = (*lang).as_bytes();
let rlang = str::from_utf8(rlang).unwrap();
let parse_result = LangString::parse(rlang);
compile_fail = parse_result.compile_fail;
ignore = parse_result.ignore;
if !parse_result.rust {
(my_opaque.dfltblk)(ob, orig_text, lang,
opaque as *const hoedown_renderer_data,
line);
true
} else {
false
}
};
let lines = origtext.lines().filter_map(|l| map_line(l).for_html());
let text = lines.collect::<Vec<&str>>().join("\n");
if rendered { return }
PLAYGROUND.with(|play| {
// insert newline to clearly separate it from the
// previous block so we can shorten the html output
let mut s = String::from("\n");
let playground_button = play.borrow().as_ref().and_then(|&(ref krate, ref url)| {
if url.is_empty() {
return None;
}
let test = origtext.lines()
.map(|l| map_line(l).for_code())
.collect::<Vec<&str>>().join("\n");
let krate = krate.as_ref().map(|s| &**s);
let (test, _) = test::make_test(&test, krate, false,
&Default::default());
let channel = if test.contains("#![feature(") {
"&amp;version=nightly"
} else {
""
};
// These characters don't need to be escaped in a URI.
// FIXME: use a library function for percent encoding.
fn dont_escape(c: u8) -> bool {
(b'a' <= c && c <= b'z') ||
(b'A' <= c && c <= b'Z') ||
(b'0' <= c && c <= b'9') ||
c == b'-' || c == b'_' || c == b'.' ||
c == b'~' || c == b'!' || c == b'\'' ||
c == b'(' || c == b')' || c == b'*'
}
let mut test_escaped = String::new();
for b in test.bytes() {
if dont_escape(b) {
test_escaped.push(char::from(b));
} else {
write!(test_escaped, "%{:02X}", b).unwrap();
}
}
Some(format!(
r#"<a class="test-arrow" target="_blank" href="{}?code={}{}">Run</a>"#,
url, test_escaped, channel
))
});
let tooltip = if ignore {
Some(("This example is not tested", "ignore"))
} else if compile_fail {
Some(("This example deliberately fails to compile", "compile_fail"))
} else {
None
};
s.push_str(&highlight::render_with_highlighting(
&text,
Some(&format!("rust-example-rendered{}",
if ignore { " ignore" }
else if compile_fail { " compile_fail" }
else { "" })),
None,
playground_button.as_ref().map(String::as_str),
tooltip));
hoedown_buffer_put(ob, s.as_ptr(), s.len());
})
}
}
extern fn header(ob: *mut hoedown_buffer, text: *const hoedown_buffer,
level: libc::c_int, data: *const hoedown_renderer_data,
_: libc::size_t) {
// hoedown does this, we may as well too
unsafe { hoedown_buffer_put(ob, "\n".as_ptr(), 1); }
// Extract the text provided
let s = if text.is_null() {
"".to_owned()
} else {
let s = unsafe { (*text).as_bytes() };
str::from_utf8(&s).unwrap().to_owned()
};
// Discard '<em>', '<code>' tags and some escaped characters,
// transform the contents of the header into a hyphenated string
// without non-alphanumeric characters other than '-' and '_'.
//
// This is a terrible hack working around how hoedown gives us rendered
// html for text rather than the raw text.
let mut id = s.clone();
let repl_sub = vec!["<em>", "</em>", "<code>", "</code>",
"<strong>", "</strong>",
"&lt;", "&gt;", "&amp;", "&#39;", "&quot;"];
for sub in repl_sub {
id = id.replace(sub, "");
}
let id = id.chars().filter_map(|c| {
if c.is_alphanumeric() || c == '-' || c == '_' {
if c.is_ascii() {
Some(c.to_ascii_lowercase())
} else {
Some(c)
}
} else if c.is_whitespace() && c.is_ascii() {
Some('-')
} else {
None
}
}).collect::<String>();
let opaque = unsafe { (*data).opaque as *mut hoedown_html_renderer_state };
let opaque = unsafe { &mut *((*opaque).opaque as *mut MyOpaque) };
let id = derive_id(id);
let sec = opaque.toc_builder.as_mut().map_or("".to_owned(), |builder| {
format!("{} ", builder.push(level as u32, s.clone(), id.clone()))
});
// Render the HTML
let text = format!("<h{lvl} id='{id}' class='section-header'>\
<a href='#{id}'>{sec}{}</a></h{lvl}>",
s, lvl = level, id = id, sec = sec);
unsafe { hoedown_buffer_put(ob, text.as_ptr(), text.len()); }
}
extern fn codespan(
ob: *mut hoedown_buffer,
text: *const hoedown_buffer,
_: *const hoedown_renderer_data,
_: libc::size_t
) -> libc::c_int {
let content = if text.is_null() {
"".to_owned()
} else {
let bytes = unsafe { (*text).as_bytes() };
let s = str::from_utf8(bytes).unwrap();
collapse_whitespace(s)
};
let content = format!("<code>{}</code>", Escape(&content));
unsafe {
hoedown_buffer_put(ob, content.as_ptr(), content.len());
}
// Return anything except 0, which would mean "also print the code span verbatim".
1
}
unsafe {
let ob = hoedown_buffer_new(DEF_OUNIT);
let renderer = hoedown_html_renderer_new(html_flags, 0);
let mut opaque = MyOpaque {
dfltblk: (*renderer).blockcode.unwrap(),
toc_builder: if print_toc {Some(TocBuilder::new())} else {None}
toc_builder: if print_toc {Some(TocBuilder::new())} else {None},
links: None,
};
(*((*renderer).opaque as *mut hoedown_html_renderer_state)).opaque
= &mut opaque as *mut _ as *mut libc::c_void;
(*renderer).blockcode = Some(block);
(*renderer).header = Some(header);
(*renderer).codespan = Some(codespan);
(*renderer).blockcode = Some(hoedown_block);
(*renderer).header = Some(hoedown_header);
(*renderer).codespan = Some(hoedown_codespan);
let document = hoedown_document_new(renderer, HOEDOWN_EXTENSIONS, 16);
hoedown_document_render(document, ob, s.as_ptr(),
@ -1140,6 +1143,86 @@ pub fn plain_summary_line(md: &str) -> String {
s
}
pub fn markdown_links(md: &str, render_type: RenderType) -> Vec<String> {
if md.is_empty() {
return vec![];
}
match render_type {
RenderType::Hoedown => {
extern fn hoedown_link(
_ob: *mut hoedown_buffer,
_content: *const hoedown_buffer,
link: *const hoedown_buffer,
_title: *const hoedown_buffer,
data: *const hoedown_renderer_data,
_line: libc::size_t
) -> libc::c_int {
if link.is_null() {
return 0;
}
let opaque = unsafe { (*data).opaque as *mut hoedown_html_renderer_state };
let opaque = unsafe { &mut *((*opaque).opaque as *mut MyOpaque) };
if let Some(ref mut links) = opaque.links {
let s = unsafe { (*link).as_bytes() };
let s = str::from_utf8(&s).unwrap().to_owned();
links.push(s);
}
//returning 0 here means "emit the span verbatim", but we're not using the output
//anyway so we don't really care
0
}
unsafe {
let ob = hoedown_buffer_new(DEF_OUNIT);
let renderer = hoedown_html_renderer_new(0, 0);
let mut opaque = MyOpaque {
dfltblk: (*renderer).blockcode.unwrap(),
toc_builder: None,
links: Some(vec![]),
};
(*((*renderer).opaque as *mut hoedown_html_renderer_state)).opaque
= &mut opaque as *mut _ as *mut libc::c_void;
(*renderer).blockcode = Some(hoedown_block);
(*renderer).header = Some(hoedown_header);
(*renderer).codespan = Some(hoedown_codespan);
(*renderer).link = Some(hoedown_link);
let document = hoedown_document_new(renderer, HOEDOWN_EXTENSIONS, 16);
hoedown_document_render(document, ob, md.as_ptr(),
md.len() as libc::size_t);
hoedown_document_free(document);
hoedown_html_renderer_free(renderer);
opaque.links.unwrap()
}
}
RenderType::Pulldown => {
let mut opts = Options::empty();
opts.insert(OPTION_ENABLE_TABLES);
opts.insert(OPTION_ENABLE_FOOTNOTES);
let p = Parser::new_ext(md, opts);
let iter = Footnotes::new(CodeBlocks::new(HeadingLinks::new(p, None)));
let mut links = vec![];
for ev in iter {
if let Event::Start(Tag::Link(dest, _)) = ev {
links.push(dest.into_owned());
}
}
links
}
}
}
#[cfg(test)]
mod tests {
use super::{LangString, Markdown, MarkdownHtml};