orgize/src/parsers.rs

831 lines
26 KiB
Rust
Raw Normal View History

2019-08-06 07:03:16 +01:00
// parser related functions
use std::borrow::Cow;
use std::iter::once;
2019-08-10 13:17:39 +01:00
use std::marker::PhantomData;
2019-08-06 07:03:16 +01:00
use indextree::{Arena, NodeId};
use jetscii::{bytes, BytesConst};
use memchr::{memchr, memchr_iter};
2019-10-02 09:50:36 +01:00
use nom::{bytes::complete::take_while1, combinator::verify, error::ParseError, IResult};
2019-08-06 07:03:16 +01:00
use crate::config::ParseConfig;
2019-10-02 09:50:36 +01:00
use crate::elements::{
block::parse_block_element, emphasis::parse_emphasis, keyword::parse_keyword,
2019-10-28 05:33:18 +00:00
radio_target::parse_radio_target, table::parse_table_el, BabelCall, CenterBlock, Clock,
Comment, CommentBlock, Cookie, Drawer, DynBlock, Element, ExampleBlock, ExportBlock,
FixedWidth, FnDef, FnRef, InlineCall, InlineSrc, Keyword, Link, List, ListItem, Macros,
QuoteBlock, Rule, Snippet, SourceBlock, SpecialBlock, Table, TableRow, Target, Timestamp,
Title, VerseBlock,
2019-10-02 09:50:36 +01:00
};
2019-08-06 07:03:16 +01:00
pub trait ElementArena<'a> {
2019-10-30 03:31:37 +00:00
fn append<T>(&mut self, element: T, parent: NodeId) -> NodeId
where
T: Into<Element<'a>>;
fn insert_before_last_child<T>(&mut self, element: T, parent: NodeId) -> NodeId
where
T: Into<Element<'a>>;
fn set<T>(&mut self, node: NodeId, element: T)
where
T: Into<Element<'a>>;
}
impl<'a> ElementArena<'a> for Arena<Element<'a>> {
2019-10-30 03:31:37 +00:00
fn append<T>(&mut self, element: T, parent: NodeId) -> NodeId
where
T: Into<Element<'a>>,
{
let node = self.new_node(element.into());
parent.append(node, self);
node
}
2019-10-30 03:31:37 +00:00
fn insert_before_last_child<T>(&mut self, element: T, parent: NodeId) -> NodeId
where
T: Into<Element<'a>>,
{
if let Some(child) = self[parent].last_child() {
let node = self.new_node(element.into());
child.insert_before(node, self);
node
} else {
2019-10-30 03:31:37 +00:00
self.append(element, parent)
}
}
2019-10-30 03:31:37 +00:00
fn set<T>(&mut self, node: NodeId, element: T)
where
T: Into<Element<'a>>,
{
*self[node].get_mut() = element.into();
}
}
2019-08-11 04:40:52 +01:00
pub struct OwnedArena<'a, 'b, 'c> {
arena: &'b mut Arena<Element<'c>>,
phantom: PhantomData<&'a ()>,
}
impl<'a, 'b, 'c> OwnedArena<'a, 'b, 'c> {
pub fn new(arena: &'b mut Arena<Element<'c>>) -> OwnedArena<'a, 'b, 'c> {
OwnedArena {
arena,
phantom: PhantomData,
}
}
}
impl<'a> ElementArena<'a> for OwnedArena<'a, '_, '_> {
2019-10-30 03:31:37 +00:00
fn append<T>(&mut self, element: T, parent: NodeId) -> NodeId
where
T: Into<Element<'a>>,
{
self.arena.append(element.into().into_owned(), parent)
2019-08-11 04:40:52 +01:00
}
2019-10-30 03:31:37 +00:00
fn insert_before_last_child<T>(&mut self, element: T, parent: NodeId) -> NodeId
where
T: Into<Element<'a>>,
{
2019-10-02 09:50:36 +01:00
self.arena
.insert_before_last_child(element.into().into_owned(), parent)
2019-08-11 04:40:52 +01:00
}
2019-10-30 03:31:37 +00:00
fn set<T>(&mut self, node: NodeId, element: T)
where
T: Into<Element<'a>>,
{
self.arena.set(node, element.into().into_owned());
}
2019-08-11 04:40:52 +01:00
}
#[derive(Debug)]
2019-08-06 07:03:16 +01:00
pub enum Container<'a> {
// Block, List Item
2019-10-30 03:31:37 +00:00
Block { content: &'a str, node: NodeId },
2019-08-06 07:03:16 +01:00
// Pargraph, Inline Markup
2019-10-30 03:31:37 +00:00
Inline { content: &'a str, node: NodeId },
2019-08-06 07:03:16 +01:00
// Headline
2019-10-30 03:31:37 +00:00
Headline { content: &'a str, node: NodeId },
2019-08-06 07:03:16 +01:00
// Document
2019-10-30 03:31:37 +00:00
Document { content: &'a str, node: NodeId },
2019-08-06 07:03:16 +01:00
}
2019-08-10 13:17:39 +01:00
pub fn parse_container<'a, T: ElementArena<'a>>(
arena: &mut T,
container: Container<'a>,
config: &ParseConfig,
) {
let containers = &mut vec![container];
while let Some(container) = containers.pop() {
match container {
Container::Document { content, node } => {
parse_section_and_headlines(arena, content, node, containers);
}
Container::Headline { content, node } => {
parse_headline_content(arena, content, node, containers, config);
}
Container::Block { content, node } => {
parse_blocks(arena, content, node, containers);
}
Container::Inline { content, node } => {
parse_inlines(arena, content, node, containers);
}
}
}
}
pub fn parse_headline_content<'a, T: ElementArena<'a>>(
arena: &mut T,
2019-08-06 07:03:16 +01:00
content: &'a str,
parent: NodeId,
containers: &mut Vec<Container<'a>>,
config: &ParseConfig,
) {
let (tail, (title, content)) = Title::parse(content, config).unwrap();
2019-10-30 03:31:37 +00:00
let node = arena.append(title, parent);
2019-08-06 07:03:16 +01:00
containers.push(Container::Inline { content, node });
parse_section_and_headlines(arena, tail, parent, containers);
2019-08-06 07:03:16 +01:00
}
pub fn parse_section_and_headlines<'a, T: ElementArena<'a>>(
arena: &mut T,
2019-08-06 07:03:16 +01:00
content: &'a str,
parent: NodeId,
containers: &mut Vec<Container<'a>>,
) {
let content = skip_empty_lines(content);
if content.is_empty() {
return;
}
let mut last_end = 0;
for i in memchr_iter(b'\n', content.as_bytes()).chain(once(content.len())) {
2019-10-02 09:50:36 +01:00
if let Some((mut tail, (headline_content, level))) = parse_headline(&content[last_end..]) {
2019-08-06 07:03:16 +01:00
if last_end != 0 {
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Section, parent);
let content = &content[0..last_end];
containers.push(Container::Block { content, node });
2019-08-06 07:03:16 +01:00
}
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Headline { level }, parent);
2019-08-06 07:03:16 +01:00
containers.push(Container::Headline {
content: headline_content,
node,
});
2019-10-02 09:50:36 +01:00
while let Some((new_tail, (content, level))) = parse_headline(tail) {
debug_assert_ne!(tail, new_tail);
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Headline { level }, parent);
2019-08-06 07:03:16 +01:00
containers.push(Container::Headline { content, node });
tail = new_tail;
}
return;
}
last_end = i + 1;
}
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Section, parent);
2019-08-06 07:03:16 +01:00
containers.push(Container::Block { content, node });
}
pub fn parse_blocks<'a, T: ElementArena<'a>>(
arena: &mut T,
2019-08-06 07:03:16 +01:00
content: &'a str,
parent: NodeId,
containers: &mut Vec<Container<'a>>,
) {
let mut tail = skip_empty_lines(content);
if let Some(new_tail) = parse_block(content, arena, parent, containers) {
2019-08-06 07:03:16 +01:00
tail = skip_empty_lines(new_tail);
}
let mut text = tail;
let mut pos = 0;
while !tail.is_empty() {
let i = memchr(b'\n', tail.as_bytes())
.map(|i| i + 1)
.unwrap_or_else(|| tail.len());
if tail.as_bytes()[0..i].iter().all(u8::is_ascii_whitespace) {
2019-10-28 05:33:18 +00:00
let (tail_, blank) = blank_lines(&tail[i..]);
debug_assert_ne!(tail, tail_);
tail = tail_;
2019-10-30 03:31:37 +00:00
let node = arena.append(
2019-10-28 05:33:18 +00:00
Element::Paragraph {
// including current line (&tail[0..i])
post_blank: blank + 1,
},
parent,
);
containers.push(Container::Inline {
2019-10-28 05:33:18 +00:00
content: &text[0..pos].trim_end(),
node,
});
pos = 0;
2019-08-06 07:03:16 +01:00
text = tail;
} else if let Some(new_tail) = parse_block(tail, arena, parent, containers) {
2019-08-06 07:03:16 +01:00
if pos != 0 {
2019-10-28 05:33:18 +00:00
let node =
arena.insert_before_last_child(Element::Paragraph { post_blank: 0 }, parent);
containers.push(Container::Inline {
2019-10-28 05:33:18 +00:00
content: &text[0..pos].trim_end(),
node,
});
2019-08-06 07:03:16 +01:00
pos = 0;
}
debug_assert_ne!(tail, skip_empty_lines(new_tail));
2019-08-06 07:03:16 +01:00
tail = skip_empty_lines(new_tail);
text = tail;
} else {
debug_assert_ne!(tail, &tail[i..]);
2019-08-06 07:03:16 +01:00
tail = &tail[i..];
pos += i;
}
}
if !text.is_empty() {
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Paragraph { post_blank: 0 }, parent);
containers.push(Container::Inline {
2019-10-28 05:33:18 +00:00
content: &text[0..pos].trim_end(),
node,
});
2019-08-06 07:03:16 +01:00
}
}
pub fn parse_block<'a, T: ElementArena<'a>>(
2019-08-06 07:03:16 +01:00
contents: &'a str,
arena: &mut T,
parent: NodeId,
2019-08-06 07:03:16 +01:00
containers: &mut Vec<Container<'a>>,
) -> Option<&'a str> {
2019-10-28 05:33:18 +00:00
// footnote definitions must be start at column 0
2019-10-02 09:50:36 +01:00
if let Some((tail, (fn_def, content))) = FnDef::parse(contents) {
2019-10-30 03:31:37 +00:00
let node = arena.append(fn_def, parent);
2019-08-06 07:03:16 +01:00
containers.push(Container::Block { content, node });
return Some(tail);
2019-10-28 05:33:18 +00:00
}
2019-10-30 03:31:37 +00:00
if let Some(tail) = parse_list(arena, contents, parent, containers) {
return Some(tail);
2019-08-06 07:03:16 +01:00
}
let contents = contents.trim_start();
2019-08-06 07:03:16 +01:00
match contents.as_bytes().get(0)? {
b'C' => {
2019-10-02 09:50:36 +01:00
let (tail, clock) = Clock::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(clock, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
2019-08-06 07:03:16 +01:00
}
b'\'' => {
// TODO: LaTeX environment
2019-10-02 09:50:36 +01:00
None
2019-08-06 07:03:16 +01:00
}
b'-' => {
2019-10-28 05:33:18 +00:00
let (tail, rule) = Rule::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(rule, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
2019-08-06 07:03:16 +01:00
}
b':' => {
2019-10-02 09:50:36 +01:00
if let Some((tail, (drawer, content))) = Drawer::parse(contents) {
2019-10-30 03:31:37 +00:00
let node = arena.append(drawer, parent);
containers.push(Container::Block { content, node });
2019-10-02 09:50:36 +01:00
Some(tail)
} else {
2019-10-28 05:33:18 +00:00
let (tail, fixed_width) = FixedWidth::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(fixed_width, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
2019-08-06 07:03:16 +01:00
}
}
b'|' => {
2019-10-02 09:50:36 +01:00
let tail = parse_table(arena, contents, containers, parent)?;
Some(tail)
}
b'#' => {
2019-10-28 05:33:18 +00:00
if let Some((tail, (name, args, content, blank))) = parse_block_element(contents) {
match_block(
arena,
parent,
containers,
name.into(),
args.map(Into::into),
content,
2019-10-28 05:33:18 +00:00
blank,
);
2019-10-02 09:50:36 +01:00
Some(tail)
} else if let Some((tail, (dyn_block, content))) = DynBlock::parse(contents) {
2019-10-30 03:31:37 +00:00
let node = arena.append(dyn_block, parent);
containers.push(Container::Block { content, node });
2019-10-02 09:50:36 +01:00
Some(tail)
2019-10-28 05:33:18 +00:00
} else if let Some((tail, (key, optional, value, blank))) = parse_keyword(contents) {
if (&*key).eq_ignore_ascii_case("CALL") {
2019-10-30 03:31:37 +00:00
arena.append(
BabelCall {
value: value.into(),
2019-10-28 05:33:18 +00:00
post_blank: blank,
},
parent,
);
} else {
2019-10-30 03:31:37 +00:00
arena.append(
Keyword {
key: key.into(),
optional: optional.map(Into::into),
value: value.into(),
2019-10-28 05:33:18 +00:00
post_blank: blank,
},
parent,
);
}
2019-10-02 09:50:36 +01:00
Some(tail)
} else {
2019-10-28 05:33:18 +00:00
let (tail, comment) = Comment::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(comment, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
}
2019-10-02 09:50:36 +01:00
_ => None,
2019-08-06 07:03:16 +01:00
}
}
pub fn match_block<'a, T: ElementArena<'a>>(
arena: &mut T,
parent: NodeId,
containers: &mut Vec<Container<'a>>,
name: Cow<'a, str>,
parameters: Option<Cow<'a, str>>,
content: &'a str,
2019-10-28 05:33:18 +00:00
post_blank: usize,
) {
match &*name.to_uppercase() {
"CENTER" => {
2019-10-28 05:33:18 +00:00
let (content, pre_blank) = blank_lines(content);
2019-10-30 03:31:37 +00:00
let node = arena.append(
2019-10-28 05:33:18 +00:00
CenterBlock {
parameters,
pre_blank,
post_blank,
},
parent,
);
containers.push(Container::Block { content, node });
}
"QUOTE" => {
2019-10-28 05:33:18 +00:00
let (content, pre_blank) = blank_lines(content);
2019-10-30 03:31:37 +00:00
let node = arena.append(
2019-10-28 05:33:18 +00:00
QuoteBlock {
parameters,
pre_blank,
post_blank,
},
parent,
);
containers.push(Container::Block { content, node });
}
"VERSE" => {
let (content, pre_blank) = blank_lines(content);
2019-10-30 03:31:37 +00:00
let node = arena.append(
2019-10-28 05:33:18 +00:00
VerseBlock {
parameters,
pre_blank,
post_blank,
},
parent,
);
containers.push(Container::Block { content, node });
}
"COMMENT" => {
2019-10-30 03:31:37 +00:00
arena.append(
CommentBlock {
data: parameters,
contents: content.into(),
2019-10-28 05:33:18 +00:00
post_blank,
},
parent,
);
}
"EXAMPLE" => {
2019-10-30 03:31:37 +00:00
arena.append(
ExampleBlock {
data: parameters,
contents: content.into(),
2019-10-28 05:33:18 +00:00
post_blank,
},
parent,
);
}
"EXPORT" => {
2019-10-30 03:31:37 +00:00
arena.append(
ExportBlock {
data: parameters.unwrap_or_default(),
contents: content.into(),
2019-10-28 05:33:18 +00:00
post_blank,
},
parent,
);
}
"SRC" => {
let (language, arguments) = match &parameters {
Some(Cow::Borrowed(args)) => {
let (language, arguments) =
args.split_at(args.find(' ').unwrap_or_else(|| args.len()));
(language.into(), arguments.into())
2019-08-06 07:03:16 +01:00
}
None => (Cow::Borrowed(""), Cow::Borrowed("")),
_ => unreachable!("`parse_block_element` returns `Some(Cow::Borrowed)` or `None`"),
};
2019-10-30 03:31:37 +00:00
arena.append(
SourceBlock {
arguments,
language,
contents: content.into(),
2019-10-28 05:33:18 +00:00
post_blank,
},
parent,
);
}
_ => {
2019-10-28 05:33:18 +00:00
let (content, pre_blank) = blank_lines(content);
2019-10-30 03:31:37 +00:00
let node = arena.append(
2019-10-28 05:33:18 +00:00
SpecialBlock {
parameters,
name,
pre_blank,
post_blank,
},
parent,
);
2019-08-06 07:03:16 +01:00
containers.push(Container::Block { content, node });
}
}
}
struct InlinePositions<'a> {
bytes: &'a [u8],
pos: usize,
next: Option<usize>,
}
impl InlinePositions<'_> {
fn new(bytes: &[u8]) -> InlinePositions {
InlinePositions {
bytes,
pos: 0,
next: Some(0),
}
}
}
impl Iterator for InlinePositions<'_> {
type Item = usize;
fn next(&mut self) -> Option<Self::Item> {
lazy_static::lazy_static! {
static ref PRE_BYTES: BytesConst =
bytes!(b'@', b'<', b'[', b' ', b'(', b'{', b'\'', b'"', b'\n');
}
self.next.take().or_else(|| {
PRE_BYTES.find(&self.bytes[self.pos..]).map(|i| {
self.pos += i + 1;
match self.bytes[self.pos - 1] {
b'{' => {
self.next = Some(self.pos);
self.pos - 1
}
b' ' | b'(' | b'\'' | b'"' | b'\n' => self.pos,
_ => self.pos - 1,
}
})
})
}
}
pub fn parse_inlines<'a, T: ElementArena<'a>>(
arena: &mut T,
2019-08-06 07:03:16 +01:00
content: &'a str,
parent: NodeId,
containers: &mut Vec<Container<'a>>,
) {
let mut tail = content;
if let Some(tail_) = parse_inline(tail, arena, containers, parent) {
tail = tail_;
2019-08-06 07:03:16 +01:00
}
while let Some((tail_, i)) = InlinePositions::new(tail.as_bytes())
.filter_map(|i| parse_inline(&tail[i..], arena, containers, parent).map(|tail| (tail, i)))
.next()
{
if i != 0 {
2019-08-10 13:17:39 +01:00
arena.insert_before_last_child(
Element::Text {
value: tail[0..i].into(),
2019-08-10 13:17:39 +01:00
},
parent,
);
2019-08-06 07:03:16 +01:00
}
tail = tail_;
2019-08-06 07:03:16 +01:00
}
if !tail.is_empty() {
2019-10-30 03:31:37 +00:00
arena.append(Element::Text { value: tail.into() }, parent);
2019-08-06 07:03:16 +01:00
}
}
pub fn parse_inline<'a, T: ElementArena<'a>>(
2019-08-06 07:03:16 +01:00
contents: &'a str,
arena: &mut T,
2019-08-06 07:03:16 +01:00
containers: &mut Vec<Container<'a>>,
parent: NodeId,
) -> Option<&'a str> {
2019-08-06 07:03:16 +01:00
if contents.len() < 3 {
return None;
}
2019-10-02 09:50:36 +01:00
match contents.as_bytes()[0] {
b'@' => {
2019-10-02 09:50:36 +01:00
let (tail, snippet) = Snippet::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(snippet, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
b'{' => {
2019-10-02 09:50:36 +01:00
let (tail, macros) = Macros::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(macros, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
b'<' => {
2019-10-02 09:50:36 +01:00
if let Some((tail, _content)) = parse_radio_target(contents) {
2019-10-30 03:31:37 +00:00
arena.append(Element::RadioTarget, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
} else if let Some((tail, target)) = Target::parse(contents) {
2019-10-30 03:31:37 +00:00
arena.append(target, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
} else if let Some((tail, timestamp)) = Timestamp::parse_active(contents) {
2019-10-30 03:31:37 +00:00
arena.append(timestamp, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
} else {
let (tail, timestamp) = Timestamp::parse_diary(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(timestamp, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
}
2019-08-06 07:03:16 +01:00
b'[' => {
2019-10-02 09:50:36 +01:00
if let Some((tail, fn_ref)) = FnRef::parse(contents) {
2019-10-30 03:31:37 +00:00
arena.append(fn_ref, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
} else if let Some((tail, link)) = Link::parse(contents) {
2019-10-30 03:31:37 +00:00
arena.append(link, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
} else if let Some((tail, cookie)) = Cookie::parse(contents) {
2019-10-30 03:31:37 +00:00
arena.append(cookie, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
} else {
let (tail, timestamp) = Timestamp::parse_inactive(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(timestamp, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
2019-08-06 07:03:16 +01:00
}
}
b'*' => {
2019-10-02 09:50:36 +01:00
let (tail, content) = parse_emphasis(contents, b'*')?;
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Bold, parent);
2019-10-02 09:50:36 +01:00
containers.push(Container::Inline { content, node });
Some(tail)
2019-08-06 07:03:16 +01:00
}
b'+' => {
2019-10-02 09:50:36 +01:00
let (tail, content) = parse_emphasis(contents, b'+')?;
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Strike, parent);
2019-10-02 09:50:36 +01:00
containers.push(Container::Inline { content, node });
Some(tail)
2019-08-06 07:03:16 +01:00
}
b'/' => {
2019-10-02 09:50:36 +01:00
let (tail, content) = parse_emphasis(contents, b'/')?;
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Italic, parent);
2019-10-02 09:50:36 +01:00
containers.push(Container::Inline { content, node });
Some(tail)
2019-08-06 07:03:16 +01:00
}
b'_' => {
2019-10-02 09:50:36 +01:00
let (tail, content) = parse_emphasis(contents, b'_')?;
2019-10-30 03:31:37 +00:00
let node = arena.append(Element::Underline, parent);
2019-10-02 09:50:36 +01:00
containers.push(Container::Inline { content, node });
Some(tail)
}
b'=' => {
2019-10-02 09:50:36 +01:00
let (tail, value) = parse_emphasis(contents, b'=')?;
let value = value.into();
2019-10-30 03:31:37 +00:00
arena.append(Element::Verbatim { value }, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
b'~' => {
2019-10-02 09:50:36 +01:00
let (tail, value) = parse_emphasis(contents, b'~')?;
let value = value.into();
2019-10-30 03:31:37 +00:00
arena.append(Element::Code { value }, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
b's' => {
2019-10-02 09:50:36 +01:00
let (tail, inline_src) = InlineSrc::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(inline_src, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
}
b'c' => {
2019-10-02 09:50:36 +01:00
let (tail, inline_call) = InlineCall::parse(contents)?;
2019-10-30 03:31:37 +00:00
arena.append(inline_call, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
2019-08-06 07:03:16 +01:00
}
2019-10-02 09:50:36 +01:00
_ => None,
2019-08-06 07:03:16 +01:00
}
}
2019-10-30 03:31:37 +00:00
pub fn parse_list<'a, T: ElementArena<'a>>(
arena: &mut T,
2019-10-30 03:31:37 +00:00
contents: &'a str,
2019-08-06 07:03:16 +01:00
parent: NodeId,
containers: &mut Vec<Container<'a>>,
2019-10-30 03:31:37 +00:00
) -> Option<&'a str> {
let (mut tail, (first_item, content)) = ListItem::parse(contents)?;
let first_item_indent = first_item.indent;
let first_item_ordered = first_item.ordered;
let parent = arena.append(Element::Document { pre_blank: 0 }, parent); // placeholder
let node = arena.append(first_item, parent);
containers.push(Container::Block { content, node });
while let Some((tail_, (item, content))) = ListItem::parse(tail) {
if item.indent == first_item_indent {
let node = arena.append(item, parent);
containers.push(Container::Block { content, node });
debug_assert_ne!(tail, tail_);
tail = tail_;
} else {
break;
}
2019-08-06 07:03:16 +01:00
}
2019-10-30 03:31:37 +00:00
let (tail, blank) = blank_lines(tail);
arena.set(
parent,
List {
indent: first_item_indent,
ordered: first_item_ordered,
post_blank: blank,
},
);
Some(tail)
2019-08-06 07:03:16 +01:00
}
pub fn parse_table<'a, T: ElementArena<'a>>(
arena: &mut T,
2019-08-06 07:03:16 +01:00
contents: &'a str,
containers: &mut Vec<Container<'a>>,
parent: NodeId,
) -> Option<&'a str> {
2019-08-06 07:03:16 +01:00
if contents.trim_start().starts_with('|') {
2019-10-30 03:31:37 +00:00
let table_node = arena.append(Table::Org { tblfm: None }, parent);
2019-08-06 07:03:16 +01:00
let mut last_end = 0;
for start in memchr_iter(b'\n', contents.as_bytes()).chain(once(contents.len())) {
2019-08-06 07:03:16 +01:00
let line = contents[last_end..start].trim();
match TableRow::parse(line) {
Some(TableRow::Standard) => {
2019-10-30 03:31:37 +00:00
let row_node = arena.append(TableRow::Standard, table_node);
2019-08-06 07:03:16 +01:00
for cell in line[1..].split_terminator('|') {
2019-10-30 03:31:37 +00:00
let cell_node = arena.append(Element::TableCell, row_node);
2019-08-06 07:03:16 +01:00
containers.push(Container::Inline {
content: cell.trim(),
node: cell_node,
});
}
}
Some(TableRow::Rule) => {
2019-10-30 03:31:37 +00:00
arena.append(TableRow::Rule, table_node);
2019-08-06 07:03:16 +01:00
}
None => return Some(&contents[last_end..]),
2019-08-06 07:03:16 +01:00
}
last_end = start + 1;
}
Some("")
2019-08-06 07:03:16 +01:00
} else {
2019-10-02 09:50:36 +01:00
let (tail, value) = parse_table_el(contents)?;
let value = value.into();
2019-10-30 03:31:37 +00:00
arena.append(Table::TableEl { value }, parent);
2019-10-02 09:50:36 +01:00
Some(tail)
2019-08-06 07:03:16 +01:00
}
}
pub fn line<'a, E: ParseError<&'a str>>(input: &'a str) -> IResult<&str, &str, E> {
if let Some(i) = memchr(b'\n', input.as_bytes()) {
if i > 0 && input.as_bytes()[i - 1] == b'\r' {
Ok((&input[i + 1..], &input[0..i - 1]))
} else {
Ok((&input[i + 1..], &input[0..i]))
}
} else {
Ok(("", input))
}
}
pub fn eol<'a, E: ParseError<&'a str>>(input: &'a str) -> IResult<&str, &str, E> {
verify(line, |s: &str| {
s.as_bytes().iter().all(|c| c.is_ascii_whitespace())
})(input)
}
2019-10-02 09:50:36 +01:00
pub fn take_lines_while(predicate: impl Fn(&str) -> bool) -> impl Fn(&str) -> (&str, &str) {
move |input| {
let mut last_end = 0;
for i in memchr_iter(b'\n', input.as_bytes()) {
if i > 0 && input.as_bytes()[i - 1] == b'\r' {
if !predicate(&input[last_end..i - 1]) {
2019-10-02 09:50:36 +01:00
return (&input[last_end..], &input[0..last_end]);
}
} else if !predicate(&input[last_end..i]) {
2019-10-02 09:50:36 +01:00
return (&input[last_end..], &input[0..last_end]);
}
last_end = i + 1;
}
if !predicate(&input[last_end..]) {
2019-10-02 09:50:36 +01:00
(&input[last_end..], &input[0..last_end])
} else {
2019-10-02 09:50:36 +01:00
("", input)
}
}
}
pub fn skip_empty_lines(input: &str) -> &str {
take_lines_while(|line| line.as_bytes().iter().all(|c| c.is_ascii_whitespace()))(input).0
}
2019-10-02 09:50:36 +01:00
pub fn parse_headline(input: &str) -> Option<(&str, (&str, usize))> {
2019-08-11 04:40:52 +01:00
let (input_, level) = parse_headline_level(input)?;
let (input_, content) = take_lines_while(move |line| {
2019-10-02 09:50:36 +01:00
parse_headline_level(line)
.map(|(_, l)| l > level)
.unwrap_or(true)
})(input_);
Some((input_, (&input[0..level + content.len()], level)))
}
2019-10-02 09:50:36 +01:00
pub fn parse_headline_level(input: &str) -> Option<(&str, usize)> {
let (input, stars) = take_while1::<_, _, ()>(|c: char| c == '*')(input).ok()?;
if input.starts_with(' ') || input.starts_with('\n') || input.is_empty() {
2019-10-02 09:50:36 +01:00
Some((input, stars.len()))
} else {
2019-10-02 09:50:36 +01:00
None
}
}
pub fn take_one_word<'a, E: ParseError<&'a str>>(input: &'a str) -> IResult<&str, &str, E> {
take_while1(|c: char| !c.is_ascii_whitespace())(input)
}
#[test]
2019-08-10 13:17:39 +01:00
pub fn test_skip_empty_lines() {
assert_eq!(skip_empty_lines("foo"), "foo");
assert_eq!(skip_empty_lines(" foo"), " foo");
assert_eq!(skip_empty_lines(" \nfoo\n"), "foo\n");
assert_eq!(skip_empty_lines(" \n\n\nfoo\n"), "foo\n");
assert_eq!(skip_empty_lines(" \n \n\nfoo\n"), "foo\n");
assert_eq!(skip_empty_lines(" \n \n\n foo\n"), " foo\n");
}
2019-10-28 05:33:18 +00:00
pub fn blank_lines(input: &str) -> (&str, usize) {
let bytes = input.as_bytes();
let mut blank = 0;
let mut last_end = 0;
for i in memchr_iter(b'\n', bytes) {
if bytes[last_end..i].iter().all(u8::is_ascii_whitespace) {
blank += 1;
} else {
break;
}
last_end = 1 + i;
}
(&input[last_end..], blank)
}
#[test]
pub fn test_blank_lines() {
assert_eq!(blank_lines("foo"), ("foo", 0));
assert_eq!(blank_lines(" foo"), (" foo", 0));
assert_eq!(blank_lines(" \t\nfoo\n"), ("foo\n", 1));
assert_eq!(blank_lines("\n \r\n\nfoo\n"), ("foo\n", 3));
assert_eq!(blank_lines("\r\n \n \r\n foo\n"), (" foo\n", 3));
}