feat: update list node parsing
This commit is contained in:
PoiScript
parent
ed987d468a
commit
b7ddc0f076
10 changed files with 411 additions and 242 deletions
|
|
@ -17,6 +17,7 @@ use super::{
|
|||
input::Input,
|
||||
keyword::affiliated_keyword_nodes,
|
||||
object::object_nodes,
|
||||
paragraph::paragraph_nodes,
|
||||
SyntaxKind::*,
|
||||
};
|
||||
|
||||
|
|
@ -28,24 +29,35 @@ pub fn list_node(input: Input) -> IResult<Input, GreenElement, ()> {
|
|||
fn list_node_base(input: Input) -> IResult<Input, GreenElement, ()> {
|
||||
let (input, affiliated_keywords) = affiliated_keyword_nodes(input)?;
|
||||
let (input, first_indent) = space0(input)?;
|
||||
let (input, first_item) = list_item_node(first_indent, input)?;
|
||||
let (input, (ends_with_empty_blank_lines, first_item)) = list_item_node(first_indent, input)?;
|
||||
|
||||
let mut children = vec![];
|
||||
children.extend(affiliated_keywords);
|
||||
children.push(first_item);
|
||||
|
||||
let mut input = input;
|
||||
while !input.is_empty() {
|
||||
while !ends_with_empty_blank_lines && !input.is_empty() {
|
||||
let (input_, indent) = space0(input)?;
|
||||
|
||||
if indent.input_len() != first_indent.input_len() {
|
||||
break;
|
||||
}
|
||||
|
||||
if let Ok((input_, list_item)) = list_item_node(indent, input_) {
|
||||
children.push(list_item);
|
||||
input = input_;
|
||||
} else {
|
||||
let Ok((input_, (ends_with_empty_blank_lines, list_item))) = list_item_node(indent, input_)
|
||||
else {
|
||||
break;
|
||||
};
|
||||
|
||||
children.push(list_item);
|
||||
debug_assert!(
|
||||
input.input_len() > input_.input_len(),
|
||||
"{} > {}",
|
||||
input.input_len(),
|
||||
input_.input_len(),
|
||||
);
|
||||
input = input_;
|
||||
|
||||
if ends_with_empty_blank_lines {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
|
@ -57,7 +69,11 @@ fn list_node_base(input: Input) -> IResult<Input, GreenElement, ()> {
|
|||
Ok((input, node(LIST, children)))
|
||||
}
|
||||
|
||||
fn list_item_node<'a>(indent: Input<'a>, input: Input<'a>) -> IResult<Input<'a>, GreenElement, ()> {
|
||||
#[tracing::instrument(level = "debug", skip(input, indent), fields(input = input.s))]
|
||||
fn list_item_node<'a>(
|
||||
indent: Input<'a>,
|
||||
input: Input<'a>,
|
||||
) -> IResult<Input<'a>, (bool, GreenElement), ()> {
|
||||
let (input, bullet) = recognize(tuple((
|
||||
alt((
|
||||
tag("+"),
|
||||
|
|
@ -91,7 +107,9 @@ fn list_item_node<'a>(indent: Input<'a>, input: Input<'a>) -> IResult<Input<'a>,
|
|||
let (input, counter) = opt(list_item_counter)(input)?;
|
||||
let (input, checkbox) = opt(list_item_checkbox)(input)?;
|
||||
let (input, tag) = cond(!is_ordered, opt(list_item_tag))(input)?;
|
||||
let (input, content) = list_item_content_node(input, indent.input_len())?;
|
||||
let (input, (ends_with_empty_blank_lines, content)) =
|
||||
list_item_content_node(input, indent.input_len())?;
|
||||
let (input, post_blank) = cond(!ends_with_empty_blank_lines, blank_lines)(input)?;
|
||||
|
||||
let mut children = vec![
|
||||
indent.token(LIST_ITEM_INDENT),
|
||||
|
|
@ -109,10 +127,17 @@ fn list_item_node<'a>(indent: Input<'a>, input: Input<'a>) -> IResult<Input<'a>,
|
|||
}
|
||||
|
||||
children.push(content);
|
||||
if let Some(post_blank) = post_blank {
|
||||
children.extend(post_blank);
|
||||
}
|
||||
|
||||
Ok((input, node(LIST_ITEM, children)))
|
||||
Ok((
|
||||
input,
|
||||
(ends_with_empty_blank_lines, node(LIST_ITEM, children)),
|
||||
))
|
||||
}
|
||||
|
||||
#[tracing::instrument(level = "debug", skip(input), fields(input = input.s))]
|
||||
fn list_item_counter(input: Input) -> IResult<Input, (GreenElement, Input), ()> {
|
||||
let (input, node) = map(
|
||||
tuple((l_bracket_token, at_token, alphanumeric1, r_bracket_token)),
|
||||
|
|
@ -129,6 +154,7 @@ fn list_item_counter(input: Input) -> IResult<Input, (GreenElement, Input), ()>
|
|||
Ok((input, (node, ws)))
|
||||
}
|
||||
|
||||
#[tracing::instrument(level = "debug", skip(input), fields(input = input.s))]
|
||||
fn list_item_checkbox(input: Input) -> IResult<Input, (GreenElement, Input), ()> {
|
||||
let (input, node) = map(
|
||||
tuple((
|
||||
|
|
@ -151,6 +177,7 @@ fn list_item_checkbox(input: Input) -> IResult<Input, (GreenElement, Input), ()>
|
|||
Ok((input, (node, ws)))
|
||||
}
|
||||
|
||||
#[tracing::instrument(level = "debug", skip(input), fields(input = input.s))]
|
||||
fn list_item_tag(input: Input) -> IResult<Input, (GreenElement, Input), ()> {
|
||||
let bytes = input.as_bytes();
|
||||
|
||||
|
|
@ -167,19 +194,23 @@ fn list_item_tag(input: Input) -> IResult<Input, (GreenElement, Input), ()> {
|
|||
Ok((input, (node(LIST_ITEM_TAG, children), ws)))
|
||||
}
|
||||
|
||||
fn list_item_content_node(input: Input, indent: usize) -> IResult<Input, GreenElement, ()> {
|
||||
#[tracing::instrument(level = "debug", skip(input), fields(input = input.s))]
|
||||
fn list_item_content_node(input: Input, indent: usize) -> IResult<Input, (bool, GreenElement), ()> {
|
||||
if memchr(b'\n', input.as_bytes()).is_none() {
|
||||
return Ok((
|
||||
input.of(""),
|
||||
node(LIST_ITEM_CONTENT, [node(PARAGRAPH, object_nodes(input))]),
|
||||
(
|
||||
false,
|
||||
node(LIST_ITEM_CONTENT, [node(PARAGRAPH, object_nodes(input))]),
|
||||
),
|
||||
));
|
||||
};
|
||||
|
||||
let mut skip_one = true;
|
||||
let mut i = input;
|
||||
let mut children = vec![];
|
||||
let mut previous_line_is_blank = false;
|
||||
'l: loop {
|
||||
let mut previous_blank_line: Option<(Input, Input)> = None;
|
||||
'l: while !i.is_empty() {
|
||||
for (input, head) in line_starts_iter(i.as_str())
|
||||
// the first line in list item content will always be a paragraph
|
||||
// so we need to skip it in the first iteration
|
||||
|
|
@ -188,50 +219,51 @@ fn list_item_content_node(input: Input, indent: usize) -> IResult<Input, GreenEl
|
|||
{
|
||||
match get_line_indent(input.as_str()) {
|
||||
Some(next_indent) => {
|
||||
previous_line_is_blank = false;
|
||||
|
||||
if next_indent <= indent {
|
||||
let (input, head) = previous_blank_line.unwrap_or_else(|| (input, head));
|
||||
if !head.is_empty() {
|
||||
children.push(node(PARAGRAPH, object_nodes(head)));
|
||||
children.extend(paragraph_nodes(head)?);
|
||||
}
|
||||
return Ok((input, node(LIST_ITEM_CONTENT, children)));
|
||||
return Ok((input, (false, node(LIST_ITEM_CONTENT, children))));
|
||||
}
|
||||
|
||||
previous_blank_line = None;
|
||||
|
||||
if let Ok((input, element)) = element_node(input) {
|
||||
if !head.is_empty() {
|
||||
children.push(node(PARAGRAPH, object_nodes(head)));
|
||||
children.extend(paragraph_nodes(head)?);
|
||||
}
|
||||
children.push(element);
|
||||
debug_assert!(
|
||||
input.input_len() < i.input_len(),
|
||||
"{} < {}",
|
||||
input.input_len(),
|
||||
i.input_len()
|
||||
);
|
||||
i = input;
|
||||
skip_one = false;
|
||||
continue 'l;
|
||||
}
|
||||
}
|
||||
_ if previous_line_is_blank => {
|
||||
// list item ends at two consecutive empty lines
|
||||
if !head.is_empty() {
|
||||
children.push(node(PARAGRAPH, object_nodes(head)));
|
||||
}
|
||||
let (input, post_blank) = blank_lines(input)?;
|
||||
|
||||
children.extend(post_blank);
|
||||
|
||||
return Ok((input, node(LIST_ITEM_CONTENT, children)));
|
||||
}
|
||||
_ => {
|
||||
previous_line_is_blank = true;
|
||||
// list item ends at two consecutive empty lines
|
||||
if let Some((input, head)) = previous_blank_line {
|
||||
if !head.is_empty() {
|
||||
children.extend(paragraph_nodes(head)?);
|
||||
}
|
||||
|
||||
return Ok((input, (true, node(LIST_ITEM_CONTENT, children))));
|
||||
} else {
|
||||
previous_blank_line = Some((input, head))
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
children.extend(paragraph_nodes(i)?);
|
||||
break;
|
||||
}
|
||||
|
||||
if !i.is_empty() {
|
||||
children.push(node(PARAGRAPH, object_nodes(i)));
|
||||
}
|
||||
|
||||
Ok((input.of(""), node(LIST_ITEM_CONTENT, children)))
|
||||
Ok((input.of(""), (false, node(LIST_ITEM_CONTENT, children))))
|
||||
}
|
||||
|
||||
fn get_line_indent(input: &str) -> Option<usize> {
|
||||
|
|
@ -247,9 +279,8 @@ fn parse() {
|
|||
|
||||
let to_list = to_ast::<List>(list_node);
|
||||
|
||||
let list = to_list("1)");
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("1)").syntax,
|
||||
@r###"
|
||||
LIST@0..2
|
||||
LIST_ITEM@0..2
|
||||
|
|
@ -260,9 +291,8 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list("+ ");
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("+ ").syntax,
|
||||
@r###"
|
||||
LIST@0..2
|
||||
LIST_ITEM@0..2
|
||||
|
|
@ -273,9 +303,8 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list("-\n");
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("-\n").syntax,
|
||||
@r###"
|
||||
LIST@0..2
|
||||
LIST_ITEM@0..2
|
||||
|
|
@ -283,14 +312,12 @@ fn parse() {
|
|||
LIST_ITEM_BULLET@0..1 "-"
|
||||
LIST_ITEM_CONTENT@1..2
|
||||
PARAGRAPH@1..2
|
||||
TEXT@1..2 "\n"
|
||||
BLANK_LINE@1..2 "\n"
|
||||
"###
|
||||
);
|
||||
|
||||
let list = to_list("+ 1");
|
||||
assert!(!list.is_ordered());
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("+ 1").syntax,
|
||||
@r###"
|
||||
LIST@0..3
|
||||
LIST_ITEM@0..3
|
||||
|
|
@ -302,9 +329,8 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list("+ 1\n");
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("+ 1\n").syntax,
|
||||
@r###"
|
||||
LIST@0..4
|
||||
LIST_ITEM@0..4
|
||||
|
|
@ -316,12 +342,13 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list("+ [@A] 1\n\n\n+ 2");
|
||||
// list ends with two consecutive blank lines, and these blank lines
|
||||
// will be the post_blank of list node
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("+ [@A] 1\n\n\n+ 2").syntax,
|
||||
@r###"
|
||||
LIST@0..14
|
||||
LIST_ITEM@0..11
|
||||
LIST@0..11
|
||||
LIST_ITEM@0..9
|
||||
LIST_ITEM_INDENT@0..0 ""
|
||||
LIST_ITEM_BULLET@0..2 "+ "
|
||||
LIST_ITEM_COUNTER@2..6
|
||||
|
|
@ -330,25 +357,21 @@ fn parse() {
|
|||
TEXT@4..5 "A"
|
||||
R_BRACKET@5..6 "]"
|
||||
WHITESPACE@6..7 " "
|
||||
LIST_ITEM_CONTENT@7..11
|
||||
PARAGRAPH@7..10
|
||||
TEXT@7..10 "1\n\n"
|
||||
BLANK_LINE@10..11 "\n"
|
||||
LIST_ITEM@11..14
|
||||
LIST_ITEM_INDENT@11..11 ""
|
||||
LIST_ITEM_BULLET@11..13 "+ "
|
||||
LIST_ITEM_CONTENT@13..14
|
||||
PARAGRAPH@13..14
|
||||
TEXT@13..14 "2"
|
||||
LIST_ITEM_CONTENT@7..9
|
||||
PARAGRAPH@7..9
|
||||
TEXT@7..9 "1\n"
|
||||
BLANK_LINE@9..10 "\n"
|
||||
BLANK_LINE@10..11 "\n"
|
||||
"###
|
||||
);
|
||||
|
||||
let list = to_list("+ *TAG* :: item1\n+ [X] item2");
|
||||
// empty line between list item, the empty line will be
|
||||
// the post_blank of first item
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("+ *TAG* :: item1\n\n+ [X] item2").syntax,
|
||||
@r###"
|
||||
LIST@0..28
|
||||
LIST_ITEM@0..17
|
||||
LIST@0..29
|
||||
LIST_ITEM@0..18
|
||||
LIST_ITEM_INDENT@0..0 ""
|
||||
LIST_ITEM_BULLET@0..2 "+ "
|
||||
LIST_ITEM_TAG@2..10
|
||||
|
|
@ -362,20 +385,22 @@ fn parse() {
|
|||
LIST_ITEM_CONTENT@10..17
|
||||
PARAGRAPH@10..17
|
||||
TEXT@10..17 " item1\n"
|
||||
LIST_ITEM@17..28
|
||||
LIST_ITEM_INDENT@17..17 ""
|
||||
LIST_ITEM_BULLET@17..19 "+ "
|
||||
LIST_ITEM_CHECK_BOX@19..22
|
||||
L_BRACKET@19..20 "["
|
||||
TEXT@20..21 "X"
|
||||
R_BRACKET@21..22 "]"
|
||||
WHITESPACE@22..23 " "
|
||||
LIST_ITEM_CONTENT@23..28
|
||||
PARAGRAPH@23..28
|
||||
TEXT@23..28 "item2"
|
||||
BLANK_LINE@17..18 "\n"
|
||||
LIST_ITEM@18..29
|
||||
LIST_ITEM_INDENT@18..18 ""
|
||||
LIST_ITEM_BULLET@18..20 "+ "
|
||||
LIST_ITEM_CHECK_BOX@20..23
|
||||
L_BRACKET@20..21 "["
|
||||
TEXT@21..22 "X"
|
||||
R_BRACKET@22..23 "]"
|
||||
WHITESPACE@23..24 " "
|
||||
LIST_ITEM_CONTENT@24..29
|
||||
PARAGRAPH@24..29
|
||||
TEXT@24..29 "item2"
|
||||
"###
|
||||
);
|
||||
|
||||
// nested list
|
||||
let list = to_list(
|
||||
r#"+ item1
|
||||
+ item2"#,
|
||||
|
|
@ -400,9 +425,8 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list("* item1\nitem2");
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("* item1\nitem2").syntax,
|
||||
@r###"
|
||||
LIST@0..8
|
||||
LIST_ITEM@0..8
|
||||
|
|
@ -414,13 +438,8 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list(
|
||||
r#"* item1
|
||||
|
||||
still item 1"#,
|
||||
);
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list("* item1\n\n still item 1").syntax,
|
||||
@r###"
|
||||
LIST@0..23
|
||||
LIST_ITEM@0..23
|
||||
|
|
@ -428,7 +447,8 @@ fn parse() {
|
|||
LIST_ITEM_BULLET@0..2 "* "
|
||||
LIST_ITEM_CONTENT@2..23
|
||||
PARAGRAPH@2..9
|
||||
TEXT@2..9 "item1\n\n"
|
||||
TEXT@2..8 "item1\n"
|
||||
BLANK_LINE@8..9 "\n"
|
||||
PARAGRAPH@9..23
|
||||
TEXT@9..23 " still item 1"
|
||||
"###
|
||||
|
|
@ -455,7 +475,8 @@ fn parse() {
|
|||
LIST_ITEM_BULLET@14..16 "+ "
|
||||
LIST_ITEM_CONTENT@16..26
|
||||
PARAGRAPH@16..26
|
||||
TEXT@16..26 "item2\n "
|
||||
TEXT@16..22 "item2\n"
|
||||
BLANK_LINE@22..26 " "
|
||||
"###
|
||||
);
|
||||
|
||||
|
|
@ -476,14 +497,16 @@ fn parse() {
|
|||
LIST_ITEM_BULLET@0..3 "1. "
|
||||
LIST_ITEM_CONTENT@3..23
|
||||
PARAGRAPH@3..10
|
||||
TEXT@3..10 "item1\n\n"
|
||||
TEXT@3..9 "item1\n"
|
||||
BLANK_LINE@9..10 "\n"
|
||||
LIST@10..23
|
||||
LIST_ITEM@10..23
|
||||
LIST_ITEM_INDENT@10..14 " "
|
||||
LIST_ITEM_BULLET@14..16 "- "
|
||||
LIST_ITEM_CONTENT@16..23
|
||||
PARAGRAPH@16..23
|
||||
TEXT@16..23 "item2\n\n"
|
||||
LIST_ITEM_CONTENT@16..22
|
||||
PARAGRAPH@16..22
|
||||
TEXT@16..22 "item2\n"
|
||||
BLANK_LINE@22..23 "\n"
|
||||
LIST_ITEM@23..32
|
||||
LIST_ITEM_INDENT@23..23 ""
|
||||
LIST_ITEM_BULLET@23..26 "3. "
|
||||
|
|
@ -493,21 +516,18 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list(
|
||||
r#" + item1
|
||||
|
||||
+ item2"#,
|
||||
);
|
||||
// nested list
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list(" + item1\n\n + item2").syntax,
|
||||
@r###"
|
||||
LIST@0..20
|
||||
LIST_ITEM@0..11
|
||||
LIST_ITEM_INDENT@0..2 " "
|
||||
LIST_ITEM_BULLET@2..4 "+ "
|
||||
LIST_ITEM_CONTENT@4..11
|
||||
PARAGRAPH@4..11
|
||||
TEXT@4..11 "item1\n\n"
|
||||
LIST_ITEM_CONTENT@4..10
|
||||
PARAGRAPH@4..10
|
||||
TEXT@4..10 "item1\n"
|
||||
BLANK_LINE@10..11 "\n"
|
||||
LIST_ITEM@11..20
|
||||
LIST_ITEM_INDENT@11..13 " "
|
||||
LIST_ITEM_BULLET@13..15 "+ "
|
||||
|
|
@ -517,14 +537,8 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list(
|
||||
r#" 1. item1
|
||||
2. item2
|
||||
3. item3"#,
|
||||
);
|
||||
assert!(list.is_ordered());
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list(" 1. item1\n 2. item2\n 3. item3").syntax,
|
||||
@r###"
|
||||
LIST@0..42
|
||||
LIST_ITEM@0..42
|
||||
|
|
@ -550,15 +564,9 @@ fn parse() {
|
|||
"###
|
||||
);
|
||||
|
||||
let list = to_list(
|
||||
r#" 1. item1
|
||||
#+begin_example
|
||||
hello
|
||||
#+end_example
|
||||
"#,
|
||||
);
|
||||
// Indentation of lines within other greater elements do not count
|
||||
insta::assert_debug_snapshot!(
|
||||
list.syntax,
|
||||
to_list(" 1. item1\n #+begin_example\nhello\n#+end_example\n").syntax,
|
||||
@r###"
|
||||
LIST@0..51
|
||||
LIST_ITEM@0..51
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue