talexander/organic
1
0
Fork 0
Code Issues Pull Requests Releases Activity

Fix plain list parser to not consume trailing whitespace on the last item.

Browse Source
This commit is contained in:
Tom Alexander 2023-04-14 19:24:05 -04:00
parent 665c1d70fb
commit 08fed1301e
No known key found for this signature in database
GPG Key ID: D3A179C9A53C0EDE
3 changed files with 85 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
org_mode_samples/exit_matcher_investigation/list_paragraph_nesting.org
View File

@ -6,5 +6,7 @@
lorem
ipsum
ipsum
dolar

99
src/parser/plain_list.rs
View File

@ -23,6 +23,7 @@ use nom::character::complete::one_of;
use nom::character::complete::space0;
use nom::character::complete::space1;
use nom::combinator::eof;
use nom::combinator::peek;
use nom::combinator::recognize;
use nom::combinator::verify;
use nom::multi::many1;
@ -35,26 +36,73 @@ pub fn plain_list<'r, 's>(context: Context<'r, 's>, input: &'s str) -> Res<&'s s
context.with_additional_node(ContextElement::ExitMatcherNode(ExitMatcherNode {
exit_matcher: ChainBehavior::AndParent(Some(&plain_list_end)),
}));
let (mut remaining, first_item) = plain_list_item(&parser_context, input)?;
let first_item_indentation = first_item.indentation;
let plain_list_item_matcher = parser_with_context!(plain_list_item)(&parser_context);
let exit_matcher = parser_with_context!(exit_matcher_parser)(&parser_context);
let without_consume_context =
parser_context.with_additional_node(ContextElement::ConsumeTrailingWhitespace(false));
let with_consume_context =
parser_context.with_additional_node(ContextElement::ConsumeTrailingWhitespace(true));
let without_consume_matcher = parser_with_context!(plain_list_item)(&without_consume_context);
let with_consume_matcher = parser_with_context!(plain_list_item)(&with_consume_context);
let exit_matcher = parser_with_context!(exit_matcher_parser)(&with_consume_context);
let mut children = Vec::new();
children.push(first_item);
loop {
let exit_contents = exit_matcher(remaining);
if exit_contents.is_ok() {
break;
}
let mut first_item_indentation: Option<usize> = None;
let mut remaining = input;
let next_list_item = plain_list_item_matcher(remaining);
match next_list_item {
Ok((remain, next_child)) if next_child.indentation == first_item_indentation => {
children.push(next_child);
loop {
/*
Trailing whitespace belongs to the plain list, not the plain list item
Possible outcomes:
Don't consume, yes exit matcher
Don't consume, no additional item
Consume, additional item
*/
let last_item_then_exit = tuple((without_consume_matcher, exit_matcher))(remaining);
match last_item_then_exit {
Ok((remain, (item, _exit)))
if item.indentation == *first_item_indentation.get_or_insert(item.indentation) =>
{
remaining = remain;
children.push(item);
break;
}
Ok(_) | Err(_) => break,
Ok(_) | Err(_) => {}
};
let not_last_item = tuple((with_consume_matcher, peek(without_consume_matcher)))(remaining);
match not_last_item {
Ok((remain, (item, _future_item)))
if item.indentation == *first_item_indentation.get_or_insert(item.indentation) =>
{
remaining = remain;
children.push(item);
continue;
}
Ok(_) | Err(_) => {}
};
// If its not (don't consume, exit) and its not (consume, see another item) then it must be (don't consume, no additional item)
let last_item_then_exit = without_consume_matcher(remaining);
match last_item_then_exit {
Ok((remain, item))
if item.indentation == *first_item_indentation.get_or_insert(item.indentation) =>
{
remaining = remain;
children.push(item);
break;
}
Ok(_) | Err(_) => {
return Err(nom::Err::Error(CustomError::MyError(MyError(
"Should be unreachable.",
))));
unreachable!();
}
};
}
if children.is_empty() {
return Err(nom::Err::Error(CustomError::MyError(MyError(
"Plain lists require at least one element.",
))));
}
let (remaining, _trailing_ws) =
@ -73,20 +121,27 @@ pub fn plain_list_item<'r, 's>(
let (remaining, leading_whitespace) = space0(input)?;
// It is fine that we get the indent level using the number of bytes rather than the number of characters because nom's space0 only matches space and tab (0x20 and 0x09)
let indent_level = leading_whitespace.len();
let parser_context = context
let with_consume_context = context
.with_additional_node(ContextElement::ConsumeTrailingWhitespace(true))
.with_additional_node(ContextElement::ListItem(indent_level))
.with_additional_node(ContextElement::ExitMatcherNode(ExitMatcherNode {
exit_matcher: ChainBehavior::AndParent(Some(&plain_list_item_end)),
}));
let without_consume_context = context
.with_additional_node(ContextElement::ConsumeTrailingWhitespace(false))
.with_additional_node(ContextElement::ListItem(indent_level))
.with_additional_node(ContextElement::ExitMatcherNode(ExitMatcherNode {
exit_matcher: ChainBehavior::AndParent(Some(&plain_list_item_end)),
}));
let element_matcher = parser_with_context!(element)(&parser_context);
let exit_matcher = parser_with_context!(exit_matcher_parser)(&parser_context);
let element_matcher = parser_with_context!(element)(&with_consume_context);
let exit_matcher = parser_with_context!(exit_matcher_parser)(&with_consume_context);
let (remaining, bull) =
verify(bullet, |bull: &str| bull != "*" || indent_level > 0)(remaining)?;
let maybe_contentless_item: Res<&str, &str> = alt((eof, line_ending))(remaining);
match maybe_contentless_item {
Ok((rem, _ws)) => {
// TODO: do we need to consume if this isn't the last item?
let source = get_consumed(input, rem);
return Ok((
rem,
@ -329,15 +384,17 @@ baz"#;
lorem
ipsum
ipsum"#;
dolar"#;
let initial_context: ContextTree<'_, '_> = ContextTree::new();
let document_context =
initial_context.with_additional_node(ContextElement::DocumentRoot(input));
let plain_list_matcher = parser_with_context!(plain_list)(&document_context);
let (remaining, result) =
plain_list_matcher(input).expect("Should parse the plain list successfully.");
assert_eq!(remaining, "ipsum");
assert_eq!(remaining, "dolar");
assert_eq!(
result.get_source(),
r#"1. foo
@ -348,6 +405,8 @@ ipsum"#;
lorem
ipsum
"#
);

4
toy_language.txt
View File

@ -6,5 +6,7 @@
lorem
ipsum
ipsum
dolar