Require either eof or whitespace to line ending for valueless items.

We are properly handling this currently, but it is good to have more test coverage.

org_mode_samples/greater_element/plain_list/description_list_empty_value.org

@@ -0,0 +1,6 @@
+- foo ::
+
+- bar ::
+
+
+baz

org_mode_samples/greater_element/plain_list/mixed_types.org

@@ -0,0 +1,3 @@
+1. foo
+- bar
+- lorem :: ipsum

org_mode_samples/greater_element/plain_list/ordered_list_with_fake_tag.org

@@ -0,0 +1,2 @@
+# Since this is an ordered list, the text before the " :: " is NOT parsed as a tag.
+1. foo :: bar

org_mode_samples/object/plain_link/with_parenthesis.org

@@ -1 +1,11 @@
+# Should be a link:
 https://en.wikipedia.org/wiki/Shebang_(Unix)
+
+# No closing parenthesis, so link ends at underscore.
+https://en.wikipedia.org/wiki/Shebang_(Unix
+
+# Parenthesis only allowed to depth of 2 so link ends at underscore.
+https://en.wikipedia.org/wiki/Shebang_(((Unix)))
+
+# Even though they eventually become balanced, we hit negative parenthesis depth so link ends at )
+https://en.wikipedia.org/wiki/Shebang)Unix(

org_mode_samples/object/regular_link/with_parenthesis.org

@@ -0,0 +1 @@
+[[https://en.wikipedia.org/wiki/Shebang_(Unix)]]

org_mode_samples/object/subscript_and_superscript/with_parenthesis.org

@@ -0,0 +1,13 @@
+foo_(bar)
+
+foo_(b(ar)
+
+foo_(b{ar)
+
+foo_{b(ar}
+
+foo_(b(a)r)
+
+foo_b(a)r
+
+foo_(b+ar)

src/compare/diff.rs

@@ -732,6 +732,10 @@ fn compare_plain_list<'s>(
         Ok(_) => {}
     };
 
+    // TODO compare :type
+    //
+    // :type is an unquoted atom of either descriptive, ordered, or unordered
+
     for (emacs_child, rust_child) in children.iter().skip(2).zip(rust.children.iter()) {
         child_status.push(compare_plain_list_item(source, emacs_child, rust_child)?);
     }

src/parser/plain_link.rs

@@ -5,17 +5,24 @@ use nom::character::complete::anychar;
 use nom::character::complete::none_of;
 use nom::character::complete::one_of;
 use nom::combinator::eof;
+use nom::combinator::not;
 use nom::combinator::peek;
 use nom::combinator::recognize;
 use nom::combinator::verify;
+use nom::multi::many0;
+use nom::multi::many1;
 use nom::multi::many_till;
+use nom::sequence::tuple;
 
+use super::org_source::BracketDepth;
 use super::org_source::OrgSource;
 use super::util::maybe_consume_object_trailing_whitespace_if_not_exiting;
 use crate::context::parser_with_context;
 use crate::context::ContextElement;
+use crate::context::ContextMatcher;
 use crate::context::ExitClass;
 use crate::context::ExitMatcherNode;
+use crate::context::Matcher;
 use crate::context::RefContext;
 use crate::error::CustomError;
 use crate::error::MyError;
@@ -130,17 +137,77 @@ fn path_plain<'b, 'g, 'r, 's>(
     context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    // TODO: "optionally containing parenthesis-wrapped non-whitespace non-bracket substrings up to a depth of two. The string must end with either a non-punctation non-whitespace character, a forwards slash, or a parenthesis-wrapped substring"
+    let path_plain_end = path_plain_end(input.get_parenthesis_depth());
     let parser_context = ContextElement::ExitMatcherNode(ExitMatcherNode {
         class: ExitClass::Gamma,
         exit_matcher: &path_plain_end,
     });
     let parser_context = context.with_additional_node(&parser_context);
 
-    let exit_matcher = parser_with_context!(exit_matcher_parser)(&parser_context);
+    let (remaining, _components) = many1(alt((
+        parser_with_context!(path_plain_no_parenthesis)(&parser_context),
+        parser_with_context!(path_plain_parenthesis)(&parser_context),
+    )))(input)?;
 
+    let source = get_consumed(input, remaining);
+    Ok((remaining, source))
+}
+
+fn path_plain_end(starting_parenthesis_depth: BracketDepth) -> impl ContextMatcher {
+    move |context, input: OrgSource<'_>| _path_plain_end(context, input, starting_parenthesis_depth)
+}
+
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+fn _path_plain_end<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+    starting_parenthesis_depth: BracketDepth,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    let (remaining, _leading_punctuation) = many0(verify(anychar, |c| {
+        !" \t\r\n[]<>()/".contains(*c) && c.is_ascii_punctuation()
+    }))(input)?;
+
+    let disallowed_character = recognize(one_of(" \t\r\n[]<>"))(remaining);
+    if disallowed_character.is_ok() {
+        return disallowed_character;
+    }
+
+    let current_depth = remaining.get_parenthesis_depth() - starting_parenthesis_depth;
+    if current_depth == 0 {
+        let close_parenthesis =
+            tag::<&str, OrgSource<'_>, CustomError<OrgSource<'_>>>(")")(remaining);
+        if close_parenthesis.is_ok() {
+            return close_parenthesis;
+        }
+
+        let open_parenthesis_without_match = recognize(tuple((
+            peek(tag("(")),
+            not(parser_with_context!(path_plain_parenthesis)(context)),
+        )))(remaining);
+        if open_parenthesis_without_match.is_ok() {
+            return open_parenthesis_without_match;
+        }
+    }
+
+    // many0 punctuation
+    Err(nom::Err::Error(CustomError::MyError(MyError(
+        "No path plain end".into(),
+    ))))
+}
+
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+fn path_plain_no_parenthesis<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
     let (remaining, path) = recognize(verify(
-        many_till(anychar, peek(exit_matcher)),
+        many_till(
+            anychar,
+            alt((
+                peek(path_plain_no_parenthesis_disallowed_character),
+                parser_with_context!(exit_matcher_parser)(context),
+            )),
+        ),
         |(children, _exit_contents)| !children.is_empty(),
     ))(input)?;
 
@@ -148,14 +215,65 @@ fn path_plain<'b, 'g, 'r, 's>(
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
-fn path_plain_end<'b, 'g, 'r, 's>(
+fn path_plain_no_parenthesis_disallowed_character<'s>(
-    _context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    recognize(many_till(
+    recognize(verify(anychar, |c| {
-        verify(anychar, |c| {
+        c.is_whitespace() || "()[]<>".contains(*c)
-            *c != '/' && (c.is_ascii_punctuation() || c.is_whitespace())
+    }))(input)
-        }),
+}
-        one_of(" \t\r\n()[]<>"),
+
-    ))(input)
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+fn path_plain_parenthesis<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    let (remaining, _opening) = tag("(")(input)?;
+    let starting_depth = remaining.get_parenthesis_depth();
+
+    let (remaining, _path) = recognize(verify(
+        many_till(
+            anychar,
+            alt((
+                peek(path_plain_parenthesis_end(starting_depth)),
+                parser_with_context!(exit_matcher_parser)(context),
+            )),
+        ),
+        |(children, _exit_contents)| !children.is_empty(),
+    ))(remaining)?;
+    let (remaining, _opening) = tag(")")(remaining)?;
+    let source = get_consumed(input, remaining);
+
+    Ok((remaining, source))
+}
+
+fn path_plain_parenthesis_end(starting_parenthesis_depth: BracketDepth) -> impl Matcher {
+    move |input: OrgSource<'_>| _path_plain_parenthesis_end(input, starting_parenthesis_depth)
+}
+
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+fn _path_plain_parenthesis_end<'s>(
+    input: OrgSource<'s>,
+    starting_parenthesis_depth: BracketDepth,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    let current_depth = input.get_parenthesis_depth() - starting_parenthesis_depth;
+    if current_depth < 0 {
+        // This shouldn't be possible because if depth is 0 then a closing parenthesis should end the link.
+        unreachable!("Exceeded plain link parenthesis depth.")
+    }
+    if current_depth == 0 {
+        let close_parenthesis = tag::<&str, OrgSource<'_>, CustomError<OrgSource<'_>>>(")")(input);
+        if close_parenthesis.is_ok() {
+            return close_parenthesis;
+        }
+    }
+    if current_depth == 1 {
+        let open_parenthesis = tag::<&str, OrgSource<'_>, CustomError<OrgSource<'_>>>("(")(input);
+        if open_parenthesis.is_ok() {
+            return open_parenthesis;
+        }
+    }
+    Err(nom::Err::Error(CustomError::MyError(MyError(
+        "No closing parenthesis".into(),
+    ))))
 }

src/parser/plain_list.rs

@@ -57,7 +57,7 @@ pub(crate) fn detect_plain_list<'b, 'g, 'r, 's>(
             parser_with_context!(bullet)(context),
             alt((space1, line_ending, eof)),
         )),
-        |(_start, indent, bull, _after_whitespace)| {
+        |(_start, indent, (_bullet_type, bull), _after_whitespace)| {
             Into::<&str>::into(bull) != "*" || indent.len() > 0
         },
     )(input)
@@ -151,9 +151,9 @@ fn plain_list_item<'b, 'g, 'r, 's>(
 ) -> Res<OrgSource<'s>, PlainListItem<'s>> {
     start_of_line(input)?;
     let (remaining, (indent_level, _leading_whitespace)) = indentation_level(context, input)?;
-    let (remaining, bull) = verify(
+    let (remaining, (bullet_type, bull)) = verify(
         parser_with_context!(bullet)(context),
-        |bull: &OrgSource<'_>| Into::<&str>::into(bull) != "*" || indent_level > 0,
+        |(_bullet_type, bull)| Into::<&str>::into(bull) != "*" || indent_level > 0,
     )(remaining)?;
 
     let (remaining, _maybe_counter_set) = opt(tuple((
@@ -165,15 +165,33 @@ fn plain_list_item<'b, 'g, 'r, 's>(
 
     let (remaining, maybe_checkbox) = opt(tuple((space1, item_checkbox)))(remaining)?;
 
-    let (remaining, maybe_tag) =
+    let (remaining, maybe_tag) = if let BulletType::Unordered = bullet_type {
-        opt(tuple((space1, parser_with_context!(item_tag)(context))))(remaining)?;
+        opt(tuple((space1, parser_with_context!(item_tag)(context))))(remaining)?
+    } else {
+        (remaining, None)
+    };
+
+    let exit_matcher = plain_list_item_end(indent_level);
+    let contexts = [
+        ContextElement::ConsumeTrailingWhitespace(true),
+        ContextElement::ExitMatcherNode(ExitMatcherNode {
+            class: ExitClass::Beta,
+            exit_matcher: &exit_matcher,
+        }),
+    ];
+    let parser_context = context.with_additional_node(&contexts[0]);
+    let parser_context = parser_context.with_additional_node(&contexts[1]);
 
     let maybe_contentless_item: Res<OrgSource<'_>, ()> = peek(parser_with_context!(
         detect_contentless_item_contents
-    )(context))(remaining);
+    )(&parser_context))(remaining);
     match maybe_contentless_item {
         Ok((_rem, _ws)) => {
-            let (remaining, _trailing_ws) = opt(blank_line)(remaining)?;
+            let (remaining, _trailing_ws) = if context.should_consume_trailing_whitespace() {
+                recognize(alt((recognize(many1(blank_line)), eof)))(remaining)?
+            } else {
+                recognize(alt((blank_line, eof)))(remaining)?
+            };
             let source = get_consumed(input, remaining);
             return Ok((
                 remaining,
@@ -191,17 +209,7 @@ fn plain_list_item<'b, 'g, 'r, 's>(
         }
         Err(_) => {}
     };
-    let (remaining, _ws) = item_tag_post_gap(context, remaining)?;
+    let (remaining, _ws) = item_tag_post_gap(&parser_context, remaining)?;
-    let exit_matcher = plain_list_item_end(indent_level);
-    let contexts = [
-        ContextElement::ConsumeTrailingWhitespace(true),
-        ContextElement::ExitMatcherNode(ExitMatcherNode {
-            class: ExitClass::Beta,
-            exit_matcher: &exit_matcher,
-        }),
-    ];
-    let parser_context = context.with_additional_node(&contexts[0]);
-    let parser_context = parser_context.with_additional_node(&contexts[1]);
 
     let (mut remaining, (mut children, _exit_contents)) = many_till(
         include_input(parser_with_context!(element(true))(&parser_context)),
@@ -241,19 +249,28 @@ fn plain_list_item<'b, 'g, 'r, 's>(
     ));
 }
 
+#[derive(Debug)]
+enum BulletType {
+    Ordered,
+    Unordered,
+}
+
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
 fn bullet<'b, 'g, 'r, 's>(
     context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, OrgSource<'s>> {
+) -> Res<OrgSource<'s>, (BulletType, OrgSource<'s>)> {
     alt((
-        tag("*"),
+        map(tag("*"), |bull| (BulletType::Unordered, bull)),
-        tag("-"),
+        map(tag("-"), |bull| (BulletType::Unordered, bull)),
-        tag("+"),
+        map(tag("+"), |bull| (BulletType::Unordered, bull)),
+        map(
             recognize(tuple((
                 parser_with_context!(counter)(context),
                 alt((tag("."), tag(")"))),
             ))),
+            |bull| (BulletType::Ordered, bull),
+        ),
     ))(input)
 }
 

src/parser/subscript_and_superscript.rs

@@ -23,6 +23,7 @@ use crate::context::ContextElement;
 use crate::context::ContextMatcher;
 use crate::context::ExitClass;
 use crate::context::ExitMatcherNode;
+use crate::context::Matcher;
 use crate::context::RefContext;
 use crate::error::CustomError;
 use crate::error::MyError;
@@ -112,6 +113,10 @@ fn script_body<'b, 'g, 'r, 's>(
         map(parser_with_context!(script_with_braces)(context), |body| {
             ScriptBody::WithBraces(body.into())
         }),
+        map(
+            parser_with_context!(script_with_parenthesis)(context),
+            |body| ScriptBody::Braceless(body.into()),
+        ),
     ))(input)
 }
 
@@ -199,3 +204,49 @@ fn _script_with_braces_end<'b, 'g, 'r, 's>(
     }
     tag("}")(input)
 }
+
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+fn script_with_parenthesis<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    let (remaining, _) = tag("(")(input)?;
+    let exit_with_depth = script_with_parenthesis_end(remaining.get_parenthesis_depth());
+
+    let (remaining, _) = many_till(
+        anychar,
+        alt((
+            peek(exit_with_depth),
+            parser_with_context!(exit_matcher_parser)(context),
+        )),
+    )(remaining)?;
+
+    let (remaining, _) = tag(")")(remaining)?;
+    let source = get_consumed(input, remaining);
+    Ok((remaining, source))
+}
+
+fn script_with_parenthesis_end(starting_parenthesis_depth: BracketDepth) -> impl Matcher {
+    move |input: OrgSource<'_>| _script_with_parenthesis_end(input, starting_parenthesis_depth)
+}
+
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+fn _script_with_parenthesis_end<'s>(
+    input: OrgSource<'s>,
+    starting_parenthesis_depth: BracketDepth,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    let current_depth = input.get_parenthesis_depth() - starting_parenthesis_depth;
+    if current_depth < 0 {
+        // This shouldn't be possible because if depth is 0 then a closing bracket should end the citation.
+        unreachable!("Exceeded citation key suffix bracket depth.")
+    }
+    if current_depth == 0 {
+        let close_parenthesis = tag::<&str, OrgSource<'_>, CustomError<OrgSource<'_>>>(")")(input);
+        if close_parenthesis.is_ok() {
+            return close_parenthesis;
+        }
+    }
+    Err(nom::Err::Error(CustomError::MyError(MyError(
+        "No script parenthesis end.".into(),
+    ))))
+}

src/parser/text_markup.rs

@@ -312,7 +312,7 @@ fn post<'b, 'g, 'r, 's>(
     _context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, ()> {
-    let (remaining, _) = alt((recognize(one_of(" \r\n\t-.,;:!?')}[\"")), line_ending))(input)?;
+    let (remaining, _) = alt((recognize(one_of(" \r\n\t-.,;:!?')}[\"\\")), line_ending))(input)?;
     Ok((remaining, ()))
 }