Publish version 0.1.13.

Cargo.toml

@@ -2,7 +2,7 @@
 
 [package]
 name = "organic"
-version = "0.1.12"
+version = "0.1.13"
 authors = ["Tom Alexander <tom@fizz.buzz>"]
 description = "An org-mode parser."
 edition = "2021"

org_mode_samples/document/only_line_breaks.org

@@ -0,0 +1,4 @@
+
+
+
+

org_mode_samples/document/post_blank.org

@@ -0,0 +1,5 @@
+* foo
+
+
+
+

org_mode_samples/lesser_element/fixed_width_area/blank_line_in_middle.org

@@ -0,0 +1,3 @@
+: foo
+:
+: bar

org_mode_samples/lesser_element/fixed_width_area/list_post_blank.org

@@ -0,0 +1,6 @@
+1. foo
+   #+begin_src text
+
+   #+end_src
+
+2. baz

org_mode_samples/object/target/simple.org

@@ -0,0 +1,3 @@
+<<FOO>> bar
+
+[[FOO][baz]]

org_mode_samples/sections_and_headings/heading_with_subheading_post_blank.org

@@ -0,0 +1,5 @@
+* foo
+
+** bar
+
+* baz

src/compare/util.rs

@@ -285,7 +285,7 @@ where
 pub(crate) fn compare_children<'b, 's, 'x, RC>(
     source: &'s str,
     emacs: &'b Token<'s>,
-    rust_children: &'x Vec<RC>,
+    rust_children: &'x [RC],
     child_status: &mut Vec<DiffEntry<'b, 's>>,
     this_status: &mut DiffStatus,
     message: &mut Option<String>,

src/event_count/database.rs

@@ -27,7 +27,7 @@ pub(crate) fn record_event(event_type: EventType, input: OrgSource<'_>) {
 pub fn report(original_document: &str) {
     let mut db = GLOBAL_DATA.lock().unwrap();
     let db = db.get_or_insert_with(HashMap::new);
-    let mut results: Vec<_> = db.iter().map(|(k, v)| (k, v)).collect();
+    let mut results: Vec<_> = db.iter().collect();
     results.sort_by_key(|(_k, v)| *v);
     // This would put the most common at the top, but that is a pain when there is already a lot of output from the parser.
     // results.sort_by(|(_ak, av), (_bk, bv)| bv.cmp(av));

src/parser/bullshitium.rs

@@ -63,6 +63,7 @@ pub(crate) fn broken_end<'b, 'g, 'r, 's>(
         match paragraph.children.first_mut() {
             Some(Object::PlainText(plain_text)) => {
                 plain_text.source = input.get_until_end_of_str(plain_text.source).into();
+                paragraph.contents = Some(input.get_until_end_of_str(plain_text.source).into());
             }
             Some(obj) => {
                 panic!("Unhandled first object type inside bullshitium {:?}", obj);
@@ -80,10 +81,10 @@ pub(crate) fn broken_end<'b, 'g, 'r, 's>(
 
         Ok((
             remaining,
-            Paragraph::of_text_full(
+            Paragraph::of_text(
                 input.get_until(remaining).into(),
                 body,
-                if body.len() > 0 { Some(body) } else { None },
+                if !body.is_empty() { Some(body) } else { None },
                 post_blank.map(Into::<&str>::into),
             ),
         ))
@@ -123,6 +124,7 @@ pub(crate) fn broken_dynamic_block<'b, 'g, 'r, 's>(
         match paragraph.children.first_mut() {
             Some(Object::PlainText(plain_text)) => {
                 plain_text.source = input.get_until_end_of_str(plain_text.source).into();
+                paragraph.contents = Some(input.get_until_end_of_str(plain_text.source).into());
             }
             Some(obj) => {
                 panic!("Unhandled first object type inside bullshitium {:?}", obj);
@@ -133,14 +135,18 @@ pub(crate) fn broken_dynamic_block<'b, 'g, 'r, 's>(
         };
         Ok((remaining, paragraph))
     } else {
-        let (remaining, _trailing_ws) =
+        let (remaining, post_blank) =
             maybe_consume_trailing_whitespace_if_not_exiting(context, lead_in_remaining)?;
 
+        let body = Into::<&str>::into(input.get_until(lead_in_remaining));
+
         Ok((
             remaining,
             Paragraph::of_text(
                 input.get_until(remaining).into(),
-                input.get_until(lead_in_remaining).into(),
+                body,
+                if !body.is_empty() { Some(body) } else { None },
+                post_blank.map(Into::<&str>::into),
             ),
         ))
     }

src/parser/document.rs

@@ -3,6 +3,7 @@ use std::path::Path;
 use nom::combinator::all_consuming;
 use nom::combinator::opt;
 use nom::multi::many0;
+use nom::InputTake;
 
 use super::headline::heading;
 use super::in_buffer_settings::apply_in_buffer_settings;
@@ -195,9 +196,9 @@ fn _document<'b, 'g, 'r, 's>(
             zeroth_section,
             children,
             contents: if contents.len() > 0 {
-                Some(Into::<&str>::into(contents))
+                Into::<&str>::into(contents)
             } else {
-                None
+                Into::<&str>::into(remaining.take(0))
             },
         },
     ))

src/parser/dynamic_block.rs

@@ -6,20 +6,20 @@ use nom::character::complete::anychar;
 use nom::character::complete::line_ending;
 use nom::character::complete::space0;
 use nom::character::complete::space1;
-use nom::combinator::consumed;
 use nom::combinator::eof;
 use nom::combinator::not;
 use nom::combinator::opt;
 use nom::combinator::peek;
 use nom::combinator::recognize;
-use nom::multi::many0;
 use nom::multi::many_till;
-use nom::sequence::preceded;
 use nom::sequence::tuple;
 
 use super::affiliated_keyword::parse_affiliated_keywords;
+use super::greater_block::leading_blank_lines_end;
 use super::org_source::OrgSource;
+use super::paragraph::empty_paragraph;
 use super::util::maybe_consume_trailing_whitespace_if_not_exiting;
+use crate::context::bind_context;
 use crate::context::parser_with_context;
 use crate::context::ContextElement;
 use crate::context::ExitClass;
@@ -28,7 +28,6 @@ use crate::context::RefContext;
 use crate::error::CustomError;
 use crate::error::Res;
 use crate::parser::element_parser::element;
-use crate::parser::util::blank_line;
 use crate::parser::util::exit_matcher_parser;
 use crate::parser::util::get_consumed;
 use crate::parser::util::immediate_in_section;
@@ -36,7 +35,6 @@ use crate::parser::util::start_of_line;
 use crate::types::DynamicBlock;
 use crate::types::Element;
 use crate::types::Keyword;
-use crate::types::Paragraph;
 
 #[cfg_attr(
     feature = "tracing",
@@ -81,23 +79,25 @@ where
     let element_matcher = parser_with_context!(element(true))(&parser_context);
     let exit_matcher = parser_with_context!(exit_matcher_parser)(&parser_context);
     not(exit_matcher)(remaining)?;
-    let (remaining, leading_blank_lines) = opt(consumed(tuple((
-        blank_line,
-        many0(preceded(not(exit_matcher), blank_line)),
-    ))))(remaining)?;
-    let leading_blank_lines =
-        leading_blank_lines.map(|(source, (first_line, _remaining_lines))| {
-            Element::Paragraph(Paragraph::of_text(source.into(), first_line.into()))
-        });
+    let contents_begin = remaining;
+    let blank_line_context = ContextElement::ExitMatcherNode(ExitMatcherNode {
+        class: ExitClass::Alpha,
+        exit_matcher: &leading_blank_lines_end,
+    });
+    let blank_line_context = parser_context.with_additional_node(&blank_line_context);
+
+    let (remaining, leading_blank_lines) =
+        opt(bind_context!(empty_paragraph, &blank_line_context))(remaining)?;
     let (remaining, (mut children, _exit_contents)) =
         many_till(element_matcher, exit_matcher)(remaining)?;
     if let Some(lines) = leading_blank_lines {
-        children.insert(0, lines);
+        children.insert(0, Element::Paragraph(lines));
     }
+    let contents = get_consumed(contents_begin, remaining);
 
     let (remaining, _end) = dynamic_block_end(&parser_context, remaining)?;
 
-    let (remaining, _trailing_ws) =
+    let (remaining, post_blank) =
         maybe_consume_trailing_whitespace_if_not_exiting(context, remaining)?;
     let source = get_consumed(input, remaining);
     Ok((
@@ -111,6 +111,12 @@ where
             block_name: name.into(),
             parameters: parameters.map(|val| val.into()),
             children,
+            contents: if contents.len() > 0 {
+                Some(Into::<&str>::into(contents))
+            } else {
+                None
+            },
+            post_blank: post_blank.map(Into::<&str>::into),
         },
     ))
 }

src/parser/fixed_width_area.rs

@@ -14,7 +14,7 @@ use nom::InputTake;
 
 use super::affiliated_keyword::parse_affiliated_keywords;
 use super::org_source::OrgSource;
-use super::util::maybe_consume_trailing_whitespace_if_not_exiting_mid_line;
+use super::util::maybe_consume_trailing_whitespace_if_not_exiting;
 use super::util::org_line_ending;
 use crate::context::parser_with_context;
 use crate::context::RefContext;
@@ -48,8 +48,11 @@ where
         ),
     ))(remaining)?;
 
-    let (remaining, post_blank) =
-        maybe_consume_trailing_whitespace_if_not_exiting_mid_line(context, remaining)?;
+    let post_blank_begin = remaining;
+    let (remaining, _first_line_break) = org_line_ending(remaining)?;
+    let (remaining, _additional_post_blank) =
+        maybe_consume_trailing_whitespace_if_not_exiting(context, remaining)?;
+    let post_blank = get_consumed(post_blank_begin, remaining);
     let source = get_consumed(input, remaining);
     let mut value = Vec::with_capacity(remaining_lines.len() + 1);
     value.push(Into::<&str>::into(first_line));
@@ -63,7 +66,11 @@ where
                 affiliated_keywords,
             ),
             value,
-            post_blank: post_blank.map(Into::<&str>::into),
+            post_blank: if post_blank.len() > 0 {
+                Some(Into::<&str>::into(post_blank))
+            } else {
+                None
+            },
         },
     ))
 }
@@ -72,8 +79,8 @@ where
 fn fixed_width_area_line<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, OrgSource<'s>> {
     start_of_line(input)?;
     let (remaining, _) = tuple((space0, tag(":")))(input)?;
-    if let Ok((remain, _line_break)) = org_line_ending(remaining) {
-        return Ok((remain, remaining.take(0)));
+    if let Ok((_remain, _line_break)) = org_line_ending(remaining) {
+        return Ok((remaining, remaining.take(0)));
     }
     let (remaining, _) = tag(" ")(remaining)?;
     let (remaining, value) = recognize(many_till(anychar, peek(org_line_ending)))(remaining)?;

src/parser/greater_block.rs

@@ -339,7 +339,7 @@ fn _greater_block_end<'b, 'g, 'r, 's, 'c>(
     feature = "tracing",
     tracing::instrument(ret, level = "debug", skip(_context))
 )]
-fn leading_blank_lines_end<'b, 'g, 'r, 's, 'c>(
+pub(crate) fn leading_blank_lines_end<'b, 'g, 'r, 's, 'c>(
     _context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {

src/parser/headline.rs

@@ -67,6 +67,7 @@ fn _heading<'b, 'g, 'r, 's>(
     let section_matcher = bind_context!(section, context);
     let heading_matcher = bind_context!(heading(pre_headline.star_count), context);
     let (contents_begin, _) = opt(many0(blank_line))(remaining)?;
+    let maybe_post_blank = get_consumed(remaining, contents_begin);
     let (remaining, maybe_section) =
         opt(map(section_matcher, DocumentElement::Section))(remaining)?;
     let (remaining, _ws) = opt(tuple((start_of_line, many0(blank_line))))(remaining)?;
@@ -83,7 +84,8 @@ fn _heading<'b, 'g, 'r, 's>(
         }
         children.insert(0, section);
     }
-    let remaining = if children.is_empty() {
+    let has_children = !children.is_empty();
+    let remaining = if !has_children {
         // Support empty headings
         let (remain, _ws) = many0(blank_line)(remaining)?;
         remain
@@ -119,6 +121,11 @@ fn _heading<'b, 'g, 'r, 's>(
             } else {
                 None
             },
+            post_blank: if has_children {
+                None
+            } else {
+                Some(Into::<&str>::into(maybe_post_blank))
+            },
         },
     ))
 }

src/parser/paragraph.rs

@@ -101,7 +101,7 @@ pub(crate) fn empty_paragraph<'b, 'g, 'r, 's>(
         let source = get_consumed(input, remaining);
         Ok((
             remaining,
-            Paragraph::of_text_full(
+            Paragraph::of_text(
                 Into::<&str>::into(source),
                 Into::<&str>::into(first_line_with_spaces),
                 Some(Into::<&str>::into(first_line_with_spaces)),
@@ -116,7 +116,7 @@ pub(crate) fn empty_paragraph<'b, 'g, 'r, 's>(
         let source = get_consumed(input, remaining);
         Ok((
             remaining,
-            Paragraph::of_text_full(
+            Paragraph::of_text(
                 Into::<&str>::into(source),
                 Into::<&str>::into(first_line),
                 Some(Into::<&str>::into(first_line)),

src/parser/target.rs

@@ -46,7 +46,7 @@ pub(crate) fn target<'b, 'g, 'r, 's>(
         )));
     }
     let (remaining, _) = tag(">>")(remaining)?;
-    let (remaining, _trailing_whitespace) =
+    let (remaining, post_blank) =
         maybe_consume_object_trailing_whitespace_if_not_exiting(context, remaining)?;
     let source = get_consumed(input, remaining);
 
@@ -55,6 +55,7 @@ pub(crate) fn target<'b, 'g, 'r, 's>(
         Target {
             source: source.into(),
             value: body.into(),
+            post_blank: post_blank.map(Into::<&str>::into),
         },
     ))
 }

src/parser/util.rs

@@ -72,13 +72,6 @@ fn element_trailing_whitespace<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, O
     alt((eof, recognize(many0(blank_line))))(input)
 }
 
-#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
-fn element_trailing_whitespace_mid_line<'s>(
-    input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    alt((eof, recognize(many0(blank_line))))(input)
-}
-
 #[cfg_attr(
     feature = "tracing",
     tracing::instrument(ret, level = "debug", skip(context))
@@ -121,22 +114,6 @@ pub(crate) fn maybe_consume_trailing_whitespace_if_not_exiting<'b, 'g, 'r, 's>(
     }
 }
 
-#[cfg_attr(
-    feature = "tracing",
-    tracing::instrument(ret, level = "debug", skip(context))
-)]
-pub(crate) fn maybe_consume_trailing_whitespace_if_not_exiting_mid_line<'b, 'g, 'r, 's>(
-    context: RefContext<'b, 'g, 'r, 's>,
-    input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, Option<OrgSource<'s>>> {
-    if context.should_consume_trailing_whitespace() && exit_matcher_parser(context, input).is_err()
-    {
-        Ok(opt(element_trailing_whitespace_mid_line)(input)?)
-    } else {
-        Ok((input, None))
-    }
-}
-
 #[cfg_attr(
     feature = "tracing",
     tracing::instrument(ret, level = "debug", skip(context))

src/types/document.rs

@@ -17,7 +17,7 @@ pub struct Document<'s> {
     pub path: Option<PathBuf>,
     pub zeroth_section: Option<Section<'s>>,
     pub children: Vec<Heading<'s>>,
-    pub contents: Option<&'s str>,
+    pub contents: &'s str,
 }
 
 #[derive(Debug)]
@@ -36,6 +36,7 @@ pub struct Heading<'s> {
     pub deadline: Option<Timestamp<'s>>,
     pub closed: Option<Timestamp<'s>>,
     pub contents: Option<&'s str>,
+    pub post_blank: Option<&'s str>,
 }
 
 #[derive(Debug)]
@@ -63,14 +64,11 @@ impl<'s> StandardProperties<'s> for Document<'s> {
     }
 
     fn get_contents<'b>(&'b self) -> Option<&'s str> {
-        self.contents
+        Some(self.contents)
     }
 
     fn get_post_blank(&self) -> PostBlank {
-        self.into_iter()
-            .last()
-            .map(|child| child.get_post_blank())
-            .unwrap_or(0)
+        0
     }
 }
 
@@ -102,10 +100,11 @@ impl<'s> StandardProperties<'s> for Heading<'s> {
     }
 
     fn get_post_blank(&self) -> PostBlank {
-        self.children
-            .last()
-            .map(|child| child.get_post_blank())
+        self.post_blank
+            .map(|text| text.lines().count())
             .unwrap_or(0)
+            .try_into()
+            .expect("Too much post-blank to fit into a PostBlank.")
     }
 }
 

src/types/greater_element.rs

@@ -87,6 +87,8 @@ pub struct DynamicBlock<'s> {
     pub block_name: &'s str,
     pub parameters: Option<&'s str>,
     pub children: Vec<Element<'s>>,
+    pub contents: Option<&'s str>,
+    pub post_blank: Option<&'s str>,
 }
 
 #[derive(Debug)]
@@ -243,11 +245,15 @@ impl<'s> StandardProperties<'s> for DynamicBlock<'s> {
     }
 
     fn get_contents<'b>(&'b self) -> Option<&'s str> {
-        todo!()
+        self.contents
     }
 
     fn get_post_blank(&self) -> PostBlank {
-        todo!()
+        self.post_blank
+            .map(|text| text.lines().count())
+            .unwrap_or(0)
+            .try_into()
+            .expect("Too much post-blank to fit into a PostBlank.")
     }
 }
 

src/types/lesser_element.rs

@@ -204,21 +204,7 @@ impl<'s> Paragraph<'s> {
     /// Generate a paragraph of the passed in text with no additional properties.
     ///
     /// This is used for elements that support an "empty" content like greater blocks.
-    pub(crate) fn of_text(source: &'s str, body: &'s str) -> Self {
-        // TODO: This should be replaced with of_text_full.
-        Paragraph {
-            source,
-            contents: None,
-            post_blank: None,
-            affiliated_keywords: AffiliatedKeywords::default(),
-            children: vec![Object::PlainText(PlainText { source: body })],
-        }
-    }
-
-    /// Generate a paragraph of the passed in text with no additional properties.
-    ///
-    /// This is used for elements that support an "empty" content like greater blocks.
-    pub(crate) fn of_text_full(
+    pub(crate) fn of_text(
         source: &'s str,
         body: &'s str,
         contents: Option<&'s str>,

src/types/object.rs

@@ -266,6 +266,7 @@ pub struct LineBreak<'s> {
 pub struct Target<'s> {
     pub source: &'s str,
     pub value: &'s str,
+    pub post_blank: Option<&'s str>,
 }
 
 #[derive(Debug)]
@@ -923,11 +924,15 @@ impl<'s> StandardProperties<'s> for Target<'s> {
     }
 
     fn get_contents<'b>(&'b self) -> Option<&'s str> {
-        todo!()
+        None
     }
 
     fn get_post_blank(&self) -> PostBlank {
-        todo!()
+        self.post_blank
+            .map(|post_blank| post_blank.chars().count())
+            .unwrap_or(0)
+            .try_into()
+            .expect("Too much post-blank to fit into a PostBlank.")
     }
 }
 
@@ -1009,11 +1014,13 @@ impl<'s> StandardProperties<'s> for PlainText<'s> {
     }
 
     fn get_contents<'b>(&'b self) -> Option<&'s str> {
-        todo!()
+        // This field does not actually exist in emacs for plaintext
+        Some(self.source)
     }
 
     fn get_post_blank(&self) -> PostBlank {
-        todo!()
+        // This field does not actually exist in emacs for plaintext
+        0
     }
 }