Publish version 0.1.12.

Now that entities are stored in the settings struct, these variables are massive which makes them balloon trace sizes while being mostly unreadable. This removes Debug from them to serve as a static-analysis check that context is ALWAYS ignored in tracing calls.

Cargo.toml

@@ -2,7 +2,7 @@
 
 [package]
 name = "organic"
-version = "0.1.11"
+version = "0.1.12"
 authors = ["Tom Alexander <tom@fizz.buzz>"]
 description = "An org-mode parser."
 edition = "2021"

build.rs

@@ -66,10 +66,6 @@ fn write_test(test_file: &mut File, test: &walkdir::DirEntry) {
 }
 
 #[cfg(feature = "compare")]
-fn is_expect_fail(name: &str) -> Option<&str> {
-    match name {
-        "greater_element_drawer_drawer_with_headline_inside" => Some("Apparently lines with :end: become their own paragraph. This odd behavior needs to be investigated more."),
-        "element_container_priority_footnote_definition_dynamic_block" => Some("Apparently broken begin lines become their own paragraph."),
-        _ => None,
-    }
+fn is_expect_fail(_name: &str) -> Option<&str> {
+    None
 }

org_mode_samples/bullshitium/fake_paragraph/broken_end.org

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

org_mode_samples/bullshitium/fake_paragraph/broken_end_isolated.org

@@ -0,0 +1,2 @@
+foo
+:end:

scripts/dump_ast.bash

@@ -0,0 +1,58 @@
+#!/usr/bin/env bash
+#
+# Dump the AST of an org-mode document from emacs
+set -euo pipefail
+IFS=$'\n\t'
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+
+REALPATH=$(command -v uu-realpath || command -v realpath)
+MAKE=$(command -v gmake || command -v make)
+
+############## Setup #########################
+
+function die {
+    local status_code="$1"
+    shift
+    (>&2 echo "${@}")
+    exit "$status_code"
+}
+
+function log {
+    (>&2 echo "${@}")
+}
+
+############## Program #########################
+
+function main {
+    if [ $# -eq 0 ]; then
+        dump_ast_stdin "${@}"
+    else
+        dump_ast_file "${@}"
+    fi
+}
+
+function dump_ast_stdin {
+    # Until we can find a good way to encode stdin as an elisp string in bash, I cannot operate on stdin.
+    die 1 "This script only works on files."
+}
+
+function dump_ast_file {
+    local target_file mounted_file elisp_script
+    target_file=$($REALPATH "$1")
+    mounted_file="/input${target_file}"
+    elisp_script=$(cat <<EOF
+(progn
+     (erase-buffer)
+     (require 'org)
+     (defun org-table-align () t)
+     (find-file-read-only "${mounted_file}")
+     (org-mode)
+     (message "%s" (pp-to-string (org-element-parse-buffer)))
+)
+EOF
+          )
+    exec docker run --init --rm -i --mount type=tmpfs,destination=/tmp -v "/:/input:ro" --entrypoint "" organic-test emacs -q --no-site-file --no-splash --batch --eval "$elisp_script"
+}
+
+
+main "${@}"

src/context/constants.rs

@@ -1,15 +1,27 @@
 use super::global_settings::EntityDefinition;
 
-pub(crate) const DEFAULT_ORG_ELEMENT_PARSED_KEYWORDS: [&str; 1] = ["CAPTION"];
+/// Keywords that contain the standard set of objects (excluding footnote references).
+///
+/// Corresponds to org-element-parsed-keywords elisp variable.
+pub(crate) const ORG_ELEMENT_PARSED_KEYWORDS: [&str; 1] = ["CAPTION"];
 
-pub(crate) const DEFAULT_ORG_ELEMENT_DUAL_KEYWORDS: [&str; 2] = ["CAPTION", "RESULTS"];
+/// Keywords that can have a secondary value in square brackets.
+///
+/// Corresponds to org-element-dual-keywords elisp variable.
+pub(crate) const ORG_ELEMENT_DUAL_KEYWORDS: [&str; 2] = ["CAPTION", "RESULTS"];
 
-pub(crate) const DEFAULT_ORG_ELEMENT_AFFILIATED_KEYWORDS: [&str; 13] = [
+/// Keywords that can be affiliated with an element.
+///
+/// Corresponds to org-element-affiliated-keywords elisp variable.
+pub(crate) const ORG_ELEMENT_AFFILIATED_KEYWORDS: [&str; 13] = [
     "CAPTION", "DATA", "HEADER", "HEADERS", "LABEL", "NAME", "PLOT", "RESNAME", "RESULT",
     "RESULTS", "SOURCE", "SRCNAME", "TBLNAME",
 ];
 
-pub(crate) const DEFAULT_ORG_ELEMENT_KEYWORD_TRANSLATION_ALIST: [(&str, &str); 8] = [
+/// Mapping of keyword names.
+///
+/// Corresponds to org-element-keyword-translation-alist elisp variable.
+pub(crate) const ORG_ELEMENT_KEYWORD_TRANSLATION_ALIST: [(&str, &str); 8] = [
     ("DATA", "NAME"),
     ("LABEL", "NAME"),
     ("RESNAME", "NAME"),

src/context/context.rs

@@ -12,7 +12,6 @@ use crate::error::CustomError;
 use crate::error::Res;
 use crate::parser::OrgSource;
 
-#[derive(Debug)]
 pub(crate) enum ContextElement<'r, 's> {
     /// Stores a parser that indicates that children should exit upon matching an exit matcher.
     ExitMatcherNode(ExitMatcherNode<'r>),
@@ -34,15 +33,6 @@ pub(crate) struct ExitMatcherNode<'r> {
     pub(crate) class: ExitClass,
 }
 
-impl<'r> std::fmt::Debug for ExitMatcherNode<'r> {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        let mut formatter = f.debug_struct("ExitMatcherNode");
-        formatter.field("class", &self.class.to_string());
-        formatter.finish()
-    }
-}
-
-#[derive(Debug)]
 pub(crate) struct Context<'g, 'r, 's> {
     global_settings: &'g GlobalSettings<'g, 's>,
     tree: List<'r, &'r ContextElement<'r, 's>>,

src/context/exiting.rs

@@ -1,13 +1,7 @@
-#[derive(Debug, Copy, Clone)]
+#[derive(Copy, Clone)]
 pub(crate) enum ExitClass {
     Document = 1,
     Alpha = 2,
     Beta = 3,
     Gamma = 4,
 }
-
-impl std::fmt::Display for ExitClass {
-    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
-        write!(f, "{:?}", self)
-    }
-}

src/context/file_access_interface.rs

@@ -1,17 +1,16 @@
-use std::fmt::Debug;
 use std::path::PathBuf;
 
 #[cfg(any(feature = "compare", feature = "foreign_document_test"))]
-pub trait FileAccessInterface: Sync + Debug {
+pub trait FileAccessInterface: Sync {
     fn read_file(&self, path: &str) -> Result<String, std::io::Error>;
 }
 
 #[cfg(not(any(feature = "compare", feature = "foreign_document_test")))]
-pub trait FileAccessInterface: Debug {
+pub trait FileAccessInterface {
     fn read_file(&self, path: &str) -> Result<String, std::io::Error>;
 }
 
-#[derive(Debug, Clone)]
+#[derive(Clone)]
 pub struct LocalFileAccessInterface {
     pub working_directory: Option<PathBuf>,
 }

src/context/global_settings.rs

@@ -5,16 +5,12 @@ use super::constants::DEFAULT_ORG_ENTITIES;
 use super::constants::DEFAULT_ORG_LINK_PARAMETERS;
 use super::FileAccessInterface;
 use super::LocalFileAccessInterface;
-use crate::context::constants::DEFAULT_ORG_ELEMENT_AFFILIATED_KEYWORDS;
-use crate::context::constants::DEFAULT_ORG_ELEMENT_DUAL_KEYWORDS;
-use crate::context::constants::DEFAULT_ORG_ELEMENT_KEYWORD_TRANSLATION_ALIST;
-use crate::context::constants::DEFAULT_ORG_ELEMENT_PARSED_KEYWORDS;
 use crate::types::IndentationLevel;
 use crate::types::Object;
 
 // TODO: Ultimately, I think we'll need most of this: https://orgmode.org/manual/In_002dbuffer-Settings.html
 
-#[derive(Debug, Clone)]
+#[derive(Clone)]
 pub struct GlobalSettings<'g, 's> {
     pub radio_targets: Vec<&'g Vec<Object<'s>>>,
     pub file_access: &'g dyn FileAccessInterface,
@@ -58,26 +54,6 @@ pub struct GlobalSettings<'g, 's> {
     ///
     /// Corresponds to org-entities elisp variable.
     pub entities: &'g [EntityDefinition<'s>],
-
-    /// Keywords that contain the standard set of objects (excluding footnote references).
-    ///
-    /// Corresponds to org-element-parsed-keywords elisp variable.
-    pub element_parsed_keywords: &'g [&'s str],
-
-    /// Keywords that can have a secondary value in square brackets.
-    ///
-    /// Corresponds to org-element-dual-keywords elisp variable.
-    pub element_dual_keywords: &'g [&'s str],
-
-    /// Keywords that can be affiliated with an element.
-    ///
-    /// Corresponds to org-element-affiliated-keywords elisp variable.
-    pub element_affiliated_keywords: &'g [&'s str],
-
-    /// Mapping of keyword names.
-    ///
-    /// Corresponds to org-element-keyword-translation-alist elisp variable.
-    pub element_keyword_translation_alist: &'g [(&'s str, &'s str)],
 }
 
 pub const DEFAULT_TAB_WIDTH: IndentationLevel = 8;
@@ -112,10 +88,6 @@ impl<'g, 's> GlobalSettings<'g, 's> {
             link_parameters: &DEFAULT_ORG_LINK_PARAMETERS,
             link_templates: BTreeMap::new(),
             entities: &DEFAULT_ORG_ENTITIES,
-            element_parsed_keywords: &DEFAULT_ORG_ELEMENT_PARSED_KEYWORDS,
-            element_dual_keywords: &DEFAULT_ORG_ELEMENT_DUAL_KEYWORDS,
-            element_affiliated_keywords: &DEFAULT_ORG_ELEMENT_AFFILIATED_KEYWORDS,
-            element_keyword_translation_alist: &DEFAULT_ORG_ELEMENT_KEYWORD_TRANSLATION_ALIST,
         }
     }
 }
@@ -126,7 +98,7 @@ impl<'g, 's> Default for GlobalSettings<'g, 's> {
     }
 }
 
-#[derive(Debug, Clone, PartialEq, Default)]
+#[derive(Clone, PartialEq, Default)]
 pub enum HeadlineLevelFilter {
     Odd,
 

src/context/mod.rs

@@ -1,7 +1,7 @@
 use crate::error::Res;
 use crate::parser::OrgSource;
 
-mod constants;
+pub(crate) mod constants;
 #[allow(clippy::module_inception)]
 mod context;
 mod exiting;

src/parser/affiliated_keyword.rs

@@ -19,29 +19,27 @@ use super::object_parser::standard_set_object;
 use super::util::confine_context;
 use super::OrgSource;
 use crate::context::bind_context;
+use crate::context::constants::ORG_ELEMENT_DUAL_KEYWORDS;
+use crate::context::constants::ORG_ELEMENT_KEYWORD_TRANSLATION_ALIST;
+use crate::context::constants::ORG_ELEMENT_PARSED_KEYWORDS;
 use crate::context::Context;
 use crate::context::ContextElement;
 use crate::context::GlobalSettings;
 use crate::context::List;
-use crate::context::RefContext;
 use crate::error::Res;
 use crate::types::AffiliatedKeywordValue;
 use crate::types::AffiliatedKeywords;
 use crate::types::Keyword;
 
-#[cfg_attr(
-    feature = "tracing",
-    tracing::instrument(ret, level = "debug", skip(context))
-)]
-pub(crate) fn affiliated_keywords<'b, 'g, 'r, 's>(
-    context: RefContext<'b, 'g, 'r, 's>,
+#[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
+pub(crate) fn affiliated_keywords<'s>(
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, Vec<Keyword<'s>>> {
     let mut ret = Vec::new();
     let mut remaining = input;
 
     loop {
-        let result = affiliated_keyword(context, remaining);
+        let result = affiliated_keyword(remaining);
         match result {
             Ok((remain, kw)) => {
                 remaining = remain;
@@ -65,8 +63,8 @@ where
 {
     let mut ret = BTreeMap::new();
     for kw in input {
-        let translated_name = translate_name(global_settings, kw.key);
-        let keyword_type = identify_keyword_type(global_settings, translated_name.as_str());
+        let translated_name = translate_name(kw.key);
+        let keyword_type = identify_keyword_type(translated_name.as_str());
         match keyword_type {
             AffiliatedKeywordType::SingleString => {
                 ret.insert(
@@ -151,12 +149,12 @@ where
     AffiliatedKeywords { keywords: ret }
 }
 
-fn translate_name<'g, 's>(global_settings: &'g GlobalSettings<'g, 's>, name: &'s str) -> String {
+fn translate_name(name: &str) -> String {
     let name_until_optval = name
         .split_once('[')
         .map(|(before, _after)| before)
         .unwrap_or(name);
-    for (src, dst) in global_settings.element_keyword_translation_alist {
+    for (src, dst) in ORG_ELEMENT_KEYWORD_TRANSLATION_ALIST {
         if name_until_optval.eq_ignore_ascii_case(src) {
             return dst.to_lowercase();
         }
@@ -171,20 +169,15 @@ enum AffiliatedKeywordType {
     ObjectTree,
 }
 
-fn identify_keyword_type<'g, 's>(
-    global_settings: &'g GlobalSettings<'g, 's>,
-    name: &'s str,
-) -> AffiliatedKeywordType {
+fn identify_keyword_type(name: &str) -> AffiliatedKeywordType {
     let is_multiple = ["CAPTION", "HEADER"]
         .into_iter()
         .any(|candidate| name.eq_ignore_ascii_case(candidate))
         || name.to_lowercase().starts_with("attr_");
-    let is_parsed = global_settings
-        .element_parsed_keywords
+    let is_parsed = ORG_ELEMENT_PARSED_KEYWORDS
         .iter()
         .any(|candidate| name.eq_ignore_ascii_case(candidate));
-    let can_have_optval = global_settings
-        .element_dual_keywords
+    let can_have_optval = ORG_ELEMENT_DUAL_KEYWORDS
         .iter()
         .any(|candidate| name.eq_ignore_ascii_case(candidate));
     match (is_multiple, is_parsed, can_have_optval) {

src/parser/bullshitium.rs

@@ -0,0 +1,165 @@
+use nom::branch::alt;
+use nom::bytes::complete::tag_no_case;
+use nom::character::complete::anychar;
+use nom::character::complete::space0;
+use nom::multi::many_till;
+use nom::sequence::tuple;
+
+use super::paragraph::paragraph;
+use super::util::maybe_consume_trailing_whitespace_if_not_exiting;
+use super::util::org_line_ending;
+use super::util::start_of_line;
+use super::OrgSource;
+use crate::context::bind_context;
+use crate::context::RefContext;
+use crate::error::CustomError;
+use crate::error::Res;
+use crate::parser::macros::element;
+use crate::types::AffiliatedKeywords;
+use crate::types::Object;
+use crate::types::Paragraph;
+use crate::types::PlainText;
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
+pub(crate) fn bullshitium<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, Paragraph<'s>> {
+    alt((
+        bind_context!(broken_end, context),
+        bind_context!(broken_dynamic_block, context),
+    ))(input)
+}
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
+pub(crate) fn detect_bullshitium<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, ()> {
+    element!(detect_broken_end, context, input);
+    element!(detect_broken_dynamic_block, context, input);
+    Err(nom::Err::Error(CustomError::Static("No bullshitium.")))
+}
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
+pub(crate) fn broken_end<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, Paragraph<'s>> {
+    start_of_line(input)?;
+    let (remaining, _) = space0(input)?;
+    let (remaining, _) = tag_no_case(":end:")(remaining)?;
+    let (lead_in_remaining, _) = tuple((space0, org_line_ending))(remaining)?;
+    if let Ok((remaining, mut paragraph)) =
+        paragraph(std::iter::empty(), lead_in_remaining, context, input)
+    {
+        match paragraph.children.first_mut() {
+            Some(Object::PlainText(plain_text)) => {
+                plain_text.source = input.get_until_end_of_str(plain_text.source).into();
+            }
+            Some(obj) => {
+                panic!("Unhandled first object type inside bullshitium {:?}", obj);
+            }
+            None => {
+                unreachable!("Paragraph must have children.");
+            }
+        };
+        Ok((remaining, paragraph))
+    } else {
+        let (remaining, _trailing_ws) =
+            maybe_consume_trailing_whitespace_if_not_exiting(context, lead_in_remaining)?;
+
+        Ok((
+            remaining,
+            Paragraph {
+                source: input.get_until(remaining).into(),
+                affiliated_keywords: AffiliatedKeywords::default(),
+                children: vec![Object::PlainText(PlainText {
+                    source: input.get_until(lead_in_remaining).into(),
+                })],
+            },
+        ))
+    }
+}
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(_context))
+)]
+pub(crate) fn detect_broken_end<'b, 'g, 'r, 's>(
+    _context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, ()> {
+    start_of_line(input)?;
+    let (remaining, _) = space0(input)?;
+    let (remaining, _) = tag_no_case(":end:")(remaining)?;
+    let (_remaining, _) = tuple((space0, org_line_ending))(remaining)?;
+    Ok((input, ()))
+}
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
+pub(crate) fn broken_dynamic_block<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, Paragraph<'s>> {
+    start_of_line(input)?;
+    let (remaining, _) = space0(input)?;
+    let (remaining, _) = tag_no_case("#+BEGIN:")(remaining)?;
+    let (lead_in_remaining, _) = many_till(anychar, org_line_ending)(remaining)?;
+    if let Ok((remaining, mut paragraph)) =
+        paragraph(std::iter::empty(), lead_in_remaining, context, input)
+    {
+        match paragraph.children.first_mut() {
+            Some(Object::PlainText(plain_text)) => {
+                plain_text.source = input.get_until_end_of_str(plain_text.source).into();
+            }
+            Some(obj) => {
+                panic!("Unhandled first object type inside bullshitium {:?}", obj);
+            }
+            None => {
+                unreachable!("Paragraph must have children.");
+            }
+        };
+        Ok((remaining, paragraph))
+    } else {
+        let (remaining, _trailing_ws) =
+            maybe_consume_trailing_whitespace_if_not_exiting(context, lead_in_remaining)?;
+
+        Ok((
+            remaining,
+            Paragraph {
+                source: input.get_until(remaining).into(),
+                affiliated_keywords: AffiliatedKeywords::default(),
+                children: vec![Object::PlainText(PlainText {
+                    source: input.get_until(lead_in_remaining).into(),
+                })],
+            },
+        ))
+    }
+}
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(_context))
+)]
+pub(crate) fn detect_broken_dynamic_block<'b, 'g, 'r, 's>(
+    _context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, ()> {
+    start_of_line(input)?;
+    let (remaining, _) = space0(input)?;
+    let (_remaining, _) = tag_no_case("#+BEGIN:")(remaining)?;
+    Ok((input, ()))
+}

src/parser/element_parser.rs

@@ -32,6 +32,8 @@ use crate::event_count::record_event;
 #[cfg(feature = "event_count")]
 use crate::event_count::EventType;
 use crate::parser::affiliated_keyword::affiliated_keywords;
+use crate::parser::bullshitium::bullshitium;
+use crate::parser::bullshitium::detect_bullshitium;
 use crate::parser::macros::ak_element;
 use crate::parser::macros::element;
 use crate::parser::table::org_mode_table;
@@ -57,8 +59,7 @@ fn _element<'b, 'g, 'r, 's>(
 ) -> Res<OrgSource<'s>, Element<'s>> {
     #[cfg(feature = "event_count")]
     record_event(EventType::ElementStart, input);
-    let (post_affiliated_keywords_input, affiliated_keywords) =
-        affiliated_keywords(context, input)?;
+    let (post_affiliated_keywords_input, affiliated_keywords) = affiliated_keywords(input)?;
 
     let mut affiliated_keywords = affiliated_keywords.into_iter();
 
@@ -242,6 +243,9 @@ fn _element<'b, 'g, 'r, 's>(
     );
 
     if can_be_paragraph {
+        // Fake paragraphs
+        element!(bullshitium, context, input, Element::Paragraph);
+
         // Paragraph without affiliated keyword
         ak_element!(
             paragraph,
@@ -272,8 +276,7 @@ fn _detect_element<'b, 'g, 'r, 's>(
     input: OrgSource<'s>,
     can_be_paragraph: bool,
 ) -> Res<OrgSource<'s>, ()> {
-    let (post_affiliated_keywords_input, affiliated_keywords) =
-        affiliated_keywords(context, input)?;
+    let (post_affiliated_keywords_input, affiliated_keywords) = affiliated_keywords(input)?;
 
     let mut affiliated_keywords = affiliated_keywords.into_iter();
 
@@ -319,6 +322,11 @@ fn _detect_element<'b, 'g, 'r, 's>(
         input
     );
 
+    // Fake paragraphs
+    if !can_be_paragraph {
+        element!(detect_bullshitium, context, input);
+    }
+
     if _element(context, input, can_be_paragraph).is_ok() {
         return Ok((input, ()));
     }

src/parser/entity.rs

@@ -1,11 +1,11 @@
 use nom::branch::alt;
 use nom::bytes::complete::tag;
 use nom::character::complete::satisfy;
+use nom::combinator::cond;
 use nom::combinator::eof;
 use nom::combinator::map;
 use nom::combinator::peek;
 use nom::combinator::recognize;
-use nom::combinator::verify;
 use nom::sequence::tuple;
 
 use super::org_source::OrgSource;
@@ -58,13 +58,16 @@ fn name<'b, 'g, 'r, 's>(
     for entity in context.get_global_settings().entities {
         let result = tuple((
             tag::<_, _, CustomError>(entity.name),
+            cond(
+                !entity.name.ends_with(' '),
                 alt((
-                verify(map(tag("{}"), |_| true), |_| !entity.name.ends_with(' ')),
+                    map(tag("{}"), |_| true),
                     map(peek(recognize(entity_end)), |_| false),
                 )),
+            ),
         ))(input);
         if let Ok((remaining, (ent, use_brackets))) = result {
-            return Ok((remaining, (entity, ent, use_brackets)));
+            return Ok((remaining, (entity, ent, use_brackets.unwrap_or(false))));
         }
     }
 

src/parser/in_buffer_settings.rs

@@ -84,7 +84,10 @@ fn in_buffer_settings_key<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, OrgSou
     ))(input)
 }
 
-#[cfg_attr(feature = "tracing", tracing::instrument(level = "debug"))]
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(level = "debug", skip(original_settings))
+)]
 pub(crate) fn apply_in_buffer_settings<'g, 's, 'sf>(
     keywords: Vec<Keyword<'sf>>,
     original_settings: &'g GlobalSettings<'g, 's>,

src/parser/keyword.rs

@@ -21,7 +21,8 @@ use super::org_source::OrgSource;
 use super::util::get_consumed;
 use super::util::maybe_consume_trailing_whitespace_if_not_exiting;
 use super::util::org_line_ending;
-use crate::context::bind_context;
+use crate::context::constants::ORG_ELEMENT_AFFILIATED_KEYWORDS;
+use crate::context::constants::ORG_ELEMENT_DUAL_KEYWORDS;
 use crate::context::RefContext;
 use crate::error::CustomError;
 use crate::error::Res;
@@ -98,11 +99,8 @@ where
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
-pub(crate) fn affiliated_keyword<'b, 'g, 'r, 's>(
-    context: RefContext<'b, 'g, 'r, 's>,
-    input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, Keyword<'s>> {
-    filtered_keyword(bind_context!(affiliated_key, context))(input)
+pub(crate) fn affiliated_keyword<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, Keyword<'s>> {
+    filtered_keyword(affiliated_key)(input)
 }
 
 #[cfg_attr(
@@ -137,24 +135,18 @@ fn regular_keyword_key<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, OrgSource
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
-fn affiliated_key<'b, 'g, 'r, 's>(
-    context: RefContext<'b, 'g, 'r, 's>,
-    input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    element!(dual_affiliated_key, context, input);
-    element!(plain_affiliated_key, context, input);
+fn affiliated_key<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    element!(dual_affiliated_key, input);
+    element!(plain_affiliated_key, input);
     element!(export_keyword, input);
     Err(nom::Err::Error(CustomError::Static("No affiliated key.")))
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
-fn plain_affiliated_key<'b, 'g, 'r, 's>(
-    context: RefContext<'b, 'g, 'r, 's>,
-    input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    for keyword in context.get_global_settings().element_affiliated_keywords {
+fn plain_affiliated_key<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    for keyword in ORG_ELEMENT_AFFILIATED_KEYWORDS {
         let result = map(
-            tuple((tag_no_case::<_, _, CustomError>(*keyword), peek(tag(":")))),
+            tuple((tag_no_case::<_, _, CustomError>(keyword), peek(tag(":")))),
             |(key, _)| key,
         )(input);
         if let Ok((remaining, ent)) = result {
@@ -166,13 +158,10 @@ fn plain_affiliated_key<'b, 'g, 'r, 's>(
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
-fn dual_affiliated_key<'b, 'g, 'r, 's>(
-    context: RefContext<'b, 'g, 'r, 's>,
-    input: OrgSource<'s>,
-) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    for keyword in context.get_global_settings().element_dual_keywords {
+fn dual_affiliated_key<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    for keyword in ORG_ELEMENT_DUAL_KEYWORDS {
         let result = recognize(tuple((
-            tag_no_case::<_, _, CustomError>(*keyword),
+            tag_no_case::<_, _, CustomError>(keyword),
             tag("["),
             optval,
             tag("]"),
@@ -232,19 +221,12 @@ mod tests {
     use test::Bencher;
 
     use super::*;
-    use crate::context::Context;
-    use crate::context::ContextElement;
-    use crate::context::GlobalSettings;
-    use crate::context::List;
     use crate::parser::OrgSource;
 
     #[bench]
     fn bench_affiliated_keyword(b: &mut Bencher) {
         let input = OrgSource::new("#+CAPTION[*foo*]: bar *baz*");
-        let global_settings = GlobalSettings::default();
-        let initial_context = ContextElement::document_context();
-        let initial_context = Context::new(&global_settings, List::new(&initial_context));
 
-        b.iter(|| assert!(affiliated_keyword(&initial_context, input).is_ok()));
+        b.iter(|| assert!(affiliated_keyword(input).is_ok()));
     }
 }

src/parser/mod.rs

@@ -1,6 +1,7 @@
 mod affiliated_keyword;
 mod angle_link;
 mod babel_call;
+mod bullshitium;
 mod citation;
 mod citation_reference;
 mod clock;

src/parser/org_source.rs

@@ -82,6 +82,15 @@ impl<'s> OrgSource<'s> {
         self.slice(..(other.end - self.start))
     }
 
+    pub(crate) fn get_until_end_of_str(&self, other: &'s str) -> OrgSource<'s> {
+        let full_source_start = self.full_source.as_ptr() as usize;
+        let other_start = other.as_ptr() as usize - full_source_start;
+        let other_end = other_start + other.len();
+        debug_assert!(other_start >= self.start);
+        debug_assert!(other_end <= self.end);
+        self.slice(..(other_end - self.start))
+    }
+
     pub(crate) fn get_start_of_line(&self) -> OrgSource<'s> {
         let skipped_text = self.text_since_line_break();
         let mut bracket_depth = self.bracket_depth;

src/parser/plain_list.rs

@@ -3,6 +3,7 @@ use nom::bytes::complete::tag;
 use nom::character::complete::anychar;
 use nom::character::complete::digit1;
 use nom::character::complete::line_ending;
+use nom::character::complete::multispace1;
 use nom::character::complete::one_of;
 use nom::character::complete::space0;
 use nom::character::complete::space1;
@@ -17,6 +18,7 @@ use nom::multi::many0;
 use nom::multi::many1;
 use nom::multi::many_till;
 use nom::sequence::tuple;
+use nom::InputTake;
 
 use super::affiliated_keyword::parse_affiliated_keywords;
 use super::element_parser::element;
@@ -25,6 +27,7 @@ use super::org_source::OrgSource;
 use super::util::include_input;
 use super::util::indentation_level;
 use super::util::non_whitespace_character;
+use crate::context::bind_context;
 use crate::context::parser_with_context;
 use crate::context::ContextElement;
 use crate::context::ContextMatcher;
@@ -80,6 +83,46 @@ where
     return Err(nom::Err::Error(CustomError::Static("No element detected.")));
 }
 
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
+pub(crate) fn detect_not_plain_list_item_indent<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, (u16, OrgSource<'s>)> {
+    if let Ok((_remaining, (_, indent, _))) = tuple((
+        start_of_line,
+        parser_with_context!(indentation_level)(context),
+        not(tuple((
+            parser_with_context!(bullet)(context),
+            alt((space1, line_ending, eof)),
+        ))),
+    ))(input)
+    {
+        return Ok((input, indent));
+    }
+
+    // Headlines are not plain list items.
+    if let Ok((_remaining, (_, indent, _))) = verify(
+        tuple((
+            start_of_line,
+            parser_with_context!(indentation_level)(context),
+            tuple((
+                parser_with_context!(bullet)(context),
+                alt((space1, line_ending, eof)),
+            )),
+        )),
+        |(_, (depth, _), ((_, bullet), _))| {
+            *depth == 0 && Into::<&str>::into(bullet).starts_with('*')
+        },
+    )(input)
+    {
+        return Ok((input, indent));
+    }
+    return Err(nom::Err::Error(CustomError::Static("No element detected.")));
+}
+
 #[cfg_attr(
     feature = "tracing",
     tracing::instrument(ret, level = "debug", skip(context, affiliated_keywords))
@@ -120,7 +163,7 @@ where
     // While #3 is the most slow, it also seems to cleanest and involves the least manual mutation of already-parsed objects so I am going with #3 for now, but we should revisit #1 or #2 when the parser is more developed.
 
     loop {
-        let list_item = parser_with_context!(plain_list_item)(&parser_context)(remaining);
+        let list_item = plain_list_item(&parser_context, remaining);
         match (&first_item_list_type, &list_item) {
             (None, Ok((_remain, (list_type, _item)))) => {
                 let _ = first_item_list_type.insert(*list_type);
@@ -140,25 +183,17 @@ where
             }
         };
 
-        let maybe_exit = parser_with_context!(exit_matcher_parser)(&parser_context)(remaining);
+        let maybe_exit = exit_matcher_parser(&parser_context, remaining);
         if maybe_exit.is_ok() {
             break;
         }
     }
 
-    let (final_child_start, _final_item_first_parse) = match children.pop() {
-        Some(final_child) => final_child,
-        None => {
+    if children.is_empty() {
         return Err(nom::Err::Error(CustomError::Static(
             "Plain lists require at least one element.",
         )));
     }
-    };
-    let final_item_context = ContextElement::ConsumeTrailingWhitespace(false);
-    let final_item_context = parser_context.with_additional_node(&final_item_context);
-    let (remaining, (_, reparsed_final_item)) =
-        parser_with_context!(plain_list_item)(&final_item_context)(final_child_start)?;
-    children.push((final_child_start, reparsed_final_item));
 
     let (remaining, _trailing_ws) =
         maybe_consume_trailing_whitespace_if_not_exiting(context, remaining)?;
@@ -187,10 +222,10 @@ fn plain_list_item<'b, 'g, 'r, 's>(
 ) -> Res<OrgSource<'s>, (PlainListType, PlainListItem<'s>)> {
     start_of_line(input)?;
     let (remaining, (indent_level, _leading_whitespace)) = indentation_level(context, input)?;
-    let (remaining, (bullet_type, bull)) = verify(
-        parser_with_context!(bullet)(context),
-        |(_bullet_type, bull)| !Into::<&str>::into(bull).starts_with('*') || indent_level > 0,
-    )(remaining)?;
+    let (remaining, (bullet_type, bull)) =
+        verify(bind_context!(bullet, context), |(_bullet_type, bull)| {
+            !Into::<&str>::into(bull).starts_with('*') || indent_level > 0
+        })(remaining)?;
 
     let (remaining, maybe_counter_set) =
         opt(tuple((space1, tag("[@"), counter_set_value, tag("]"))))(remaining)?;
@@ -199,7 +234,7 @@ fn plain_list_item<'b, 'g, 'r, 's>(
     let (remaining, maybe_checkbox) = opt(tuple((space1, item_checkbox)))(remaining)?;
 
     let (remaining, maybe_tag) = if let BulletType::Unordered = bullet_type {
-        opt(tuple((space1, parser_with_context!(item_tag)(context))))(remaining)?
+        opt(tuple((space1, bind_context!(item_tag, context))))(remaining)?
     } else {
         (remaining, None)
     };
@@ -211,6 +246,12 @@ fn plain_list_item<'b, 'g, 'r, 's>(
     };
 
     let exit_matcher = plain_list_item_end(indent_level);
+    let final_item_whitespace_cutoff = final_item_whitespace_cutoff(indent_level);
+    let final_whitespace_context = ContextElement::ExitMatcherNode(ExitMatcherNode {
+        class: ExitClass::Beta,
+        exit_matcher: &final_item_whitespace_cutoff,
+    });
+    let final_whitespace_context = context.with_additional_node(&final_whitespace_context);
     let contexts = [
         ContextElement::ConsumeTrailingWhitespace(true),
         ContextElement::ExitMatcherNode(ExitMatcherNode {
@@ -218,17 +259,21 @@ fn plain_list_item<'b, 'g, 'r, 's>(
             exit_matcher: &exit_matcher,
         }),
     ];
-    let parser_context = context.with_additional_node(&contexts[0]);
+    let parser_context = final_whitespace_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
-    )(&parser_context))(remaining);
+    let maybe_contentless_item: Res<OrgSource<'_>, ()> =
+        detect_contentless_item_contents(&parser_context, remaining);
     if let Ok((_rem, _ws)) = maybe_contentless_item {
-        let (remaining, _trailing_ws) = if context.should_consume_trailing_whitespace() {
-            recognize(alt((recognize(many1(blank_line)), eof)))(remaining)?
-        } else {
+        let (remaining, _trailing_ws) = if tuple((
+            blank_line,
+            bind_context!(final_item_whitespace_cutoff, context),
+        ))(remaining)
+        .is_ok()
+        {
             recognize(alt((blank_line, eof)))(remaining)?
+        } else {
+            recognize(alt((recognize(many1(blank_line)), eof)))(remaining)?
         };
         let source = get_consumed(input, remaining);
         return Ok((
@@ -256,26 +301,14 @@ fn plain_list_item<'b, 'g, 'r, 's>(
         .filter(|b| *b == b'\n')
         .count();
 
-    let (mut remaining, (mut children, _exit_contents)) = many_till(
-        include_input(parser_with_context!(element(true))(&parser_context)),
-        parser_with_context!(exit_matcher_parser)(&parser_context),
+    let (remaining, (children, _exit_contents)) = many_till(
+        include_input(bind_context!(element(true), &parser_context)),
+        bind_context!(exit_matcher_parser, &parser_context),
     )(remaining)?;
 
-    if !children.is_empty() && !context.should_consume_trailing_whitespace() {
-        let final_item_context = ContextElement::ConsumeTrailingWhitespace(false);
-        let final_item_context = parser_context.with_additional_node(&final_item_context);
-        let (final_child_start, _original_final_child) = children
-            .pop()
-            .expect("if-statement already checked that children was non-empty.");
-        let (remain, reparsed_final_element) = include_input(parser_with_context!(element(true))(
-            &final_item_context,
-        ))(final_child_start)?;
-        remaining = remain;
-        children.push(reparsed_final_element);
-    }
-
+    // We have to use the parser_context here to include the whitespace cut-off
     let (remaining, _trailing_ws) =
-        maybe_consume_trailing_whitespace_if_not_exiting(context, remaining)?;
+        maybe_consume_trailing_whitespace_if_not_exiting(&final_whitespace_context, remaining)?;
 
     let source = get_consumed(input, remaining);
     return Ok((
@@ -322,7 +355,7 @@ fn bullet<'b, 'g, 'r, 's>(
             map(tag("+"), |bull| (BulletType::Unordered, bull)),
             map(
                 recognize(tuple((
-                    parser_with_context!(counter)(context),
+                    bind_context!(counter, context),
                     alt((tag("."), tag(")"))),
                 ))),
                 |bull| (BulletType::Ordered, bull),
@@ -377,6 +410,52 @@ fn counter_set_value<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PlainListIt
     ))(input)
 }
 
+const fn final_item_whitespace_cutoff(indent_level: IndentationLevel) -> impl ContextMatcher {
+    move |context, input: OrgSource<'_>| {
+        impl_final_item_whitespace_cutoff(context, input, indent_level)
+    }
+}
+
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
+fn impl_final_item_whitespace_cutoff<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+    indent_level: IndentationLevel,
+) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    start_of_line(input)?;
+    // element!(plain_list_end, context, input);
+
+    if let Ok((_remaining, _)) = verify(
+        tuple((
+            opt(blank_line),
+            bind_context!(indentation_level, context),
+            not(multispace1),
+        )),
+        |(_, (depth, _stars), _not_whitespace)| *depth < indent_level,
+    )(input)
+    {
+        return Ok((input, input.take(0)));
+    }
+
+    if let Ok((_remaining, _)) = tuple((
+        opt(blank_line),
+        verify(
+            bind_context!(detect_not_plain_list_item_indent, context),
+            |(depth, _)| *depth == indent_level,
+        ),
+    ))(input)
+    {
+        return Ok((input, input.take(0)));
+    }
+
+    Err(nom::Err::Error(CustomError::Static(
+        "No whitespace cut-off.",
+    )))
+}
+
 #[cfg_attr(
     feature = "tracing",
     tracing::instrument(ret, level = "debug", skip(_context))
@@ -412,7 +491,7 @@ fn _plain_list_item_end<'b, 'g, 'r, 's>(
     start_of_line(input)?;
     recognize(tuple((
         opt(blank_line),
-        parser_with_context!(line_indented_lte_matcher)(context),
+        bind_context!(line_indented_lte_matcher, context),
     )))(input)
 }
 
@@ -431,7 +510,7 @@ fn _line_indented_lte<'b, 'g, 'r, 's>(
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
     let matched = recognize(verify(
         tuple((
-            parser_with_context!(indentation_level)(context),
+            bind_context!(indentation_level, context),
             non_whitespace_character,
         )),
         // 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)
@@ -457,8 +536,8 @@ fn item_tag<'b, 'g, 'r, 's>(
     let (remaining, (children, _exit_contents)) = verify(
         many_till(
             // TODO: Should this be using a different set like the minimal set?
-            parser_with_context!(standard_set_object)(&parser_context),
-            parser_with_context!(exit_matcher_parser)(&parser_context),
+            bind_context!(standard_set_object, &parser_context),
+            bind_context!(exit_matcher_parser, &parser_context),
         ),
         |(children, _exit_contents)| !children.is_empty(),
     )(input)?;
@@ -508,7 +587,7 @@ fn item_tag_post_gap<'b, 'g, 'r, 's>(
                 alt((
                     peek(recognize(not(blank_line))),
                     peek(recognize(tuple((many0(blank_line), eof)))),
-                    parser_with_context!(exit_matcher_parser)(context),
+                    bind_context!(exit_matcher_parser, context),
                 )),
             ),
         ))),
@@ -538,7 +617,7 @@ fn detect_contentless_item_contents<'b, 'g, 'r, 's>(
 ) -> Res<OrgSource<'s>, ()> {
     let (remaining, _) = recognize(many_till(
         blank_line,
-        parser_with_context!(exit_matcher_parser)(context),
+        bind_context!(exit_matcher_parser, context),
     ))(input)?;
     Ok((remaining, ()))
 }

src/parser/util.rs

@@ -243,6 +243,10 @@ pub(crate) fn org_line_ending(input: OrgSource<'_>) -> Res<OrgSource<'_>, OrgSou
 }
 
 /// Match the whitespace at the beginning of a line and give it an indentation level.
+#[cfg_attr(
+    feature = "tracing",
+    tracing::instrument(ret, level = "debug", skip(context))
+)]
 pub(crate) fn indentation_level<'s>(
     context: RefContext<'_, '_, '_, 's>,
     input: OrgSource<'s>,

src/types/util.rs

@@ -515,4 +515,25 @@ mod tests {
         assert!(matches!(output, Cow::Borrowed(_)));
         Ok(())
     }
+
+    #[test]
+    fn test_coalesce_whitespace_if_line_break() -> Result<(), Box<dyn std::error::Error>> {
+        assert_eq!(coalesce_whitespace_if_line_break("foo  bar"), "foo  bar");
+        assert_eq!(coalesce_whitespace_if_line_break("foo \n bar"), "foo bar");
+        Ok(())
+    }
+
+    #[test]
+    fn test_remove_whitespace_if_line_break() -> Result<(), Box<dyn std::error::Error>> {
+        assert_eq!(remove_whitespace_if_line_break("foo  bar"), "foo  bar");
+        assert_eq!(remove_whitespace_if_line_break("foo \n bar"), "foobar");
+        Ok(())
+    }
+
+    #[test]
+    fn test_remove_line_break() -> Result<(), Box<dyn std::error::Error>> {
+        assert_eq!(remove_line_break("foo  bar"), "foo  bar");
+        assert_eq!(remove_line_break("foo \n bar"), "foo  bar");
+        Ok(())
+    }
 }