Handle puncuation-only search options.

org_mode_samples/object/plain_link/all_default_links_with_search_option.org

@@ -0,0 +1,25 @@
+non-link text::foo
+eww://foo::foo
+rmail://foo::foo
+mhe://foo::foo
+irc://foo::foo
+info://foo::foo
+gnus://foo::foo
+docview://foo::foo
+bibtex://foo::foo
+bbdb://foo::foo
+w3m://foo::foo
+doi://foo::foo
+file+sys://foo::foo
+file+emacs://foo::foo
+shell://foo::foo
+news://foo::foo
+mailto://foo::foo
+https://foo::foo
+http://foo::foo
+ftp://foo::foo
+help://foo::foo
+file://foo::foo
+elisp://foo::foo
+randomfakeprotocl://foo::foo
+non-link text::foo

org_mode_samples/object/plain_link/relative_path.org

@@ -0,0 +1,4 @@
+# These do not become links
+./simple.org::foo
+../simple.org::foo
+/simple.org::foo

org_mode_samples/object/plain_link/search_option.org

@@ -0,0 +1,16 @@
+file:simple.org::foo
+
+file:simple.org::#foo
+file:simple.org::foo  bar
+file:simple.org::foo
+bar
+file:simple.org::foo::bar
+file:simple.org::/foo/
+
+# Does not become a search option because it is inside parenthesis.
+https://en.wikipedia.org/wiki/Shebang_(Uni::x)
+
+
+file:simple.org::* foo
+file:simple.org::*bar
+file:simple.org::b*az

org_mode_samples/object/plain_link/template.org

@@ -0,0 +1,8 @@
+# None of these end up as links.
+
+#+LINK:       foo https://foo.bar/baz#%s
+foo::lorem
+
+cat::bat
+#+LINK:       cat https://dog%s
+cat:bat

org_mode_samples/object/regular_link/application.org

@@ -0,0 +1,2 @@
+[[file+sys://foo]]
+[[file+emacs://foo]]

org_mode_samples/object/regular_link/search_option.org

@@ -0,0 +1,2 @@
+# Does not become a search option because it is inside parenthesis.
+[[https://en.wikipedia.org/wiki/Shebang_(Uni::x)]]

src/compare/diff.rs

@@ -2802,8 +2802,8 @@ fn compare_regular_link<'b, 's>(
         ),
         (
             EmacsField::Required(":application"),
-            compare_identity,
-            compare_property_always_nil
+            |r| r.application.as_ref(),
+            compare_property_quoted_string
         ),
         (
             EmacsField::Required(":search-option"),
@@ -2860,7 +2860,7 @@ fn compare_radio_link<'b, 's>(
         ),
         (
             EmacsField::Required(":raw-link"),
-            |r| Some(&r.raw_link),
+            |r| Some(r.get_raw_link()),
             compare_property_quoted_string
         ),
         (
@@ -2936,10 +2936,55 @@ fn compare_plain_link<'b, 's>(
     emacs: &'b Token<'s>,
     rust: &'b PlainLink<'s>,
 ) -> Result<DiffEntry<'b, 's>, Box<dyn std::error::Error>> {
-    let this_status = DiffStatus::Good;
-    let message = None;
+    let mut this_status = DiffStatus::Good;
+    let mut message = None;
 
-    // TODO: Compare :type :path :format :raw-link :application :search-option
+    if let Some((new_status, new_message)) = compare_properties!(
+        emacs,
+        rust,
+        (
+            EmacsField::Required(":type"),
+            |r| {
+                match &r.link_type {
+                    LinkType::File => Some(Cow::Borrowed("file")),
+                    LinkType::Protocol(protocol) => Some(protocol.clone()),
+                    LinkType::Id => Some(Cow::Borrowed("id")),
+                    LinkType::CustomId => Some(Cow::Borrowed("custom-id")),
+                    LinkType::CodeRef => Some(Cow::Borrowed("coderef")),
+                    LinkType::Fuzzy => Some(Cow::Borrowed("fuzzy")),
+                }
+            },
+            compare_property_quoted_string
+        ),
+        (
+            EmacsField::Required(":path"),
+            |r| Some(r.path),
+            compare_property_quoted_string
+        ),
+        (
+            EmacsField::Required(":format"),
+            |_| Some("plain"),
+            compare_property_unquoted_atom
+        ),
+        (
+            EmacsField::Required(":raw-link"),
+            |r| Some(r.raw_link),
+            compare_property_quoted_string
+        ),
+        (
+            EmacsField::Required(":application"),
+            |r| r.application,
+            compare_property_quoted_string
+        ),
+        (
+            EmacsField::Required(":search-option"),
+            |r| r.search_option,
+            compare_property_quoted_string
+        )
+    )? {
+        this_status = new_status;
+        message = new_message;
+    }
 
     Ok(DiffResult {
         status: this_status,

src/parser/plain_link.rs

@@ -1,13 +1,19 @@
 use nom::branch::alt;
+use nom::bytes::complete::is_not;
 use nom::bytes::complete::tag;
 use nom::bytes::complete::tag_no_case;
 use nom::character::complete::anychar;
 use nom::character::complete::none_of;
 use nom::character::complete::one_of;
+use nom::combinator::consumed;
 use nom::combinator::eof;
+use nom::combinator::map;
+use nom::combinator::map_parser;
 use nom::combinator::not;
+use nom::combinator::opt;
 use nom::combinator::peek;
 use nom::combinator::recognize;
+use nom::combinator::rest;
 use nom::combinator::verify;
 use nom::multi::many0;
 use nom::multi::many1;
@@ -30,6 +36,7 @@ use crate::error::Res;
 use crate::parser::util::exit_matcher_parser;
 use crate::parser::util::get_consumed;
 use crate::parser::util::WORD_CONSTITUENT_CHARACTERS;
+use crate::types::LinkType;
 use crate::types::PlainLink;
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
@@ -38,9 +45,7 @@ pub(crate) fn plain_link<'b, 'g, 'r, 's>(
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, PlainLink<'s>> {
     let (remaining, _) = pre(context, input)?;
-    let (remaining, proto) = protocol(context, remaining)?;
-    let (remaining, _separator) = tag(":")(remaining)?;
-    let (remaining, path) = path_plain(context, remaining)?;
+    let (remaining, path_plain) = parse_path_plain(context, remaining)?;
     peek(parser_with_context!(post)(context))(remaining)?;
     let (remaining, _trailing_whitespace) =
         maybe_consume_object_trailing_whitespace_if_not_exiting(context, remaining)?;
@@ -49,12 +54,24 @@ pub(crate) fn plain_link<'b, 'g, 'r, 's>(
         remaining,
         PlainLink {
             source: source.into(),
-            link_type: proto.into(),
-            path: path.into(),
+            link_type: path_plain.link_type,
+            path: path_plain.path,
+            raw_link: path_plain.raw_link,
+            search_option: path_plain.search_option,
+            application: path_plain.application,
         },
     ))
 }
 
+#[derive(Debug)]
+struct PathPlain<'s> {
+    link_type: LinkType<'s>,
+    path: &'s str,
+    raw_link: &'s str,
+    search_option: Option<&'s str>,
+    application: Option<&'s str>,
+}
+
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
 fn pre<'b, 'g, 'r, 's>(
     _context: RefContext<'b, 'g, 'r, 's>,
@@ -84,6 +101,95 @@ fn post<'b, 'g, 'r, 's>(
     Ok((remaining, ()))
 }
 
+fn parse_path_plain<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, PathPlain<'s>> {
+    alt((
+        parser_with_context!(file_path_plain)(context),
+        parser_with_context!(protocol_path_plain)(context),
+    ))(input)
+}
+
+pub(crate) fn parse_file_and_application<'s>(
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, Option<OrgSource<'s>>> {
+    let (remaining, _) = tag("file")(input)?;
+    let (remaining, application) =
+        opt(map(tuple((tag("+"), rest)), |(_, application)| application))(remaining)?;
+    // Assert we consumed the entire protocol.
+    not(anychar)(remaining)?;
+    Ok((remaining, application))
+}
+
+fn file_path_plain<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, PathPlain<'s>> {
+    let path_plain_end = path_plain_end(input.get_parenthesis_depth(), true);
+    let parser_context = ContextElement::ExitMatcherNode(ExitMatcherNode {
+        class: ExitClass::Gamma,
+        exit_matcher: &path_plain_end,
+    });
+    let parser_context = context.with_additional_node(&parser_context);
+    let (remaining, (raw_link, (application, _, path, search_option))) = consumed(tuple((
+        map_parser(
+            parser_with_context!(protocol)(&parser_context),
+            parse_file_and_application,
+        ),
+        tag(":"),
+        parser_with_context!(path_plain)(&parser_context),
+        opt(map(
+            tuple((
+                tag("::"),
+                verify(is_not(" \t\r\n"), |search_option| {
+                    Into::<&str>::into(search_option)
+                        .chars()
+                        .any(char::is_alphanumeric)
+                }),
+            )),
+            |(_, search_option)| search_option,
+        )),
+    )))(input)?;
+    Ok((
+        remaining,
+        PathPlain {
+            link_type: LinkType::File,
+            path: path.into(),
+            raw_link: raw_link.into(),
+            search_option: search_option.map(Into::<&str>::into),
+            application: application.map(Into::<&str>::into),
+        },
+    ))
+}
+
+fn protocol_path_plain<'b, 'g, 'r, 's>(
+    context: RefContext<'b, 'g, 'r, 's>,
+    input: OrgSource<'s>,
+) -> Res<OrgSource<'s>, PathPlain<'s>> {
+    let path_plain_end = path_plain_end(input.get_parenthesis_depth(), false);
+    let parser_context = ContextElement::ExitMatcherNode(ExitMatcherNode {
+        class: ExitClass::Gamma,
+        exit_matcher: &path_plain_end,
+    });
+    let parser_context = context.with_additional_node(&parser_context);
+    let (remaining, (raw_link, (protocol, _, path))) = consumed(tuple((
+        parser_with_context!(protocol)(&parser_context),
+        tag(":"),
+        parser_with_context!(path_plain)(&parser_context),
+    )))(input)?;
+    Ok((
+        remaining,
+        PathPlain {
+            link_type: LinkType::Protocol(protocol.into()),
+            path: path.into(),
+            raw_link: raw_link.into(),
+            search_option: None,
+            application: None,
+        },
+    ))
+}
+
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
 pub(crate) fn protocol<'b, 'g, 'r, 's>(
     context: RefContext<'b, 'g, 'r, 's>,
@@ -109,24 +215,28 @@ fn path_plain<'b, 'g, 'r, 's>(
     context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
-    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);
-
+    // The caller needs to put an instance of path_plain_end on the context before calling this.
     let (remaining, _components) = many1(alt((
-        parser_with_context!(path_plain_no_parenthesis)(&parser_context),
-        parser_with_context!(path_plain_parenthesis)(&parser_context),
+        parser_with_context!(path_plain_no_parenthesis)(context),
+        parser_with_context!(path_plain_parenthesis)(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)
+fn path_plain_end(
+    starting_parenthesis_depth: BracketDepth,
+    enable_search_option: bool,
+) -> impl ContextMatcher {
+    move |context, input: OrgSource<'_>| {
+        _path_plain_end(
+            context,
+            input,
+            starting_parenthesis_depth,
+            enable_search_option,
+        )
+    }
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
@@ -134,7 +244,16 @@ fn _path_plain_end<'b, 'g, 'r, 's>(
     context: RefContext<'b, 'g, 'r, 's>,
     input: OrgSource<'s>,
     starting_parenthesis_depth: BracketDepth,
+    enable_search_option: bool,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
+    let current_depth = input.get_parenthesis_depth() - starting_parenthesis_depth;
+    if enable_search_option && current_depth == 0 {
+        let search_option = peek(tag("::"))(input);
+        if search_option.is_ok() {
+            return search_option;
+        }
+    }
+
     let (remaining, _leading_punctuation) = many0(verify(anychar, |c| {
         !" \t\r\n[]<>()/".contains(*c) && c.is_ascii_punctuation()
     }))(input)?;
@@ -144,7 +263,6 @@ fn _path_plain_end<'b, 'g, 'r, 's>(
         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);
@@ -161,7 +279,6 @@ fn _path_plain_end<'b, 'g, 'r, 's>(
         }
     }
 
-    // many0 punctuation
     Err(nom::Err::Error(CustomError::MyError(MyError(
         "No path plain end".into(),
     ))))

src/parser/radio_link.rs

@@ -41,7 +41,6 @@ pub(crate) fn radio_link<'b, 'g, 'r, 's>(
                     source: source.into(),
                     children: rematched_target,
                     path: path.into(),
-                    raw_link: path.into(),
                 },
             ));
         }
@@ -194,8 +193,7 @@ mod tests {
             &Object::RadioLink(RadioLink {
                 source: "bar ",
                 children: vec![Object::PlainText(PlainText { source: "bar" })],
-                path: "bar".into(),
-                raw_link: "bar".into()
+                path: "bar".into()
             })
         );
     }
@@ -238,8 +236,7 @@ mod tests {
                     source: "*bar* ",
                     children: vec![Object::PlainText(PlainText { source: "bar" })]
                 })],
-                path: "*bar* ".into(),
-                raw_link: "*bar* ".into()
+                path: "*bar* ".into()
             })
         );
     }

src/parser/regular_link.rs

@@ -69,6 +69,7 @@ fn regular_link_without_description<'b, 'g, 'r, 's>(
             raw_link: path.raw_link,
             search_option: path.search_option,
             children: Vec::new(),
+            application: path.application,
         },
     ))
 }
@@ -95,6 +96,7 @@ fn regular_link_with_description<'b, 'g, 'r, 's>(
             raw_link: path.raw_link,
             search_option: path.search_option,
             children: description,
+            application: path.application,
         },
     ))
 }
@@ -105,6 +107,7 @@ struct PathReg<'s> {
     path: Cow<'s, str>,
     raw_link: Cow<'s, str>,
     search_option: Option<Cow<'s, str>>,
+    application: Option<Cow<'s, str>>,
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
@@ -161,6 +164,7 @@ fn parse_path_reg<'b, 'g, 'r, 's>(
                 path: link.path.into_owned().into(),
                 raw_link: link.raw_link.into_owned().into(),
                 search_option: link.search_option.map(|s| s.into_owned().into()),
+                application: link.application.map(|s| s.into_owned().into()),
             },
         ))
     } else {
@@ -240,6 +244,8 @@ fn apply_link_templates<'b, 'g, 'r, 's>(
 }
 
 fn file_path_reg<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PathReg<'s>> {
+    // TODO: Update this to parse out the application like plain link does.
+    // TODO: Handle parenthesis like plain link does.
     let (remaining, (raw_link, (_, path, search_option))) = consumed(tuple((
         alt((tag("file:"), peek(tag(".")), peek(tag("/")))),
         recognize(many_till(anychar, alt((peek(tag("::")), eof)))),
@@ -256,6 +262,7 @@ fn file_path_reg<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PathReg<'s>> {
             search_option: search_option
                 .map(Into::<&str>::into)
                 .map(Into::<Cow<str>>::into),
+            application: None, // TODO
         },
     ))
 }
@@ -269,6 +276,7 @@ fn id_path_reg<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PathReg<'s>> {
             path: path.into(),
             raw_link: raw_link.into(),
             search_option: None,
+            application: None,
         },
     ))
 }
@@ -282,6 +290,7 @@ fn custom_id_path_reg<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PathReg<'s
             path: path.into(),
             raw_link: raw_link.into(),
             search_option: None,
+            application: None,
         },
     ))
 }
@@ -299,6 +308,7 @@ fn code_ref_path_reg<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PathReg<'s>
             path: path.into(),
             raw_link: raw_link.into(),
             search_option: None,
+            application: None,
         },
     ))
 }
@@ -319,6 +329,7 @@ fn protocol_path_reg<'b, 'g, 'r, 's>(
             path: path.into(),
             raw_link: raw_link.into(),
             search_option: None,
+            application: None,
         },
     ))
 }
@@ -332,6 +343,7 @@ fn fuzzy_path_reg<'s>(input: OrgSource<'s>) -> Res<OrgSource<'s>, PathReg<'s>> {
             path: body.into(),
             raw_link: body.into(),
             search_option: None,
+            application: None,
         },
     ))
 }

src/types/object.rs

@@ -85,6 +85,7 @@ pub struct RegularLink<'s> {
     pub raw_link: Cow<'s, str>,
     pub search_option: Option<Cow<'s, str>>,
     pub children: Vec<Object<'s>>,
+    pub application: Option<Cow<'s, str>>,
 }
 
 #[derive(Debug, PartialEq)]
@@ -98,15 +99,17 @@ pub struct RadioTarget<'s> {
 pub struct RadioLink<'s> {
     pub source: &'s str,
     pub path: &'s str,
-    pub raw_link: &'s str, // TODO: Is this always the same as path? If so, this could be a function instead of a field.
     pub children: Vec<Object<'s>>,
 }
 
 #[derive(Debug, PartialEq)]
 pub struct PlainLink<'s> {
     pub source: &'s str,
-    pub link_type: &'s str,
+    pub link_type: LinkType<'s>,
     pub path: &'s str,
+    pub raw_link: &'s str,
+    pub search_option: Option<&'s str>,
+    pub application: Option<&'s str>,
 }
 
 #[derive(Debug, PartialEq)]
@@ -720,3 +723,9 @@ impl<'s> RegularLink<'s> {
         })
     }
 }
+
+impl<'s> RadioLink<'s> {
+    pub fn get_raw_link(&self) -> &'s str {
+        self.path
+    }
+}