From 56865c68fc8633df7425720389cac01367438001 Mon Sep 17 00:00:00 2001
From: Tom Alexander <tom@fizz.buzz>
Date: Tue, 29 Aug 2023 15:44:04 -0400
Subject: [PATCH] Do not allow plain links without a path.

---
 .../object/plain_link/empty_links.org         | 52 +++++++++++++
 src/parser/entity.rs                          |  6 +-
 src/parser/plain_link.rs                      | 76 +++++++++++--------
 3 files changed, 100 insertions(+), 34 deletions(-)
 create mode 100644 org_mode_samples/object/plain_link/empty_links.org

diff --git a/org_mode_samples/object/plain_link/empty_links.org b/org_mode_samples/object/plain_link/empty_links.org
new file mode 100644
index 00000000..73c94a8d
--- /dev/null
+++ b/org_mode_samples/object/plain_link/empty_links.org
@@ -0,0 +1,52 @@
+non-link text
+eww://
+rmail://
+mhe://
+irc://
+info://
+gnus://
+docview://
+bibtex://
+bbdb://
+w3m://
+doi://
+file+sys://
+file+emacs://
+shell://
+news://
+mailto://
+https://
+http://
+ftp://
+help://
+file://
+elisp://
+randomfakeprotocl://
+non-link text
+
+
+non-link text
+eww:
+rmail:
+mhe:
+irc:
+info:
+gnus:
+docview:
+bibtex:
+bbdb:
+w3m:
+doi:
+file+sys:
+file+emacs:
+shell:
+news:
+mailto:
+https:
+http:
+ftp:
+help:
+file:
+elisp:
+randomfakeprotocl:
+non-link text
diff --git a/src/parser/entity.rs b/src/parser/entity.rs
index 8bd7186e..7621e59f 100644
--- a/src/parser/entity.rs
+++ b/src/parser/entity.rs
@@ -15,7 +15,8 @@ use crate::error::Res;
 use crate::parser::object::Entity;
 use crate::parser::util::get_consumed;
 
-const ENTITIES: [&'static str; 413] = [
+// TODO: Make this a user-provided variable corresponding to elisp's org-entities
+const ORG_ENTITIES: [&'static str; 413] = [
     "Agrave",
     "agrave",
     "Aacute",
@@ -457,8 +458,7 @@ fn name<'r, 's>(
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
     // TODO: This should be defined by org-entities and optionally org-entities-user
-    for entity in ENTITIES {
-        // foo
+    for entity in ORG_ENTITIES {
         let result = tag_no_case::<_, _, CustomError<_>>(entity)(input);
         match result {
             Ok((remaining, ent)) => {
diff --git a/src/parser/plain_link.rs b/src/parser/plain_link.rs
index 58617ea6..8518ae6c 100644
--- a/src/parser/plain_link.rs
+++ b/src/parser/plain_link.rs
@@ -7,6 +7,7 @@ use nom::character::complete::one_of;
 use nom::combinator::eof;
 use nom::combinator::peek;
 use nom::combinator::recognize;
+use nom::combinator::verify;
 use nom::multi::many_till;
 
 use super::org_source::OrgSource;
@@ -23,6 +24,33 @@ use crate::parser::util::exit_matcher_parser;
 use crate::parser::util::get_consumed;
 use crate::parser::util::WORD_CONSTITUENT_CHARACTERS;
 
+// TODO: Make this a user-provided variable corresponding to elisp's org-link-parameters
+const ORG_LINK_PARAMETERS: [&'static str; 23] = [
+    "id",
+    "eww",
+    "rmail",
+    "mhe",
+    "irc",
+    "info",
+    "gnus",
+    "docview",
+    "bibtex",
+    "bbdb",
+    "w3m",
+    "doi",
+    "file+sys",
+    "file+emacs",
+    "shell",
+    "news",
+    "mailto",
+    "https",
+    "http",
+    "ftp",
+    "help",
+    "file",
+    "elisp",
+];
+
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
 pub fn plain_link<'r, 's>(
     context: Context<'r, 's>,
@@ -73,36 +101,19 @@ pub fn protocol<'r, 's>(
     input: OrgSource<'s>,
 ) -> Res<OrgSource<'s>, OrgSource<'s>> {
     // TODO: This should be defined by org-link-parameters
-    let (remaining, proto) = alt((
-        alt((
-            tag_no_case("id"),
-            tag_no_case("eww"),
-            tag_no_case("rmail"),
-            tag_no_case("mhe"),
-            tag_no_case("irc"),
-            tag_no_case("info"),
-            tag_no_case("gnus"),
-            tag_no_case("docview"),
-            tag_no_case("bibtex"),
-            tag_no_case("bbdb"),
-            tag_no_case("w3m"),
-        )),
-        alt((
-            tag_no_case("doi"),
-            tag_no_case("file+sys"),
-            tag_no_case("file+emacs"),
-            tag_no_case("shell"),
-            tag_no_case("news"),
-            tag_no_case("mailto"),
-            tag_no_case("https"),
-            tag_no_case("http"),
-            tag_no_case("ftp"),
-            tag_no_case("help"),
-            tag_no_case("file"),
-            tag_no_case("elisp"),
-        )),
-    ))(input)?;
-    Ok((remaining, proto))
+    for link_parameter in ORG_LINK_PARAMETERS {
+        let result = tag_no_case::<_, _, CustomError<_>>(link_parameter)(input);
+        match result {
+            Ok((remaining, ent)) => {
+                return Ok((remaining, ent));
+            }
+            Err(_) => {}
+        }
+    }
+
+    Err(nom::Err::Error(CustomError::MyError(MyError(
+        "NoLinkProtocol".into(),
+    ))))
 }
 
 #[cfg_attr(feature = "tracing", tracing::instrument(ret, level = "debug"))]
@@ -119,7 +130,10 @@ fn path_plain<'r, 's>(
 
     let exit_matcher = parser_with_context!(exit_matcher_parser)(&parser_context);
 
-    let (remaining, path) = recognize(many_till(anychar, peek(exit_matcher)))(input)?;
+    let (remaining, path) = recognize(verify(
+        many_till(anychar, peek(exit_matcher)),
+        |(children, _exit_contents)| !children.is_empty(),
+    ))(input)?;
     Ok((remaining, path))
 }