From 4d27e054dad551258ba007c915fa84028cd7b72d Mon Sep 17 00:00:00 2001
From: trivernis <trivernis@protonmail.com>
Date: Thu, 13 Aug 2020 19:07:43 +0200
Subject: [PATCH] Add html character escape codes

---
 Cargo.lock             |  2 +-
 Cargo.toml             |  2 +-
 src/elements/mod.rs    |  6 ++++++
 src/elements/tokens.rs |  4 ++++
 src/format/html.rs     |  7 +++++++
 src/parser/inline.rs   | 16 ++++++++++++++++
 6 files changed, 35 insertions(+), 2 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index ad58059..c33b2b2 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1117,7 +1117,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 
 [[package]]
 name = "snekdown"
-version = "0.21.3"
+version = "0.22.0"
 dependencies = [
  "asciimath-rs 0.5.1 (registry+https://github.com/rust-lang/crates.io-index)",
  "base64 0.12.3 (registry+https://github.com/rust-lang/crates.io-index)",
diff --git a/Cargo.toml b/Cargo.toml
index 41e7831..bb70444 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "snekdown"
-version = "0.21.3"
+version = "0.22.0"
 authors = ["trivernis <trivernis@protonmail.com>"]
 edition = "2018"
 license-file = "LICENSE"
diff --git a/src/elements/mod.rs b/src/elements/mod.rs
index b333f11..aa86622 100644
--- a/src/elements/mod.rs
+++ b/src/elements/mod.rs
@@ -173,6 +173,7 @@ pub enum Inline {
     Math(Math),
     BibReference(Arc<RwLock<BibReference>>),
     TemplateVar(Arc<RwLock<TemplateVariable>>),
+    CharacterCode(CharacterCode),
     LineBreak,
 }
 
@@ -268,6 +269,11 @@ pub struct MathBlock {
     pub(crate) expression: Expression,
 }
 
+#[derive(Clone, Debug)]
+pub struct CharacterCode {
+    pub(crate) code: String,
+}
+
 // implementations
 
 impl Document {
diff --git a/src/elements/tokens.rs b/src/elements/tokens.rs
index 1b717de..4572326 100644
--- a/src/elements/tokens.rs
+++ b/src/elements/tokens.rs
@@ -33,6 +33,7 @@ pub(crate) const L_BRACE: char = '}';
 pub(crate) const PERCENT: char = '%';
 pub(crate) const COMMA: char = ',';
 pub(crate) const MATH: char = '$';
+pub(crate) const AMPERSAND: char = '&';
 
 // aliases
 
@@ -75,6 +76,9 @@ pub(crate) const EMOJI: char = COLON;
 pub(crate) const MATH_INLINE: &'static [char] = &[MATH, MATH];
 pub(crate) const BOLD: [char; 2] = [ASTERISK, ASTERISK];
 
+pub(crate) const CHARACTER_START: char = AMPERSAND;
+pub(crate) const CHARACTER_STOP: char = SEMICOLON;
+
 // groups
 
 pub(crate) const QUOTES: [char; 2] = [SINGLE_QUOTE, DOUBLE_QUOTE];
diff --git a/src/format/html.rs b/src/format/html.rs
index 98acc6a..3b94a58 100644
--- a/src/format/html.rs
+++ b/src/format/html.rs
@@ -68,6 +68,7 @@ impl ToHtml for Inline {
             Inline::TemplateVar(var) => var.read().unwrap().to_html(),
             Inline::Math(m) => m.to_html(),
             Inline::LineBreak => "<br>".to_string(),
+            Inline::CharacterCode(code) => code.to_html(),
         }
     }
 }
@@ -608,3 +609,9 @@ impl ToHtml for TemplateVariable {
         }
     }
 }
+
+impl ToHtml for CharacterCode {
+    fn to_html(&self) -> String {
+        format!("&{};", encode_minimal(self.code.as_str()))
+    }
+}
diff --git a/src/parser/inline.rs b/src/parser/inline.rs
index a47fdb2..58ca8ee 100644
--- a/src/parser/inline.rs
+++ b/src/parser/inline.rs
@@ -31,6 +31,7 @@ pub(crate) trait ParseInline {
     fn parse_metadata_pair(&mut self) -> ParseResult<(String, MetadataValue)>;
     fn parse_placeholder(&mut self) -> ParseResult<Arc<RwLock<Placeholder>>>;
     fn parse_template(&mut self) -> ParseResult<Template>;
+    fn parse_character_code(&mut self) -> ParseResult<CharacterCode>;
 }
 
 impl ParseInline for Parser {
@@ -87,6 +88,8 @@ impl ParseInline for Parser {
             Ok(Inline::BibReference(bibref))
         } else if let Ok(math) = self.parse_math() {
             Ok(Inline::Math(math))
+        } else if let Ok(char_code) = self.parse_character_code() {
+            Ok(Inline::CharacterCode(char_code))
         } else {
             Ok(Inline::Plain(self.parse_plain()?))
         }
@@ -507,4 +510,17 @@ impl ParseInline for Parser {
             variables: vars,
         })
     }
+
+    /// parses a character code &code; like a html character code
+    fn parse_character_code(&mut self) -> ParseResult<CharacterCode> {
+        let start_index = self.ctm.get_index();
+        self.ctm.assert_char(&CHARACTER_START, None)?;
+        self.ctm.seek_one()?;
+        let code =
+            self.ctm
+                .get_string_until_any_or_rewind(&[CHARACTER_STOP], &[LB], start_index)?;
+        self.ctm.seek_one()?;
+
+        Ok(CharacterCode { code })
+    }
 }