Hashtag parsing

magnetar_mmm_parser/src/lib.rs

@@ -68,6 +68,7 @@ pub enum Token<'a> {
     },
     UnicodeEmoji(Cow<'a, str>),
     ShortcodeEmoji(Cow<'a, str>),
+    Hashtag(Cow<'a, str>),
 }
 
 impl Token<'_> {
@@ -128,6 +129,7 @@ impl Token<'_> {
             Token::ShortcodeEmoji(shortcode) => {
                 Token::ShortcodeEmoji(Cow::Owned(shortcode.clone().into_owned()))
             }
+            Token::Hashtag(url) => Token::Hashtag(Cow::Owned(url.clone().into_owned())),
         }
     }
 }
@@ -703,6 +705,33 @@ impl Context {
             },
         ))
     }
+
+    fn hashtag<'a>(&self, input: Span<'a>) -> IResult<Span<'a>, Token<'a>> {
+        // TODO: Skip when preceded by alphanumerics
+
+        let (input, _) = tag("#")(input)?;
+
+        let (input, hashtag_text) =
+            map(recognize(many1(hashtag_chars)), Span::into_fragment)(input)?;
+
+        Ok((input, Token::Hashtag(hashtag_text.into())))
+    }
+}
+
+#[inline]
+fn hashtag_chars(input: Span) -> IResult<Span, Span> {
+    recognize(alt((
+        recognize(tuple((tag("("), hashtag_chars, tag(")")))),
+        recognize(tuple((tag("["), hashtag_chars, tag("]")))),
+        recognize(tuple((tag("「"), hashtag_chars, tag("」")))),
+        recognize(tuple((tag("（"), hashtag_chars, tag("）")))),
+        recognize(tuple((
+            not(space1),
+            not_line_ending,
+            not(one_of(".,:;!?#?/[]【】()「」（）<>")),
+            anychar,
+        ))),
+    )))(input)
 }
 
 #[inline]