Ignore mentions and hashtags inside code blocks

src/models/posts/hashtags.rs

@@ -2,6 +2,7 @@ use regex::{Captures, Regex};
 
 use crate::errors::ValidationError;
 use crate::frontend::get_tag_page_url;
+use super::links::is_inside_code_block;
 
 const HASHTAG_RE: &str = r"(?m)(?P<before>^|\s|>|[\(])#(?P<tag>[^\s<]+)";
 const HASHTAG_SECONDARY_RE: &str = r"^(?P<tag>[0-9A-Za-z]+)(?P<after>[\.,:?\)]?)$";
@@ -13,6 +14,11 @@ pub fn find_hashtags(text: &str) -> Vec<String> {
     let hashtag_secondary_re = Regex::new(HASHTAG_SECONDARY_RE).unwrap();
     let mut tags = vec![];
     for caps in hashtag_re.captures_iter(text) {
+        let tag_match = caps.name("tag").expect("should have tag group");
+        if is_inside_code_block(&tag_match, text) {
+            // Ignore hashtags inside code blocks
+            continue;
+        };
         if let Some(secondary_caps) = hashtag_secondary_re.captures(&caps["tag"]) {
             let tag_name = secondary_caps["tag"].to_string().to_lowercase();
             if !tags.contains(&tag_name) {
@@ -28,6 +34,11 @@ pub fn replace_hashtags(instance_url: &str, text: &str, tags: &[String]) -> Stri
     let hashtag_re = Regex::new(HASHTAG_RE).unwrap();
     let hashtag_secondary_re = Regex::new(HASHTAG_SECONDARY_RE).unwrap();
     let result = hashtag_re.replace_all(text, |caps: &Captures| {
+        let tag_match = caps.name("tag").expect("should have tag group");
+        if is_inside_code_block(&tag_match, text) {
+            // Don't replace hashtags inside code blocks
+            return caps[0].to_string();
+        };
         if let Some(secondary_caps) = hashtag_secondary_re.captures(&caps["tag"]) {
             let before = caps["before"].to_string();
             let tag = secondary_caps["tag"].to_string();

src/models/posts/links.rs

@@ -1,6 +1,6 @@
 use std::collections::HashMap;
 
-use regex::{Captures, Regex};
+use regex::{Captures, Match, Regex};
 use tokio_postgres::GenericClient;
 
 use crate::database::DatabaseError;
@@ -10,10 +10,10 @@ use super::types::Post;
 // MediaWiki-like syntax: [[url|text]]
 const OBJECT_LINK_SEARCH_RE: &str = r"(?m)\[\[(?P<url>[^\s\|]+)(\|(?P<text>.+?))?\]\]";
 
-fn is_inside_code_block(caps: &Captures, text: &str) -> bool {
+pub fn is_inside_code_block(match_: &Match, text: &str) -> bool {
     // TODO: remove workaround.
     // Perform replacement only inside text nodes during markdown parsing
-    let text_before = &text[0..caps.name("url").unwrap().start()];
+    let text_before = &text[0..match_.start()];
     let code_open = text_before.matches("<code>").count();
     let code_closed = text_before.matches("</code>").count();
     code_open > code_closed
@@ -24,10 +24,12 @@ fn find_object_links(text: &str) -> Vec<String> {
     let link_re = Regex::new(OBJECT_LINK_SEARCH_RE).unwrap();
     let mut links = vec![];
     for caps in link_re.captures_iter(text) {
-        let url = caps["url"].to_string();
-        if is_inside_code_block(&caps, text) {
+        let url_match = caps.name("url").expect("should have url group");
+        if is_inside_code_block(&url_match, text) {
+            // Ignore links inside code blocks
             continue;
         };
+        let url = caps["url"].to_string();
         if !links.contains(&url) {
             links.push(url);
         };
@@ -60,12 +62,17 @@ pub fn replace_object_links(
 ) -> String {
     let mention_re = Regex::new(OBJECT_LINK_SEARCH_RE).unwrap();
     let result = mention_re.replace_all(text, |caps: &Captures| {
+        let url_match = caps.name("url").expect("should have url group");
+        if is_inside_code_block(&url_match, text) {
+            // Don't replace inside code blocks
+            return caps[0].to_string();
+        };
         let url = caps["url"].to_string();
         let link_text = caps.name("text")
             .map(|match_| match_.as_str())
             .unwrap_or(&url)
             .to_string();
-        if link_map.contains_key(&url) && !is_inside_code_block(caps, text) {
+        if link_map.contains_key(&url) {
             return format!(r#"<a href="{0}">{1}</a>"#, url, link_text);
         };
         // Leave unchanged if post does not exist
@@ -84,6 +91,16 @@ mod tests {
         "test ([[https://example.org/2]])",
     );
 
+    #[test]
+    fn test_is_inside_code_block() {
+        let text = "abc<code>&&</code>xyz";
+        let regexp = Regex::new("&&").unwrap();
+        let mat = regexp.find(text).unwrap();
+        assert_eq!(mat.start(), 9);
+        let result = is_inside_code_block(&mat, text);
+        assert_eq!(result, true);
+    }
+
     #[test]
     fn test_find_object_links() {
         let results = find_object_links(TEXT_WITH_OBJECT_LINKS);

src/models/posts/mentions.rs

@@ -8,6 +8,7 @@ use crate::database::DatabaseError;
 use crate::errors::ValidationError;
 use crate::models::profiles::queries::get_profiles_by_accts;
 use crate::models::profiles::types::DbActorProfile;
+use super::links::is_inside_code_block;
 
 // See also: ACTOR_ADDRESS_RE in activitypub::actors::types
 const MENTION_RE: &str = r"@?(?P<username>[\w\.-]+)@(?P<hostname>.+)";
@@ -23,6 +24,11 @@ fn find_mentions(
     let mention_secondary_re = Regex::new(MENTION_SEARCH_SECONDARY_RE).unwrap();
     let mut mentions = vec![];
     for caps in mention_re.captures_iter(text) {
+        let mention_match = caps.name("mention").expect("should have mention group");
+        if is_inside_code_block(&mention_match, text) {
+            // No mentions inside code blocks
+            continue;
+        };
         if let Some(secondary_caps) = mention_secondary_re.captures(&caps["mention"]) {
             let username = secondary_caps["username"].to_string();
             let hostname = secondary_caps.name("hostname")
@@ -62,6 +68,11 @@ pub fn replace_mentions(
     let mention_re = Regex::new(MENTION_SEARCH_RE).unwrap();
     let mention_secondary_re = Regex::new(MENTION_SEARCH_SECONDARY_RE).unwrap();
     let result = mention_re.replace_all(text, |caps: &Captures| {
+        let mention_match = caps.name("mention").expect("should have mention group");
+        if is_inside_code_block(&mention_match, text) {
+            // Don't replace mentions inside code blocks
+            return caps[0].to_string();
+        };
         if let Some(secondary_caps) = mention_secondary_re.captures(&caps["mention"]) {
             let username = secondary_caps["username"].to_string();
             let hostname = secondary_caps.name("hostname")