Merge branch '0.1.1' into main

.github/actions-rs/grcov.yml

@@ -9,3 +9,4 @@ ignore:
   - "src/ui/*"
   - src/main.rs
   - src/helpers/open.rs
+  - src/helpers/strings/lookup.rs

CHANGELOG.md

@@ -1,10 +1,17 @@
 # Changelog
 
 - [Changelog](#changelog)
+  - [0.1.1](#011)
   - [0.1.0](#010)
 
 ---
 
+## 0.1.1
+
+Released on 17/11/2021
+
+- Escape HTML entities from text (e.g. `&` or `“`)
+
 ## 0.1.0
 
 Released on 15/11/2021

Cargo.toml

@@ -11,7 +11,7 @@ license = "MIT"
 name = "tuifeed"
 readme = "README.md"
 repository = "https://github.com/veeso/tuifeed"
-version = "0.1.0"
+version = "0.1.1"
 
 [[bin]]
 name = "tuifeed"

README.md

@@ -1,7 +1,7 @@
 # tuifeed
 
 <p align="center">Developed by <a href="https://veeso.github.io/" target="_blank">@veeso</a></p>
-<p align="center">Current version: 0.1.0 (15/11/2021)</p>
+<p align="center">Current version: 0.1.1 (17/11/2021)</p>
 <p align="center">~ A terminal news feed reader with a fancy ui ~</p>
 
 <p align="center">

install.sh

@@ -8,7 +8,7 @@
 #   -f, -y, --force, --yes
 #     Skip the confirmation prompt during installation
 
-TUIFEED_VERSION="0.1.0"
+TUIFEED_VERSION="0.1.1"
 GITHUB_URL="https://github.com/veeso/tuifeed/releases/download/v${TUIFEED_VERSION}"
 DEB_URL="${GITHUB_URL}/tuifeed_${TUIFEED_VERSION}_amd64.deb"
 RPM_URL="${GITHUB_URL}/tuifeed-${TUIFEED_VERSION}-1.x86_64.rpm"

src/feed/mod.rs

@@ -82,11 +82,13 @@ impl From<RssFeed> for Feed {
 impl From<RssEntry> for Article {
     fn from(entry: RssEntry) -> Self {
         Self {
-            title: entry.title.map(|x| x.content),
+            title: entry
+                .title
+                .map(|x| str_helpers::strip_html(x.content.as_str())),
             authors: entry.authors.into_iter().map(|x| x.name).collect(),
             summary: entry
                 .summary
-                .map(|x| str_helpers::strip_html_tags(x.content.as_str()))
+                .map(|x| str_helpers::strip_html(x.content.as_str()))
                 .unwrap_or_default(),
             url: entry
                 .links

src/helpers/strings.rs → src/helpers/strings/mod.rs

@@ -25,11 +25,20 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
+mod lookup;
+
 use regex::Regex;
 use unicode_truncate::UnicodeTruncateStr;
 
 lazy_static! {
     static ref HTML_TAG_REGEX: Regex = Regex::new(r"<[^>]+>").unwrap();
+    /**
+     * Matches HTML entities in string
+     *
+     * - group 2: decimal (maybe)
+     * - group 3: literal (e.g. amp, gt, ...) (maybe)
+     */
+    static ref HTML_ENTITIES_REGEX: Regex = Regex::new(r"&(#([0-9]+))?([a-z]+)?;").unwrap();
     static ref REPEATED_NEWLINES_REGEX: Regex = Regex::new(r"(\r?\n|\r)\d*(\r?\n|\r)").unwrap();
 }
 
@@ -51,18 +60,31 @@ pub fn replace_multiple_newlines(s: &str, with: &str) -> String {
     REPEATED_NEWLINES_REGEX.replace_all(s, with).to_string()
 }
 
-/// strip_html_tags
+/// strip_html
 ///
-/// Strip html tags from string
-pub fn strip_html_tags(s: &str) -> String {
-    HTML_TAG_REGEX
-        .replace_all(s, "")
-        .to_string()
-        .replace("&nbsp;", " ")
-        .replace("&amp;", "&")
-        .replace("&lt;", "<")
-        .replace("&gt;", ">")
-        .replace("&quot;", "\"")
+/// Strip html tags and entities from string
+pub fn strip_html(s: &str) -> String {
+    let mut escaped = HTML_TAG_REGEX.replace_all(s, "").to_string();
+    let copy = escaped.clone();
+    for group in HTML_ENTITIES_REGEX.captures_iter(copy.as_str()) {
+        if let Some(mtch) = group.get(2) {
+            // Convert mtch to u32
+            let replace_with = match mtch.as_str().parse::<u32>() {
+                Err(_) => '�',
+                Ok(val) => char::from_u32(val).unwrap_or('�'),
+            };
+            // Get char from decimal
+            escaped = escaped.replace(&group[0], replace_with.to_string().as_str());
+        } else if let Some(mtch) = group.get(3) {
+            let replace_with = lookup::HTML_ENTITIES_TABLE
+                .iter()
+                .find(|(repr, _)| *repr == mtch.as_str())
+                .map(|(_, code)| code)
+                .unwrap_or(&"�");
+            escaped = escaped.replace(&group[0], replace_with);
+        }
+    }
+    escaped
 }
 
 #[cfg(test)]
@@ -94,13 +116,17 @@ mod test {
     }
 
     #[test]
-    fn should_strip_html_tags() {
+    fn should_strip_html() {
         assert_eq!(
-            strip_html_tags(
+            strip_html(
                 r#"<p><img src="https://images2.corriereobjects.it/methode_image/2021/11/09/Cultura/Foto%20Cltura%20-%20Trattate/Il%20salvataggio%20delle%20vacche%[email protected]" title="Polesine, novembre 1951,settant’anni fa l’alluvione che travolse il Veneto" alt="Polesine, novembre 1951 />Hello</p> World!"#
             ),
             "Hello World!"
         );
+        assert_eq!(
+            strip_html(r#"Hello, &lt;World&gt;&#33;"#),
+            "Hello, <World>!"
+        );
     }
 
     #[test]