Merge branch 'main'

* main: chore: Release xml-rs version 0.8.18 Replace disallowed characters
netvl · Sep 8, 2023 · 15332bd · 15332bd
2 parents 16348a4 + 617a11e
commit 15332bd
Show file tree

Hide file tree

Showing 3 changed files with 63 additions and 1 deletion.
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "xml-rs"
-version = "0.8.17"
+version = "0.8.18"
 authors = ["Vladimir Matveev <vmatveev@citrine.cc>"]
 license = "MIT"
 description = "An XML library in pure Rust"

diff --git a/src/reader/parser/inside_reference.rs b/src/reader/parser/inside_reference.rs
@@ -68,6 +68,7 @@ impl PullParser {
         };
         match char::from_u32(val) {
             Some(c) if self.is_valid_xml_char(c) => Ok(c),
+            Some(_) if self.config.c.replace_unknown_entity_references => Ok('\u{fffd}'),
             None if self.config.c.replace_unknown_entity_references => {
                 Ok('\u{fffd}')
             },

diff --git a/tests/event_reader.rs b/tests/event_reader.rs
@@ -642,6 +642,67 @@ fn issue_replacement_character_entity_reference() {
     );
 }
 
+#[test]
+fn issue_replacement_character_control_character() {
+    test(
+        br#"<doc>&#16;</doc>"#,
+        br#"
+            |StartDocument(1.0, UTF-8)
+            |StartElement(doc)
+            |1:10 Invalid character U+0010
+        "#,
+        ParserConfig::new(),
+        false,
+    );
+
+    test(
+        br#"<doc>&#x10;</doc>"#,
+        br#"
+            |StartDocument(1.0, UTF-8)
+            |StartElement(doc)
+            |1:11 Invalid character U+0010
+        "#,
+        ParserConfig::new(),
+        false,
+    );
+
+    test(
+        br#"<doc>&#16;</doc>"#,
+        format!(
+            r#"
+                |StartDocument(1.0, UTF-8)
+                |StartElement(doc)
+                |Characters("{replacement_character}")
+                |EndElement(doc)
+                |EndDocument
+            "#,
+            replacement_character = "\u{fffd}"
+        )
+        .as_bytes(),
+        ParserConfig::new()
+            .replace_unknown_entity_references(true),
+        false,
+    );
+
+    test(
+        br#"<doc>&#x10;</doc>"#,
+        format!(
+            r#"
+                |StartDocument(1.0, UTF-8)
+                |StartElement(doc)
+                |Characters("{replacement_character}")
+                |EndElement(doc)
+                |EndDocument
+            "#,
+            replacement_character = "\u{fffd}"
+        )
+        .as_bytes(),
+        ParserConfig::new()
+            .replace_unknown_entity_references(true),
+        false,
+    );
+}
+
 #[test]
 fn push_pos_issue() {
     let source = "<n><!---->L<!----><!----><!----><!----><!----><!----><!----><!----><!---->\"<!----><!---->L<!----><!----></n>";