Add Attribute.range() method.

Closes #113
RazrFalcon · May 22, 2024 · 6e7293b · 6e7293b
1 parent 0a6e7af
commit 6e7293b
Show file tree

Hide file tree

Showing 5 changed files with 37 additions and 21 deletions.
diff --git a/Cargo.toml b/Cargo.toml
@@ -20,5 +20,5 @@ exclude = ["testing-tools"]
 default = ["std", "positions"]
 std = []
 # Enables Nodes and Attributes position in the original document preserving.
-# Increases memory usage by `usize` for each Node and Attribute.
+# Increases memory usage by `Range<usize>` for each Node and Attribute.
 positions = []
diff --git a/src/lib.rs b/src/lib.rs
@@ -489,7 +489,7 @@ struct AttributeData<'input> {
     name: ExpandedNameIndexed<'input>,
     value: StringStorage<'input>,
     #[cfg(feature = "positions")]
-    pos: usize,
+    range: Range<usize>,
 }
 
 /// An attribute.
@@ -569,10 +569,23 @@ impl<'a, 'input> Attribute<'a, 'input> {
     /// ```
     ///
     /// [Document::text_pos_at]: struct.Document.html#method.text_pos_at
+    #[deprecated(note="replaced by `range`")]
     #[cfg(feature = "positions")]
     #[inline]
     pub fn position(&self) -> usize {
-        self.data.pos
+        self.data.range.start
+    }
+
+    /// Returns attribute's range in bytes in the original document.
+    ///
+    /// ```text
+    /// <e n:attr='value'/>
+    ///    ^^^^^^^^^^^^^^
+    /// ```
+    #[cfg(feature = "positions")]
+    #[inline]
+    pub fn range(&self) -> Range<usize> {
+        self.data.range.clone()
     }
 }
 

diff --git a/src/parse.rs b/src/parse.rs
@@ -352,7 +352,7 @@ struct TempAttributeData<'input> {
     prefix: &'input str,
     local: &'input str,
     value: StringStorage<'input>,
-    pos: usize,
+    range: Range<usize>,
 }
 
 impl<'input> Document<'input> {
@@ -644,8 +644,8 @@ impl<'input> tokenizer::XmlEvents<'input> for Context<'input> {
 
                 self.after_text = false;
             }
-            tokenizer::Token::Attribute(attr_start, prefix, local, value) => {
-                process_attribute(attr_start, prefix, local, value, self)?;
+            tokenizer::Token::Attribute(range, prefix, local, value) => {
+                process_attribute(range, prefix, local, value, self)?;
             }
             tokenizer::Token::ElementEnd(end, range) => {
                 process_element(end, range, self)?;
@@ -665,7 +665,7 @@ impl<'input> tokenizer::XmlEvents<'input> for Context<'input> {
 
 #[allow(clippy::too_many_arguments)]
 fn process_attribute<'input>(
-    attr_pos: usize,
+    range: Range<usize>,
     prefix: &'input str,
     local: &'input str,
     value: StrSpan<'input>,
@@ -676,7 +676,7 @@ fn process_attribute<'input>(
     if prefix == XMLNS {
         // The xmlns namespace MUST NOT be declared as the default namespace.
         if value.as_str() == NS_XMLNS_URI {
-            let pos = ctx.err_pos_at(attr_pos);
+            let pos = ctx.err_pos_at(range.start);
             return Err(Error::UnexpectedXmlnsUri(pos));
         }
 
@@ -687,13 +687,13 @@ fn process_attribute<'input>(
         // It MUST NOT be bound to any other namespace name.
         if local == NS_XML_PREFIX {
             if !is_xml_ns_uri {
-                let pos = ctx.err_pos_at(attr_pos);
+                let pos = ctx.err_pos_at(range.start);
                 return Err(Error::InvalidXmlPrefixUri(pos));
             }
         } else {
             // The xml namespace MUST NOT be bound to a non-xml prefix.
             if is_xml_ns_uri {
-                let pos = ctx.err_pos_at(attr_pos);
+                let pos = ctx.err_pos_at(range.start);
                 return Err(Error::UnexpectedXmlUri(pos));
             }
         }
@@ -704,7 +704,7 @@ fn process_attribute<'input>(
             .namespaces
             .exists(ctx.namespace_start_idx, Some(local))
         {
-            let pos = ctx.err_pos_at(attr_pos);
+            let pos = ctx.err_pos_at(range.start);
             return Err(Error::DuplicatedNamespace(local.to_string(), pos));
         }
 
@@ -715,13 +715,13 @@ fn process_attribute<'input>(
     } else if local == XMLNS {
         // The xml namespace MUST NOT be declared as the default namespace.
         if value.as_str() == NS_XML_URI {
-            let pos = ctx.err_pos_at(attr_pos);
+            let pos = ctx.err_pos_at(range.start);
             return Err(Error::UnexpectedXmlUri(pos));
         }
 
         // The xmlns namespace MUST NOT be declared as the default namespace.
         if value.as_str() == NS_XMLNS_URI {
-            let pos = ctx.err_pos_at(attr_pos);
+            let pos = ctx.err_pos_at(range.start);
             return Err(Error::UnexpectedXmlnsUri(pos));
         }
 
@@ -731,7 +731,7 @@ fn process_attribute<'input>(
             prefix,
             local,
             value,
-            pos: attr_pos,
+            range,
         });
     }
 
@@ -888,7 +888,7 @@ fn resolve_attributes(namespaces: ShortRange, ctx: &mut Context) -> Result<Short
             // always has no value.'
             None
         } else {
-            get_ns_idx_by_prefix(namespaces, attr.pos, attr.prefix, ctx)?
+            get_ns_idx_by_prefix(namespaces, attr.range.start, attr.prefix, ctx)?
         };
 
         let attr_name = ExpandedNameIndexed {
@@ -900,15 +900,15 @@ fn resolve_attributes(namespaces: ShortRange, ctx: &mut Context) -> Result<Short
         if ctx.doc.attributes[start_idx..].iter().any(|attr| {
             attr.name.as_expanded_name(&ctx.doc) == attr_name.as_expanded_name(&ctx.doc)
         }) {
-            let pos = ctx.err_pos_at(attr.pos);
+            let pos = ctx.err_pos_at(attr.range.start);
             return Err(Error::DuplicatedAttribute(attr.local.to_string(), pos));
         }
 
         ctx.doc.attributes.push(AttributeData {
             name: attr_name,
             value: attr.value,
             #[cfg(feature = "positions")]
-            pos: attr.pos,
+            range: attr.range,
         });
     }
 

diff --git a/src/tokenizer.rs b/src/tokenizer.rs
@@ -160,7 +160,7 @@ pub enum Token<'input> {
     ElementStart(&'input str, &'input str, usize),
 
     // ns:attr="value"
-    Attribute(usize, &'input str, &'input str, StrSpan<'input>),
+    Attribute(Range<usize>, &'input str, &'input str, StrSpan<'input>),
 
     ElementEnd(ElementEnd<'input>, Range<usize>),
 
@@ -561,7 +561,8 @@ fn parse_element<'input>(s: &mut Stream<'input>, events: &mut dyn XmlEvents<'inp
                 s.skip_chars(|_, c| c != quote_c && c != '<')?;
                 let value = s.slice_back_span(value_start);
                 s.consume_byte(quote)?;
-                events.token(Token::Attribute(start, prefix, local, value))?;
+                let end = s.pos();
+                events.token(Token::Attribute(start..end, prefix, local, value))?;
             }
         }
     }

diff --git a/tests/api.rs b/tests/api.rs
@@ -156,7 +156,8 @@ fn text_pos_01() {
     assert_eq!(doc.text_pos_at(node.range().end), TextPos::new(4, 5));
 
     if let Some(attr) = node.attribute_node("a") {
-        assert_eq!(doc.text_pos_at(attr.position()), TextPos::new(1, 4));
+        assert_eq!(doc.text_pos_at(attr.range().start), TextPos::new(1, 4));
+        assert_eq!(doc.text_pos_at(attr.range().end), TextPos::new(1, 9));
     }
 
     // first child is a text/whitespace, not a comment
@@ -181,7 +182,8 @@ fn text_pos_02() {
     assert_eq!(doc.text_pos_at(node.range().start), TextPos::new(1, 1));
 
     if let Some(attr) = node.attribute_node(("http://www.w3.org", "a")) {
-        assert_eq!(doc.text_pos_at(attr.position()), TextPos::new(1, 36));
+        assert_eq!(doc.text_pos_at(attr.range().start), TextPos::new(1, 36));
+        assert_eq!(doc.text_pos_at(attr.range().end), TextPos::new(1, 44));
     }
 }