Mostly complete, untested selector parser.

selectors.rs

@@ -1,15 +1,16 @@
+use std::{vec, iterator};
+use std::ascii::to_ascii_lower;
 use cssparser::*;
 use stylesheets::NamespaceMap;
 
 
 pub struct Selector {
     compound_selectors: CompoundSelector,
-    pseudo_element: PseudoElement,
-    specificity: u32,
+    pseudo_element: Option<PseudoElement>,
+//    specificity: u32,
 }
 
 pub enum PseudoElement {
-    Element, // No pseudo-element
     Before,
     After,
     FirstLine,
@@ -33,7 +34,7 @@ pub enum SimpleSelector {
     IDSelector(~str),
     ClassSelector(~str),
     LocalNameSelector{lowercase_name: ~str, cased_name: ~str},
-//    NamespaceSelector(Namespace)
+    NamespaceSelector(~str),
 
     // Attribute selectors
     AttrExists(AttrSelector),  // [foo]
@@ -45,111 +46,384 @@ pub enum SimpleSelector {
     AttrSuffixMatch(AttrSelector, ~str),  // [foo$=bar]
 
     // Pseudo-classes
-//    Empty,
-//    Root,
-//    Lang(~str),
+    Empty,
+    Root,
+    Lang(~str),
 //    NthChild(u32, u32),
 //    NthLastChild(u32, u32),
 //    NthOfType(u32, u32),
 //    NthLastOfType(u32, u32),
-//    Lang(~str),
-//    Negation(~Selector),
+    Negation(~[SimpleSelector]),
     // ...
 }
 
 pub struct AttrSelector {
     lowercase_name: ~str,
     cased_name: ~str,
-//    namespace: Option<~str>,
+    namespace: Option<~str>,
 }
 
 
-pub fn parse_selector_list(input: &[ComponentValue], namespaces: &NamespaceMap)
+type Iter = iterator::PeekableIterator<ComponentValue, vec::ConsumeIterator<ComponentValue>>;
+
+
+// None means invalid selector
+pub fn parse_selector_list(input: ~[ComponentValue], namespaces: &NamespaceMap)
                            -> Option<~[Selector]> {
-    let len = input.len();
-    let (first, pos) = match parse_selector(input, 0, namespaces) {
+    let iter = &mut input.consume_iter().peekable();
+    let first = match parse_selector(iter, namespaces) {
         None => return None,
         Some(result) => result
     };
     let mut results = ~[first];
-    let mut pos = pos;
 
     loop {
-        pos = skip_whitespace(input, pos);
-        if pos >= len { break }  // EOF
-        if input[pos] != Comma { return None }
-        pos = skip_whitespace(input, pos);
-        match parse_selector(input, pos, namespaces) {
+        skip_whitespace(iter);
+        match iter.peek() {
+            None => break,  // EOF
+            Some(&Comma) => (),
+            _ => return None,
+        }
+        match parse_selector(iter, namespaces) {
+            Some(selector) => results.push(selector),
             None => return None,
-            Some((selector, next_pos)) => {
-                results.push(selector);
-                pos = next_pos;
-            }
         }
     }
     Some(results)
 }
 
 
-fn parse_selector(input: &[ComponentValue], pos: uint, namespaces: &NamespaceMap)
-                  -> Option<(Selector, uint)> {
-    let len = input.len();
-    let (first, pos) = match parse_simple_selectors(input, pos, namespaces) {
+// None means invalid selector
+fn parse_selector(iter: &mut Iter, namespaces: &NamespaceMap)
+                  -> Option<Selector> {
+    let (first, pseudo_element) = match parse_simple_selectors(iter, namespaces) {
         None => return None,
         Some(result) => result
     };
     let mut compound = CompoundSelector{ simple_selectors: first, next: None };
-    let mut pos = pos;
+    let mut pseudo_element = pseudo_element;
 
-    loop {
-        let pre_whitespace_pos = pos;
-        pos = skip_whitespace(input, pos);
-        if pos >= len { break }  // EOF
-        let combinator = match input[pos] {
-            Delim('>') => { pos += 1; Child },
-            Delim('+') => { pos += 1; NextSibling },
-            Delim('~') => { pos += 1; LaterSibling },
-            _ => {
-                if pos > pre_whitespace_pos { Descendant }
+    while pseudo_element.is_none() {
+        let any_whitespace = skip_whitespace(iter);
+        let combinator = match iter.peek() {
+            None => break,  // EOF
+            Some(&Delim('>')) => { iter.next(); Child },
+            Some(&Delim('+')) => { iter.next(); NextSibling },
+            Some(&Delim('~')) => { iter.next(); LaterSibling },
+            Some(_) => {
+                if any_whitespace { Descendant }
                 else { return None }
             }
         };
-        pos = skip_whitespace(input, pos);
-        match parse_simple_selectors(input, pos, namespaces) {
+        match parse_simple_selectors(iter, namespaces) {
             None => return None,
-            Some((simple_selectors, next_pos)) => {
+            Some((simple_selectors, pseudo)) => {
                 compound = CompoundSelector {
                     simple_selectors: simple_selectors,
                     next: Some((~compound, combinator))
                 };
-                pos = next_pos;
+                pseudo_element = pseudo;
             }
         }
     }
     let selector = Selector{
         compound_selectors: compound,
-        pseudo_element: Element,
-        specificity: 0, // TODO
+        pseudo_element: pseudo_element,
     };
-    Some((selector, pos))
+    Some(selector)
 }
 
 
-fn parse_simple_selectors(input: &[ComponentValue], pos: uint, namespaces: &NamespaceMap)
-                           -> Option<(~[SimpleSelector], uint)> {
-    let _ = input;
-    let _ = pos;
-    let _ = namespaces;
-    None  // TODO
+// None means invalid selector
+fn parse_simple_selectors(iter: &mut Iter, namespaces: &NamespaceMap)
+                           -> Option<(~[SimpleSelector], Option<PseudoElement>)> {
+    let mut empty = true;
+    let mut simple_selectors = match parse_type_selector(iter, namespaces) {
+        None => return None,  // invalid selector
+        Some(None) => ~[],
+        Some(Some(s)) => { empty = false; s }
+    };
+
+    let mut pseudo_element = None;
+    loop {
+        match parse_one_simple_selector(iter, namespaces, /* inside_negation = */ false) {
+            None => return None, // invalid selector
+            Some(None) => break,
+            Some(Some(Left(s))) => simple_selectors.push(s),
+            Some(Some(Right(p))) => { pseudo_element = Some(p); break },
+        }
+    }
+    if empty { None }  // An empty selector is invalid
+    else { Some((simple_selectors, pseudo_element)) }
+}
+
+
+// None means invalid selector
+// Some(None) means no type selector
+// Some(Some([...])) is a type selector. Might be empty for *|*
+fn parse_type_selector(iter: &mut Iter, namespaces: &NamespaceMap)
+                       -> Option<Option<~[SimpleSelector]>> {
+    skip_whitespace(iter);
+    match parse_qualified_name(iter, /* allow_universal = */ true, namespaces) {
+        None => None,  // invalid selector
+        Some(None) => Some(None),
+        Some(Some((namespace, local_name))) => {
+            let mut simple_selectors = ~[];
+            match namespace {
+                Some(url) => simple_selectors.push(NamespaceSelector(url)),
+                None => (),
+            }
+            match local_name {
+                Some(name) => simple_selectors.push(LocalNameSelector{
+                    lowercase_name: to_ascii_lower(name),
+                    cased_name: name,
+                }),
+                None => (),
+            }
+            Some(Some(simple_selectors))
+        }
+    }
+}
+
+
+// Parse a simple selector other than a type selector
+fn parse_one_simple_selector(iter: &mut Iter, namespaces: &NamespaceMap, inside_negation: bool)
+                         -> Option<Option<Either<SimpleSelector, PseudoElement>>> {
+    match iter.peek() {
+        Some(&IDHash(_)) => match iter.next() {
+            Some(IDHash(id)) => Some(Some(Left(IDSelector(id)))),
+            _ => fail!("Implementation error, this should not happen."),
+        },
+        Some(&Delim('.')) => {
+            iter.next();
+            match iter.next() {
+                Some(Ident(class)) => Some(Some(Left(ClassSelector(class)))),
+                _ => None,  // invalid selector
+            }
+        }
+        Some(&SquareBracketBlock(_)) => match iter.next() {
+            Some(SquareBracketBlock(content))
+            => match parse_attribute_selector(content, namespaces) {
+                None => None,
+                Some(simple_selector) => Some(Some(Left(simple_selector))),
+            },
+            _ => fail!("Implementation error, this should not happen."),
+        },
+        Some(&Delim(':')) => {
+            iter.next();
+            match iter.next() {
+                Some(Ident(name)) => match parse_simple_pseudo_class(name) {
+                    None => None,
+                    Some(result) => Some(Some(result)),
+                },
+                Some(Function(name, arguments)) => match parse_functional_pseudo_class(
+                        name, arguments, namespaces, inside_negation) {
+                    None => None,
+                    Some(simple_selector) => Some(Some(Left(simple_selector))),
+                },
+                Some(Delim(':')) => {
+                    match iter.next() {
+                        Some(Ident(name)) => match parse_pseudo_element(name) {
+                            Some(pseudo_element) => Some(Some(Right(pseudo_element))),
+                            _ => None,
+                        },
+                        _ => None,
+                    }
+                }
+                _ => None,
+            }
+        }
+        _ => Some(None),
+    }
+}
+
+// None means invalid selector
+// Some(None) means not a qualified name
+// Some(Some((None, None)) means *|*
+// Some(Some((Some(url), None)) means prefix|*
+// Some(Some((None, Some(name)) means *|name
+// Some(Some((Some(url), Some(name))) means prefix|name
+// ... or equivalent
+fn parse_qualified_name(iter: &mut Iter, allow_universal: bool, namespaces: &NamespaceMap)
+                       -> Option<Option<(Option<~str>, Option<~str>)>> {
+    #[inline]
+    fn default_namespace(namespaces: &NamespaceMap, local_name: Option<~str>)
+                         -> Option<Option<(Option<~str>, Option<~str>)>> {
+        match namespaces.default {
+            None => Some(Some((None, local_name))),
+            Some(ref url) => Some(Some((Some(url.to_owned()), local_name))),
+        }
+    }
+
+    #[inline]
+    fn explicit_namespace(iter: &mut Iter, allow_universal: bool, namespace_url: Option<~str>)
+                         -> Option<Option<(Option<~str>, Option<~str>)>> {
+        assert!(iter.next() == Some(Delim('|')));
+        match iter.peek() {
+            Some(&Delim('*')) if allow_universal => {
+                iter.next();
+                Some(Some((namespace_url, None)))
+            },
+            Some(&Ident(_)) => {
+                let local_name = get_next_ident(iter);
+                Some(Some((namespace_url, Some(local_name))))
+            },
+            _ => None,  // invalid selector
+        }
+    }
+
+    match iter.peek() {
+        Some(&Ident(_)) => {
+            let value = get_next_ident(iter);
+            match iter.peek() {
+                Some(&Delim('|')) => default_namespace(namespaces, Some(value)),
+                _ => {
+                    let namespace_url = match namespaces.prefix_map.find(&value) {
+                        None => return None,  // Undeclared namespace prefix: invalid selector
+                        Some(ref url) => url.to_owned(),
+                    };
+                    explicit_namespace(iter, allow_universal, Some(namespace_url))
+                },
+            }
+        },
+        Some(&Delim('*')) => {
+            iter.next();  // Consume '*'
+            match iter.peek() {
+                Some(&Delim('|')) => {
+                    if allow_universal { default_namespace(namespaces, None) }
+                    else { None }
+                },
+                _ => explicit_namespace(iter, allow_universal, None),
+            }
+        },
+        Some(&Delim('|')) => explicit_namespace(iter, allow_universal, Some(~"")),
+        _ => return None,
+    }
+}
+
+
+fn parse_attribute_selector(content: ~[ComponentValue], namespaces: &NamespaceMap)
+                            -> Option<SimpleSelector> {
+    let iter = &mut content.consume_iter().peekable();
+    let attr = match parse_qualified_name(iter, /* allow_universal = */ false, namespaces) {
+        None => return None,  // invalid selector
+        Some(None) => return None,
+        Some(Some((_, None))) => fail!("Implementation error, this should not happen."),
+        Some(Some((namespace, Some(local_name)))) => AttrSelector {
+            namespace: namespace,
+            lowercase_name: to_ascii_lower(local_name),
+            cased_name: local_name,
+        },
+    };
+    skip_whitespace(iter);
+    macro_rules! get_value( () => {{
+        skip_whitespace(iter);
+        match iter.next() {
+            Some(Ident(value)) | Some(String(value)) => value,
+            _ => return None,
+        }
+    }};)
+    let result = match iter.next() {
+        None => AttrExists(attr),  // [foo]
+        Some(Delim('=')) => AttrEqual(attr, get_value!()),  // [foo=bar]
+        Some(IncludeMatch) => AttrIncludes(attr, get_value!()),  // [foo~=bar]
+        Some(DashMatch) => AttrDashMatch(attr, get_value!()),  // [foo|=bar]
+        Some(PrefixMatch) => AttrPrefixMatch(attr, get_value!()),  // [foo^=bar]
+        Some(SubstringMatch) => AttrSubstringMatch(attr, get_value!()),  // [foo*=bar]
+        Some(SuffixMatch) => AttrSuffixMatch(attr, get_value!()),  // [foo$=bar]
+        _ => return None
+    };
+    skip_whitespace(iter);
+    if iter.next().is_none() { Some(result) } else { None }
+}
+
+
+fn parse_simple_pseudo_class(name: ~str) -> Option<Either<SimpleSelector, PseudoElement>> {
+    let lower_name: &str = to_ascii_lower(name);
+    match lower_name {
+        "root" => Some(Left(Root)),
+        "empty" => Some(Left(Empty)),
+
+        // Supported CSS 2.1 pseudo-elements only.
+        "before" => Some(Right(Before)),
+        "after" => Some(Right(After)),
+        "first-line" => Some(Right(FirstLine)),
+        "first-letter" => Some(Right(FirstLetter)),
+        _ => None
+    }
+}
+
+
+fn parse_functional_pseudo_class(name: ~str, arguments: ~[ComponentValue],
+                                 namespaces: &NamespaceMap, inside_negation: bool)
+                                 -> Option<SimpleSelector> {
+    let lower_name: &str = to_ascii_lower(name);
+    match lower_name {
+        "lang" => parse_lang(arguments),
+        "not" => if inside_negation { None } else { parse_negation(arguments, namespaces) },
+        _ => None
+    }
+}
+
+
+fn parse_pseudo_element(name: ~str) -> Option<PseudoElement> {
+    let lower_name: &str = to_ascii_lower(name);
+    match lower_name {
+        // All supported pseudo-elements
+        "before" => Some(Before),
+        "after" => Some(After),
+        "first-line" => Some(FirstLine),
+        "first-letter" => Some(FirstLetter),
+        _ => None
+    }
+}
+
+
+fn parse_lang(arguments: ~[ComponentValue]) -> Option<SimpleSelector> {
+    let mut iter = arguments.consume_skip_whitespace();
+    match iter.next() {
+        Some(Ident(value)) => {
+            if "" == value || iter.next().is_some() { None }
+            else { Some(Lang(value)) }
+        },
+        _ => None,
+    }
+}
+
+
+// Level 3: Parse ONE simple_selector
+fn parse_negation(arguments: ~[ComponentValue], namespaces: &NamespaceMap)
+                  -> Option<SimpleSelector> {
+    let iter = &mut arguments.consume_iter().peekable();
+    Some(Negation(match parse_type_selector(iter, namespaces) {
+        None => return None,  // invalid selector
+        Some(Some(s)) => s,
+        Some(None) => {
+            match parse_one_simple_selector(iter, namespaces, /* inside_negation = */ true) {
+                Some(Some(Left(s))) => ~[s],
+                _ => return None
+            }
+        },
+    }))
+}
+
+
+/// Assuming the next token is an ident, consume it and return its value
+#[inline]
+fn get_next_ident(iter: &mut Iter) -> ~str {
+    match iter.next() {
+        Some(Ident(value)) => value,
+        _ => fail!("Implementation error, this should not happen."),
+    }
 }
 
 
 #[inline]
-fn skip_whitespace(input: &[ComponentValue], mut pos: uint) -> uint {
-    let len = input.len();
-    while pos < len {
-        if input[pos] == WhiteSpace { break }
-        pos += 1;
+fn skip_whitespace(iter: &mut Iter) -> bool {
+    let mut any_whitespace = false;
+    loop {
+        if iter.peek() != Some(&WhiteSpace) { return any_whitespace }
+        any_whitespace = true;
+        iter.next();
     }
-    pos
 }

stylesheets.rs

@@ -67,14 +67,14 @@ fn parse_stylesheet(css: &str) -> Stylesheet {
                 state = STATE_BODY;
                 log_css_error(rule.location, fmt!("Unsupported at-rule: @%s", name))
             },
-            QualifiedRule(rule) => {
+            QualifiedRule(QualifiedRule{location: location, prelude: prelude, block: block}) => {
                 state = STATE_BODY;
-                match selectors::parse_selector_list(rule.prelude, &namespaces) {
+                match selectors::parse_selector_list(prelude, &namespaces) {
                     Some(selectors) => rules.push(StyleRule{
                         selectors: selectors,
-                        declarations: properties::parse_property_declaration_list(rule.block)
+                        declarations: properties::parse_property_declaration_list(block)
                     }),
-                    None => log_css_error(rule.location, "Unsupported CSS selector."),
+                    None => log_css_error(location, "Unsupported CSS selector."),
                 }
             },
         }