Require spaces between prolog attrs

Author: kornelski

src/reader/parser.rs

@@ -249,10 +249,12 @@ pub enum DeclarationSubstate {
     InsideVersionValue,
     AfterVersionValue,
 
+    BeforeEncoding,
     InsideEncoding,
     AfterEncoding,
 
     InsideEncodingValue,
+    AfterEncodingValue,
 
     BeforeStandaloneDecl,
     InsideStandaloneDecl,
@@ -730,6 +732,24 @@ mod tests {
         expect_event!(r, p, Ok(XmlEvent::EndDocument));
     }
 
+    #[test]
+    fn malformed_declaration_attrs() {
+        let (mut r, mut p) = test_data!(r#"<?xml version x="1.0"?>"#);
+        expect_event!(r, p, Err(_));
+
+        let (mut r, mut p) = test_data!(r#"<?xml version="1.0" version="1.0"?>"#);
+        expect_event!(r, p, Err(_));
+
+        let (mut r, mut p) = test_data!(r#"<?xml version="1.0"encoding="utf-8"?>"#);
+        expect_event!(r, p, Err(_));
+
+        let (mut r, mut p) = test_data!(r#"<?xml version="1.0"standalone="yes"?>"#);
+        expect_event!(r, p, Err(_));
+
+        let (mut r, mut p) = test_data!(r#"<?xml version="1.0" encoding="utf-8"standalone="yes"?>"#);
+        expect_event!(r, p, Err(_));
+    }
+
     #[test]
     fn opening_tag_in_attribute_value() {
         use crate::reader::error::{SyntaxError, Error, ErrorKind};

src/reader/parser/inside_declaration.rs

@@ -89,6 +89,12 @@ impl PullParser {
             }),
 
             DeclarationSubstate::AfterVersionValue => match t {
+                Token::Character(c) if is_whitespace_char(c) => self.into_state_continue(State::InsideDeclaration(DeclarationSubstate::BeforeEncoding)),
+                Token::ProcessingInstructionEnd => self.emit_start_document(),
+                _ => Some(self.error(SyntaxError::UnexpectedToken(t))),
+            },
+
+            DeclarationSubstate::BeforeEncoding => match t {
                 Token::Character('e') => self.into_state_continue(State::InsideDeclaration(DeclarationSubstate::InsideEncoding)),
                 Token::Character('s') => self.into_state_continue(State::InsideDeclaration(DeclarationSubstate::InsideStandaloneDecl)),
                 Token::ProcessingInstructionEnd => self.emit_start_document(),
@@ -114,9 +120,15 @@ impl PullParser {
 
             DeclarationSubstate::InsideEncodingValue => self.read_attribute_value(t, |this, value| {
                 this.data.encoding = Some(value);
-                this.into_state_continue(State::InsideDeclaration(DeclarationSubstate::BeforeStandaloneDecl))
+                this.into_state_continue(State::InsideDeclaration(DeclarationSubstate::AfterEncodingValue))
             }),
 
+            DeclarationSubstate::AfterEncodingValue => match t {
+                Token::Character(c) if is_whitespace_char(c) => self.into_state_continue(State::InsideDeclaration(DeclarationSubstate::BeforeStandaloneDecl)),
+                Token::ProcessingInstructionEnd => self.emit_start_document(),
+                _ => Some(self.error(SyntaxError::UnexpectedToken(t))),
+            },
+
             DeclarationSubstate::BeforeStandaloneDecl => match t {
                 Token::Character('s') => self.into_state_continue(State::InsideDeclaration(DeclarationSubstate::InsideStandaloneDecl)),
                 Token::ProcessingInstructionEnd => self.emit_start_document(),

tests/oasis.fail.txt

@@ -11,7 +11,6 @@ o-p12fail6 p12fail6.xml   built-in entity refs excluded
 o-p12fail7 p12fail7.xml   The public ID has a tab character, which is disallowed  
 o-p30fail1 p30fail1.xml   An XML declaration is not the same as a TextDecl  
 o-p31fail1 p31fail1.xml   external subset excludes doctypedecl  
-o-p32fail3 p32fail3.xml   initial S is required  
 o-p40fail1 p40fail1.xml   S is required between attributes  
 o-p44fail4 p44fail4.xml   Whitespace required between attributes. 
 o-p45fail2 p45fail2.xml   S before contentspec is required.  

tests/xmltest.fail.txt

@@ -27,7 +27,6 @@ not-wf-sa-086 086.xml   Public IDs may not contain "[".
 not-wf-sa-087 087.xml   Public IDs may not contain "[". 
 not-wf-sa-089 089.xml   Parameter entities "are" always parsed; NDATA annotations  are not permitted.
 not-wf-sa-091 091.xml   Parameter entities "are" always parsed; NDATA annotations  are not permitted.
-not-wf-sa-096 096.xml   Space is required before the standalone declaration. 
 not-wf-sa-104 104.xml   Internal general parsed entities are only well formed if  they match the "content" production. 
 not-wf-sa-115 115.xml   The replacement text of this entity is an illegal character  reference, which must be rejected when it is parsed in the  context of an attribute value.
 not-wf-sa-116 116.xml   Internal general parsed entities are only well formed if  they match the "content" production. This is a partial  character reference, not a full one.