Emit actual encoding event

Author: kornelski

design.md

@@ -2,19 +2,13 @@
 
 Basic features:
  * [x] Parsing XML 1.0 documents and returning a stream of events
-   - [ ] Support reading embedded DTD schemas
-   - [ ] Support for embedded entities
+   - [x] Support for embedded entities
  * [x] Support for namespaces and emitting namespace information in events
- * [ ] \[maybe\] push-based wrapper
  * Missing XML features
-   - [ ] Support for different encodings
+   - [x] Support for different encodings
    - [ ] Attribute values normalization
    - [ ] EOL characters normalization
 
-Advanced features:
- * [ ] DTD schema validation
- * [ ] XSD schema validation
-
 # Writer
 
 Basic features:
@@ -30,8 +24,3 @@ Basic features:
     - [ ] Check for namespaces more correctly, i.e. check both for prefix and namespace URI
     - [ ] Support checking namespace prefix presence in the current namespace for events with prefix but without namespace
     - [ ] Support checking namespace prefix for events with both prefix and namespace URI
-
-# Other
-
-DOM-based API:
- * [ ] Basic support for DOM-based API

src/reader/parser.rs

@@ -57,7 +57,6 @@ mod inside_reference;
 mod outside_tag;
 
 static DEFAULT_VERSION: XmlVersion = XmlVersion::Version10;
-static DEFAULT_ENCODING: &str = "UTF-8";
 static DEFAULT_STANDALONE: Option<bool> = None;
 
 type ElementStack = Vec<OwnedName>;

src/reader/parser/inside_declaration.rs

@@ -5,7 +5,7 @@ use crate::reader::lexer::Token;
 use crate::util::Encoding;
 
 use super::{
-    DeclarationSubstate, PullParser, QualifiedNameTarget, Result, State, DEFAULT_ENCODING,
+    DeclarationSubstate, PullParser, QualifiedNameTarget, Result, State,
     DEFAULT_VERSION,
 };
 
@@ -35,9 +35,10 @@ impl PullParser {
             }
         }
 
+        let current_encoding = self.lexer.encoding();
         self.into_state_emit(State::OutsideTag, Ok(XmlEvent::StartDocument {
             version: version.unwrap_or(DEFAULT_VERSION),
-            encoding: encoding.unwrap_or(DEFAULT_ENCODING.into()),
+            encoding: encoding.unwrap_or_else(move || current_encoding.to_string()),
             standalone
         }))
     }

src/reader/parser/outside_tag.rs

@@ -5,7 +5,7 @@ use crate::reader::lexer::Token;
 
 use super::{
     ClosingTagSubstate, OpeningTagSubstate, ProcessingInstructionSubstate, PullParser, Result,
-    State, DEFAULT_ENCODING, DEFAULT_STANDALONE, DEFAULT_VERSION, DoctypeSubstate,
+    State, DEFAULT_STANDALONE, DEFAULT_VERSION, DoctypeSubstate,
 };
 
 impl PullParser {
@@ -94,7 +94,7 @@ impl PullParser {
                             self.parsed_declaration = true;
                             let sd_event = XmlEvent::StartDocument {
                                 version: DEFAULT_VERSION,
-                                encoding: DEFAULT_ENCODING.into(),
+                                encoding: self.lexer.encoding().to_string(),
                                 standalone: DEFAULT_STANDALONE
                             };
                             // next_event is always none here because we're outside of

src/util.rs

@@ -86,13 +86,13 @@ impl fmt::Display for Encoding {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         f.write_str(match self {
             Encoding::Utf8 => "UTF-8",
-            Encoding::Default => "UTF-8 (default)",
+            Encoding::Default => "UTF-8",
             Encoding::Latin1 => "ISO-8859-1",
             Encoding::Ascii => "US-ASCII",
-            Encoding::Utf16Be => "UTF-16 (BE)",
-            Encoding::Utf16Le => "UTF-16 (LE)",
+            Encoding::Utf16Be => "UTF-16",
+            Encoding::Utf16Le => "UTF-16",
             Encoding::Utf16 => "UTF-16",
-            Encoding::Unknown => "Unknown",
+            Encoding::Unknown => "(unknown)",
         })
     }
 }

tests/documents/sample_1.xml

@@ -1,4 +1,4 @@
-<?xml version="1.0" encoding="utf-8" standalone="yes"?>
+<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?>
 <project name="project-name">
     <libraries>
         <library groupId="org.example" artifactId="&lt;name&gt;" version="0.1"/>

tests/documents/sample_1_full.txt

@@ -1,4 +1,4 @@
-StartDocument(1.0, utf-8)
+StartDocument(1.0, iso-8859-1)
 StartElement(project [name="project-name"])
 Whitespace("\n    ")
 StartElement(libraries)

tests/documents/sample_1_short.txt

@@ -1,4 +1,4 @@
-StartDocument(1.0, utf-8)
+StartDocument(1.0, iso-8859-1)
 StartElement(project [name="project-name"])
 StartElement(libraries)
 StartElement(library [groupId="org.example", artifactId="<name>", version="0.1"])