1 files changed, 44 insertions, 33 deletions

diff --git a/src/parser.rs b/src/parser.rs
index 518aad4..07d48c6 100644
--- a/src/parser.rs
+++ b/src/parser.rs
@@ -2,10 +2,7 @@ use std::char;
 
 use nom::{
     branch::alt,
-    bytes::{
-        complete::take_until,
-        streaming::{is_a, tag, take},
-    },
+    bytes::streaming::{is_a, tag, take, take_till, take_until},
     character::{
         complete::one_of,
         streaming::{char, digit1, none_of, satisfy},
@@ -19,11 +16,6 @@ use nom::{
 
 // parser: parses tokens from lexer into events
 
-enum Misc<'s> {
-    Comment(Comment<'s>),
-    PI(PI<'s>),
-}
-
 type Comment<'s> = &'s str;
 
 struct PI<'s> {
@@ -34,46 +26,44 @@ struct PI<'s> {
 enum ContentItem<'s> {
     CharData(&'s str),
     Element(Element<'s>),
-    Reference(Reference<'s>),
-    CDSect(CDSect<'s>),
+    // Reference(Reference<'s>),
+    // CDSect(CDSect<'s>),
 }
 
 type Content<'s> = Option<Vec<ContentItem<'s>>>;
 
-struct Element<'s> {
-    name: &'s str,
-    attributes: Vec<Attribute<'s>>,
-    content: Content<'s>,
-}
-
 struct Attribute<'s> {
     key: &'s str,
     value: &'s str,
 }
 
-// type VersionNum<'s> = &'s str;
 /// Contains only latin characters or dash after first char
 type EncName<'s> = &'s str;
 
-// struct XMLDecl<'s> {
-//     version_info: VersionNum<'s>,
-//     encoding_decl: Option<EncName<'s>>,
-//     sd_decl: Option<bool>,
-// }
-
 struct DoctypeDecl<'s> {
     name: &'s str,
-    // TODO
+    // TODO: doctype declaration parsing
 }
-
+///
 pub fn doctypedecl(input: &str) -> IResult<&str, DoctypeDecl> {
     todo!()
 }
 
+struct Element<'s> {
+    name: &'s str,
+    attributes: Vec<Attribute<'s>>,
+    content: Content<'s>,
+}
+/// Element
 pub fn element(input: &str) -> IResult<&str, Element> {
     todo!()
 }
 
+enum Misc<'s> {
+    Comment(Comment<'s>),
+    PI(PI<'s>),
+}
+/// Misc
 pub fn misc(input: &str) -> IResult<&str, Misc> {
     todo!()
 }
@@ -210,7 +200,7 @@ pub fn pubid_char(input: &str) -> IResult<&str, PubidChar> {
 type CharData<'s> = &'s str;
 /// [14]   	CharData	   ::=   	[^<&]* - ([^<&]* ']]>' [^<&]*)
 pub fn char_data(input: &str) -> IResult<&str, CharData> {
-    take_until()(input)
+    recognize(take_until("]]>").and_then(take_till(|c| c == '<' || c == '&')))(input)
 }
 
 type Prolog<'s> = (
@@ -229,23 +219,23 @@ pub fn prolog(input: &str) -> IResult<&str, Prolog> {
 
 struct XMLDecl {
     version_info: VersionInfo,
-    encoding_decl: Option<EncodingDecl>,
-    sd_decl: Option<SDDecl>,
+    // encoding_decl: Option<EncodingDecl>,
+    // sd_decl: Option<SDDecl>,
 }
 /// [23]   	XMLDecl	   ::=   	'<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
 pub fn xml_decl(input: &str) -> IResult<&str, XMLDecl> {
     // (VersionInfo, Option<EncodingDecl>, Option<SDDecl>)
-    let (leftover, (version_info, encoding_decl, sd_decl)) = delimited(
+    let (leftover, (version_info /* encoding_decl, sd_decl */,)) = delimited(
         tag("<?xml"),
-        tuple((version_info, opt(encoding_decl), opt(sd_decl))),
+        tuple((version_info /* opt(encoding_decl), opt(sd_decl) */,)),
         tag("?>"),
     )(input)?;
     Ok((
         leftover,
         XMLDecl {
             version_info,
-            encoding_decl,
-            sd_decl,
+            // encoding_decl,
+            // sd_decl,
         },
     ))
 }
@@ -282,3 +272,24 @@ pub fn version_num(input: &str) -> IResult<&str, VersionNum> {
         )),
     )(input)
 }
+
+pub fn reference(input: &str) -> IResult<&str, char> {
+    todo!()
+}
+
+pub fn pe_reference(input: &str) -> IResult<&str, char> {
+    todo!()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_char_data() {
+        assert_eq!(Ok(("&def]]>ghi", "abc")), char_data("abc&def]]>ghi"));
+        assert_eq!(Ok(("]]>ghi", "abcdef")), char_data("abcdef]]>ghi"));
+        assert_eq!(Ok(("&defghi", "abc")), char_data("abc&defghi"));
+        assert_eq!(Ok(("", "abcdefghi")), char_data("abcdefghi"));
+    }
+}