From 3a875666a5a897d92a9c6d92a67867bcae662211 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?cel=20=F0=9F=8C=B8?= <cel@blos.sm>
Date: Fri, 21 Jun 2024 17:08:45 +0100
Subject: [PATCH] WIP: XMLDecl stuff

---
 src/parser.rs | 90 +++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 69 insertions(+), 21 deletions(-)

diff --git a/src/parser.rs b/src/parser.rs
index f882064..2acd579 100644
--- a/src/parser.rs
+++ b/src/parser.rs
@@ -30,9 +30,6 @@ struct Attribute<'s> {
     value: &'s str,
 }
 
-/// Contains only latin characters or dash after first char
-type EncName<'s> = &'s str;
-
 struct DoctypeDecl<'s> {
     name: &'s str,
     // TODO: doctype declaration parsing
@@ -52,15 +49,6 @@ 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!()
-}
-
 type Document<'s> = (Prolog<'s>, Element<'s>, Vec<Misc<'s>>);
 /// [1]   	document	   ::=   	prolog element Misc*
 pub fn document(input: &str) -> IResult<&str, Document> {
@@ -231,6 +219,7 @@ pub fn comment(input: &str) -> IResult<&str, Comment> {
     )(input)
 }
 
+#[derive(Clone)]
 struct PI<'s> {
     target: &'s str,
     instruction: Option<&'s str>,
@@ -294,7 +283,7 @@ pub fn cd_end(input: &str) -> IResult<&str, CDEnd> {
 }
 
 type Prolog<'s> = (
-    Option<XMLDecl>,
+    Option<XMLDecl<'s>>,
     Vec<Misc<'s>>,
     Option<(DoctypeDecl<'s>, Vec<Misc<'s>>)>,
 );
@@ -307,25 +296,26 @@ pub fn prolog(input: &str) -> IResult<&str, Prolog> {
     ))(input)
 }
 
-struct XMLDecl {
+struct XMLDecl<'s> {
     version_info: VersionInfo,
-    // encoding_decl: Option<EncodingDecl>,
-    // sd_decl: Option<SDDecl>,
+    encoding_decl: Option<EncodingDecl<'s>>,
+    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) */,)),
-        tag("?>"),
+        tuple((version_info, opt(encoding_decl), opt(sd_decl))),
+        pair(opt(s), tag("?>")),
     )(input)?;
+    // TODO: change to map
     Ok((
         leftover,
         XMLDecl {
             version_info,
-            // encoding_decl,
-            // sd_decl,
+            encoding_decl,
+            sd_decl,
         },
     ))
 }
@@ -363,6 +353,41 @@ pub fn version_num(input: &str) -> IResult<&str, VersionNum> {
     )(input)
 }
 
+#[derive(Clone)]
+enum Misc<'s> {
+    Comment(Comment<'s>),
+    PI(PI<'s>),
+    S,
+}
+/// [27]   	Misc	   ::=   	Comment | PI | S
+pub fn misc(input: &str) -> IResult<&str, Misc> {
+    alt((
+        map(comment, |comment| Misc::Comment(comment)),
+        map(pi, |pi| Misc::PI(pi)),
+        value(Misc::S, s),
+    ))(input)
+}
+
+type SDDecl = bool;
+/// [32]   	SDDecl	   ::=   	S 'standalone' Eq (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"'))
+pub fn sd_decl(input: &str) -> IResult<&str, SDDecl> {
+    preceded(
+        tuple((s, tag("standalone"), eq)),
+        alt((
+            delimited(
+                char('\''),
+                alt((value(true, tag("yes")), value(false, tag("no")))),
+                char('\''),
+            ),
+            delimited(
+                char('"'),
+                alt((value(true, tag("yes")), value(false, tag("no")))),
+                char('"'),
+            ),
+        )),
+    )(input)
+}
+
 pub fn reference(input: &str) -> IResult<&str, char> {
     todo!()
 }
@@ -371,6 +396,29 @@ pub fn pe_reference(input: &str) -> IResult<&str, char> {
     todo!()
 }
 
+type EncodingDecl<'s> = EncName<'s>;
+/// [80]   	EncodingDecl	   ::=   	S 'encoding' Eq ('"' EncName '"' | "'" EncName
+pub fn encoding_decl(input: &str) -> IResult<&str, EncodingDecl> {
+    preceded(
+        tuple((s, tag("encoding"), eq)),
+        alt((
+            delimited(char('"'), enc_name, char('"')),
+            delimited(char('\''), enc_name, char('\'')),
+        )),
+    )(input)
+}
+
+type EncName<'s> = &'s str;
+/// [81]   	EncName	   ::=   	[A-Za-z] ([A-Za-z0-9._] | '-')*
+pub fn enc_name(input: &str) -> IResult<&str, EncName> {
+    recognize(pair(
+        satisfy(|c| matches!(c, 'A'..='Z' | 'a'..='z' )),
+        many0(satisfy(
+            |c| matches!(c, 'A'..='Z' | 'a'..='z' | '0'..='9' | '.' | '_' | '-' ),
+        )),
+    ))(input)
+}
+
 #[cfg(test)]
 mod tests {
     use std::num::NonZero;