From 91208209d5990ff1b6c58bd23b6bc5241b27bc1e Mon Sep 17 00:00:00 2001
From: Markus Unterwaditzer <markus-tarpit+git@unterwaditzer.net>
Date: Wed, 30 Oct 2024 22:19:37 +0100
Subject: [PATCH 1/3] Relax selector parsing

Neither the proposal nor the test changes are merged yet. Also, the
tests are still failing.

https://github.com/html5lib/html5lib-tests/pull/178
https://github.com/whatwg/html/pull/10557
---
 html5ever/src/tree_builder/mod.rs      |  11 +-
 html5ever/src/tree_builder/rules.rs    | 166 ++++++-------------------
 html5ever/src/tree_builder/tag_sets.rs |   7 +-
 html5ever/src/tree_builder/types.rs    |   2 -
 rcdom/html5lib-tests                   |   2 +-
 rcdom/tests/html-tree-builder.rs       |   9 +-
 6 files changed, 47 insertions(+), 150 deletions(-)

diff --git a/html5ever/src/tree_builder/mod.rs b/html5ever/src/tree_builder/mod.rs
index 348c211d..b6d5a039 100644
--- a/html5ever/src/tree_builder/mod.rs
+++ b/html5ever/src/tree_builder/mod.rs
@@ -1146,6 +1146,7 @@ where
         n
     }
 
+    /// Pop element until an element with the given name has been popped.
     fn pop_until_named(&self, name: LocalName) -> usize {
         self.pop_until(|p| *p.ns == ns!(html) && *p.local == name)
     }
@@ -1231,16 +1232,6 @@ where
                 _ => continue,
             };
             match *name {
-                local_name!("select") => {
-                    for ancestor in self.open_elems.borrow()[0..i].iter().rev() {
-                        if self.html_elem_named(ancestor, local_name!("template")) {
-                            return InSelect;
-                        } else if self.html_elem_named(ancestor, local_name!("table")) {
-                            return InSelectInTable;
-                        }
-                    }
-                    return InSelect;
-                },
                 local_name!("td") | local_name!("th") => {
                     if !last {
                         return InCell;
diff --git a/html5ever/src/tree_builder/rules.rs b/html5ever/src/tree_builder/rules.rs
index 9bce3570..060915d1 100644
--- a/html5ever/src/tree_builder/rules.rs
+++ b/html5ever/src/tree_builder/rules.rs
@@ -624,6 +624,13 @@ where
 
                 tag @ <hr> => {
                     self.close_p_element_in_button_scope();
+                    if self.in_scope_named(default_scope, local_name!("select")) {
+                        self.generate_implied_end_except(local_name!("optgroup"));
+                        if self.in_scope_named(default_scope, local_name!("option")) {
+                            self.sink.parse_error(Borrowed("hr in option"));
+                        }
+                    }
+
                     self.insert_and_pop_element_for(tag);
                     self.frameset_ok.set(false);
                     DoneAckSelfClosing
@@ -662,28 +669,50 @@ where
                 // <noscript> handled in wildcard case below
 
                 tag @ <select> => {
+                    if self.in_scope_named(default_scope, local_name!("select")) {
+                        self.sink.parse_error(Borrowed("nested select"));
+                        self.pop_until_named(local_name!("select"));
+                    }
+
                     self.reconstruct_formatting();
                     self.insert_element_for(tag);
                     self.frameset_ok.set(false);
-                    // NB: mode == InBody but possibly self.mode != mode, if
-                    // we're processing "as in the rules for InBody".
-                    self.mode.set(match self.mode.get() {
-                        InTable | InCaption | InTableBody
-                            | InRow | InCell => InSelectInTable,
-                        _ => InSelect,
-                    });
                     Done
                 }
 
-                tag @ <optgroup> <option> => {
-                    if self.current_node_named(local_name!("option")) {
-                        self.pop();
+                tag @ <option> => {
+                    if self.in_scope_named(default_scope, local_name!("select")) {
+                        self.generate_implied_end_except(local_name!("optgroup"));
+                        if self.in_scope_named(default_scope, local_name!("option")) {
+                            self.sink.parse_error(Borrowed("nested options"));
+                        }
+                    } else {
+                        if self.current_node_named(local_name!("option")) {
+                            self.pop();
+                        }
+                        self.reconstruct_formatting();
                     }
-                    self.reconstruct_formatting();
+
                     self.insert_element_for(tag);
                     Done
                 }
 
+                tag @ <optgroup> => {
+                    if self.in_scope_named(default_scope, local_name!("select")) {
+                        self.generate_implied_end(cursory_implied_end);
+                        // XXX: perf
+                        if self.in_scope_named(default_scope, local_name!("option")) || self.in_scope_named(default_scope, local_name!("optgroup")) {
+                            self.sink.parse_error(Borrowed("nested options"));
+                        }
+                    } else {
+                        if self.current_node_named(local_name!("option")) {
+                            self.pop();
+                        }
+                        self.reconstruct_formatting();
+                    }
+                    Done
+                }
+
                 tag @ <rb> <rtc> => {
                     if self.in_scope_named(default_scope, local_name!("ruby")) {
                         self.generate_implied_end(cursory_implied_end);
@@ -1100,121 +1129,6 @@ where
                 token => self.step(InBody, token),
             }),
 
-            //§ parsing-main-inselect
-            InSelect => match_token!(token {
-                NullCharacterToken => self.unexpected(&token),
-                CharacterTokens(_, text) => self.append_text(text),
-                CommentToken(text) => self.append_comment(text),
-
-                <html> => self.step(InBody, token),
-
-                tag @ <option> => {
-                    if self.current_node_named(local_name!("option")) {
-                        self.pop();
-                    }
-                    self.insert_element_for(tag);
-                    Done
-                }
-
-                tag @ <optgroup> => {
-                    if self.current_node_named(local_name!("option")) {
-                        self.pop();
-                    }
-                    if self.current_node_named(local_name!("optgroup")) {
-                        self.pop();
-                    }
-                    self.insert_element_for(tag);
-                    Done
-                }
-
-                tag @ <hr> => {
-                    if self.current_node_named(local_name!("option")) {
-                        self.pop();
-                    }
-                    if self.current_node_named(local_name!("optgroup")) {
-                        self.pop();
-                    }
-                    self.insert_element_for(tag);
-                    self.pop();
-                    DoneAckSelfClosing
-                }
-
-                </optgroup> => {
-                    if self.open_elems.borrow().len() >= 2
-                        && self.current_node_named(local_name!("option"))
-                        && self.html_elem_named(&self.open_elems.borrow()[self.open_elems.borrow().len() - 2],
-                            local_name!("optgroup")) {
-                        self.pop();
-                    }
-                    if self.current_node_named(local_name!("optgroup")) {
-                        self.pop();
-                    } else {
-                        self.unexpected(&token);
-                    }
-                    Done
-                }
-
-                </option> => {
-                    if self.current_node_named(local_name!("option")) {
-                        self.pop();
-                    } else {
-                        self.unexpected(&token);
-                    }
-                    Done
-                }
-
-                tag @ <select> </select> => {
-                    let in_scope = self.in_scope_named(select_scope, local_name!("select"));
-
-                    if !in_scope || tag.kind == StartTag {
-                        self.unexpected(&tag);
-                    }
-
-                    if in_scope {
-                        self.pop_until_named(local_name!("select"));
-                        self.mode.set(self.reset_insertion_mode());
-                    }
-                    Done
-                }
-
-                <input> <keygen> <textarea> => {
-                    self.unexpected(&token);
-                    if self.in_scope_named(select_scope, local_name!("select")) {
-                        self.pop_until_named(local_name!("select"));
-                        Reprocess(self.reset_insertion_mode(), token)
-                    } else {
-                        Done
-                    }
-                }
-
-                <script> <template> </template> => self.step(InHead, token),
-
-                EOFToken => self.step(InBody, token),
-
-                token => self.unexpected(&token),
-            }),
-
-            //§ parsing-main-inselectintable
-            InSelectInTable => match_token!(token {
-                <caption> <table> <tbody> <tfoot> <thead> <tr> <td> <th> => {
-                    self.unexpected(&token);
-                    self.pop_until_named(local_name!("select"));
-                    Reprocess(self.reset_insertion_mode(), token)
-                }
-
-                tag @ </caption> </table> </tbody> </tfoot> </thead> </tr> </td> </th> => {
-                    self.unexpected(&tag);
-                    if self.in_scope_named(table_scope, tag.name.clone()) {
-                        self.pop_until_named(local_name!("select"));
-                        Reprocess(self.reset_insertion_mode(), TagToken(tag))
-                    } else {
-                        Done
-                    }
-                }
-
-                token => self.step(InSelect, token),
-            }),
-
             //§ parsing-main-intemplate
             InTemplate => match_token!(token {
                 CharacterTokens(_, _) => self.step(InBody, token),
diff --git a/html5ever/src/tree_builder/tag_sets.rs b/html5ever/src/tree_builder/tag_sets.rs
index 0d587973..156da246 100644
--- a/html5ever/src/tree_builder/tag_sets.rs
+++ b/html5ever/src/tree_builder/tag_sets.rs
@@ -48,13 +48,9 @@ macro_rules! declare_tag_set (
 pub(crate) fn empty_set(_: ExpandedName) -> bool {
     false
 }
-#[inline(always)]
-pub(crate) fn full_set(_: ExpandedName) -> bool {
-    true
-}
 
 declare_tag_set!(pub html_default_scope =
-    "applet" "caption" "html" "table" "td" "th" "marquee" "object" "template");
+    "applet" "caption" "html" "table" "td" "th" "marquee" "object" "select" "template");
 
 #[inline(always)]
 pub(crate) fn default_scope(name: ExpandedName) -> bool {
@@ -66,7 +62,6 @@ pub(crate) fn default_scope(name: ExpandedName) -> bool {
 declare_tag_set!(pub list_item_scope = [default_scope] + "ol" "ul");
 declare_tag_set!(pub button_scope = [default_scope] + "button");
 declare_tag_set!(pub table_scope = "html" "table" "template");
-declare_tag_set!(pub select_scope = [full_set] - "optgroup" "option");
 
 declare_tag_set!(pub table_body_context = "tbody" "tfoot" "thead" "template" "html");
 declare_tag_set!(pub table_row_context = "tr" "template" "html");
diff --git a/html5ever/src/tree_builder/types.rs b/html5ever/src/tree_builder/types.rs
index f6bb588d..ff9cf247 100644
--- a/html5ever/src/tree_builder/types.rs
+++ b/html5ever/src/tree_builder/types.rs
@@ -38,8 +38,6 @@ pub(crate) enum InsertionMode {
     InTableBody,
     InRow,
     InCell,
-    InSelect,
-    InSelectInTable,
     InTemplate,
     AfterBody,
     InFrameset,
diff --git a/rcdom/html5lib-tests b/rcdom/html5lib-tests
index c67f90ea..ee3ebba3 160000
--- a/rcdom/html5lib-tests
+++ b/rcdom/html5lib-tests
@@ -1 +1 @@
-Subproject commit c67f90eacac14e022b1f2c2e5ac559879581e9ff
+Subproject commit ee3ebba3e10c28b45f62b8e75887d5948f799aa9
diff --git a/rcdom/tests/html-tree-builder.rs b/rcdom/tests/html-tree-builder.rs
index 58b1bc6b..3bf22ee8 100644
--- a/rcdom/tests/html-tree-builder.rs
+++ b/rcdom/tests/html-tree-builder.rs
@@ -189,14 +189,13 @@ fn make_test_desc_with_scripting_flag(
     fields: &HashMap<String, String>,
     scripting_enabled: bool,
 ) -> Test {
-    let get_field = |key| {
-        let field = fields.get(key).expect("missing field");
-        field.trim_end_matches('\n').to_string()
+    let expect_field = |key| {
+        fields.get(key).unwrap_or_else(|| panic!("missing field {}, testcase: {:?}", key, fields)).to_string()
     };
 
-    let mut data = fields.get("data").expect("missing data").to_string();
+    let mut data = expect_field("data");
     data.pop();
-    let expected = get_field("document");
+    let expected = expect_field("document").trim_end_matches('\n').to_string();
     let context = fields
         .get("document-fragment")
         .map(|field| context_name(field.trim_end_matches('\n')));

From edf84995c19513e886ba6fbb97e1463f72093371 Mon Sep 17 00:00:00 2001
From: Markus Unterwaditzer <markus-tarpit+git@unterwaditzer.net>
Date: Wed, 30 Oct 2024 23:22:39 +0100
Subject: [PATCH 2/3] fmt

---
 rcdom/tests/html-tree-builder.rs | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/rcdom/tests/html-tree-builder.rs b/rcdom/tests/html-tree-builder.rs
index 3bf22ee8..f5ba4c64 100644
--- a/rcdom/tests/html-tree-builder.rs
+++ b/rcdom/tests/html-tree-builder.rs
@@ -190,7 +190,10 @@ fn make_test_desc_with_scripting_flag(
     scripting_enabled: bool,
 ) -> Test {
     let expect_field = |key| {
-        fields.get(key).unwrap_or_else(|| panic!("missing field {}, testcase: {:?}", key, fields)).to_string()
+        fields
+            .get(key)
+            .unwrap_or_else(|| panic!("missing field {}, testcase: {:?}", key, fields))
+            .to_string()
     };
 
     let mut data = expect_field("data");

From 4f8c7bb82ed90bfba018890fc01da80d0cd0f0b3 Mon Sep 17 00:00:00 2001
From: Markus Unterwaditzer <markus-tarpit+git@unterwaditzer.net>
Date: Wed, 30 Oct 2024 23:48:38 +0100
Subject: [PATCH 3/3] whoops

---
 html5ever/src/tree_builder/rules.rs | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/html5ever/src/tree_builder/rules.rs b/html5ever/src/tree_builder/rules.rs
index 060915d1..b8c5d629 100644
--- a/html5ever/src/tree_builder/rules.rs
+++ b/html5ever/src/tree_builder/rules.rs
@@ -710,6 +710,8 @@ where
                         }
                         self.reconstruct_formatting();
                     }
+
+                    self.insert_element_for(tag);
                     Done
                 }