From 4fc9496829b82b7446c208e2d893e2c863291f36 Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Sun, 8 Aug 2021 19:00:51 +1000
Subject: [PATCH] Fix references in onepass

---
 gen/codepoints.ts                 |  5 +++++
 rust/onepass/src/lib.rs           | 10 +++++-----
 rust/onepass/src/proc/mod.rs      |  7 +++++++
 rust/onepass/src/unit/attr/mod.rs |  4 ++--
 rust/onepass/src/unit/content.rs  | 10 +++++-----
 rust/onepass/src/unit/tag.rs      |  4 ++--
 6 files changed, 26 insertions(+), 14 deletions(-)

diff --git a/gen/codepoints.ts b/gen/codepoints.ts
index 0c1f1aa..cc1470c 100644
--- a/gen/codepoints.ts
+++ b/gen/codepoints.ts
@@ -46,6 +46,10 @@ const WHITESPACE_OR_SLASH_OR_EQUALS_OR_RIGHT_CHEVRON = [
 
 const DOUBLE_QUOTE = [c('"')];
 const SINGLE_QUOTE = [c("'")];
+// Official characters allowed in an attribute name.
+// NOTE: Unicode noncharacters not tested.
+// See https://html.spec.whatwg.org/multipage/syntax.html#syntax-attribute-name for spec.
+const WHATWG_ATTR_NAME_CHAR = invert([...CONTROL, c(' '), c('"'), c('\''), c('>'), c('/'), c('=')]);
 // Valid attribute quote characters.
 // See https://html.spec.whatwg.org/multipage/introduction.html#intro-early-example for spec.
 // Backtick is not a valid quote character according to spec.
@@ -88,6 +92,7 @@ impl std::ops::Index<u8> for Lookup {
     WHITESPACE_OR_SLASH,
     WHITESPACE_OR_SLASH_OR_EQUALS_OR_RIGHT_CHEVRON,
 
+    WHATWG_ATTR_NAME_CHAR,
     DOUBLE_QUOTE,
     SINGLE_QUOTE,
     ATTR_QUOTE,
diff --git a/rust/onepass/src/lib.rs b/rust/onepass/src/lib.rs
index d23e9e0..3ab0edb 100644
--- a/rust/onepass/src/lib.rs
+++ b/rust/onepass/src/lib.rs
@@ -22,7 +22,7 @@ mod unit;
 /// # Examples
 ///
 /// ```
-/// use minify_html::{Cfg, Error, in_place};
+/// use minify_html_onepass::{Cfg, Error, in_place};
 ///
 /// let mut code = b"<p>  Hello, world!  </p>".to_vec();
 /// let cfg = &Cfg {
@@ -62,7 +62,7 @@ pub fn in_place(code: &mut [u8], cfg: &Cfg) -> Result<usize, Error> {
 /// # Examples
 ///
 /// ```
-/// use minify_html::{Cfg, Error, in_place_str};
+/// use minify_html_onepass::{Cfg, Error, in_place_str};
 ///
 /// let mut code = "<p>  Hello, world!  </p>".to_string();
 /// let cfg = &Cfg {
@@ -92,7 +92,7 @@ pub fn in_place_str<'s>(code: &'s mut str, cfg: &Cfg) -> Result<&'s str, Error>
 /// # Examples
 ///
 /// ```
-/// use minify_html::{Cfg, Error, truncate};
+/// use minify_html_onepass::{Cfg, Error, truncate};
 ///
 /// let mut code = b"<p>  Hello, world!  </p>".to_vec();
 /// let cfg = &Cfg {
@@ -125,7 +125,7 @@ pub fn truncate(code: &mut Vec<u8>, cfg: &Cfg) -> Result<(), Error> {
 /// # Examples
 ///
 /// ```
-/// use minify_html::{Cfg, Error, copy};
+/// use minify_html_onepass::{Cfg, Error, copy};
 ///
 /// let mut code: &[u8] = b"<p>  Hello, world!  </p>";
 /// let cfg = &Cfg {
@@ -164,7 +164,7 @@ pub fn copy(code: &[u8], cfg: &Cfg) -> Result<Vec<u8>, Error> {
 /// # Examples
 ///
 /// ```
-/// use minify_html::{Cfg, FriendlyError, with_friendly_error};
+/// use minify_html_onepass::{Cfg, FriendlyError, with_friendly_error};
 ///
 /// let mut code = b"<p></div>".to_vec();
 /// let cfg = &Cfg {
diff --git a/rust/onepass/src/proc/mod.rs b/rust/onepass/src/proc/mod.rs
index 712347e..2baffd0 100644
--- a/rust/onepass/src/proc/mod.rs
+++ b/rust/onepass/src/proc/mod.rs
@@ -16,6 +16,7 @@ use crate::proc::range::ProcessorRange;
 use crate::proc::MatchAction::*;
 use crate::proc::MatchMode::*;
 use minify_html_common::gen::codepoints::Lookup;
+use minify_html_common::spec::tag::EMPTY_SLICE;
 
 pub mod checkpoint;
 pub mod entity;
@@ -239,6 +240,12 @@ impl<'d> Processor<'d> {
         !self._in_bounds(0)
     }
 
+    #[inline(always)]
+    pub fn get_or_empty(&self, r: Option<ProcessorRange>) -> &[u8] {
+        r.and_then(|r| self.code.get(r.start..r.end))
+            .unwrap_or(EMPTY_SLICE)
+    }
+
     #[inline(always)]
     pub fn require_not_at_end(&self) -> ProcessingResult<()> {
         if self.at_end() {
diff --git a/rust/onepass/src/unit/attr/mod.rs b/rust/onepass/src/unit/attr/mod.rs
index 8bc78cc..454b244 100644
--- a/rust/onepass/src/unit/attr/mod.rs
+++ b/rust/onepass/src/unit/attr/mod.rs
@@ -8,7 +8,7 @@ use crate::unit::attr::value::{
     process_attr_value, skip_attr_value, DelimiterType, ProcessedAttrValue,
 };
 use minify_html_common::gen::attrs::ATTRS;
-use minify_html_common::gen::codepoints::{ATTR_NAME_CHAR, WHITESPACE};
+use minify_html_common::gen::codepoints::{WHATWG_ATTR_NAME_CHAR, WHITESPACE};
 use minify_html_common::spec::tag::ns::Namespace;
 
 mod value;
@@ -34,7 +34,7 @@ pub fn process_attr(
     // It's possible to expect attribute name but not be called at an attribute, e.g. due to whitespace between name and
     // value, which causes name to be considered boolean attribute and `=` to be start of new (invalid) attribute name.
     let name = proc
-        .m(WhileInLookup(ATTR_NAME_CHAR), Keep)
+        .m(WhileInLookup(WHATWG_ATTR_NAME_CHAR), Keep)
         .require("attribute name")?;
     proc.make_lowercase(name);
     let attr_cfg = ATTRS.get(ns, &proc[element], &proc[name]);
diff --git a/rust/onepass/src/unit/content.rs b/rust/onepass/src/unit/content.rs
index 7363899..5d29250 100644
--- a/rust/onepass/src/unit/content.rs
+++ b/rust/onepass/src/unit/content.rs
@@ -64,7 +64,7 @@ pub fn process_content(
         collapse,
         destroy_whole,
         trim,
-    } = get_whitespace_minification_for_tag(parent.map(|r| &proc[r]), descendant_of_pre);
+    } = get_whitespace_minification_for_tag(proc.get_or_empty(parent), descendant_of_pre);
 
     let handle_ws = collapse || destroy_whole || trim;
 
@@ -146,7 +146,7 @@ pub fn process_content(
                     .require("tag name")?;
                 proc.make_lowercase(tag_name);
 
-                if can_omit_as_before(proc, parent, tag_name) {
+                if can_omit_as_before(proc.get_or_empty(parent), &proc[tag_name]) {
                     // TODO Is this necessary? Can a previous closing tag even exist?
                     prev_sibling_closing_tag.write_if_exists(proc);
                     tag_checkpoint.restore(proc);
@@ -169,9 +169,9 @@ pub fn process_content(
                 prev_sibling_closing_tag.replace(new_closing_tag);
             }
             ContentType::End => {
-                if prev_sibling_closing_tag
-                    .exists_and(|prev_tag| !can_omit_as_last_node(proc, parent, prev_tag))
-                {
+                if prev_sibling_closing_tag.exists_and(|prev_tag| {
+                    !can_omit_as_last_node(proc.get_or_empty(parent), &proc[prev_tag])
+                }) {
                     prev_sibling_closing_tag.write(proc);
                 };
                 break;
diff --git a/rust/onepass/src/unit/tag.rs b/rust/onepass/src/unit/tag.rs
index 05333d7..5acd641 100644
--- a/rust/onepass/src/unit/tag.rs
+++ b/rust/onepass/src/unit/tag.rs
@@ -107,7 +107,7 @@ pub fn process_tag(
     source_tag_name: ProcessorRange,
 ) -> ProcessingResult<MaybeClosingTag> {
     if prev_sibling_closing_tag
-        .exists_and(|prev_tag| !can_omit_as_before(proc, Some(prev_tag), source_tag_name))
+        .exists_and(|prev_tag| !can_omit_as_before(&proc[prev_tag], &proc[source_tag_name]))
     {
         prev_sibling_closing_tag.write(proc);
     };
@@ -237,7 +237,7 @@ pub fn process_tag(
         }
     };
 
-    let can_omit_closing_tag = can_omit_as_last_node(proc, parent, tag_name);
+    let can_omit_closing_tag = can_omit_as_last_node(proc.get_or_empty(parent), &proc[tag_name]);
     if closing_tag_omitted || proc.at_end() && can_omit_closing_tag {
         return Ok(MaybeClosingTag(None));
     };