From 9f686ebda68fdf021418aa1310f70c84e100baf2 Mon Sep 17 00:00:00 2001
From: honeybadgerdontcare <honeybadgerdontcare@users.noreply.github.com>
Date: Tue, 29 Nov 2016 23:09:26 +0100
Subject: [PATCH] Validator rollup (#6391)

* Update the set of characters which the javascript HTML parser allows in HTML tags after the first character in the tag.

* Update the set of characters which the javascript HTML parser allows in HTML tags after the first character in the tag.

* Avoid mutating the argument that's passed to validateSaxEvents.

* Fix a bug in the javascript HTML parser where we don't change the section when manufacturing a closing HEAD tag.

* Allow `amp-accordion` to have `disable-session-states` attribute.
---
 .../0.1/validator-amp-accordion.protoascii          |  1 +
 validator/engine/htmlparser.js                      | 13 ++++++++++---
 validator/engine/validator.js                       |  4 +---
 validator/validator-main.protoascii                 |  2 +-
 4 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/extensions/amp-accordion/0.1/validator-amp-accordion.protoascii b/extensions/amp-accordion/0.1/validator-amp-accordion.protoascii
index af225d3fe0f0..ec9c5b2b1fac 100644
--- a/extensions/amp-accordion/0.1/validator-amp-accordion.protoascii
+++ b/extensions/amp-accordion/0.1/validator-amp-accordion.protoascii
@@ -54,6 +54,7 @@ tags: {  # <amp-accordion>
   tag_name: "AMP-ACCORDION"
   also_requires_tag: "amp-accordion extension .js script"
   attrs: { name: "animate" value: "" }
+  attrs: { name: "disable-session-states" value: "" }
   child_tags: {
     child_tag_name_oneof: "SECTION"
   }
diff --git a/validator/engine/htmlparser.js b/validator/engine/htmlparser.js
index 0571ecc98daf..9d07f668dc9e 100644
--- a/validator/engine/htmlparser.js
+++ b/validator/engine/htmlparser.js
@@ -234,6 +234,8 @@ class TagNameStack {
             if (this.handler_.markManufacturedBody)
               this.handler_.markManufacturedBody();
             this.startTag('BODY', []);
+          } else {
+            this.region_ = TagRegion.IN_BODY;
           }
         }
         break;
@@ -836,8 +838,9 @@ amp.htmlparser.HtmlParser.HEX_ESCAPE_RE_ = /^#x([0-9A-Fa-f]+)$/;
  * @private
  */
 amp.htmlparser.HtmlParser.INSIDE_TAG_TOKEN_ = new RegExp(
-    // Don't capture space.
-    '^\\s*(?:' +
+    // Don't capture space. In this case, we don't use \s because it includes a
+    // nonbreaking space which gets included as an attribute in our validation.
+    '^[ \\t\\n\\f\\r\\v]*(?:' +
         // Capture an attribute name in group 1, and value in group 3.
         // We capture the fact that there was an attribute in group 2, since
         // interpreters are inconsistent in whether a group that matches nothing
@@ -889,7 +892,11 @@ amp.htmlparser.HtmlParser.OUTSIDE_TAG_TOKEN_ = new RegExp(
         // Comments not captured.
         '|<[!]--[\\s\\S]*?(?:-->|$)' +
         // '/' captured in group 2 for close tags, and name captured in group 3.
-        '|<(/)?([a-z!\\?][a-z0-9_:-]*)' +
+        // The first character of a tag (after possibly '/') can be A-Z, a-z,
+        // '!' or '?'. The remaining characters are more easily expressed as a
+        // negative set of: '\0', ' ', '\n', '\r', '\t', '\f', '\v', '>', or
+        // '/'.
+        '|<(/)?([a-z!\\?][^\\0 \\n\\r\\t\\f\\v>/]*)' +
         // Text captured in group 4.
         '|([^<&>]+)' +
         // Cruft captured in group 5.
diff --git a/validator/engine/validator.js b/validator/engine/validator.js
index b3c802270bd4..a2bd4cf258a5 100644
--- a/validator/engine/validator.js
+++ b/validator/engine/validator.js
@@ -4890,9 +4890,7 @@ amp.validator.validateSaxEvents = function(saxEvents, htmlFormat) {
   for (const e of saxEvents) {
     switch (e[0]) {
       case 'startTag':
-        e.shift();
-        const tagName = e.shift();
-        handler.startTag(tagName, e);
+        handler.startTag(/*tagName=*/e[1], e.slice(2));
         break;
       case 'endTag':
         handler.endTag(e[1]);
diff --git a/validator/validator-main.protoascii b/validator/validator-main.protoascii
index 0ce5ca59378a..42e1a55657d7 100644
--- a/validator/validator-main.protoascii
+++ b/validator/validator-main.protoascii
@@ -25,7 +25,7 @@ min_validator_revision_required: 189
 # newer versions of the spec file. This is currently a Google internal
 # mechanism, validator.js does not use this facility. However, any
 # change to this file (validator-main.js) requires updating this revision id.
-spec_file_revision: 315
+spec_file_revision: 316
 
 # Validator extensions.
 # =====================