[htm8] Fix AttrLexer bug with Missing value

The state wasn't reset between iterations.  Gah.

Stateful APIs need extra care.  I could probably combine ReadName() and
ReadValue(), although we would then return a 6-tuple!

And the function would be long.

data_lang/htm8.py

@@ -591,12 +591,13 @@ def ReadName(self):
                         self.next_value_is_missing = True
                         # HACK: REWIND, since we don't want to consume whitespace
                         self.pos = self.name_end
+                    else:
+                        self.next_value_is_missing = False
                     return attr_name.Ok, self.name_start, self.name_end
                 else:
                     # Reset state - e.g. you must call AttrNameEquals
                     self.name_start = -1
                     self.name_end = -1
-                    self.next_value_is_missing = False
 
                 if a == attr_name.Done:
                     return attr_name.Done, -1, -1
@@ -746,8 +747,8 @@ def AllAttrsRaw(attr_lx):
     while True:
         n, name_start, name_end = attr_lx.ReadName()
         if 0:
-            log('  AllAttrsRaw ==> ReadName %s %d %d', attr_name_str(n),
-                name_start, name_end)
+            log('  AllAttrsRaw ==> ReadName %s %d %d %r', attr_name_str(n),
+                name_start, name_end, attr_lx.s[attr_lx.pos:attr_lx.pos + 10])
         if n == attr_name.Ok:
             name = attr_lx.s[name_start:name_end]
             #log('  Name %r', name)

data_lang/htm8_test.py

@@ -391,7 +391,8 @@ def ValidTokenList(s, no_special_tags=False):
         if tok_id == h8_id.EndOfStream:
             break
         if tok_id == h8_id.Invalid:
-            raise htm8.LexError(s, start_pos)
+            raise htm8.LexError('ValidTokenList() got invalid token', s,
+                                start_pos)
         start_pos = end_pos
     return tokens
 

lazylex/html_test.py

@@ -245,6 +245,10 @@ def testValid(self):
     ('<STYLE><</STYLE>', ''),
     #'<SCRipt><</script>',
 
+    # Regression test from blog
+    ('<script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>',
+     '')
+
     # Note: Python HTMLParser.py does DYNAMIC compilation of regex with re.I
     # flag to handle this!  Gah I want something faster.
     #'<script><</SCRIPT>',