parse unquoted and singlequoted attributes

diff --git a/parse-html.coffee b/parse-html.coffee
index 204c6ff..1ef077a 100644 (file)
--- a/parse-html.coffee
+++ b/parse-html.coffee
@@ -361,11 +361,8 @@ parse_html = (txt) ->
                                tok_cur_tag = null
                                return tmp
                        else
-                               if uc_alpha.indexOf(c) > -1
-                                       tok_cur_tag[2][0][1] += c.toLowerCase()
-                               else
-                                       # Parse error if ", ` or < (that's a backtick)
-                                       tok_cur_tag[2][0][1] += c
+                               tok_cur_tag[2][0][1] += c
+                               tok_state = tok_state_attribute_value_unquoted
                return null
 
        # 8.2.4.38 http://www.w3.org/TR/html5/syntax.html#attribute-value-(double-quoted)-state
@@ -379,11 +376,45 @@ parse_html = (txt) ->
                        when "\u0000"
                                # Parse error
                                tok_cur_tag[2][0][1] += "\ufffd"
-                               tok_state = tok_state_attribute_value_unquoted
                        else
                                tok_cur_tag[2][0][1] += c
                return null
 
+       # 8.2.4.39 http://www.w3.org/TR/html5/syntax.html#attribute-value-(single-quoted)-state
+       tok_state_attribute_value_single_quoted = ->
+               switch c = txt.charAt(cur++)
+                       when "'"
+                               tok_state = tok_state_after_attribute_value_quoted
+                       when '&'
+                               tok_state = tok_state_character_reference_in_attribute_value
+                               tok_char_ref_addl_allowed = "'" # FIXME
+                       when "\u0000"
+                               # Parse error
+                               tok_cur_tag[2][0][1] += "\ufffd"
+                       else
+                               tok_cur_tag[2][0][1] += c
+               return null
+
+       # 8.2.4.40 http://www.w3.org/TR/html5/syntax.html#attribute-value-(unquoted)-state
+       tok_state_attribute_value_unquoted = ->
+               switch c = txt.charAt(cur++)
+                       when "\t", "\n", "\u000c", ' '
+                               tok_state = tok_state_before_attribute_name
+                       when '&'
+                               tok_state = tok_state_character_reference_in_attribute_value
+                               tok_char_ref_addl_allowed = '>' # FIXME
+                       when '>'
+                               tok_state = tok_state_data
+                               tmp = tok_cur_tag
+                               tok_cur_tag = null
+                               return tmp
+                       when "\u0000"
+                               tok_cur_tag[2][0][1] += "\ufffd"
+                       else
+                               # Parse Error if ', <, = or ` (backtick)
+                               tok_cur_tag[2][0][1] += c
+               return null
+
        # 8.2.4.42 http://www.w3.org/TR/html5/syntax.html#after-attribute-value-(quoted)-state
        tok_state_after_attribute_value_quoted = ->
                switch c = txt.charAt(cur++)
@@ -450,4 +481,5 @@ test_equals "named entity", html_to_json, "a&amp;1234", '[[1,"a&1234"]]'
 test_equals "broken named character references", html_to_json, "1&amp2&&amp;3&aabbcc;", '[[1,"1&2&&3&aabbcc;"]]'
 test_equals "numbered entity overrides", html_to_json, "1&#X80&#x80; &#x83", '[[1,"1€€ ƒ"]]'
 test_equals "open tag", html_to_json, "foo<span>bar", '[[1,"foo"],[0,"span",{},[[1,"bar"]]]]'
-test_equals "open tag with attributes", html_to_json, "foo<span style=\"foo: bar\">bar", '[[1,"foo"],[0,"span",{"style":"foo: bar"},[[1,"bar"]]]]'
+test_equals "open tag with attributes", html_to_json, "foo<span style=\"foo: bar\" title=\"hi\">bar", '[[1,"foo"],[0,"span",{"style":"foo: bar","title":"hi"},[[1,"bar"]]]]'
+test_equals "open tag with attributes of various quotings", html_to_json, "foo<span abc=\"def\" g=hij klm='nopqrstuv\"' autofocus>bar", '[[1,"foo"],[0,"span",{"abc":"def","g":"hij","klm":"nopqrstuv\\\"","autofocus":""},[[1,"bar"]]]]'