+ # 8.2.4.8 http://www.w3.org/TR/html5/syntax.html#tag-open-state
+ tok_state_tag_open = ->
+ switch c = txt.charAt(cur++)
+ when '!'
+ tok_state = tok_state_markup_declaration_open
+ when '/'
+ tok_state = tok_state_end_tag_open
+ when '?'
+ # Parse error
+ tok_state = tok_state_bogus_comment
+ else
+ if lc_alpha.indexOf(c) > -1
+ tok_cur_tag = [TYPE_OPEN_TAG, c, [], []]
+ tok_state = tok_state_tag_name
+ else if uc_alpha.indexOf(c) > -1
+ tok_cur_tag = [TYPE_OPEN_TAG, c.toLowerCase(), [], []]
+ tok_state = tok_state_tag_name
+ else
+ # Parse error
+ tok_state = tok_state_data
+ cur -= 1 # we didn't parse/handle the char after <
+ return [TYPE_TEXT, '<']
+ return null
+
+ # 8.2.4.10 http://www.w3.org/TR/html5/syntax.html#tag-name-state
+ tok_state_tag_name = ->
+ switch c = txt.charAt(cur++)
+ when "\t", "\n", "\u000c", ' '
+ tok_state = tok_state_before_attribute_name
+ when '/'
+ tok_state = tok_state_self_closing_start_tag
+ when '>'
+ tok_state = tok_state_data
+ tmp = tok_cur_tag
+ tok_cur_tag = null
+ return tmp
+ when "\u0000"
+ # Parse error
+ tok_cur_tag[1] += "\ufffd"
+ else
+ if uc_alpha.indexOf(c) > -1
+ tok_cur_tag[1] += c.toLowerCase()
+ else
+ tok_cur_tag[1] += c
+ return null
+
+ # 8.2.4.34 http://www.w3.org/TR/html5/syntax.html#before-attribute-name-state
+ tok_state_before_attribute_name = ->
+ attr_name = null
+ switch c = txt.charAt(cur++)
+ when "\t", "\n", "\u000c", ' '
+ return null
+ when '/'
+ tok_state = tok_state_self_closing_start_tag
+ return null
+ when '>'
+ tok_state = tok_state_data
+ tmp = tok_cur_tag
+ tok_cur_tag = null
+ return tmp
+ when "\u0000"
+ # Parse error
+ attr_name = "\ufffd"
+ when '"', "'", '<', '='
+ # Parse error
+ attr_name = c
+ else
+ if uc_alpha.indexOf(c) > -1
+ attr_name = c.toLowerCase()
+ else
+ attr_name = c
+ if attr_name?
+ tok_cur_tag[2].unshift [attr_name, '']
+ tok_state = tok_state_attribute_name
+ return null
+
+ # 8.2.4.35 http://www.w3.org/TR/html5/syntax.html#attribute-name-state
+ tok_state_attribute_name = ->
+ switch c = txt.charAt(cur++)
+ when "\t", "\n", "\u000c", ' '
+ tok_state = tok_state_after_attribute_name
+ when '/'
+ tok_state = tok_state_self_closing_start_tag
+ when '='
+ tok_state = tok_state_before_attribute_value
+ when '>'
+ tok_state = tok_state_data
+ tmp = tok_cur_tag
+ tok_cur_tag = null
+ return tmp
+ when "\u0000"
+ # Parse error
+ tok_cur_tag[2][0][0] += "\ufffd"
+ else
+ if uc_alpha.indexOf(c) > -1
+ tok_cur_tag[2][0][0] += c.toLowerCase()
+ else
+ # Parse error if ", ' or <
+ tok_cur_tag[2][0][0] += c
+ return null
+
+ # 8.2.4.37 http://www.w3.org/TR/html5/syntax.html#before-attribute-value-state
+ tok_state_before_attribute_value = ->
+ switch c = txt.charAt(cur++)
+ when "\t", "\n", "\u000c", ' '
+ return null
+ when '"'
+ tok_state = tok_state_attribute_value_double_quoted
+ when '&'
+ tok_state = tok_state_attribute_value_unquoted
+ cur -= 1
+ when "'"
+ tok_state = tok_state_attribute_value_single_quoted
+ when "\u0000"
+ # Parse error
+ tok_cur_tag[2][0][1] += "\ufffd"
+ tok_state = tok_state_attribute_value_unquoted
+ when '>'
+ # Parse error
+ tok_state = tok_state_data
+ tmp = tok_cur_tag
+ tok_cur_tag = null
+ return tmp
+ else
+ 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
+ tok_state_attribute_value_double_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
+
+ # 8.2.4.39 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++)
+ when "\t", "\n", "\u000c", ' '
+ tok_state = tok_state_before_attribute_name
+ when '/'
+ tok_state = tok_state_self_closing_start_tag
+ when '>'
+ tok_state = tok_state_data
+ tmp = tok_cur_tag
+ tok_cur_tag = null
+ return tmp
+ else
+ # Parse Error
+ tok_state = tok_state_before_attribute_name
+ cur -= 1 # we didn't handle that char
+ return null
+