From: Jason Woofenden <jason@jasonwoof.com>
Date: Mon, 14 Dec 2015 20:21:24 +0000 (-0500)
Subject: parse end tags, close tags with proper nesting
X-Git-Url: https://jasonwoof.com/gitweb/?a=commitdiff_plain;h=50c30e8b7c0663890369776992b1f57876e6226e;p=peach-html5-editor.git

parse end tags, close tags with proper nesting
---

diff --git a/parse-html.coffee b/parse-html.coffee
index a9e1ce0..db1837b 100644
--- a/parse-html.coffee
+++ b/parse-html.coffee
@@ -32,7 +32,8 @@ TYPE_WHITESPACE = 2
 TYPE_COMMENT = 3
 # the following types are emited by the tokenizer, but shouldn't end up in the tree:
 TYPE_OPEN_TAG = 4 # name, [attributes ([key,value]...) in reverse order], [children]
-TYPE_EOF = 5
+TYPE_CLOSE_TAG = 5 # name
+TYPE_EOF = 6
 
 lc_alpha = "abcdefghijklmnopqrstuvwxqz"
 uc_alpha = "ABCDEFGHIJKLMNOPQRSTUVWXQZ"
@@ -210,6 +211,28 @@ parse_html = (txt) ->
 					return [TYPE_TEXT, '<']
 		return null
 
+	# 8.2.4.9 http://www.w3.org/TR/html5/syntax.html#end-tag-open-state
+	tok_state_end_tag_open = ->
+		switch c = txt.charAt(cur++)
+			when '>'
+				parse_error()
+				tok_state = tok_state_data
+			when '' # EOF
+				parse_error()
+				tok_state = tok_state_data
+				return [TYPE_TEXT, '</']
+			else
+				if uc_alpha.indexOf(c) > -1
+					tok_cur_tag = [TYPE_CLOSE_TAG, c.toLowerCase(), [], []]
+					tok_state = tok_state_tag_name
+				else if lc_alpha.indexOf(c) > -1
+					tok_cur_tag = [TYPE_CLOSE_TAG, c, [], []]
+					tok_state = tok_state_tag_name
+				else
+					parse_error()
+					tok_state = tok_state_bogus_comment
+		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++)
@@ -503,12 +526,22 @@ parse_html = (txt) ->
 				open_tags[0][3].push t
 				open_tags.unshift t
 				# TODO implement formatting elements thing
+			when TYPE_CLOSE_TAG
+				# FIXME this is just a hack for now
+				if open_tags.length < 2
+					parse_error()
+					return
+				if open_tags[0][1] isnt t[1]
+					parse_error()
+					# fall through and close something anyway
+				open_tags.shift()
 			when TYPE_EOF
 				return
 			# TODO implement close tags
 			# TODO implement self-closing tags
 			else
 				console.log "UNIMPLEMENTED tag type: #{t[0]}"
+		return
 
 	# tree constructor initialization
 	# see comments on TYPE_TAG/etc for the structure of this data
@@ -550,3 +583,4 @@ test_equals "open tag with attributes of various quotings", html_to_json, "foo<s
 test_equals "attribute entity exceptions dq", html_to_json, "foo<a href=\"foo?t=1&amp=2&ampo=3&amp;lt=foo\">bar", '[[1,"foo"],[0,"a",{"href":"foo?t=1&amp=2&ampo=3&lt=foo"},[[1,"bar"]]]]'
 test_equals "attribute entity exceptions sq", html_to_json, "foo<a href='foo?t=1&amp=2&ampo=3&amp;lt=foo'>bar", '[[1,"foo"],[0,"a",{"href":"foo?t=1&amp=2&ampo=3&lt=foo"},[[1,"bar"]]]]'
 test_equals "attribute entity exceptions uq", html_to_json, "foo<a href=foo?t=1&amp=2&ampo=3&amp;lt=foo>bar", '[[1,"foo"],[0,"a",{"href":"foo?t=1&amp=2&ampo=3&lt=foo"},[[1,"bar"]]]]'
+test_equals "matching closing tags", html_to_json, "foo<a href=\"hi\">hi</a><div>1<div>foo</div>2</div>bar", '[[1,"foo"],[0,"a",{"href":"hi"},[[1,"hi"]]],[0,"div",{},[[1,"1"],[0,"div",{},[[1,"foo"]]],[1,"2"]]],[1,"bar"]]'