improved tpl-resisting and testing

.travis.yml

@@ -4,12 +4,13 @@ sudo: false
 
 env:
   global:
-    - LUAROCKS=2.3.0
+    - LUAROCKS=3.1.3
   matrix:
     - LUA=lua5.1
     - LUA=lua5.2
     - LUA=lua5.3
-    - LUA=luajit     # latest stable version (2.0.4)
+    - LUA=lua5.4
+    - LUA=luajit     # latest stable version
     - LUA=luajit2.0  # current head of 2.0 branch
     - LUA=luajit2.1  # current head of 2.1 branch
 

.travis/setup_lua.sh

@@ -2,13 +2,13 @@
 
 # A script for setting up environment for travis-ci testing.
 # Sets up Lua and Luarocks.
-# LUA must be "lua5.1", "lua5.2" or "luajit".
+# LUA must be "lua5.x" or "luajit".
 # luajit2.0 - master v2.0
 # luajit2.1 - master v2.1
 
 set -eufo pipefail
 
-LUAJIT_VERSION="2.0.4"
+LUAJIT_VERSION="2.0.5"
 LUAJIT_BASE="LuaJIT-$LUAJIT_VERSION"
 
 source .travis/platform.sh
@@ -62,14 +62,14 @@ else
 
   if [ "$LUA" == "lua5.1" ]; then
     curl http://www.lua.org/ftp/lua-5.1.5.tar.gz | tar xz
-    cd lua-5.1.5;
   elif [ "$LUA" == "lua5.2" ]; then
     curl http://www.lua.org/ftp/lua-5.2.4.tar.gz | tar xz
-    cd lua-5.2.4;
   elif [ "$LUA" == "lua5.3" ]; then
-    curl http://www.lua.org/ftp/lua-5.3.2.tar.gz | tar xz
-    cd lua-5.3.2;
+    curl http://www.lua.org/ftp/lua-5.3.5.tar.gz | tar xz
+  elif [ "$LUA" == "lua5.4" ]; then
+    curl http://www.lua.org/work/lua-5.4.0-alpha-rc2.tar.gz | tar xz
   fi
+  cd lua-5*
 
   # Build Lua without backwards compatibility for testing
   perl -i -pe 's/-DLUA_COMPAT_(ALL|5_2)//' src/Makefile

src/htmlparser.lua

@@ -109,40 +109,36 @@ local function parse(text,limit) -- {{{
 		end
 		-- g }}}
 
-		-- templaters {{{
-		text=text:gsub(
-				[=[(=[%s]-)(['"])]=].. -- only match attr.values, and not random strings between two random quoting marks
-				[=[([^%2<>]+)]=]..
-				[=[([^%2>]-)]=]..
-				[=[(%2)]=],
-				function(...)return g(4,...)end
-			) -- Escape "<" inside attr.values (see issue #50)
-		text=text:gsub(
-				[=[(=[%s]-)(['"])]=].. -- only match attr.values, and not random strings between two random quoting marks
-				[=[([^%2<>]+)]=]..
-				[=[([^%2<]-)]=]..
-				[=[(%2)]=],
-				function(...)return g(4,...)end
-			) -- Escape ">" inside attr.values (see issue #50)
---[[
-]]
-		text = text:gsub(
-				"(<[^!])".. -- Comments aren't templaters placeholders
-				"([^>]-)"..
+		-- tpl-placeholders and attributes {{{
+		text=text
+			:gsub(
+				"(=[%s]-)".. -- only match attr.values, and not random strings between two random apostrophs
+				"(%b'')",
+				function(...)return g(2,...)end
+			)
+			:gsub(
+				"(=[%s]-)".. -- same for "
+				'(%b"")',
+				function(...)return g(2,...)end
+			) -- Escape "<"/">" inside attr.values (see issue #50)
+			:gsub(
+				"(<".. -- Match "<",
+				(opts.tpl_skip_pattern or "[^!]").. -- with exclusion pattern (for example, to ignore comments, which aren't template placeholders, but can legally contain "<"/">" inside.
+				")([^>]+)".. -- If matched, we want to escape '<'s if we meet them inside tag
 				"(>)",
 				function(...)return g(2,...)end
-		) -- scan for a second "<", inside "<>" (if it shows before ">"), until it inside the comment or CDATA
-		text=text:gsub(
-				"("..tpr["<"]..")"..
-				"([^%w%s])"..
-				"([^%2]-)"..
-				"(%2)"..
-				"(>)"..
-				"([^>]-)"..
-				"(>)", -- Comments and CDATA aren't templaters placeholders
+			)
+			:gsub(
+				"("..
+				(tpr["<"] or "__FAILED__").. -- Here we search for "<", we escaped in previous gsub (and don't break things if we have no escaping replacement)
+				")("..
+				(opts.tpl_marker_pattern or "[^%w%s]").. -- Capture templating symbol
+				")([%g%s]-)".. -- match placeholder's content
+				"(%2)(>)".. -- placeholder's tail
+				"([^>]*>)", -- remainings
 				function(...)return g(5,...)end
-			) -- try to find matching ">" for previous replace
-		-- templaters }}}
+			)
+		-- }}}
 	end -- }}}
 
 	local index = 0

tst/init.lua

@@ -309,8 +309,13 @@ end
 
 function test_loop_limit()
 	local tree = htmlparser.parse([[
+		<a id='a >b'>moo</a>
+		<a id='c> d'>moo</a>
+		<a id='e > f'>moo</a>
+		<a id="g >h">moo</a>
+		<a id="i> j">moo</a>
+		<a id="k > l">moo</a>
 		<a id='1>2'>moo</a>
-		<a id='2>3'>moo</a>
 		<b id='foo<bar'>moo</b>
 		<img <%tpl%> foo=bar></img>
 		<img <%tpl%> />
@@ -319,10 +324,8 @@ function test_loop_limit()
 		<i <=moo=>>k</i>
 		<s <-foo->>o</s>
 		<div <*bar*>></div>
-		<p>
 		<a id="unclosed>Element"> with unclosed attribute</a>
-		</p>
 		<div data-pic="aa<%=image_url%>bb" ></div>
 	]]) -- issue#42
-	assert(1==1)
+	assert(#tree.nodes==17)
 end