Merge pull request #30 from ljie-PI/master

make it work when there's a dot in the attribute selector

src/htmlparser/ElementNode.lua

@@ -149,11 +149,25 @@ local function select(self, s)
     childrenonly = false
     if part == "*" then goto nextpart end
     local excludes, filter = Set:new()
-    for t, w in string.gmatch(part,
+    local halfword = ""
+    for t, w, c in string.gmatch(part,
       "([:%[#.]?)" ..        -- t = an optional :, [, #, or .
       "([^:%(%[#.%]%)]+)" .. -- w = 1 or more of anything not :, (, [, #, ., ], or )
-      "%]?%)?"               -- followed by an uncaptured optional ] and/or )
+      "(%]?)%)?"               -- followed by an uncaptured optional ] and/or )
     ) do
+      -- this if..elseif.. block will match the pattern like "[src='aaa.jpg']"
+      if t == "[" and c ~= "]" then
+        halfword = t .. w
+        goto nextw
+      elseif c == "" and halfword ~= "" then
+        halfword = halfword .. t .. w
+        goto nextw
+      elseif t ~= "[" and c == "]" then
+        halfword = halfword .. t .. w .. c
+        t, w = "[", string.sub(halfword, 2, -2)
+        halfword = ""
+      end
+
       if t == ":" then filter = w goto nextw end
       local matched = match(t, w)
       if filter == "not" then

tst/init.lua

@@ -85,7 +85,7 @@ function test_attr_equal()
 		<n a1 a2= a3='' a4=""
 			a5='a"5"' a6="a'6'" a7='#.[] :()' a8='|*+-=?$^%&/'
 			a9=a9
-		a10></n>
+		a10 a11="a11.js.jpg"></n>
 	]])
 	assert_equal(1, #tree.nodes, "top level")
 	assert(tree("[a1='']")[1], "a1=''")
@@ -102,6 +102,8 @@ function test_attr_equal()
 	assert(tree("[a9='a9']")[1], "a9='a9'")
 	assert(tree("[a10='']")[1], "a10=''")
 	assert(tree("[a10=]")[1], "a10=")
+	-- An excepton for a7. Some times we may select javascript or img nodes with attr selector [src="a.js"] or [src="a.jpg"]
+	assert(tree("[a11='a11.js.jpg']")[1], "a11=")
 end
 
 function test_attr_notequal()