git push origin masterMerge branch 'ChenNingCong-fixselector'

Project.toml

@@ -0,0 +1,16 @@
+name = "Cascadia"
+uuid = "54eefc05-d75b-58de-a785-1a3403f0919f"
+
+[deps]
+AbstractTrees = "1520ce14-60c1-5f80-bbc7-55ef81b5835c"
+Gumbo = "708ec375-b3d6-5a57-a7ce-8257bf98657a"
+
+[compat]
+julia = "≥ 0.7.0"
+
+[extras]
+JSON = "682c06a0-de6a-54ab-a142-c8b1cf79cde6"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["JSON", "Test"]

src/selector.jl

@@ -54,7 +54,13 @@ end
 #// found, or false if no match is found.
 function hasDescendantMatch(n::HTMLNode, a::Selector)
     for c in Gumbo.children(n)
-        if typeof(c) == HTMLText; return a(c); end
+        if typeof(c) == HTMLText
+            if a(c)
+                return true
+            else
+                continue
+            end
+        end
         for cc in PostOrderDFS(c)
             if a(cc); return true; end
         end
@@ -139,6 +145,7 @@ function typeSelector(tg) #->Selector
     tg=lowercase(tg)
     return Selector() do n::HTMLNode
         if !(typeof(n) <: HTMLElement); return false; end
+        #@info tg n string(tag(n))
         lowercase(string(tag(n))) == tg
     end
 end
@@ -244,7 +251,7 @@ end
 #// the attribute named key matches the regular expression rx
 function attributeRegexSelector(key::AbstractString, rx::Regex) #->Selector
     return attributeSelector(key) do s
-        return occursin(rx, key)
+        return occursin(rx, s)
     end
 end
 
@@ -384,14 +391,15 @@ function nthChildSelector(a::Int, b::Int, last::Bool, ofType::Bool) #->Selector
         if parent == NullNode; return false; end
         i=-1
         count=0
+        temp = nothing
         for c in Gumbo.children(parent)
-            if typeof(c) != HTMLElement || (ofType && tag(c) != tag(n))
+            if !(typeof(c) <: HTMLElement) || (ofType && tag(c) != tag(n))
                 continue
             end
             count += 1
             if c == n
                 i=count
-                if c !== children(parent)[end]
+                if !last
                     break
                 end
             end
@@ -400,7 +408,7 @@ function nthChildSelector(a::Int, b::Int, last::Bool, ofType::Bool) #->Selector
             # This shouldn't happen, since n should always be one of its parent's children.
             return false
         end
-        if c===children(parent)[end]
+        if last
             i = count -i + 1
         end
 
@@ -419,12 +427,13 @@ end
 #// If ofType is true, it implements :only-of-type instead.
 function onlyChildSelector(ofType::Bool) #-> Selector
     return Selector() do n::HTMLNode
+        #@info n
         if !(typeof(n) <: HTMLElement); return false; end
         parent = n.parent
         if parent == nothing  || parent == NullNode; return false; end
         count = 0
         for c in Gumbo.children(parent)
-            if typeof(c) != HTMLElement || (oftype && tag(c) != tag(n))
+            if !(typeof(c) <: HTMLElement) || (ofType && tag(c) != tag(n))
                 continue
             end
             count += 1

test/runtests.jl

@@ -3,12 +3,18 @@ using Test
 using JSON
 using Gumbo
 
-# write your own tests here
-@test 1 == 1
+function checkstring(elem::HTMLElement{T}) where {T}
+    opentag = "<$T"
+    for (name,value) in sort(collect(elem.attributes), by=x->x.first)
+        opentag *= " $name=\"$value\""
+    end
+    opentag *= ">"
+end
 
 #A function to simplify test artifact creation
 P(x) = Cascadia.Parser(x)
 
+@testset "Basic Tests" begin
 @test Cascadia.parseName(P("abc")) == "abc"
 @test Cascadia.parseName(P("x")) == "x"
 @test Cascadia.parseIdentifier(P("abc")) == "abc"
@@ -29,26 +35,24 @@ P(x) = Cascadia.Parser(x)
 @test Cascadia.parseString(P("'x\\\r\nx'")) == "xx"
 @test Cascadia.parseString(P("\"a\\\"b\"")) == "a\"b"
 
-Cascadia.parseInteger(P("90:")) == 90
+@test Cascadia.parseInteger(P("90:")) == 90
+
+end
 
 ###Selector tests. Load data from file.
 
 selectorTests=JSON.parsefile(joinpath(dirname(@__FILE__), "selectorTests.json"))
 
 cnt = 0
-for (i, d) in enumerate(selectorTests)
+@testset "Selector Test $(d["Selector"])" for d in selectorTests
     c = Selector(d["Selector"])
     @test typeof(c) == Selector
     n=parsehtml(d["HTML"])
     r=eachmatch(c, n.root)
-    l=length(r)
-    e = length(d["Results"])
-    if l != e
-        global cnt += 1
-        println("Test Failure (known) for $(d["Selector"]) Expected $e, got $l")
-    else
-        println("Test Success         for $(d["Selector"])")
+    e = d["Results"]
+    @test length(r) == length(e)
+
+    for i in 1:length(r)
+        @test lowercase(checkstring(r[i])) == e[i]
     end
 end
-@test cnt <= 19
-println("Total test failures: $cnt / $(length(selectorTests))")

test/selectorTests.json

@@ -437,23 +437,23 @@
     "HTML": "<ul>\n\t\t\t<li><a id=\"a1\" href=\"http://www.google.com/finance\"/>\n\t\t\t<li><a id=\"a2\" href=\"http://finance.yahoo.com/\"/>\n\t\t\t<li><a id=\"a2\" href=\"http://finance.untrusted.com/\"/>\n\t\t\t<li><a id=\"a3\" href=\"https://www.google.com/news\"/>\n\t\t\t<li><a id=\"a4\" href=\"http://news.yahoo.com\"/>\n\t\t</ul>",
     "Selector": "[href#=(fina)]:not([href#=(\\/\\/[^\\/]+untrusted)])",
     "Results": [
-      "<a id=\"a1\" href=\"http://www.google.com/finance\">",
-      "<a id=\"a2\" href=\"http://finance.yahoo.com/\">"
+      "<a href=\"http://www.google.com/finance\" id=\"a1\">",
+      "<a href=\"http://finance.yahoo.com/\" id=\"a2\">"
     ]
   },
   {
     "HTML": "<ul>\n\t\t\t<li><a id=\"a1\" href=\"http://www.google.com/finance\"/>\n\t\t\t<li><a id=\"a2\" href=\"http://finance.yahoo.com/\"/>\n\t\t\t<li><a id=\"a3\" href=\"https://www.google.com/news\"/>\n\t\t\t<li><a id=\"a4\" href=\"http://news.yahoo.com\"/>\n\t\t</ul>",
     "Selector": "[href#=(^https:\\/\\/[^\\/]*\\/?news)]",
     "Results": [
-      "<a id=\"a3\" href=\"https://www.google.com/news\">"
+      "<a href=\"https://www.google.com/news\" id=\"a3\">"
     ]
   },
   {
     "HTML": "<form>\n\t\t\t<label>Username <input type=\"text\" name=\"username\" /></label>\n\t\t\t<label>Password <input type=\"password\" name=\"password\" /></label>\n\t\t\t<label>Country\n\t\t\t\t<select name=\"country\">\n\t\t\t\t\t<option value=\"ca\">Canada</option>\n\t\t\t\t\t<option value=\"us\">United States</option>\n\t\t\t\t</select>\n\t\t\t</label>\n\t\t\t<label>Bio <textarea name=\"bio\"></textarea></label>\n\t\t\t<button>Sign up</button>\n\t\t</form>",
     "Selector": ":input",
     "Results": [
-      "<input type=\"text\" name=\"username\">",
-      "<input type=\"password\" name=\"password\">",
+      "<input name=\"username\" type=\"text\">",
+      "<input name=\"password\" type=\"password\">",
       "<select name=\"country\">",
       "<textarea name=\"bio\">",
       "<button>"