Make htmlparser parse unquoted attrib values (#9537)

Fixes #6154
This commit is contained in:
rec
2018-10-29 11:10:00 +01:00
committed by Andreas Rumpf
parent 680f5eeb15
commit 9fd0a71e4d
3 changed files with 65 additions and 2 deletions

View File

@@ -78,3 +78,48 @@ block t2814:
echo "case " & ltype[0] & " failed !"
quit(2)
echo "true"
block t6154:
let foo = """
<!DOCTYPE html>
<html>
<head>
<title> foobar </title>
</head>
<body>
<p class=foo id=bar></p>
<p something=&#9;foo&#9;bar&#178;></p>
<p something= &#9;foo&#9;bar&#178; foo =bloo></p>
<p class="foo2" id="bar2"></p>
<p wrong= ></p>
</body>
</html>
"""
var errors: seq[string] = @[]
let html = parseHtml(newStringStream(foo), "statichtml", errors=errors)
doAssert "statichtml(11, 18) Error: attribute value expected" in errors
let ps = html.findAll("p")
doAssert ps.len == 5
doAssert ps[0].attrsLen == 2
doAssert ps[0].attr("class") == "foo"
doAssert ps[0].attr("id") == "bar"
doassert ps[0].len == 0
doAssert ps[1].attrsLen == 1
doAssert ps[1].attr("something") == "\tfoo\tbar²"
doassert ps[1].len == 0
doAssert ps[2].attrsLen == 2
doAssert ps[2].attr("something") == "\tfoo\tbar²"
doAssert ps[2].attr("foo") == "bloo"
doassert ps[2].len == 0
doAssert ps[3].attrsLen == 2
doAssert ps[3].attr("class") == "foo2"
doAssert ps[3].attr("id") == "bar2"
doassert ps[3].len == 0
doAssert ps[4].attrsLen == 1
doAssert ps[4].attr("wrong") == ""