Commit6879f6a

Greg Guthe

committed

html5lib_shim: validate unicode points for convert_entity

1 parent90cb80b commit6879f6aCopy full SHA for 6879f6a

File tree

+25

-2

lines changed

+25

-2

lines changed

Lines changed: 15 additions & 2 deletions

Original file line number	Diff line number	Diff line change
`@@ -459,9 +459,22 @@ def convert_entity(value):`
`459`	`459`	`ifvalue[0]=="#":`
`460`	`460`	`iflen(value)<2:`
`461`	`461`	`returnNone`
	`462`	`+`
`462`	`463`	`ifvalue[1]in ("x","X"):`
`463`		`-returnsix.unichr(int(value[2:],16))`
`464`		`-returnsix.unichr(int(value[1:],10))`
	`464`	`+# hex-encoded code point`
	`465`	`+int_as_string,base=value[2:],16`
	`466`	`+else:`
	`467`	`+# decimal code point`
	`468`	`+int_as_string,base=value[1:],10`
	`469`	`+`
	`470`	`+ifint_as_string=="":`
	`471`	`+returnNone`
	`472`	`+`
	`473`	`+code_point=int(int_as_string,base)`
	`474`	`+if0<code_point<0x110000:`
	`475`	`+returnsix.unichr(code_point)`
	`476`	`+else:`
	`477`	`+returnNone`
`465`	`478`
`466`	`479`	`returnENTITIES.get(value,None)`
`467`	`480`

Lines changed: 10 additions & 0 deletions

Original file line number	Diff line number	Diff line change
`@@ -19,6 +19,16 @@`
`19`	`19`	`("&xx;","&xx;"),`
`20`	`20`	`# Handles multiple entities in the same string`
`21`	`21`	`("this & that & that","this & that & that"),`
	`22`	`+# Handles empty decimal and hex encoded code points`
	`23`	`+ ("&#x;","&#x;"),`
	`24`	`+ ("&#;","&#;"),`
	`25`	`+# Handles too high unicode points`
	`26`	`+ ("&#x110000;","&#x110000;"),`
	`27`	`+ ("&#x110111;","&#x110111;"),`
	`28`	`+ ("&#9277809;","&#9277809;"),`
	`29`	`+# Handles negative unicode points`
	`30`	`+ ("&#-1;","&#-1;"),`
	`31`	`+ ("&#x-1;","&#x-1;"),`
`22`	`32`	`],`
`23`	`33`	`)`
`24`	`34`	`deftest_convert_entities(data,expected):`

Comments

(0)