| Index: components/test/data/dom_distiller/derived_features.json
|
| diff --git a/components/test/data/dom_distiller/derived_features.json b/components/test/data/dom_distiller/derived_features.json
|
| new file mode 100644
|
| index 0000000000000000000000000000000000000000..40071b330b9705a0d962698b457773ea58f89ac0
|
| --- /dev/null
|
| +++ b/components/test/data/dom_distiller/derived_features.json
|
| @@ -0,0 +1,194 @@
|
| +[
|
| + {
|
| + "url": "https://zh-tw.facebook.com/permalink.php",
|
| + "index": 0,
|
| + "features": [
|
| + "opengraph",
|
| + false,
|
| + "forum",
|
| + false,
|
| + "index",
|
| + false,
|
| + "view",
|
| + false,
|
| + "asp",
|
| + false,
|
| + "phpbb",
|
| + false,
|
| + "php",
|
| + true,
|
| + "pathlength",
|
| + 14,
|
| + "domain",
|
| + false,
|
| + "pathcomponents",
|
| + 1,
|
| + "slugdetector",
|
| + 1,
|
| + "pathnumbers",
|
| + 0,
|
| + "lastSegmentLength",
|
| + 13,
|
| + "formcount",
|
| + 1,
|
| + "anchorcount",
|
| + 26,
|
| + "elementcount",
|
| + 120,
|
| + "anchorratio",
|
| + 0.21666666666666667,
|
| + "innertextlength",
|
| + 492,
|
| + "textcontentlength",
|
| + 12832,
|
| + "innerhtmllength",
|
| + 19715,
|
| + "innertextlengthratio",
|
| + 0.024955617550088766,
|
| + "textcontentlengthratio",
|
| + 0.6508749682982501,
|
| + "innertexttextcontentlengthratio",
|
| + 0.03834164588528678,
|
| + "innertextwordcount",
|
| + 5,
|
| + "textcontentwordcount",
|
| + 1470,
|
| + "innerhtmlwordcount",
|
| + 2313,
|
| + "innertextwordcountratio",
|
| + 0.00216169476869866,
|
| + "textcontentwordcountratio",
|
| + 0.6355382619974059,
|
| + "innertexttextcontentwordcountratio",
|
| + 0.003401360544217687
|
| + ]
|
| + },
|
| + {
|
| + "url": "https://zb.getitfree.us/myaccount/survey.html",
|
| + "index": 1,
|
| + "features": [
|
| + "opengraph",
|
| + false,
|
| + "forum",
|
| + false,
|
| + "index",
|
| + false,
|
| + "view",
|
| + false,
|
| + "asp",
|
| + false,
|
| + "phpbb",
|
| + false,
|
| + "php",
|
| + false,
|
| + "pathlength",
|
| + 22,
|
| + "domain",
|
| + false,
|
| + "pathcomponents",
|
| + 2,
|
| + "slugdetector",
|
| + 1,
|
| + "pathnumbers",
|
| + 0,
|
| + "lastSegmentLength",
|
| + 11,
|
| + "formcount",
|
| + 0,
|
| + "anchorcount",
|
| + 5,
|
| + "elementcount",
|
| + 92,
|
| + "anchorratio",
|
| + 0.05434782608695652,
|
| + "innertextlength",
|
| + 1137,
|
| + "textcontentlength",
|
| + 25037,
|
| + "innerhtmllength",
|
| + 33602,
|
| + "innertextlengthratio",
|
| + 0.03383727159097673,
|
| + "textcontentlengthratio",
|
| + 0.7451044580679721,
|
| + "innertexttextcontentlengthratio",
|
| + 0.045412789072173186,
|
| + "innertextwordcount",
|
| + 181,
|
| + "textcontentwordcount",
|
| + 2811,
|
| + "innerhtmlwordcount",
|
| + 3926,
|
| + "innertextwordcountratio",
|
| + 0.04610290371879776,
|
| + "textcontentwordcountratio",
|
| + 0.7159959246051961,
|
| + "innertexttextcontentwordcountratio",
|
| + 0.06438989683386695
|
| + ]
|
| + },
|
| + {
|
| + "url": "https://yande.re/pool",
|
| + "index": 2,
|
| + "features": [
|
| + "opengraph",
|
| + false,
|
| + "forum",
|
| + false,
|
| + "index",
|
| + false,
|
| + "view",
|
| + false,
|
| + "asp",
|
| + false,
|
| + "phpbb",
|
| + false,
|
| + "php",
|
| + false,
|
| + "pathlength",
|
| + 5,
|
| + "domain",
|
| + false,
|
| + "pathcomponents",
|
| + 1,
|
| + "slugdetector",
|
| + 0,
|
| + "pathnumbers",
|
| + 0,
|
| + "lastSegmentLength",
|
| + 4,
|
| + "formcount",
|
| + 10,
|
| + "anchorcount",
|
| + 112,
|
| + "elementcount",
|
| + 492,
|
| + "anchorratio",
|
| + 0.22764227642276422,
|
| + "innertextlength",
|
| + 1676,
|
| + "textcontentlength",
|
| + 41089,
|
| + "innerhtmllength",
|
| + 56900,
|
| + "innertextlengthratio",
|
| + 0.02945518453427065,
|
| + "textcontentlengthratio",
|
| + 0.7221265377855888,
|
| + "innertexttextcontentlengthratio",
|
| + 0.04078950570712356,
|
| + "innertextwordcount",
|
| + 303,
|
| + "textcontentwordcount",
|
| + 3874,
|
| + "innerhtmlwordcount",
|
| + 6502,
|
| + "innertextwordcountratio",
|
| + 0.04660104583205168,
|
| + "textcontentwordcountratio",
|
| + 0.5958166717932943,
|
| + "innertexttextcontentwordcountratio",
|
| + 0.07821373257614868
|
| + ]
|
| + }
|
| +]
|
|
|