1 | /*
|
---|
2 |
|
---|
3 | All the websites that have some MRI detected AND which are either in NZ or with NZ TLD
|
---|
4 | or (so if they're from overseas) don't contain /mi or mi.* in URL path.
|
---|
5 | We'll include Australia, to get the valid "kiwiproperty.com" website,
|
---|
6 | otherwise the sole exception, included in the result list.
|
---|
7 |
|
---|
8 |
|
---|
9 | db.getCollection('Websites').find({$and: [
|
---|
10 | {numPagesContainingMRI: {$gt: 0}},
|
---|
11 | {$or: [{geoLocationCountryCode: /(NZ|AU)/}, {domain: /\.nz$/}, {urlContainsLangCodeInPath: false}]}
|
---|
12 | ]}).count()
|
---|
13 |
|
---|
14 | 397
|
---|
15 |
|
---|
16 | Aggregate results by a count of country codes. Also have a domain listing in the output.
|
---|
17 |
|
---|
18 | However, we want to group nz TLD with websites that originate in NZ.
|
---|
19 | So we also want the converse: to remove websites with .nz TLD from any originating
|
---|
20 | country codes that are from outside NZ.
|
---|
21 |
|
---|
22 | db.getCollection('Websites').find({$and: [
|
---|
23 | {geoLocationCountryCode: {$ne: "NZ"}},
|
---|
24 | {domain: {$not: /\.nz/}},
|
---|
25 | {numPagesContainingMRI: {$gt: 0}},
|
---|
26 | {$or: [{geoLocationCountryCode: "AU"}, {urlContainsLangCodeInPath: false}]}
|
---|
27 | ]}).count()
|
---|
28 |
|
---|
29 | 221 websites
|
---|
30 |
|
---|
31 | db.getCollection('Websites').find({$and: [
|
---|
32 | {numPagesContainingMRI: {$gt: 0}},
|
---|
33 | {$or: [{geoLocationCountryCode:"NZ"},{domain: /\.nz/}]}
|
---|
34 | ]}).count()
|
---|
35 |
|
---|
36 | 176
|
---|
37 |
|
---|
38 | (221 + 176 = 397, which adds up to above.)
|
---|
39 |
|
---|
40 |
|
---|
41 | Counts by country code excluding NZ related sites
|
---|
42 | db.Websites.aggregate([
|
---|
43 | {
|
---|
44 | $match: {
|
---|
45 | $and: [
|
---|
46 | {geoLocationCountryCode: {$ne: "NZ"}},
|
---|
47 | {domain: {$not: /\.nz/}},
|
---|
48 | {numPagesContainingMRI: {$gt: 0}},
|
---|
49 | {$or: [{geoLocationCountryCode: "AU"}, {urlContainsLangCodeInPath: false}]}
|
---|
50 | ]
|
---|
51 | }
|
---|
52 | },
|
---|
53 | { $unwind: "$geoLocationCountryCode" },
|
---|
54 | {
|
---|
55 | $group: {
|
---|
56 | _id: {$toLower: '$geoLocationCountryCode'},
|
---|
57 | count: { $sum: 1 },
|
---|
58 | domain: { $addToSet: '$domain' }
|
---|
59 | }
|
---|
60 | },
|
---|
61 | { $sort : { count : -1} }
|
---|
62 | ]);
|
---|
63 |
|
---|
64 | Count of NZ related sites - output put under a hardcoded _id of "nz"
|
---|
65 | and once again requesting a domain listing in output:
|
---|
66 |
|
---|
67 | db.Websites.aggregate([
|
---|
68 | {
|
---|
69 | $match: {
|
---|
70 | $and: [
|
---|
71 | {numPagesContainingMRI: {$gt: 0}},
|
---|
72 | {$or: [{geoLocationCountryCode:"NZ"},{domain: /\.nz/}]}
|
---|
73 | ]
|
---|
74 | }
|
---|
75 | },
|
---|
76 | { $unwind: "$geoLocationCountryCode" },
|
---|
77 | {
|
---|
78 | $group: {
|
---|
79 | _id: "nz",
|
---|
80 | count: { $sum: 1 },
|
---|
81 | domain: { $addToSet: '$domain' }
|
---|
82 | }
|
---|
83 | },
|
---|
84 | { $sort : { count : -1} }
|
---|
85 | ]);
|
---|
86 |
|
---|
87 | */
|
---|
88 |
|
---|
89 | /* 0 */
|
---|
90 | {
|
---|
91 | "_id" : "us",
|
---|
92 | "count" : 117.0,
|
---|
93 | "domain" : [
|
---|
94 | "http://shangrilapress.net",
|
---|
95 | "https://www.terakau.org",
|
---|
96 | "http://dannykahei.tripod.com",
|
---|
97 | "https://in.pinterest.com",
|
---|
98 | "http://takethatvacation.com",
|
---|
99 | "http://malecek.com",
|
---|
100 | "http://word-dialect.blogspot.com",
|
---|
101 | "https://www.blue-frontiers.com",
|
---|
102 | "https://static-promote.weebly.com",
|
---|
103 | "http://www.thesalmons.org",
|
---|
104 | "http://ngarangatahi.tripod.com",
|
---|
105 | "http://tkrow.tripod.com",
|
---|
106 | "http://niken8media.logdown.com",
|
---|
107 | "https://www.vaihaunui.net",
|
---|
108 | "https://www.podrozeady.com",
|
---|
109 | "https://www.nccri.ie",
|
---|
110 | "http://georgegi.tripod.com",
|
---|
111 | "http://www.lunar-occultations.com",
|
---|
112 | "http://frontrowphotos.com",
|
---|
113 | "http://linkvip.top",
|
---|
114 | "http://rangiwewehi.com",
|
---|
115 | "http://www.pressreader.com",
|
---|
116 | "http://anglicanhistory.org",
|
---|
117 | "http://www.unicode.org",
|
---|
118 | "https://wol.jw.org",
|
---|
119 | "http://pumanawawhangara.blogspot.com",
|
---|
120 | "http://hannas-reiseblog.blogspot.com",
|
---|
121 | "http://seapixonline.com",
|
---|
122 | "http://ww25.milfsplease.com",
|
---|
123 | "http://www.wikitree.com",
|
---|
124 | "http://ritusehji.blogspot.com",
|
---|
125 | "https://www.pinterest.it",
|
---|
126 | "http://naturalfatburner.net",
|
---|
127 | "http://kiaorahola.blogspot.com",
|
---|
128 | "http://www.hudl.com",
|
---|
129 | "http://shuttersportnelson.photoshelter.com",
|
---|
130 | "https://www.pinterest.ca",
|
---|
131 | "http://precious-testimonies.com",
|
---|
132 | "http://www.muhammad.com",
|
---|
133 | "http://www.gotquestions.org",
|
---|
134 | "https://www.pinterest.co.uk",
|
---|
135 | "https://biblehub.com",
|
---|
136 | "http://tuhua2010.blogspot.com",
|
---|
137 | "http://www.precious-testimonies.com",
|
---|
138 | "http://www.huapala.org",
|
---|
139 | "https://nl.pinterest.com",
|
---|
140 | "http://www.whoisthatr.com",
|
---|
141 | "https://www.oemsec.com",
|
---|
142 | "https://www.dbnames.net",
|
---|
143 | "https://www.myadsclassified.com",
|
---|
144 | "http://mikebonnice.com",
|
---|
145 | "http://fhr.kiwicelts.com",
|
---|
146 | "https://www.webwiki.com",
|
---|
147 | "https://www.pinterest.jp",
|
---|
148 | "https://kjohnsonnz.blogspot.com",
|
---|
149 | "http://svenskadress.net",
|
---|
150 | "http://www.godrules.net",
|
---|
151 | "https://www.pinterest.fr",
|
---|
152 | "http://rhymebrain.com",
|
---|
153 | "http://www.frogsonline.com",
|
---|
154 | "http://www.v3whois.com",
|
---|
155 | "http://piripi.blogspot.com",
|
---|
156 | "http://www.twttoa.com",
|
---|
157 | "http://wikiedit.org",
|
---|
158 | "https://livestream.com",
|
---|
159 | "http://burkekm001.tripod.com",
|
---|
160 | "https://maorinews.com",
|
---|
161 | "http://www.geni.com",
|
---|
162 | "http://www.waimate.com",
|
---|
163 | "http://m.biblepub.com",
|
---|
164 | "http://wowwars.net",
|
---|
165 | "https://www.natekore2018.com",
|
---|
166 | "http://tatai09.blogspot.com",
|
---|
167 | "https://ebible.org",
|
---|
168 | "http://capsuraotearoa.blogspot.com",
|
---|
169 | "http://bahaiprayers.net",
|
---|
170 | "https://www.breaker.audio",
|
---|
171 | "https://www.pipirikiapapatuanuku.org",
|
---|
172 | "http://www.the-naked.com",
|
---|
173 | "https://phet.colorado.edu",
|
---|
174 | "http://manateina.blogspot.com",
|
---|
175 | "http://tkkpipipaopao.blogspot.com",
|
---|
176 | "https://www.hidroponia.org.mx",
|
---|
177 | "http://mrshamiltonskoolkidz.blogspot.com",
|
---|
178 | "http://aclhokiangarocks.blogspot.com",
|
---|
179 | "http://www.eyecontactsite.com",
|
---|
180 | "http://www.hiroa.pf",
|
---|
181 | "http://www.forensicfashion.com",
|
---|
182 | "http://www.code-postal.com",
|
---|
183 | "http://lianzaconference2012.blogspot.com",
|
---|
184 | "http://mahoraroom8.blogspot.com",
|
---|
185 | "http://korora.econ.yale.edu",
|
---|
186 | "https://docs.google.com",
|
---|
187 | "https://www.indexmundi.com",
|
---|
188 | "https://www.seapixonline.com",
|
---|
189 | "https://www.bible.com",
|
---|
190 | "https://www.knowatom.com",
|
---|
191 | "https://chromium.googlesource.com",
|
---|
192 | "http://www.krassotkin.ru",
|
---|
193 | "http://www.roadsmile.com",
|
---|
194 | "https://www.code-postal.com",
|
---|
195 | "http://blogdepasopor.blogspot.com",
|
---|
196 | "http://eartheum.com",
|
---|
197 | "http://www.steve-wheeler.co.uk",
|
---|
198 | "http://www.mkiwi.com",
|
---|
199 | "http://maaori.com",
|
---|
200 | "https://www.kaifineart.com",
|
---|
201 | "https://png.bible",
|
---|
202 | "https://www.poehalisnami.ua",
|
---|
203 | "http://atopeconlostopes.blogspot.com",
|
---|
204 | "http://www.whoisentry.com",
|
---|
205 | "http://loquevendra318.com",
|
---|
206 | "https://za.pinterest.com",
|
---|
207 | "http://www.namesdir.com",
|
---|
208 | "https://drive.google.com",
|
---|
209 | "http://worldradiomap.com"
|
---|
210 | ]
|
---|
211 | }
|
---|
212 |
|
---|
213 | /* 1 */
|
---|
214 | {
|
---|
215 | "_id" : "nz",
|
---|
216 | "count" : 176.0,
|
---|
217 | "domain" : [
|
---|
218 | "http://tmoa.tki.org.nz",
|
---|
219 | "http://www.rotoruanz.com",
|
---|
220 | "https://admin.teara.govt.nz",
|
---|
221 | "http://www.tekura.school.nz",
|
---|
222 | "http://www.tetaurawhiri.govt.nz",
|
---|
223 | "https://www.whanau-tahi.school.nz",
|
---|
224 | "http://auturoa.nz",
|
---|
225 | "http://www.ngamanawainc.co.nz",
|
---|
226 | "http://southerntribes.co.nz",
|
---|
227 | "https://player.vimeo.com",
|
---|
228 | "https://www.components-mart.nz",
|
---|
229 | "http://www.cs.waikato.ac.nz",
|
---|
230 | "https://www.terakipaewhenua.school.nz",
|
---|
231 | "http://oilcrash.com",
|
---|
232 | "https://manawatuheritage.pncc.govt.nz",
|
---|
233 | "http://maori.tki.org.nz",
|
---|
234 | "http://kaupare.co.nz",
|
---|
235 | "http://ngatiporoukiponeke.org.nz",
|
---|
236 | "http://cms.sunsmartschools.co.nz",
|
---|
237 | "http://philipbeadle.co.nz",
|
---|
238 | "http://waitarahistory.org.nz",
|
---|
239 | "http://hangaraumatihiko.tki.org.nz",
|
---|
240 | "https://sexualviolence.victimsinfo.govt.nz",
|
---|
241 | "http://www.kura-porirua.school.nz",
|
---|
242 | "http://www.rakaumanga.school.nz",
|
---|
243 | "http://www.huri-translations.pf",
|
---|
244 | "https://kotahimiriona.co.nz",
|
---|
245 | "http://ngarauhuia.ngatiapakiterato.iwi.nz",
|
---|
246 | "http://videos.e-agent.nz",
|
---|
247 | "http://kurakokiri.maori.nz",
|
---|
248 | "http://kuraaiwi.maori.nz",
|
---|
249 | "http://www.tewikiotereomaori.co.nz",
|
---|
250 | "http://arataua.nz",
|
---|
251 | "http://www.brettgraham.co.nz",
|
---|
252 | "http://anglicanprayerbook.nz",
|
---|
253 | "https://e-ako-pangarau.nzmaths.co.nz",
|
---|
254 | "https://www.pinterest.nz",
|
---|
255 | "http://www.tasteofplenty.co.nz",
|
---|
256 | "http://www.nzpcn.org.nz",
|
---|
257 | "https://www.puau.school.nz",
|
---|
258 | "https://www.rereahu.maori.nz",
|
---|
259 | "http://blog.teara.govt.nz",
|
---|
260 | "http://www.ruralfind.co.nz",
|
---|
261 | "https://www.korokikahukura.co.nz",
|
---|
262 | "http://givealittle.co.nz",
|
---|
263 | "http://tewikiotereomaori.nz",
|
---|
264 | "http://dev.nzpcn.org.nz",
|
---|
265 | "http://www.firstworldwar.tki.org.nz",
|
---|
266 | "http://rsnz.natlib.govt.nz",
|
---|
267 | "http://biketorqueyamaha.co.nz",
|
---|
268 | "http://conference.tpwt.maori.nz",
|
---|
269 | "http://myfathersworld.net.nz",
|
---|
270 | "http://whatonga.school.nz",
|
---|
271 | "https://teaomaori.news",
|
---|
272 | "https://www.ashtangatauranga.co.nz",
|
---|
273 | "http://www.eventcinemas.co.nz",
|
---|
274 | "http://artizani.co.nz",
|
---|
275 | "https://www.stats.govt.nz",
|
---|
276 | "https://keepourmoneyclean.govt.nz",
|
---|
277 | "http://www.teipukarea.maori.nz",
|
---|
278 | "http://kuraproductions.co.nz",
|
---|
279 | "http://www.otepoti.school.nz",
|
---|
280 | "https://register.tpota.org.nz",
|
---|
281 | "http://www.tewhanake.maori.nz",
|
---|
282 | "https://office.e-agent.nz",
|
---|
283 | "http://community.nzdl.org",
|
---|
284 | "https://www.blushandbrows.nz",
|
---|
285 | "https://cdn.tehiku.nz",
|
---|
286 | "http://www.oag.govt.nz",
|
---|
287 | "http://tmmkkm.school.nz",
|
---|
288 | "http://www.tetaumuturunanga.iwi.nz",
|
---|
289 | "http://teaohou.natlib.govt.nz",
|
---|
290 | "http://www.kmk.maori.nz",
|
---|
291 | "https://www.maoritelevision.com",
|
---|
292 | "https://sooty.nz",
|
---|
293 | "http://hana.co.nz",
|
---|
294 | "http://waiata.maori.nz",
|
---|
295 | "http://www.pakanae.maori.nz",
|
---|
296 | "http://www.w3vietnam.org.nz",
|
---|
297 | "http://www.zoomin.co.nz",
|
---|
298 | "http://www.hrc.co.nz",
|
---|
299 | "https://www.wingspan.co.nz",
|
---|
300 | "https://www.cruisetourstauranga.co.nz",
|
---|
301 | "http://kurataiao.tki.org.nz",
|
---|
302 | "http://punareo.co.nz",
|
---|
303 | "http://www.finlaysonpark.school.nz",
|
---|
304 | "http://www.kurakokiri.maori.nz",
|
---|
305 | "https://rapuatearatika.education.govt.nz",
|
---|
306 | "https://www.lcds-display.nz",
|
---|
307 | "http://www.livingheritage.org.nz",
|
---|
308 | "http://www.heartland.co.nz",
|
---|
309 | "http://www.biketorqueyamaha.co.nz",
|
---|
310 | "https://2019.nethui.nz",
|
---|
311 | "http://archerpix.com",
|
---|
312 | "http://www.tkkmmokopuna.school.nz",
|
---|
313 | "http://www.wcl.govt.nz",
|
---|
314 | "https://tiritiowaitangi.govt.nz",
|
---|
315 | "http://rakaumanga.school.nz",
|
---|
316 | "http://holyspirit.nz",
|
---|
317 | "http://crimson.co.nz",
|
---|
318 | "https://www.ngamanawainc.co.nz",
|
---|
319 | "http://rexedra.gen.nz",
|
---|
320 | "http://www.kupengahao.co.nz",
|
---|
321 | "https://www.tematawai.maori.nz",
|
---|
322 | "http://tiritiowaitangi.govt.nz",
|
---|
323 | "http://rurued.school.nz",
|
---|
324 | "http://w3vietnam.org.nz",
|
---|
325 | "https://www.rotorua-rafting.co.nz",
|
---|
326 | "https://www.e-agent.nz",
|
---|
327 | "http://reoora.co.nz",
|
---|
328 | "http://archive.stats.govt.nz",
|
---|
329 | "https://www.dnc.org.nz",
|
---|
330 | "https://liveresults.co.nz",
|
---|
331 | "https://www.taitokerautrust.org.nz",
|
---|
332 | "https://www.infinite-electronic.nz",
|
---|
333 | "https://kaiiwicamp.nz",
|
---|
334 | "http://www.tereowrap.nz",
|
---|
335 | "https://m.wairarapatv.co.nz",
|
---|
336 | "http://ngatiwhakaue.iwi.nz",
|
---|
337 | "http://www.kkmmaungarongo.co.nz",
|
---|
338 | "https://rehuamarae.co.nz",
|
---|
339 | "http://www.tmoa.tki.org.nz",
|
---|
340 | "http://www.gans.co.nz",
|
---|
341 | "http://www.topomap.co.nz",
|
---|
342 | "http://www.electionresults.govt.nz",
|
---|
343 | "http://archive.electionresults.govt.nz",
|
---|
344 | "http://satellites.co.nz",
|
---|
345 | "https://haereheikaiako.co.nz",
|
---|
346 | "http://www.twtop.school.nz",
|
---|
347 | "http://www.waiata.maori.nz",
|
---|
348 | "http://www.temarareo.org",
|
---|
349 | "http://tetaurawhiri.govt.nz",
|
---|
350 | "http://www.28maoribattalion.org.nz",
|
---|
351 | "https://ttw1.cwp.govt.nz",
|
---|
352 | "http://www.methodist.org.nz",
|
---|
353 | "http://avonside.net",
|
---|
354 | "https://www.takitimu.ac.nz",
|
---|
355 | "https://www.terito.school.nz",
|
---|
356 | "https://www.electionresults.org.nz",
|
---|
357 | "http://firstworldwar.tki.org.nz",
|
---|
358 | "http://animations.tewhanake.maori.nz",
|
---|
359 | "https://hepatakakupu.nz",
|
---|
360 | "https://www.zenbu.co.nz",
|
---|
361 | "https://www.sporty.co.nz",
|
---|
362 | "https://www.tasteofplenty.co.nz",
|
---|
363 | "http://otorohanga.directorybusiness.co.nz",
|
---|
364 | "https://www.puhaandpakeha.co.nz",
|
---|
365 | "http://kete.wcl.govt.nz",
|
---|
366 | "https://interactives.stuff.co.nz",
|
---|
367 | "http://maori.livingheritage.org.nz",
|
---|
368 | "https://www.hapuhauora.health.nz",
|
---|
369 | "http://kaiiwicamp.nz",
|
---|
370 | "http://talkingtothecan.com",
|
---|
371 | "http://www.tuwharetoa.iwi.nz",
|
---|
372 | "http://nzpostcard.co.nz",
|
---|
373 | "https://paekupu.co.nz",
|
---|
374 | "http://www.runanga.co.nz",
|
---|
375 | "https://curriculumtool.education.govt.nz",
|
---|
376 | "http://www.matarikifestival.org.nz",
|
---|
377 | "http://www.jeremybaker.nz",
|
---|
378 | "http://ngatipahauwera.co.nz",
|
---|
379 | "http://pukapuka.nz",
|
---|
380 | "http://www.writersfestival.co.nz",
|
---|
381 | "http://temahurehure.maori.nz",
|
---|
382 | "http://pukoro.co.nz",
|
---|
383 | "http://tehauora.org.nz",
|
---|
384 | "http://pukekohe.directorybusiness.co.nz",
|
---|
385 | "http://kmpmusic.co.nz",
|
---|
386 | "http://www.maoriinvestments.co.nz",
|
---|
387 | "https://www.komako.org.nz",
|
---|
388 | "https://www.tuiatematangi.ac.nz"
|
---|
389 | ]
|
---|
390 | }
|
---|
391 |
|
---|
392 | /* 2 */
|
---|
393 | {
|
---|
394 | "_id" : "de",
|
---|
395 | "count" : 19.0,
|
---|
396 | "domain" : [
|
---|
397 | "http://www.cartogiraffe.com",
|
---|
398 | "http://etymologie.info",
|
---|
399 | "http://arts.mythologica.fr",
|
---|
400 | "http://svenkirsten.com",
|
---|
401 | "http://weltderberge.de",
|
---|
402 | "https://www.you-fly.com",
|
---|
403 | "http://klaaskoehne.de",
|
---|
404 | "http://www.nierstrasz.org",
|
---|
405 | "https://www.tvteile.de",
|
---|
406 | "http://vulkane.ch",
|
---|
407 | "http://etoile-de-lune.net",
|
---|
408 | "http://www.stephe.de",
|
---|
409 | "http://insecta.pro",
|
---|
410 | "http://m.distanta.1km.net",
|
---|
411 | "https://ersatzteile-fachversand.de",
|
---|
412 | "https://laskar02cinta.page.tl",
|
---|
413 | "http://www.behlig.de",
|
---|
414 | "https://www.cartogiraffe.com",
|
---|
415 | "http://www.udhr.de"
|
---|
416 | ]
|
---|
417 | }
|
---|
418 |
|
---|
419 | /* 3 */
|
---|
420 | {
|
---|
421 | "_id" : "fr",
|
---|
422 | "count" : 16.0,
|
---|
423 | "domain" : [
|
---|
424 | "http://rapanui.fr",
|
---|
425 | "http://splaf.free.fr",
|
---|
426 | "https://www.lexilogos.com",
|
---|
427 | "http://mahajana.net",
|
---|
428 | "http://www.gif.ovh",
|
---|
429 | "http://baladeornithologique.com",
|
---|
430 | "http://www.gaudry.be",
|
---|
431 | "http://kihikihi.fr",
|
---|
432 | "http://www.blueheavenisland.com",
|
---|
433 | "http://www.gototahiti.net",
|
---|
434 | "http://www.maraamusurfskirace.com",
|
---|
435 | "http://www.rongo-rongo.com",
|
---|
436 | "http://chantsdeluttes.free.fr",
|
---|
437 | "http://pt.city-usa.net",
|
---|
438 | "https://www.manualscat.com",
|
---|
439 | "http://blueheavenisland.com"
|
---|
440 | ]
|
---|
441 | }
|
---|
442 |
|
---|
443 | /* 4 */
|
---|
444 | {
|
---|
445 | "_id" : "nl",
|
---|
446 | "count" : 16.0,
|
---|
447 | "domain" : [
|
---|
448 | "https://www.arrowhead.eu",
|
---|
449 | "http://tonhut.nl",
|
---|
450 | "http://nielsonboutique.co.uk",
|
---|
451 | "http://longhornlaw.net",
|
---|
452 | "http://tetsubo.org",
|
---|
453 | "https://arrowheadproject.azurewebsites.net",
|
---|
454 | "http://hidsonphoto.com",
|
---|
455 | "http://www.gouvernante.info",
|
---|
456 | "http://gouvernante.info",
|
---|
457 | "http://diverosa.com",
|
---|
458 | "https://arrowhead.eu",
|
---|
459 | "http://www.nonlinear.demon.nl",
|
---|
460 | "http://www.encyclo.co.uk",
|
---|
461 | "https://www.henrifloor.nl",
|
---|
462 | "http://skimap.info",
|
---|
463 | "http://wearehomework.com"
|
---|
464 | ]
|
---|
465 | }
|
---|
466 |
|
---|
467 | /* 5 */
|
---|
468 | {
|
---|
469 | "_id" : "dk",
|
---|
470 | "count" : 8.0,
|
---|
471 | "domain" : [
|
---|
472 | "http://akona.ngapuhitelevision.com",
|
---|
473 | "http://waiatarangatiratanga.ngapuhitelevision.com",
|
---|
474 | "http://jazz.ngapuhitelevision.com",
|
---|
475 | "http://ngapuhitelevision.com",
|
---|
476 | "http://ngapuhiradio.com",
|
---|
477 | "http://www.rennertweb.de",
|
---|
478 | "http://powhiri.ngapuhitelevision.com",
|
---|
479 | "http://komisch.ngapuhitelevision.com"
|
---|
480 | ]
|
---|
481 | }
|
---|
482 |
|
---|
483 | /* 6 */
|
---|
484 | {
|
---|
485 | "_id" : "ca",
|
---|
486 | "count" : 7.0,
|
---|
487 | "domain" : [
|
---|
488 | "http://bcmarina.com",
|
---|
489 | "http://www.myrasplace.net",
|
---|
490 | "http://00.gs",
|
---|
491 | "http://aguadilla.airport-authority.com",
|
---|
492 | "http://bckayak.com",
|
---|
493 | "https://articles.imperialtometric.com",
|
---|
494 | "http://daandehn.com"
|
---|
495 | ]
|
---|
496 | }
|
---|
497 |
|
---|
498 | /* 7 */
|
---|
499 | {
|
---|
500 | "_id" : "au",
|
---|
501 | "count" : 5.0,
|
---|
502 | "domain" : [
|
---|
503 | "http://theunderwaterworld.com",
|
---|
504 | "https://www.kiwiproperty.com",
|
---|
505 | "http://fionajack.net",
|
---|
506 | "https://infogram.com",
|
---|
507 | "https://koreromaori.com"
|
---|
508 | ]
|
---|
509 | }
|
---|
510 |
|
---|
511 | /* 8 */
|
---|
512 | {
|
---|
513 | "_id" : "gb",
|
---|
514 | "count" : 4.0,
|
---|
515 | "domain" : [
|
---|
516 | "https://omniatlas.com",
|
---|
517 | "http://www.wordsearchfun.com",
|
---|
518 | "http://www.woolrych.org",
|
---|
519 | "http://mikestephens.co.uk"
|
---|
520 | ]
|
---|
521 | }
|
---|
522 |
|
---|
523 | /* 9 */
|
---|
524 | {
|
---|
525 | "_id" : "es",
|
---|
526 | "count" : 4.0,
|
---|
527 | "domain" : [
|
---|
528 | "https://www.uv.es",
|
---|
529 | "https://www.reclamaciondevuelos.com",
|
---|
530 | "http://www.info-hoteles.com",
|
---|
531 | "http://www.cruceros-princess.mx"
|
---|
532 | ]
|
---|
533 | }
|
---|
534 |
|
---|
535 | /* 10 */
|
---|
536 | {
|
---|
537 | "_id" : "cz",
|
---|
538 | "count" : 4.0,
|
---|
539 | "domain" : [
|
---|
540 | "http://www.henryklahola.nazory.cz",
|
---|
541 | "https://www.fipojobs.com",
|
---|
542 | "http://about.ilikeyou.com",
|
---|
543 | "http://henryklahola.nazory.cz"
|
---|
544 | ]
|
---|
545 | }
|
---|
546 |
|
---|
547 | /* 11 */
|
---|
548 | {
|
---|
549 | "_id" : "it",
|
---|
550 | "count" : 3.0,
|
---|
551 | "domain" : [
|
---|
552 | "http://oipaz.net",
|
---|
553 | "http://www.marcosanti.it",
|
---|
554 | "http://www.pegasoesmicamion.com"
|
---|
555 | ]
|
---|
556 | }
|
---|
557 |
|
---|
558 | /* 12 */
|
---|
559 | {
|
---|
560 | "_id" : "at",
|
---|
561 | "count" : 3.0,
|
---|
562 | "domain" : [
|
---|
563 | "http://www.tmtmm.net",
|
---|
564 | "http://www.petit-prince.at",
|
---|
565 | "http://petit-prince.at"
|
---|
566 | ]
|
---|
567 | }
|
---|
568 |
|
---|
569 | /* 13 */
|
---|
570 | {
|
---|
571 | "_id" : "ro",
|
---|
572 | "count" : 2.0,
|
---|
573 | "domain" : [
|
---|
574 | "http://parohiauceadesus.ro",
|
---|
575 | "http://www.parohiauceadesus.ro"
|
---|
576 | ]
|
---|
577 | }
|
---|
578 |
|
---|
579 | /* 14 */
|
---|
580 | {
|
---|
581 | "_id" : "ch",
|
---|
582 | "count" : 2.0,
|
---|
583 | "domain" : [
|
---|
584 | "https://nicoledidi.ch",
|
---|
585 | "https://photos.axelebert.org"
|
---|
586 | ]
|
---|
587 | }
|
---|
588 |
|
---|
589 | /* 15 */
|
---|
590 | {
|
---|
591 | "_id" : "il",
|
---|
592 | "count" : 2.0,
|
---|
593 | "domain" : [
|
---|
594 | "https://www.hitiaotera.com",
|
---|
595 | "http://www.daat.ac.il"
|
---|
596 | ]
|
---|
597 | }
|
---|
598 |
|
---|
599 | /* 16 */
|
---|
600 | {
|
---|
601 | "_id" : "ru",
|
---|
602 | "count" : 1.0,
|
---|
603 | "domain" : [
|
---|
604 | "https://www.gismeteo.lv"
|
---|
605 | ]
|
---|
606 | }
|
---|
607 |
|
---|
608 | /* 17 */
|
---|
609 | {
|
---|
610 | "_id" : "bg",
|
---|
611 | "count" : 1.0,
|
---|
612 | "domain" : [
|
---|
613 | "http://anitra.net"
|
---|
614 | ]
|
---|
615 | }
|
---|
616 |
|
---|
617 | /* 18 */
|
---|
618 | {
|
---|
619 | "_id" : "mx",
|
---|
620 | "count" : 1.0,
|
---|
621 | "domain" : [
|
---|
622 | "http://www.gelbukh.com"
|
---|
623 | ]
|
---|
624 | }
|
---|
625 |
|
---|
626 | /* 19 */
|
---|
627 | {
|
---|
628 | "_id" : "unknown",
|
---|
629 | "count" : 1.0,
|
---|
630 | "domain" : [
|
---|
631 | "https://www.viveipcl.com"
|
---|
632 | ]
|
---|
633 | }
|
---|
634 |
|
---|
635 | /* 20 */
|
---|
636 | {
|
---|
637 | "_id" : "jp",
|
---|
638 | "count" : 1.0,
|
---|
639 | "domain" : [
|
---|
640 | "http://yutaka.it-n.jp"
|
---|
641 | ]
|
---|
642 | }
|
---|
643 |
|
---|
644 | /* 21 */
|
---|
645 | {
|
---|
646 | "_id" : "cn",
|
---|
647 | "count" : 1.0,
|
---|
648 | "domain" : [
|
---|
649 | "http://kiwi2china.com"
|
---|
650 | ]
|
---|
651 | }
|
---|
652 |
|
---|
653 | /* 22 */
|
---|
654 | {
|
---|
655 | "_id" : "ie",
|
---|
656 | "count" : 1.0,
|
---|
657 | "domain" : [
|
---|
658 | "https://coggle.it"
|
---|
659 | ]
|
---|
660 | }
|
---|
661 |
|
---|
662 | /* 23 */
|
---|
663 | {
|
---|
664 | "_id" : "ir",
|
---|
665 | "count" : 1.0,
|
---|
666 | "domain" : [
|
---|
667 | "https://www.dideo.ir"
|
---|
668 | ]
|
---|
669 | }
|
---|
670 |
|
---|
671 | /* 24 */
|
---|
672 | {
|
---|
673 | "_id" : "fi",
|
---|
674 | "count" : 1.0,
|
---|
675 | "domain" : [
|
---|
676 | "http://pertti.com"
|
---|
677 | ]
|
---|
678 | }
|
---|