diff --git a/dbms/src/Functions/domain.h b/dbms/src/Functions/domain.h index 10251f355b6..37c5a6fe5cd 100644 --- a/dbms/src/Functions/domain.h +++ b/dbms/src/Functions/domain.h @@ -102,7 +102,7 @@ inline StringRef getURLHost(const char * data, size_t size) return StringRef{}; has_dot_delimiter = true; } - else if (isEndOfUrl(*pos)) + else if (isCharEndOfUrl(*pos)) break; else if (isUnsafeCharUrl(*pos) || isReservedCharUrl(*pos)) return StringRef{}; diff --git a/dbms/src/Functions/tl.sql b/dbms/src/Functions/tl.sql new file mode 100644 index 00000000000..3f9d241d654 --- /dev/null +++ b/dbms/src/Functions/tl.sql @@ -0,0 +1,26 @@ +CREATE TABLE ip( + a FixedString(16) +) ENGINE = Memory; + + + +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); +INSERT INTO ip(a) VALUES(IPv6StringToNum('2001:0db8:0000:85a3:0000:0000:ac1f:8001')); + +WITH IPv6CIDRtoIPv6Range(IPv6StringToNum('2001:0DB8:AC10:FE01:FEED:BABE:CAFE:F00D'), 40) as test SELECT IPv6NumToString(a) FROM ip WHERE a BETWEEN tupleElement(test, 1) AND tupleElement(test, 2) ; diff --git a/dbms/tests/queries/0_stateless/00398_url_functions.reference b/dbms/tests/queries/0_stateless/00398_url_functions.reference index e5cead07b9c..23390c199f0 100644 --- a/dbms/tests/queries/0_stateless/00398_url_functions.reference +++ b/dbms/tests/queries/0_stateless/00398_url_functions.reference @@ -20,7 +20,7 @@ example.com com ru -ru + com com com diff --git a/dbms/tests/queries/1_stateful/00037_uniq_state_merge1.reference b/dbms/tests/queries/1_stateful/00037_uniq_state_merge1.reference index f0a48df0f91..3bedecd267b 100644 --- a/dbms/tests/queries/1_stateful/00037_uniq_state_merge1.reference +++ b/dbms/tests/queries/1_stateful/00037_uniq_state_merge1.reference @@ -1,15 +1,14 @@ - 85185 85185 -yandex.ru 25106 25106 + 89348 89348 +yandex.ru 25105 25105 avito.ru 16523 16523 -mail.yandex.ru 13662 13662 +mail.yandex.ru 13659 13659 mail.ru 7643 7643 auto.ru 7350 7350 hurpass.com 6395 6395 best.ru 5477 5477 tv.yandex.ru 5341 5341 korer.ru 4967 4967 -mail.yandsearch 4244 4244 -main=hurriyet.com 3806 3806 +mail.yandsearch 4237 4237 yandex.ua 3802 3802 korablitz.ru 3717 3717 uyelik.hurriyet.com 3584 3584 @@ -20,14 +19,14 @@ coccoc.com 2707 2707 rutube.ru 2699 2699 rbc.ru 2644 2644 mamba.ru 2598 2598 -mail.yandex 2443 2443 +mail.yandex 2441 2441 pikabu.ru 2130 2130 yandex.php 2057 2057 e.mail.yandex.ru 1971 1971 brandex.ru 1969 1969 market.ru 1913 1913 mynet.ru 1881 1881 -mail.yandex.ua 1824 1824 +mail.yandex.ua 1823 1823 rutube.com 1821 1821 news.rambler.com 1787 1787 hurpass.com.tr 1763 1763 @@ -61,16 +60,14 @@ haber.com 1043 1043 megogo.net 993 993 sozcu.com 991 991 yandex.by 938 938 -fast-golove.mail.ru_Mobile=0&at=35&text=производств 927 927 gazeta.ru 892 892 -yandex.ru;yandex.ru 891 891 fotki.yandex.ru 875 875 fast-golove.mail.yandex.php 842 842 lenta.ru 820 820 publicdaroglundai_anketa.ru 813 813 mail.yandex.kz 810 810 censor.net 807 807 -mail.yandex.by 805 805 +mail.yandex.by 804 804 nnn.ru 796 796 maxi.su 788 788 rambler.ru 755 755 @@ -98,3 +95,6 @@ ria.ru 591 591 marina_prezideniz.hurriyet.com 578 578 youtube.ru 575 575 cars.autochno.ru 570 570 +a2.stars.auto.yandsearch 566 566 +love.mail.ru 560 560 +mail.rambler.ru 553 553 diff --git a/dbms/tests/queries/1_stateful/00038_uniq_state_merge2.reference b/dbms/tests/queries/1_stateful/00038_uniq_state_merge2.reference index 0ea3994f453..575d19b2ebf 100644 --- a/dbms/tests/queries/1_stateful/00038_uniq_state_merge2.reference +++ b/dbms/tests/queries/1_stateful/00038_uniq_state_merge2.reference @@ -1,100 +1,100 @@ -ru 282356 70396 - 96315 85424 -com 72459 33535 -html 33195 17665 -ua 31469 18290 -tr 18197 12501 -net 17633 12318 -php 16416 11358 -yandsearch 12917 9992 -by 8690 7254 -yandex 8004 6661 -org 5397 4920 -tv 4957 4360 -kz 4915 4388 -phtml 2754 2529 -tr&callback_url=http 2742 2556 -su 2731 2505 -xml 2731 2520 -aspx 2593 2394 -search 1854 1844 -shtml 1788 1688 -hurriyet 1517 1468 -tr&user 1469 1419 -tr&users 1361 1299 -tr&callback 1197 1157 -jpg 1129 1094 -jsp 1028 1000 -ru_Mobile=0&at=35&text=производств 927 927 -lv 897 892 -tr&user_page=http 859 831 -tr&user_page 858 834 -exe 853 837 -me 791 754 -tr&used 761 747 -pro 757 753 -htm 756 736 -do 747 730 +ru 262911 69218 + 92101 89421 +com 63297 30285 +ua 29037 17475 +html 25077 15037 +tr 16770 11857 +net 16387 11686 +php 14373 10307 +yandsearch 12024 9484 +by 8192 6915 +yandex 7211 6124 +org 4890 4514 +kz 4677 4209 +tv 4400 3928 +su 2602 2396 +phtml 2409 2226 +xml 2322 2182 +aspx 1959 1848 +search 1835 1827 +hurriyet 1385 1345 +shtml 995 966 +lv 879 875 +jsp 855 845 +exe 814 798 +pro 737 734 airway 724 724 -mail 632 618 -biz 623 612 -info 525 515 +me 675 647 +jpg 662 647 +do 625 611 +mail 593 581 +biz 537 530 bstatistik_dlja-dlya-naches 521 521 -tr&callback_url=https 508 501 -news 469 464 -sx 465 464 -pogoda 460 460 -hu 432 429 +info 461 453 +pogoda 459 459 +sx 450 449 +news 448 444 sportlibrary 431 431 -aspx&referer 407 395 -auto 406 398 -tr&callbusiness 398 395 -az 393 385 -fm 387 386 -online 382 380 +hu 396 393 +htm 393 385 +fm 379 378 +online 374 372 bstatistic 366 366 -tr&callbusines 365 362 -wbp 344 344 -tr&user_page=https 326 325 -ru;yandsearch 322 321 -tr&callback_url 307 297 +auto 363 355 +az 356 350 +wbp 343 343 bstana 304 304 -am 292 290 -ru;yandex 279 276 -blog 274 274 -pl 273 271 -diary 267 265 -jpg,http 267 261 -slando 260 258 -eu 256 253 -im 248 245 -to 242 240 -tj 231 231 -mynet 226 225 -sberbank 224 224 -_2544 222 222 +blog 268 268 +diary 262 261 +am 260 258 +slando 254 252 +im 238 235 +eu 237 234 liveinteria 218 218 -mamba 217 217 -jpg,https 215 215 -tr&user_pts=&states 213 213 -jpg&pos 212 210 -html&lang 211 207 +to 215 213 +mamba 214 214 auto-supers 208 208 -yandsearchplus 205 204 -asp 201 198 -aspx&referera 201 201 +sberbank 207 207 +tj 205 205 bstatistik_dlja-dlya_avia 201 201 bstanii_otryasam 200 200 +pl 200 198 wroad_5d 200 200 -com&callback_url=http 198 198 -ru&pos=3_0 198 198 -haber 196 191 -ws 194 193 -kg 191 190 -video 190 190 -co 188 188 +mynet 191 190 bstan 187 187 -swf 186 186 -cc 185 182 +yandsearchplus 186 186 +haber 184 179 +jpg,https 184 184 turkasovki 183 183 -wssp 176 176 +co 177 177 +video 177 177 +gif","photos 175 175 +mgshared_zone 172 172 +wssp 172 172 +jpg,http 170 168 +swf 167 167 +cc 166 164 +ws 164 164 +kg 157 156 +mobili_s_probegom 154 153 +cgi 153 152 +yandsearcher 152 151 +uz 150 150 +nsf 149 149 +adriver 147 144 +slandsearch 143 142 +korrez 140 140 +bstatistik_dlja-dlja-putin 139 139 +rambler 133 132 +mvideo 132 132 +asp 129 128 +vc 127 127 +md 121 121 +jpg","photo 119 119 +mp4 118 117 +ee 116 115 +loveplaceOfSearchplus 111 111 +nl 111 111 +bstatistika 107 107 +br 102 102 +sport 99 99 diff --git a/dbms/tests/queries/1_stateful/00044_any_left_join_string.reference b/dbms/tests/queries/1_stateful/00044_any_left_join_string.reference index f5ef97e01d7..05e97417263 100644 --- a/dbms/tests/queries/1_stateful/00044_any_left_join_string.reference +++ b/dbms/tests/queries/1_stateful/00044_any_left_join_string.reference @@ -1,10 +1,10 @@ - 3959563 550936 + 4508175 712434 auto.ru 576845 8935 -yandex.ru 410783 111278 +yandex.ru 410776 111278 korer.ru 277987 0 avito.ru 163820 15556 -mail.yandex.ru 152468 1046 -main=hurriyet.com 152096 259 +mail.yandex.ru 152447 1046 mail.ru 87949 22225 best.ru 58537 55 korablitz.ru 51844 0 +hurpass.com 49671 1251 diff --git a/dbms/tests/queries/1_stateful/00089_position_functions_with_non_constant_arg.reference b/dbms/tests/queries/1_stateful/00089_position_functions_with_non_constant_arg.reference index 530d00668a4..4d0ba2b70f3 100644 --- a/dbms/tests/queries/1_stateful/00089_position_functions_with_non_constant_arg.reference +++ b/dbms/tests/queries/1_stateful/00089_position_functions_with_non_constant_arg.reference @@ -2,5 +2,5 @@ 0 0 0 -http://topicId=323145-EXC=1-PG=10&from=distriruyu-redakciy-lakovora-dalgames.mail@mail.yandex.ru/yandex.ru/news.mail -http://plugin_sd=1&ie=UTF-8&l=vi&p=AhY_cQZSQQ5JBlUEZVcJG1F4XldSeWNjVEdhen83@mail.yandsearch +https://povary_dlya-511-gemotedDynamo_accoshyutoy-s-kortosh@bk.ru/yandsearch?text=simages%2F8%2F10544998#posts%2Fkartofeleri +https://povary_dlya-511-gemotedDynamo_accoshyutoy-s-kortosh@bk.ru/yandsearch?text=simages%2F8%2F10544998#posts%2Fkartofeleri