write regex captures to parquet arrays.

[mediawiki_dump_tools.git] / test / baseline_output / basic_regextest_1.tsv

diff --git a/test/baseline_output/basic_regextest_1.tsv b/test/baseline_output/basic_regextest_1.tsv
index b1201fb2bb0c16f1f14ea3d137301d161f2c012d..085070ee56993099ce752039c0f6737be5117f98 100644 (file)
--- a/test/baseline_output/basic_regextest_1.tsv
+++ b/test/baseline_output/basic_regextest_1.tsv
@@ -1,27 +1,27 @@
-anon   articleid       date_time       deleted editor  editorid        minor   namespace       page_word       revert  reverteds       revid   sha1    testcases       text_chars      title
-FALSE  56237363        2018-01-07 10:40:58     FALSE   "NinjaRobotPirate"      3742946 FALSE   3       page, page      FALSE           819091731       135nz8q6lfam6cojla7azb7k5alx3t3         1141    "User talk:86.139.142.254"
-FALSE  56237364        2018-01-07 10:41:10     FALSE   "Kavin kavitha" 32792125        FALSE   3               FALSE           819091755       0pwezjc6yopz0smc8al6ogc4fax5bwo         663     "User talk:Kavin kavitha"
-FALSE  56237365        2018-01-07 10:41:26     FALSE   "Amicable always"       32621254        FALSE   3               FALSE           819091788       sz3t2ap7z8bpkdvdvi195f3i35949bv TestCase, TestCase      399     "User talk:Dr.vivek163"
-FALSE  56237366        2018-01-07 10:41:31     FALSE   "ClueBot NG"    13286072        FALSE   3       page    FALSE           819091796       r6s5j8j3iykenrhuhpnkpsmmd71vubf         1260    "User talk:Twistorl"
-FALSE  56237368        2018-01-07 10:41:51     FALSE   "Khruner"       8409334 FALSE   0       page    FALSE           819091825       tf5qz2yaswx61zrlm9ovxzuhl7r2dc4 TestCase        2249    "Kom Firin"
-FALSE  56237368        2018-01-27 12:16:02     FALSE   "Khruner"       8409334 TRUE    0       page    FALSE           822610647       e6oa4g0qv64icdaq26uu1zzbyr5hcbh         2230    "Kom Firin"
-FALSE  56237369        2018-01-07 10:42:05     FALSE   "Editingaccount1994"    32794215        FALSE   2       page, page      FALSE           819091844       0fyvyh2a8xu41gt8obr34oba0bfixj6         27840   "User:Editingaccount1994/sandbox"
-FALSE  56237369        2018-01-07 11:09:52     FALSE   "AnomieBOT"     7611264 TRUE    2       page, page      FALSE           819093984       8gy52aolt5rg3eaketwj5v7eiw0apv2         27787   "User:Editingaccount1994/sandbox"
-FALSE  56237369        2018-01-12 21:45:50     FALSE   "SporkBot"      12406635        TRUE    2       page, page      FALSE           820064189       he8ydemaanxlrpftqxkez8jfpge1fsj         27784   "User:Editingaccount1994/sandbox"
-FALSE  56237369        2018-01-12 23:28:11     FALSE   "SporkBot"      12406635        TRUE    2       page, page      FALSE           820078679       0to17w9rth3url8n7gvucdtobybdq5h         27783   "User:Editingaccount1994/sandbox"
-FALSE  56237369        2018-01-12 23:28:39     FALSE   "SporkBot"      12406635        TRUE    2       page, page      FALSE           820078733       531dizmmloyxffbkdr5vph7owh921eg         27782   "User:Editingaccount1994/sandbox"
-FALSE  56237369        2018-01-13 13:45:33     FALSE   "Frietjes"      13791031        FALSE   2       page, page      FALSE           820177382       nik9p2u2fuk4yazjxt8ymbicxv5qid9         27757   "User:Editingaccount1994/sandbox"
-FALSE  56237369        2018-01-24 01:35:22     FALSE   "CommonsDelinker"       2304267 FALSE   2       page, page      FALSE           822038928       gwk6pampl8si1v5pv3kwgteg710sfw3         27667   "User:Editingaccount1994/sandbox"
-FALSE  56237370        2018-01-07 10:42:20     FALSE   "PamD"  1368779 FALSE   0               FALSE           819091874       n4ozbsgle13p9yywtfrz982ccj8woc9         25      "Anita del Rey"
-FALSE  56237371        2018-01-07 10:42:27     FALSE   "ClueBot NG"    13286072        FALSE   3       page    FALSE           819091883       ksohnvsbeuzwpl5vb8a3v8m18hva0a7         1274    "User talk:119.94.96.157"
-FALSE  56237372        2018-01-07 10:42:50     FALSE   "Underbar dk"   677153  FALSE   14              FALSE           819091914       je7aw21fedbwyqsyofpisdrynsu7olr         113     "Category:Ohmi Railway"
-FALSE  56237375        2018-01-07 10:43:32     FALSE   "TastyPoutine"  882433  FALSE   3               FALSE           819091968       cpm4tkzcx4hc6irr9ukbi06ogud8dtq         199     "User talk:92.226.219.222"
-FALSE  56237375        2018-01-07 11:10:24     FALSE   "AnomieBOT"     7611264 TRUE    3       page, page, page, page  FALSE           819094036       artmfz8b2gxhb3pp8a5p4ksplxqfkpg         1840    "User talk:92.226.219.222"
-FALSE  56237375        2018-01-07 14:33:36     FALSE   "Only"  702940  FALSE   3       page, page, page, page, page, page      FALSE           819112363       dn9wj0n8d8pdd5lqe56uw5xamupowr1         2949    "User talk:92.226.219.222"
-FALSE  56237376        2018-01-07 10:44:01     FALSE   "Dipayanacharya"        32794237        FALSE   2               FALSE           819092004       ofueugwatmmn7u73isw732neuza57gk         28      "User:Dipayanacharya"
-FALSE  56237376        2018-01-07 10:49:08     FALSE   "Dipayanacharya"        32794237        FALSE   2               FALSE           819092390       dsz55xv96ec2uv6w9c1z7c52ipfovbw         38      "User:Dipayanacharya"
-FALSE  56237378        2018-01-07 10:44:56     FALSE   "Vinegarymass911"       21516552        FALSE   0               FALSE           819092066       9ma38hak0ef1ew4fpiutxpnzd8oz1wd         65      "BSCIC"
-FALSE  56237379        2018-01-07 10:45:21     FALSE   "BrownHairedGirl"       754619  FALSE   14              FALSE           819092102       4dvakoat58bzyf5hmtthxukt29hip6n         285     "Category:Women government ministers of Yemen"
-FALSE  56237381        2018-01-07 10:45:54     FALSE   "PRehse"        410898  FALSE   1               FALSE           819092135       2sjrxsc7os9k9pg4su2t4rk2j8nn0h7         103     "Talk:List of Morning Glories Characters"
-FALSE  56237382        2018-01-07 10:45:56     FALSE   "ClueBot NG"    13286072        FALSE   3       page    FALSE           819092138       3y9t5wpk6ur5jhone75rhm4wjf01fgi         1330    "User talk:106.207.126.114"
-FALSE  56237382        2018-01-07 10:50:22     FALSE   "HindWIKI"      31190506        FALSE   3       page    FALSE           819092495       8wvn6vh3isyt0dorpe89lztrburgupe         2355    "User talk:106.207.126.114"
+revid  date_time       articleid       editorid        title   namespace       deleted text_chars      revert  reverteds       sha1    minor   editor  anon    testcases       page_word
+819091731      2018-01-07 10:40:58     56237363        3742946 "User talk:86.139.142.254"      3       FALSE   1141    FALSE           135nz8q6lfam6cojla7azb7k5alx3t3 FALSE   "NinjaRobotPirate"      FALSE           "page,page"
+819091755      2018-01-07 10:41:10     56237364        32792125        "User talk:Kavin kavitha"       3       FALSE   663     FALSE           0pwezjc6yopz0smc8al6ogc4fax5bwo FALSE   "Kavin kavitha" FALSE           
+819091788      2018-01-07 10:41:26     56237365        32621254        "User talk:Dr.vivek163" 3       FALSE   399     FALSE           sz3t2ap7z8bpkdvdvi195f3i35949bv FALSE   "Amicable always"       FALSE   "TestCase,TestCase"     
+819091796      2018-01-07 10:41:31     56237366        13286072        "User talk:Twistorl"    3       FALSE   1260    FALSE           r6s5j8j3iykenrhuhpnkpsmmd71vubf FALSE   "ClueBot NG"    FALSE           "page"
+819091825      2018-01-07 10:41:51     56237368        8409334 "Kom Firin"     0       FALSE   2249    FALSE           tf5qz2yaswx61zrlm9ovxzuhl7r2dc4 FALSE   "Khruner"       FALSE   "TestCase"      "page"
+822610647      2018-01-27 12:16:02     56237368        8409334 "Kom Firin"     0       FALSE   2230    FALSE           e6oa4g0qv64icdaq26uu1zzbyr5hcbh TRUE    "Khruner"       FALSE           "page"
+819091844      2018-01-07 10:42:05     56237369        32794215        "User:Editingaccount1994/sandbox"       2       FALSE   27840   FALSE           0fyvyh2a8xu41gt8obr34oba0bfixj6 FALSE   "Editingaccount1994"    FALSE           "page,page"
+819093984      2018-01-07 11:09:52     56237369        7611264 "User:Editingaccount1994/sandbox"       2       FALSE   27787   FALSE           8gy52aolt5rg3eaketwj5v7eiw0apv2 TRUE    "AnomieBOT"     FALSE           "page,page"
+820064189      2018-01-12 21:45:50     56237369        12406635        "User:Editingaccount1994/sandbox"       2       FALSE   27784   FALSE           he8ydemaanxlrpftqxkez8jfpge1fsj TRUE    "SporkBot"      FALSE           "page,page"
+820078679      2018-01-12 23:28:11     56237369        12406635        "User:Editingaccount1994/sandbox"       2       FALSE   27783   FALSE           0to17w9rth3url8n7gvucdtobybdq5h TRUE    "SporkBot"      FALSE           "page,page"
+820078733      2018-01-12 23:28:39     56237369        12406635        "User:Editingaccount1994/sandbox"       2       FALSE   27782   FALSE           531dizmmloyxffbkdr5vph7owh921eg TRUE    "SporkBot"      FALSE           "page,page"
+820177382      2018-01-13 13:45:33     56237369        13791031        "User:Editingaccount1994/sandbox"       2       FALSE   27757   FALSE           nik9p2u2fuk4yazjxt8ymbicxv5qid9 FALSE   "Frietjes"      FALSE           "page,page"
+822038928      2018-01-24 01:35:22     56237369        2304267 "User:Editingaccount1994/sandbox"       2       FALSE   27667   FALSE           gwk6pampl8si1v5pv3kwgteg710sfw3 FALSE   "CommonsDelinker"       FALSE           "page,page"
+819091874      2018-01-07 10:42:20     56237370        1368779 "Anita del Rey" 0       FALSE   25      FALSE           n4ozbsgle13p9yywtfrz982ccj8woc9 FALSE   "PamD"  FALSE           
+819091883      2018-01-07 10:42:27     56237371        13286072        "User talk:119.94.96.157"       3       FALSE   1274    FALSE           ksohnvsbeuzwpl5vb8a3v8m18hva0a7 FALSE   "ClueBot NG"    FALSE           "page"
+819091914      2018-01-07 10:42:50     56237372        677153  "Category:Ohmi Railway" 14      FALSE   113     FALSE           je7aw21fedbwyqsyofpisdrynsu7olr FALSE   "Underbar dk"   FALSE           
+819091968      2018-01-07 10:43:32     56237375        882433  "User talk:92.226.219.222"      3       FALSE   199     FALSE           cpm4tkzcx4hc6irr9ukbi06ogud8dtq FALSE   "TastyPoutine"  FALSE           
+819094036      2018-01-07 11:10:24     56237375        7611264 "User talk:92.226.219.222"      3       FALSE   1840    FALSE           artmfz8b2gxhb3pp8a5p4ksplxqfkpg TRUE    "AnomieBOT"     FALSE           "page,page,page,page"
+819112363      2018-01-07 14:33:36     56237375        702940  "User talk:92.226.219.222"      3       FALSE   2949    FALSE           dn9wj0n8d8pdd5lqe56uw5xamupowr1 FALSE   "Only"  FALSE           "page,page,page,page,page,page"
+819092004      2018-01-07 10:44:01     56237376        32794237        "User:Dipayanacharya"   2       FALSE   28      FALSE           ofueugwatmmn7u73isw732neuza57gk FALSE   "Dipayanacharya"        FALSE           
+819092390      2018-01-07 10:49:08     56237376        32794237        "User:Dipayanacharya"   2       FALSE   38      FALSE           dsz55xv96ec2uv6w9c1z7c52ipfovbw FALSE   "Dipayanacharya"        FALSE           
+819092066      2018-01-07 10:44:56     56237378        21516552        "BSCIC" 0       FALSE   65      FALSE           9ma38hak0ef1ew4fpiutxpnzd8oz1wd FALSE   "Vinegarymass911"       FALSE           
+819092102      2018-01-07 10:45:21     56237379        754619  "Category:Women government ministers of Yemen"  14      FALSE   285     FALSE           4dvakoat58bzyf5hmtthxukt29hip6n FALSE   "BrownHairedGirl"       FALSE           
+819092135      2018-01-07 10:45:54     56237381        410898  "Talk:List of Morning Glories Characters"       1       FALSE   103     FALSE           2sjrxsc7os9k9pg4su2t4rk2j8nn0h7 FALSE   "PRehse"        FALSE           
+819092138      2018-01-07 10:45:56     56237382        13286072        "User talk:106.207.126.114"     3       FALSE   1330    FALSE           3y9t5wpk6ur5jhone75rhm4wjf01fgi FALSE   "ClueBot NG"    FALSE           "page"
+819092495      2018-01-07 10:50:22     56237382        31190506        "User talk:106.207.126.114"     3       FALSE   2355    FALSE           8wvn6vh3isyt0dorpe89lztrburgupe FALSE   "HindWIKI"      FALSE           "page"