Viewing abuse filter 8: Repeating characters – Travel guide at Travelerzom

Editing filter 8

Filter parameters

Filter ID:	8
Description: (publicly viewable)
Statistics:	Of the last 414 actions, this filter has matched 0 (0%). On average, its run time is 0.25 ms, and it consumes 2.1 conditions of the condition limit.
Conditions:	page_namespace = 0 & !("confirmed" in user_groups) &(rmwhitespace(added_lines) rlike "([^_:.'\|=}{0 -]{1,9})\1{7}"\| added_lines rlike "[bcdfghjklmnpqrstvwxz]{9,}") & !(rmwhitespace(removed_lines) rlike "([^_:.'\|=}{-]{1,9})\1{6}") & !(removed_lines rlike "[bcdfghjklmnpqrstvwxz]{8,}") & !(added_lines rlike "https?:\/\/\S([^_:'\|=}{-]{1,9})\1{7}\|https?:\/\/\S[bcdfghjklmnpqrstvwxz]{7,}\|(^\|\n)[\|!]") & !(lcase(added_lines) rlike "number\|spell\|math\|string\|<br\b") & !(lcase(page_title) rlike "number") page_namespace = 0 & !("confirmed" in user_groups) &(rmwhitespace(added_lines) rlike "([^_:.'\|=}{0 -]{1,9})\1{7}"\| added_lines rlike "[bcdfghjklmnpqrstvwxz]{9,}") & !(rmwhitespace(removed_lines) rlike "([^_:.'\|=}{-]{1,9})\1{6}") & !(removed_lines rlike "[bcdfghjklmnpqrstvwxz]{8,}") & !(added_lines rlike "https?:\/\/\S([^_:'\|=}{-]{1,9})\1{7}\|https?:\/\/\S[bcdfghjklmnpqrstvwxz]{7,}\|(^\|\n)[\|!]") & !(lcase(added_lines) rlike "number\|spell\|math\|string\|<br\b") & !(lcase(page_title) rlike "number")
Notes:	Catch copy/paste vandalism like hahahahahahahahahahahahaha. --Wknight94 This doesn't need to be private. -BJ This has far too many false positives to be set to disallow... Not to mention that it seems to block people from using "..." and "'''''" (for bold/italic), amongst other things. --Conti It can at least warn. "'''''''" problem is fixed now. Added curly brace exception too. Need example of "..." since at least 7 occurrences are needed to trigger. --Wknight94 Changed title to something more neutral. --Conti Too many false positives - mostly in reference URLs. Unless someone has a clever way to fix, this is causing more harm than good IMHO. Agreed, please do not enable this filter without substantial changes, and substantial log only testing. -Prodego removing the quote, it was triggered by '''xxxx''' in certain cases -- Luk The filter should ignore urls. - Ruslik Ignoring all edits with URLs in them isn't a good idea. I've made it a bit more complex, actually checking if a link contains repeating chars. If any added link contains added chars, it stops. Theoretically we could do all sorts of counting to make sure there isn't a link with repeating chars AND other repeating chars, but I think that would just eat away at needless conditions, considering it is likely very rare. -Prodego Turning out not to be too useful and the filter is overwrought now. Disabling. -Wknight94 Merged testing edit summary from 292 - Shirik 12 Feb 2010 Optimized summary block - Shirik 12 Feb 2010 ignoring if repeated chars was already there, per request -- Jun 18, Soap Merge filter 366 into this. – Oct 28, EdoDodo Filter tripped the rate limit. I reset it; I didn't see any FPs. -- Shirik 7 Nov 2010 Reset rate limit. - KoH 10 Mar 2011 Simplify. -Sole Soul Ignore zero - FPs. --Reaper 2012-05-25 fix syntax --SHB2000 2024-11-14 Changing '!(lcase(article_text) rlike "number")' to "…page_title…" is a functional change. I don't know why pages that include "number" would be excluded though – that's too common a word. If it's about phone numbers, those should be allowed regardless, by the regular expression (which my head cannot parse right now). –LPfi 2024-11-14
Flags:	Hide details of this filter from public view Enable this filter Mark as deleted
Filter last modified:	09:22, 14 November 2024 by LPfi (talk \| contribs)
History:	View this filter's history
Tools:	Export this filter to another wiki {"data":{"rules":"page_namespace = 0 \u0026 !(\"confirmed\" in user_groups)\r\n\u0026(rmwhitespace(added_lines) rlike \"([^_:.'\|=}{0 -]{1,9})\\1{7}\"\|\r\nadded_lines rlike \"[bcdfghjklmnpqrstvwxz]{9,}\")\r\n\u0026 !(rmwhitespace(removed_lines) rlike \"([^_:.'\|=}{-]{1,9})\\1{6}\")\r\n\u0026 !(removed_lines rlike \"[bcdfghjklmnpqrstvwxz]{8,}\")\r\n\u0026 !(added_lines rlike \"https?:\\/\\/\\S([^_:'\|=}{-]{1,9})\\1{7}\|https?:\\/\\/\\S*[bcdfghjklmnpqrstvwxz]{7,}\|(^\|\\n)[\|!]\")\r\n\u0026 !(lcase(added_lines) rlike \"number\|spell\|math\|string\|\u003Cbr\\b\")\r\n\u0026 !(lcase(page_title) rlike \"number\")","name":"Repeating characters","comments":"Catch copy/paste vandalism like hahahahahahahahahahahahaha. --Wknight94\r\n\r\nThis doesn't need to be private. -BJ\r\n\r\nThis has far too many false positives to be set to disallow... Not to mention that it seems to block people from using \"...\" and \"'''''\" (for bold/italic), amongst other things. --Conti\r\n\r\nIt can at least warn. \"'''''''\" problem is fixed now. Added curly brace exception too. Need example of \"...\" since at least 7 occurrences are needed to trigger. --Wknight94\r\n\r\nChanged title to something more neutral. --Conti\r\n\r\nToo many false positives - mostly in reference URLs. Unless someone has a clever way to fix, this is causing more harm than good IMHO.\r\n\r\nAgreed, please do not enable this filter without substantial changes, and substantial log only testing. -Prodego\r\n\r\nremoving the quote, it was triggered by '''xxxx''' in certain cases -- Luk\r\nThe filter should ignore urls. - Ruslik\r\n\r\nIgnoring all edits with URLs in them isn't a good idea. I've made it a bit more complex, actually checking if a link contains repeating chars. If any added link contains added chars, it stops. Theoretically we could do all sorts of counting to make sure there isn't a link with repeating chars AND other repeating chars, but I think that would just eat away at needless conditions, considering it is likely very rare. -Prodego\r\n\r\nTurning out not to be too useful and the filter is overwrought now. Disabling. -Wknight94\r\nMerged testing edit summary from 292 - Shirik 12 Feb 2010\r\nOptimized summary block - Shirik 12 Feb 2010\r\nignoring if repeated chars was already there, per request -- Jun 18, Soap\r\n\r\nMerge filter 366 into this. \u2013\u00a0Oct 28, EdoDodo\r\nFilter tripped the rate limit. I reset it; I didn't see any FPs. -- Shirik 7 Nov 2010\r\nReset rate limit. - KoH 10 Mar 2011\r\n\r\nSimplify. -Sole Soul\r\n\r\nIgnore zero - FPs. --Reaper 2012-05-25\r\n\r\nfix syntax --SHB2000 2024-11-14\r\n\r\nChanging '!(lcase(article_text) rlike \"number\")' to \"\u2026page_title\u2026\" is a functional change. I don't know why pages that include \"number\" would be excluded though \u2013 that's too common a word. If it's about phone numbers, those should be allowed regardless, by the regular expression (which my head cannot parse right now). \u2013LPfi 2024-11-14","group":"default","actions":{"tag":["repeating characters"]},"enabled":true,"deleted":false,"privacylevel":0,"global":false},"actions":{"tag":["repeating characters"]}}