Author: jhardin
Date: Fri Jun 5 23:08:19 2020
New Revision: 1878527
URL: http://svn.apache.org/viewvc?rev=1878527&view=rev
Log:
LOTSA_MONEY family: fix FP on dates, don't trigger on number followed by word beginning with "o"
Modified:
spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf
Modified: spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf
URL: http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf?rev=1878527&r1=1878526&r2=1878527&view=diff
==============================================================================
--- spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf (original)
+++ spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf Fri Jun 5 23:08:19 2020
@@ -7,13 +7,14 @@
ifplugin Mail::SpamAssassin::Plugin::ReplaceTags
replace_tag CURRENCY [\(\[]?(?:\bU[Ss][D\$]{0,2}|\$(?:US)?|usd|CAD|GBP|=[Aa][34]|\xa3|&\#16[34];|(?i:pounds\ssterling)|\xa4|EUR(?:OS)?|(?:d')?[Ee]uro?s?|(?i:eur)\sde|CHF|FCFA|d[\xf3]lares\sde\slos\sE+\.\s?U+\.)[\]\)]?
replace_tag GB_UK \b(?:U\.?K\.?|(?:Great\s)?Brit(?:ain|ish)|G\.?B\.?)\b
+ replace_tag NUM_NOT_DATE [1-9](?!\d\d\d\.\d\d\.\d\d\s)(?!\d\.\d\d\.\d\d\d\d\s)
- body __LOTSA_MONEY_00 /<CURRENCY>[\s\.]?[1-9][\dOo][,\.][\dOo]{3}(?:(?!\d)|\b)/
- body __LOTSA_MONEY_01 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)[\s\.]?[1-9][\d.,\sOo]{5,20}[\dOo](?<!\.00)/
- body __LOTSA_MONEY_02 /(?<!\d)[1-9][\d.,\sOo]{5,20}[\dOo][\)\]\(]?\s?(?:<CURRENCY>|Pounds|(?i:dollars?|bucks))\b/
- body __LOTSA_MONEY_03 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)[1-9][\d.,\sOo]{0,5}[\)\]]?\s?(?i:M(?i:il)?\b|mil+(?i:io|<O>)n|hund?[re]+a?[dt]|thousand|tausend|milh[\xf5]es)/
- body __LOTSA_MONEY_04 /(?:(?<!\d)[1-9][\d\.,]{0,4}(?:M|\smilli?one?s|\s?mln)|million(?!s)|mill<O>n|hund?rea?d(?!s)[^\.]{1,25}thousand(?!s)|cents?[^\.]{1,25}mille|hundert[^\.]{1,30}tausend|ientos?[^\.]{1,20}mil|cent[a-z\s]{1,20}mil\s[a-z]{1,20}centos)[^\.\$]{0,50}?(?:(?:U\.?\s?S\.?\s?(?:A\.?\s?)?|united\s?states\s|E\.\s?U\.\s|canad(?:ian|a)\s|(?:ia\s)?de\s)?d(?:[o\xf3]|[\xc3][\xb3])l+are?s?|\bbucks|USD|GBP|<GB_UK>\spounds?|(?:<GB_UK>\s)?pounds?\ssterling|pounds(?!\sof)|(?:d'\s?)?euros?|francs?)\b/i
- body __LOTSA_MONEY_05 /(?:(?:sum|value|amount)\sof\s)[1-9][\d.,\sO]{7,20}[\dO\.][\)\]\(\s]{0,3}(?:pounds?|dollars?|euros?|bucks)\b/i
+ body __LOTSA_MONEY_00 /<CURRENCY>[\s\.]?<NUM_NOT_DATE>[\dOo][,\.][\dOo]{3}(?:(?!\d)|\b)/
+ body __LOTSA_MONEY_01 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)[\s\.]?<NUM_NOT_DATE>[\d.,\sOo]{5,20}[\dOo](?<!\.00)\b/
+ body __LOTSA_MONEY_02 /(?<!\d)<NUM_NOT_DATE>[\d.,\sOo]{5,20}[\dOo][\)\]\(]?\s?(?:<CURRENCY>|Pounds|(?i:dollars?|bucks))\b/
+ body __LOTSA_MONEY_03 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)<NUM_NOT_DATE>[\d.,\sOo]{0,5}[\)\]]?\s?(?i:M(?i:il)?\b|mil+(?i:io|<O>)n|hund?[re]+a?[dt]|thousand|tausend|milh[\xf5]es)/
+ body __LOTSA_MONEY_04 /(?:(?<!\d)<NUM_NOT_DATE>[\d\.,]{0,4}(?:M|\smilli?one?s|\s?mln)|million(?!s)|mill<O>n|hund?rea?d(?!s)[^\.]{1,25}thousand(?!s)|cents?[^\.]{1,25}mille|hundert[^\.]{1,30}tausend|ientos?[^\.]{1,20}mil|cent[a-z\s]{1,20}mil\s[a-z]{1,20}centos)[^\.\$]{0,50}?(?:(?:U\.?\s?S\.?\s?(?:A\.?\s?)?|united\s?states\s|E\.\s?U\.\s|canad(?:ian|a)\s|(?:ia\s)?de\s)?d(?:[o\xf3]|[\xc3][\xb3])l+are?s?|\bbucks|USD|GBP|<GB_UK>\spounds?|(?:<GB_UK>\s)?pounds?\ssterling|pounds(?!\sof)|(?:d'\s?)?euros?|francs?)\b/i
+ body __LOTSA_MONEY_05 /(?:(?:sum|value|amount)\sof\s)<NUM_NOT_DATE>[\d.,\sO]{7,20}[\dO\.][\)\]\(\s]{0,3}(?:pounds?|dollars?|euros?|bucks)\b/i
replace_rules __LOTSA_MONEY_00 __LOTSA_MONEY_01 __LOTSA_MONEY_02 __LOTSA_MONEY_03 __LOTSA_MONEY_04
# not a meta so it will show up in the report
Date: Fri Jun 5 23:08:19 2020
New Revision: 1878527
URL: http://svn.apache.org/viewvc?rev=1878527&view=rev
Log:
LOTSA_MONEY family: fix FP on dates, don't trigger on number followed by word beginning with "o"
Modified:
spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf
Modified: spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf
URL: http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf?rev=1878527&r1=1878526&r2=1878527&view=diff
==============================================================================
--- spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf (original)
+++ spamassassin/trunk/rulesrc/sandbox/jhardin/20_lotsa_money.cf Fri Jun 5 23:08:19 2020
@@ -7,13 +7,14 @@
ifplugin Mail::SpamAssassin::Plugin::ReplaceTags
replace_tag CURRENCY [\(\[]?(?:\bU[Ss][D\$]{0,2}|\$(?:US)?|usd|CAD|GBP|=[Aa][34]|\xa3|&\#16[34];|(?i:pounds\ssterling)|\xa4|EUR(?:OS)?|(?:d')?[Ee]uro?s?|(?i:eur)\sde|CHF|FCFA|d[\xf3]lares\sde\slos\sE+\.\s?U+\.)[\]\)]?
replace_tag GB_UK \b(?:U\.?K\.?|(?:Great\s)?Brit(?:ain|ish)|G\.?B\.?)\b
+ replace_tag NUM_NOT_DATE [1-9](?!\d\d\d\.\d\d\.\d\d\s)(?!\d\.\d\d\.\d\d\d\d\s)
- body __LOTSA_MONEY_00 /<CURRENCY>[\s\.]?[1-9][\dOo][,\.][\dOo]{3}(?:(?!\d)|\b)/
- body __LOTSA_MONEY_01 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)[\s\.]?[1-9][\d.,\sOo]{5,20}[\dOo](?<!\.00)/
- body __LOTSA_MONEY_02 /(?<!\d)[1-9][\d.,\sOo]{5,20}[\dOo][\)\]\(]?\s?(?:<CURRENCY>|Pounds|(?i:dollars?|bucks))\b/
- body __LOTSA_MONEY_03 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)[1-9][\d.,\sOo]{0,5}[\)\]]?\s?(?i:M(?i:il)?\b|mil+(?i:io|<O>)n|hund?[re]+a?[dt]|thousand|tausend|milh[\xf5]es)/
- body __LOTSA_MONEY_04 /(?:(?<!\d)[1-9][\d\.,]{0,4}(?:M|\smilli?one?s|\s?mln)|million(?!s)|mill<O>n|hund?rea?d(?!s)[^\.]{1,25}thousand(?!s)|cents?[^\.]{1,25}mille|hundert[^\.]{1,30}tausend|ientos?[^\.]{1,20}mil|cent[a-z\s]{1,20}mil\s[a-z]{1,20}centos)[^\.\$]{0,50}?(?:(?:U\.?\s?S\.?\s?(?:A\.?\s?)?|united\s?states\s|E\.\s?U\.\s|canad(?:ian|a)\s|(?:ia\s)?de\s)?d(?:[o\xf3]|[\xc3][\xb3])l+are?s?|\bbucks|USD|GBP|<GB_UK>\spounds?|(?:<GB_UK>\s)?pounds?\ssterling|pounds(?!\sof)|(?:d'\s?)?euros?|francs?)\b/i
- body __LOTSA_MONEY_05 /(?:(?:sum|value|amount)\sof\s)[1-9][\d.,\sO]{7,20}[\dO\.][\)\]\(\s]{0,3}(?:pounds?|dollars?|euros?|bucks)\b/i
+ body __LOTSA_MONEY_00 /<CURRENCY>[\s\.]?<NUM_NOT_DATE>[\dOo][,\.][\dOo]{3}(?:(?!\d)|\b)/
+ body __LOTSA_MONEY_01 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)[\s\.]?<NUM_NOT_DATE>[\d.,\sOo]{5,20}[\dOo](?<!\.00)\b/
+ body __LOTSA_MONEY_02 /(?<!\d)<NUM_NOT_DATE>[\d.,\sOo]{5,20}[\dOo][\)\]\(]?\s?(?:<CURRENCY>|Pounds|(?i:dollars?|bucks))\b/
+ body __LOTSA_MONEY_03 /(?:(?i:sum\sof\s)[\(\[]?|<CURRENCY>\s?)<NUM_NOT_DATE>[\d.,\sOo]{0,5}[\)\]]?\s?(?i:M(?i:il)?\b|mil+(?i:io|<O>)n|hund?[re]+a?[dt]|thousand|tausend|milh[\xf5]es)/
+ body __LOTSA_MONEY_04 /(?:(?<!\d)<NUM_NOT_DATE>[\d\.,]{0,4}(?:M|\smilli?one?s|\s?mln)|million(?!s)|mill<O>n|hund?rea?d(?!s)[^\.]{1,25}thousand(?!s)|cents?[^\.]{1,25}mille|hundert[^\.]{1,30}tausend|ientos?[^\.]{1,20}mil|cent[a-z\s]{1,20}mil\s[a-z]{1,20}centos)[^\.\$]{0,50}?(?:(?:U\.?\s?S\.?\s?(?:A\.?\s?)?|united\s?states\s|E\.\s?U\.\s|canad(?:ian|a)\s|(?:ia\s)?de\s)?d(?:[o\xf3]|[\xc3][\xb3])l+are?s?|\bbucks|USD|GBP|<GB_UK>\spounds?|(?:<GB_UK>\s)?pounds?\ssterling|pounds(?!\sof)|(?:d'\s?)?euros?|francs?)\b/i
+ body __LOTSA_MONEY_05 /(?:(?:sum|value|amount)\sof\s)<NUM_NOT_DATE>[\d.,\sO]{7,20}[\dO\.][\)\]\(\s]{0,3}(?:pounds?|dollars?|euros?|bucks)\b/i
replace_rules __LOTSA_MONEY_00 __LOTSA_MONEY_01 __LOTSA_MONEY_02 __LOTSA_MONEY_03 __LOTSA_MONEY_04
# not a meta so it will show up in the report