lucene-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sar...@apache.org
Subject [6/6] lucene-solr:branch_7x: LUCENE-8278: Some end-of-input no-scheme domain-only URL tokens are typed as <ALPHANUM> rather than <URL>
Date Mon, 11 Jun 2018 21:16:40 GMT
LUCENE-8278: Some end-of-input no-scheme domain-only URL tokens are typed as <ALPHANUM>
rather than <URL>


Project: http://git-wip-us.apache.org/repos/asf/lucene-solr/repo
Commit: http://git-wip-us.apache.org/repos/asf/lucene-solr/commit/6140d8be
Tree: http://git-wip-us.apache.org/repos/asf/lucene-solr/tree/6140d8be
Diff: http://git-wip-us.apache.org/repos/asf/lucene-solr/diff/6140d8be

Branch: refs/heads/branch_7x
Commit: 6140d8be05d1b7aad565f53dd2ee66b984b9a379
Parents: 5ef79b8
Author: Steve Rowe <sarowe@apache.org>
Authored: Mon Jun 11 17:09:30 2018 -0400
Committer: Steve Rowe <sarowe@apache.org>
Committed: Mon Jun 11 17:10:05 2018 -0400

----------------------------------------------------------------------
 lucene/CHANGES.txt                              |     6 +-
 lucene/analysis/common/build.xml                |     1 +
 .../analysis/standard/ASCIITLD.jflex-macro      |   263 +-
 .../standard/UAX29URLEmailTokenizerImpl.java    | 47679 +++++++++--------
 .../standard/UAX29URLEmailTokenizerImpl.jflex   |    26 +-
 .../standard/TestUAX29URLEmailTokenizer.java    |   158 +
 .../standard/GenerateJflexTLDMacros.java        |   148 +-
 7 files changed, 24283 insertions(+), 23998 deletions(-)
----------------------------------------------------------------------


http://git-wip-us.apache.org/repos/asf/lucene-solr/blob/6140d8be/lucene/CHANGES.txt
----------------------------------------------------------------------
diff --git a/lucene/CHANGES.txt b/lucene/CHANGES.txt
index 4412059..c160dac 100644
--- a/lucene/CHANGES.txt
+++ b/lucene/CHANGES.txt
@@ -4,7 +4,11 @@ For more information on past and future Lucene versions, please see:
 http://s.apache.org/luceneversions
 
 ======================= Lucene 7.5.0 =======================
-(No Changes)
+
+Bug Fixes
+
+* LUCENE-8278: Some end-of-input no-scheme domain-only URL tokens are typed as
+  <ALPHANUM> rather than <URL>.  (Junte Zhang, Steve Rowe)
 
 ======================= Lucene 7.4.0 =======================
 

http://git-wip-us.apache.org/repos/asf/lucene-solr/blob/6140d8be/lucene/analysis/common/build.xml
----------------------------------------------------------------------
diff --git a/lucene/analysis/common/build.xml b/lucene/analysis/common/build.xml
index 56ca4d0..6e4f118 100644
--- a/lucene/analysis/common/build.xml
+++ b/lucene/analysis/common/build.xml
@@ -111,6 +111,7 @@
       </classpath>
       <arg value="${tld.zones}"/>
       <arg value="${tld.output}"/>
+      <redirector alwayslog="true"/> <!-- stupid trick to get java class's stdout
to ant's log -->
     </java>
   </target>
 

http://git-wip-us.apache.org/repos/asf/lucene-solr/blob/6140d8be/lucene/analysis/common/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
----------------------------------------------------------------------
diff --git a/lucene/analysis/common/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
b/lucene/analysis/common/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
index 08df6c4..af88bf8 100644
--- a/lucene/analysis/common/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
+++ b/lucene/analysis/common/src/java/org/apache/lucene/analysis/standard/ASCIITLD.jflex-macro
@@ -15,10 +15,11 @@
  * limitations under the License.
  */
 // Generated from IANA Root Zone Database <http://www.internic.net/zones/root.zone>
-// file version from Friday, October 6, 2017 5:41:00 AM UTC
-// generated on Friday, October 6, 2017 4:00:48 PM UTC
+// file version from Sunday, May 27, 2018 5:16:00 PM UTC
+// generated on Sunday, May 27, 2018 7:38:07 PM UTC
 // by org.apache.lucene.analysis.standard.GenerateJflexTLDMacros
 
+// LUCENE-8278: None of the TLDs in {ASCIITLD} is a 1-character-shorter prefix of another
TLD
 ASCIITLD = "." (
 	  [aA][aA][aA]
 	| [aA][aA][rR][pP]
@@ -30,31 +31,24 @@ ASCIITLD = "." (
 	| [aA][bB][lL][eE]
 	| [aA][bB][oO][gG][aA][dD][oO]
 	| [aA][bB][uU][dD][hH][aA][bB][iI]
-	| [aA][cC]
 	| [aA][cC][aA][dD][eE][mM][yY]
 	| [aA][cC][cC][eE][nN][tT][uU][rR][eE]
-	| [aA][cC][cC][oO][uU][nN][tT][aA][nN][tT]
 	| [aA][cC][cC][oO][uU][nN][tT][aA][nN][tT][sS]
 	| [aA][cC][oO]
 	| [aA][cC][tT][iI][vV][eE]
 	| [aA][cC][tT][oO][rR]
-	| [aA][dD]
 	| [aA][dD][aA][cC]
 	| [aA][dD][sS]
 	| [aA][dD][uU][lL][tT]
-	| [aA][eE]
 	| [aA][eE][gG]
 	| [aA][eE][rR][oO]
 	| [aA][eE][tT][nN][aA]
-	| [aA][fF]
 	| [aA][fF][aA][mM][iI][lL][yY][cC][oO][mM][pP][aA][nN][yY]
 	| [aA][fF][lL]
 	| [aA][fF][rR][iI][cC][aA]
 	| [aA][gG]
 	| [aA][gG][aA][kK][hH][aA][nN]
 	| [aA][gG][eE][nN][cC][yY]
-	| [aA][iI]
-	| [aA][iI][gG]
 	| [aA][iI][gG][oO]
 	| [aA][iI][rR][bB][uU][sS]
 	| [aA][iI][rR][fF][oO][rR][cC][eE]
@@ -80,20 +74,17 @@ ASCIITLD = "." (
 	| [aA][nN][dD][rR][oO][iI][dD]
 	| [aA][nN][qQ][uU][aA][nN]
 	| [aA][nN][zZ]
-	| [aA][oO]
 	| [aA][oO][lL]
 	| [aA][pP][aA][rR][tT][mM][eE][nN][tT][sS]
 	| [aA][pP][pP]
 	| [aA][pP][pP][lL][eE]
 	| [aA][qQ]
 	| [aA][qQ][uU][aA][rR][eE][lL][lL][eE]
-	| [aA][rR]
 	| [aA][rR][aA][bB]
 	| [aA][rR][aA][mM][cC][oO]
 	| [aA][rR][cC][hH][iI]
 	| [aA][rR][mM][yY]
 	| [aA][rR][pP][aA]
-	| [aA][rR][tT]
 	| [aA][rR][tT][eE]
 	| [aA][sS]
 	| [aA][sS][dD][aA]
@@ -104,21 +95,16 @@ ASCIITLD = "." (
 	| [aA][tT][tT][oO][rR][nN][eE][yY]
 	| [aA][uU]
 	| [aA][uU][cC][tT][iI][oO][nN]
-	| [aA][uU][dD][iI]
 	| [aA][uU][dD][iI][bB][lL][eE]
 	| [aA][uU][dD][iI][oO]
 	| [aA][uU][sS][pP][oO][sS][tT]
 	| [aA][uU][tT][hH][oO][rR]
-	| [aA][uU][tT][oO]
 	| [aA][uU][tT][oO][sS]
 	| [aA][vV][iI][aA][nN][cC][aA]
-	| [aA][wW]
 	| [aA][wW][sS]
-	| [aA][xX]
 	| [aA][xX][aA]
 	| [aA][zZ]
 	| [aA][zZ][uU][rR][eE]
-	| [bB][aA]
 	| [bB][aA][bB][yY]
 	| [bB][aA][iI][dD][uU]
 	| [bB][aA][nN][aA][mM][eE][xX]
@@ -135,14 +121,12 @@ ASCIITLD = "." (
 	| [bB][aA][sS][kK][eE][tT][bB][aA][lL][lL]
 	| [bB][aA][uU][hH][aA][uU][sS]
 	| [bB][aA][yY][eE][rR][nN]
-	| [bB][bB]
 	| [bB][bB][cC]
 	| [bB][bB][tT]
 	| [bB][bB][vV][aA]
 	| [bB][cC][gG]
 	| [bB][cC][nN]
 	| [bB][dD]
-	| [bB][eE]
 	| [bB][eE][aA][tT][sS]
 	| [bB][eE][aA][uU][tT][yY]
 	| [bB][eE][eE][rR]
@@ -155,11 +139,9 @@ ASCIITLD = "." (
 	| [bB][gG]
 	| [bB][hH]
 	| [bB][hH][aA][rR][tT][iI]
-	| [bB][iI]
 	| [bB][iI][bB][lL][eE]
 	| [bB][iI][dD]
 	| [bB][iI][kK][eE]
-	| [bB][iI][nN][gG]
 	| [bB][iI][nN][gG][oO]
 	| [bB][iI][oO]
 	| [bB][iI][zZ]
@@ -171,22 +153,17 @@ ASCIITLD = "." (
 	| [bB][lL][oO][gG]
 	| [bB][lL][oO][oO][mM][bB][eE][rR][gG]
 	| [bB][lL][uU][eE]
-	| [bB][mM]
 	| [bB][mM][sS]
 	| [bB][mM][wW]
-	| [bB][nN]
 	| [bB][nN][lL]
 	| [bB][nN][pP][pP][aA][rR][iI][bB][aA][sS]
-	| [bB][oO]
 	| [bB][oO][aA][tT][sS]
 	| [bB][oO][eE][hH][rR][iI][nN][gG][eE][rR]
 	| [bB][oO][fF][aA]
 	| [bB][oO][mM]
 	| [bB][oO][nN][dD]
-	| [bB][oO][oO]
 	| [bB][oO][oO][kK]
 	| [bB][oO][oO][kK][iI][nN][gG]
-	| [bB][oO][oO][tT][sS]
 	| [bB][oO][sS][cC][hH]
 	| [bB][oO][sS][tT][iI][kK]
 	| [bB][oO][sS][tT][oO][nN]
@@ -212,15 +189,11 @@ ASCIITLD = "." (
 	| [bB][vV]
 	| [bB][wW]
 	| [bB][yY]
-	| [bB][zZ]
 	| [bB][zZ][hH]
-	| [cC][aA]
 	| [cC][aA][bB]
 	| [cC][aA][fF][eE]
-	| [cC][aA][lL]
 	| [cC][aA][lL][lL]
 	| [cC][aA][lL][vV][iI][nN][kK][lL][eE][iI][nN]
-	| [cC][aA][mM]
 	| [cC][aA][mM][eE][rR][aA]
 	| [cC][aA][mM][pP]
 	| [cC][aA][nN][cC][eE][rR][rR][eE][sS][eE][aA][rR][cC][hH]
@@ -228,11 +201,9 @@ ASCIITLD = "." (
 	| [cC][aA][pP][eE][tT][oO][wW][nN]
 	| [cC][aA][pP][iI][tT][aA][lL]
 	| [cC][aA][pP][iI][tT][aA][lL][oO][nN][eE]
-	| [cC][aA][rR]
 	| [cC][aA][rR][aA][vV][aA][nN]
 	| [cC][aA][rR][dD][sS]
 	| [cC][aA][rR][eE]
-	| [cC][aA][rR][eE][eE][rR]
 	| [cC][aA][rR][eE][eE][rR][sS]
 	| [cC][aA][rR][sS]
 	| [cC][aA][rR][tT][iI][eE][rR]
@@ -254,7 +225,6 @@ ASCIITLD = "." (
 	| [cC][eE][nN][tT][eE][rR]
 	| [cC][eE][oO]
 	| [cC][eE][rR][nN]
-	| [cC][fF]
 	| [cC][fF][aA]
 	| [cC][fF][dD]
 	| [cC][gG]
@@ -265,7 +235,6 @@ ASCIITLD = "." (
 	| [cC][hH][aA][tT]
 	| [cC][hH][eE][aA][pP]
 	| [cC][hH][iI][nN][tT][aA][iI]
-	| [cC][hH][lL][oO][eE]
 	| [cC][hH][rR][iI][sS][tT][mM][aA][sS]
 	| [cC][hH][rR][oO][mM][eE]
 	| [cC][hH][rR][yY][sS][lL][eE][rR]
@@ -275,7 +244,6 @@ ASCIITLD = "." (
 	| [cC][iI][rR][cC][lL][eE]
 	| [cC][iI][sS][cC][oO]
 	| [cC][iI][tT][aA][dD][eE][lL]
-	| [cC][iI][tT][iI]
 	| [cC][iI][tT][iI][cC]
 	| [cC][iI][tT][yY]
 	| [cC][iI][tT][yY][eE][aA][tT][sS]
@@ -292,7 +260,6 @@ ASCIITLD = "." (
 	| [cC][lL][uU][bB][mM][eE][dD]
 	| [cC][mM]
 	| [cC][nN]
-	| [cC][oO]
 	| [cC][oO][aA][cC][hH]
 	| [cC][oO][dD][eE][sS]
 	| [cC][oO][fF][fF][eE][eE]
@@ -317,17 +284,14 @@ ASCIITLD = "." (
 	| [cC][oO][oO][pP]
 	| [cC][oO][rR][sS][iI][cC][aA]
 	| [cC][oO][uU][nN][tT][rR][yY]
-	| [cC][oO][uU][pP][oO][nN]
 	| [cC][oO][uU][pP][oO][nN][sS]
 	| [cC][oO][uU][rR][sS][eE][sS]
-	| [cC][rR]
 	| [cC][rR][eE][dD][iI][tT]
 	| [cC][rR][eE][dD][iI][tT][cC][aA][rR][dD]
 	| [cC][rR][eE][dD][iI][tT][uU][nN][iI][oO][nN]
 	| [cC][rR][iI][cC][kK][eE][tT]
 	| [cC][rR][oO][wW][nN]
 	| [cC][rR][sS]
-	| [cC][rR][uU][iI][sS][eE]
 	| [cC][rR][uU][iI][sS][eE][sS]
 	| [cC][sS][cC]
 	| [cC][uU]
@@ -349,8 +313,6 @@ ASCIITLD = "." (
 	| [dD][aA][yY]
 	| [dD][cC][lL][kK]
 	| [dD][dD][sS]
-	| [dD][eE]
-	| [dD][eE][aA][lL]
 	| [dD][eE][aA][lL][eE][rR]
 	| [dD][eE][aA][lL][sS]
 	| [dD][eE][gG][rR][eE][eE]
@@ -378,7 +340,6 @@ ASCIITLD = "." (
 	| [dD][kK]
 	| [dD][mM]
 	| [dD][nN][pP]
-	| [dD][oO]
 	| [dD][oO][cC][sS]
 	| [dD][oO][cC][tT][oO][rR]
 	| [dD][oO][dD][gG][eE]
@@ -400,7 +361,6 @@ ASCIITLD = "." (
 	| [dD][zZ]
 	| [eE][aA][rR][tT][hH]
 	| [eE][aA][tT]
-	| [eE][cC]
 	| [eE][cC][oO]
 	| [eE][dD][eE][kK][aA]
 	| [eE][dD][uU]
@@ -419,13 +379,11 @@ ASCIITLD = "." (
 	| [eE][rR]
 	| [eE][rR][iI][cC][sS][sS][oO][nN]
 	| [eE][rR][nN][iI]
-	| [eE][sS]
 	| [eE][sS][qQ]
 	| [eE][sS][tT][aA][tT][eE]
 	| [eE][sS][uU][rR][aA][nN][cC][eE]
 	| [eE][tT]
 	| [eE][tT][iI][sS][aA][lL][aA][tT]
-	| [eE][uU]
 	| [eE][uU][rR][oO][vV][iI][sS][iI][oO][nN]
 	| [eE][uU][sS]
 	| [eE][vV][eE][nN][tT][sS]
@@ -440,7 +398,6 @@ ASCIITLD = "." (
 	| [fF][aA][iI][rR][wW][iI][nN][dD][sS]
 	| [fF][aA][iI][tT][hH]
 	| [fF][aA][mM][iI][lL][yY]
-	| [fF][aA][nN]
 	| [fF][aA][nN][sS]
 	| [fF][aA][rR][mM]
 	| [fF][aA][rR][mM][eE][rR][sS]
@@ -450,7 +407,6 @@ ASCIITLD = "." (
 	| [fF][eE][eE][dD][bB][aA][cC][kK]
 	| [fF][eE][rR][rR][aA][rR][iI]
 	| [fF][eE][rR][rR][eE][rR][oO]
-	| [fF][iI]
 	| [fF][iI][aA][tT]
 	| [fF][iI][dD][eE][lL][iI][tT][yY]
 	| [fF][iI][dD][oO]
@@ -474,8 +430,6 @@ ASCIITLD = "." (
 	| [fF][lL][oO][wW][eE][rR][sS]
 	| [fF][lL][yY]
 	| [fF][mM]
-	| [fF][oO]
-	| [fF][oO][oO]
 	| [fF][oO][oO][dD]
 	| [fF][oO][oO][dD][nN][eE][tT][wW][oO][rR][kK]
 	| [fF][oO][oO][tT][bB][aA][lL][lL]
@@ -485,7 +439,6 @@ ASCIITLD = "." (
 	| [fF][oO][rR][uU][mM]
 	| [fF][oO][uU][nN][dD][aA][tT][iI][oO][nN]
 	| [fF][oO][xX]
-	| [fF][rR]
 	| [fF][rR][eE][eE]
 	| [fF][rR][eE][sS][eE][nN][iI][uU][sS]
 	| [fF][rR][lL]
@@ -495,25 +448,20 @@ ASCIITLD = "." (
 	| [fF][tT][rR]
 	| [fF][uU][jJ][iI][tT][sS][uU]
 	| [fF][uU][jJ][iI][xX][eE][rR][oO][xX]
-	| [fF][uU][nN]
 	| [fF][uU][nN][dD]
 	| [fF][uU][rR][nN][iI][tT][uU][rR][eE]
 	| [fF][uU][tT][bB][oO][lL]
 	| [fF][yY][iI]
-	| [gG][aA]
 	| [gG][aA][lL]
 	| [gG][aA][lL][lL][eE][rR][yY]
 	| [gG][aA][lL][lL][oO]
 	| [gG][aA][lL][lL][uU][pP]
-	| [gG][aA][mM][eE]
 	| [gG][aA][mM][eE][sS]
 	| [gG][aA][pP]
 	| [gG][aA][rR][dD][eE][nN]
 	| [gG][bB]
 	| [gG][bB][iI][zZ]
-	| [gG][dD]
 	| [gG][dD][nN]
-	| [gG][eE]
 	| [gG][eE][aA]
 	| [gG][eE][nN][tT]
 	| [gG][eE][nN][tT][iI][nN][gG]
@@ -523,17 +471,14 @@ ASCIITLD = "." (
 	| [gG][gG][eE][eE]
 	| [gG][hH]
 	| [gG][iI]
-	| [gG][iI][fF][tT]
 	| [gG][iI][fF][tT][sS]
 	| [gG][iI][vV][eE][sS]
 	| [gG][iI][vV][iI][nN][gG]
-	| [gG][lL]
 	| [gG][lL][aA][dD][eE]
 	| [gG][lL][aA][sS][sS]
 	| [gG][lL][eE]
 	| [gG][lL][oO][bB][aA][lL]
 	| [gG][lL][oO][bB][oO]
-	| [gG][mM]
 	| [gG][mM][aA][iI][lL]
 	| [gG][mM][bB][hH]
 	| [gG][mM][oO]
@@ -543,7 +488,6 @@ ASCIITLD = "." (
 	| [gG][oO][lL][dD]
 	| [gG][oO][lL][dD][pP][oO][iI][nN][tT]
 	| [gG][oO][lL][fF]
-	| [gG][oO][oO]
 	| [gG][oO][oO][dD][hH][aA][nN][dD][sS]
 	| [gG][oO][oO][dD][yY][eE][aA][rR]
 	| [gG][oO][oO][gG]
@@ -590,7 +534,6 @@ ASCIITLD = "." (
 	| [hH][iI][sS][aA][mM][iI][tT][sS][uU]
 	| [hH][iI][tT][aA][cC][hH][iI]
 	| [hH][iI][vV]
-	| [hH][kK]
 	| [hH][kK][tT]
 	| [hH][mM]
 	| [hH][nN]
@@ -616,7 +559,6 @@ ASCIITLD = "." (
 	| [hH][rR]
 	| [hH][sS][bB][cC]
 	| [hH][tT]
-	| [hH][tT][cC]
 	| [hH][uU]
 	| [hH][uU][gG][hH][eE][sS]
 	| [hH][yY][aA][tT][tT]
@@ -636,7 +578,6 @@ ASCIITLD = "." (
 	| [iI][mM][dD][bB]
 	| [iI][mM][mM][oO]
 	| [iI][mM][mM][oO][bB][iI][lL][iI][eE][nN]
-	| [iI][nN]
 	| [iI][nN][dD][uU][sS][tT][rR][iI][eE][sS]
 	| [iI][nN][fF][iI][nN][iI][tT][iI]
 	| [iI][nN][fF][oO]
@@ -655,12 +596,10 @@ ASCIITLD = "." (
 	| [iI][qQ]
 	| [iI][rR]
 	| [iI][rR][iI][sS][hH]
-	| [iI][sS]
 	| [iI][sS][eE][lL][eE][cC][tT]
 	| [iI][sS][mM][aA][iI][lL][iI]
 	| [iI][sS][tT]
 	| [iI][sS][tT][aA][nN][bB][uU][lL]
-	| [iI][tT]
 	| [iI][tT][aA][uU]
 	| [iI][tT][vV]
 	| [iI][vV][eE][cC][oO]
@@ -676,10 +615,8 @@ ASCIITLD = "." (
 	| [jJ][iI][oO]
 	| [jJ][lL][cC]
 	| [jJ][lL][lL]
-	| [jJ][mM]
 	| [jJ][mM][pP]
 	| [jJ][nN][jJ]
-	| [jJ][oO]
 	| [jJ][oO][bB][sS]
 	| [jJ][oO][bB][uU][rR][gG]
 	| [jJ][oO][tT]
@@ -698,7 +635,6 @@ ASCIITLD = "." (
 	| [kK][fF][hH]
 	| [kK][gG]
 	| [kK][hH]
-	| [kK][iI]
 	| [kK][iI][aA]
 	| [kK][iI][mM]
 	| [kK][iI][nN][dD][eE][rR]
@@ -710,10 +646,8 @@ ASCIITLD = "." (
 	| [kK][oO][eE][lL][nN]
 	| [kK][oO][mM][aA][tT][sS][uU]
 	| [kK][oO][sS][hH][eE][rR]
-	| [kK][pP]
 	| [kK][pP][mM][gG]
 	| [kK][pP][nN]
-	| [kK][rR]
 	| [kK][rR][dD]
 	| [kK][rR][eE][dD]
 	| [kK][uU][oO][kK][gG][rR][oO][uU][pP]
@@ -721,7 +655,6 @@ ASCIITLD = "." (
 	| [kK][yY]
 	| [kK][yY][oO][tT][oO]
 	| [kK][zZ]
-	| [lL][aA]
 	| [lL][aA][cC][aA][iI][xX][aA]
 	| [lL][aA][dD][bB][rR][oO][kK][eE][sS]
 	| [lL][aA][mM][bB][oO][rR][gG][hH][iI][nN][iI]
@@ -767,7 +700,7 @@ ASCIITLD = "." (
 	| [lL][iI][vV][iI][nN][gG]
 	| [lL][iI][xX][iI][lL]
 	| [lL][kK]
-	| [lL][oO][aA][nN]
+	| [lL][lL][cC]
 	| [lL][oO][aA][nN][sS]
 	| [lL][oO][cC][kK][eE][rR]
 	| [lL][oO][cC][uU][sS]
@@ -781,8 +714,6 @@ ASCIITLD = "." (
 	| [lL][pP][lL][fF][iI][nN][aA][nN][cC][iI][aA][lL]
 	| [lL][rR]
 	| [lL][sS]
-	| [lL][tT]
-	| [lL][tT][dD]
 	| [lL][tT][dD][aA]
 	| [lL][uU]
 	| [lL][uU][nN][dD][bB][eE][cC][kK]
@@ -791,7 +722,6 @@ ASCIITLD = "." (
 	| [lL][uU][xX][uU][rR][yY]
 	| [lL][vV]
 	| [lL][yY]
-	| [mM][aA]
 	| [mM][aA][cC][yY][sS]
 	| [mM][aA][dD][rR][iI][dD]
 	| [mM][aA][iI][fF]
@@ -801,7 +731,6 @@ ASCIITLD = "." (
 	| [mM][aA][nN][aA][gG][eE][mM][eE][nN][tT]
 	| [mM][aA][nN][gG][oO]
 	| [mM][aA][pP]
-	| [mM][aA][rR][kK][eE][tT]
 	| [mM][aA][rR][kK][eE][tT][iI][nN][gG]
 	| [mM][aA][rR][kK][eE][tT][sS]
 	| [mM][aA][rR][rR][iI][oO][tT][tT]
@@ -812,16 +741,13 @@ ASCIITLD = "." (
 	| [mM][cC]
 	| [mM][cC][kK][iI][nN][sS][eE][yY]
 	| [mM][dD]
-	| [mM][eE]
 	| [mM][eE][dD]
 	| [mM][eE][dD][iI][aA]
 	| [mM][eE][eE][tT]
 	| [mM][eE][lL][bB][oO][uU][rR][nN][eE]
 	| [mM][eE][mM][eE]
 	| [mM][eE][mM][oO][rR][iI][aA][lL]
-	| [mM][eE][nN]
 	| [mM][eE][nN][uU]
-	| [mM][eE][oO]
 	| [mM][eE][rR][cC][kK][mM][sS][dD]
 	| [mM][eE][tT][lL][iI][fF][eE]
 	| [mM][gG]
@@ -834,13 +760,10 @@ ASCIITLD = "." (
 	| [mM][iI][tT]
 	| [mM][iI][tT][sS][uU][bB][iI][sS][hH][iI]
 	| [mM][kK]
-	| [mM][lL]
 	| [mM][lL][bB]
 	| [mM][lL][sS]
-	| [mM][mM]
 	| [mM][mM][aA]
 	| [mM][nN]
-	| [mM][oO]
 	| [mM][oO][bB][iI]
 	| [mM][oO][bB][iI][lL][eE]
 	| [mM][oO][bB][iI][lL][yY]
@@ -863,9 +786,7 @@ ASCIITLD = "." (
 	| [mM][pP]
 	| [mM][qQ]
 	| [mM][rR]
-	| [mM][sS]
 	| [mM][sS][dD]
-	| [mM][tT]
 	| [mM][tT][nN]
 	| [mM][tT][rR]
 	| [mM][uU]
@@ -876,7 +797,6 @@ ASCIITLD = "." (
 	| [mM][xX]
 	| [mM][yY]
 	| [mM][zZ]
-	| [nN][aA]
 	| [nN][aA][bB]
 	| [nN][aA][dD][eE][xX]
 	| [nN][aA][gG][oO][yY][aA]
@@ -886,22 +806,18 @@ ASCIITLD = "." (
 	| [nN][aA][vV][yY]
 	| [nN][bB][aA]
 	| [nN][cC]
-	| [nN][eE]
 	| [nN][eE][cC]
 	| [nN][eE][tT]
 	| [nN][eE][tT][bB][aA][nN][kK]
 	| [nN][eE][tT][fF][lL][iI][xX]
 	| [nN][eE][tT][wW][oO][rR][kK]
 	| [nN][eE][uU][sS][tT][aA][rR]
-	| [nN][eE][wW]
 	| [nN][eE][wW][hH][oO][lL][lL][aA][nN][dD]
 	| [nN][eE][wW][sS]
 	| [nN][eE][xX][tT]
 	| [nN][eE][xX][tT][dD][iI][rR][eE][cC][tT]
 	| [nN][eE][xX][uU][sS]
-	| [nN][fF]
 	| [nN][fF][lL]
-	| [nN][gG]
 	| [nN][gG][oO]
 	| [nN][hH][kK]
 	| [nN][iI]
@@ -912,7 +828,6 @@ ASCIITLD = "." (
 	| [nN][iI][sS][sS][aA][nN]
 	| [nN][iI][sS][sS][aA][yY]
 	| [nN][lL]
-	| [nN][oO]
 	| [nN][oO][kK][iI][aA]
 	| [nN][oO][rR][tT][hH][wW][eE][sS][tT][eE][rR][nN][mM][uU][tT][uU][aA][lL]
 	| [nN][oO][rR][tT][oO][nN]
@@ -920,7 +835,6 @@ ASCIITLD = "." (
 	| [nN][oO][wW][rR][uU][zZ]
 	| [nN][oO][wW][tT][vV]
 	| [nN][pP]
-	| [nN][rR]
 	| [nN][rR][aA]
 	| [nN][rR][wW]
 	| [nN][tT][tT]
@@ -954,7 +868,6 @@ ASCIITLD = "." (
 	| [oO][tT][sS][uU][kK][aA]
 	| [oO][tT][tT]
 	| [oO][vV][hH]
-	| [pP][aA]
 	| [pP][aA][gG][eE]
 	| [pP][aA][nN][aA][sS][oO][nN][iI][cC]
 	| [pP][aA][nN][eE][rR][aA][iI]
@@ -966,17 +879,14 @@ ASCIITLD = "." (
 	| [pP][aA][sS][sS][aA][gG][eE][nN][sS]
 	| [pP][aA][yY]
 	| [pP][cC][cC][wW]
-	| [pP][eE]
 	| [pP][eE][tT]
 	| [pP][fF]
 	| [pP][fF][iI][zZ][eE][rR]
 	| [pP][gG]
-	| [pP][hH]
 	| [pP][hH][aA][rR][mM][aA][cC][yY]
 	| [pP][hH][dD]
 	| [pP][hH][iI][lL][iI][pP][sS]
 	| [pP][hH][oO][nN][eE]
-	| [pP][hH][oO][tT][oO]
 	| [pP][hH][oO][tT][oO][gG][rR][aA][pP][hH][yY]
 	| [pP][hH][oO][tT][oO][sS]
 	| [pP][hH][yY][sS][iI][oO]
@@ -985,7 +895,6 @@ ASCIITLD = "." (
 	| [pP][iI][cC][tT][eE][tT]
 	| [pP][iI][cC][tT][uU][rR][eE][sS]
 	| [pP][iI][dD]
-	| [pP][iI][nN]
 	| [pP][iI][nN][gG]
 	| [pP][iI][nN][kK]
 	| [pP][iI][oO][nN][eE][eE][rR]
@@ -998,19 +907,16 @@ ASCIITLD = "." (
 	| [pP][lL][uU][mM][bB][iI][nN][gG]
 	| [pP][lL][uU][sS]
 	| [pP][mM]
-	| [pP][nN]
 	| [pP][nN][cC]
 	| [pP][oO][hH][lL]
 	| [pP][oO][kK][eE][rR]
 	| [pP][oO][lL][iI][tT][iI][eE]
 	| [pP][oO][rR][nN]
 	| [pP][oO][sS][tT]
-	| [pP][rR]
 	| [pP][rR][aA][mM][eE][rR][iI][cC][aA]
 	| [pP][rR][aA][xX][iI]
 	| [pP][rR][eE][sS][sS]
 	| [pP][rR][iI][mM][eE]
-	| [pP][rR][oO]
 	| [pP][rR][oO][dD]
 	| [pP][rR][oO][dD][uU][cC][tT][iI][oO][nN][sS]
 	| [pP][rR][oO][fF]
@@ -1024,7 +930,6 @@ ASCIITLD = "." (
 	| [pP][sS]
 	| [pP][tT]
 	| [pP][uU][bB]
-	| [pP][wW]
 	| [pP][wW][cC]
 	| [pP][yY]
 	| [qQ][aA]
@@ -1035,7 +940,6 @@ ASCIITLD = "." (
 	| [rR][aA][cC][iI][nN][gG]
 	| [rR][aA][dD][iI][oO]
 	| [rR][aA][iI][dD]
-	| [rR][eE]
 	| [rR][eE][aA][dD]
 	| [rR][eE][aA][lL][eE][sS][tT][aA][tT][eE]
 	| [rR][eE][aA][lL][tT][oO][rR]
@@ -1045,11 +949,9 @@ ASCIITLD = "." (
 	| [rR][eE][dD][sS][tT][oO][nN][eE]
 	| [rR][eE][dD][uU][mM][bB][rR][eE][lL][lL][aA]
 	| [rR][eE][hH][aA][bB]
-	| [rR][eE][iI][sS][eE]
 	| [rR][eE][iI][sS][eE][nN]
 	| [rR][eE][iI][tT]
 	| [rR][eE][lL][iI][aA][nN][cC][eE]
-	| [rR][eE][nN]
 	| [rR][eE][nN][tT]
 	| [rR][eE][nN][tT][aA][lL][sS]
 	| [rR][eE][pP][aA][iI][rR]
@@ -1057,7 +959,6 @@ ASCIITLD = "." (
 	| [rR][eE][pP][uU][bB][lL][iI][cC][aA][nN]
 	| [rR][eE][sS][tT]
 	| [rR][eE][sS][tT][aA][uU][rR][aA][nN][tT]
-	| [rR][eE][vV][iI][eE][wW]
 	| [rR][eE][vV][iI][eE][wW][sS]
 	| [rR][eE][xX][rR][oO][tT][hH]
 	| [rR][iI][cC][hH]
@@ -1076,14 +977,11 @@ ASCIITLD = "." (
 	| [rR][oO][oO][mM]
 	| [rR][sS]
 	| [rR][sS][vV][pP]
-	| [rR][uU]
 	| [rR][uU][gG][bB][yY]
 	| [rR][uU][hH][rR]
 	| [rR][uU][nN]
-	| [rR][wW]
 	| [rR][wW][eE]
 	| [rR][yY][uU][kK][yY][uU]
-	| [sS][aA]
 	| [sS][aA][aA][rR][lL][aA][nN][dD]
 	| [sS][aA][fF][eE]
 	| [sS][aA][fF][eE][tT][yY]
@@ -1096,15 +994,12 @@ ASCIITLD = "." (
 	| [sS][aA][nN][dD][vV][iI][kK][cC][oO][rR][oO][mM][aA][nN][tT]
 	| [sS][aA][nN][oO][fF][iI]
 	| [sS][aA][pP]
-	| [sS][aA][pP][oO]
 	| [sS][aA][rR][lL]
 	| [sS][aA][sS]
 	| [sS][aA][vV][eE]
 	| [sS][aA][xX][oO]
-	| [sS][bB]
 	| [sS][bB][iI]
 	| [sS][bB][sS]
-	| [sS][cC]
 	| [sS][cC][aA]
 	| [sS][cC][bB]
 	| [sS][cC][hH][aA][eE][fF][fF][lL][eE][rR]
@@ -1118,7 +1013,6 @@ ASCIITLD = "." (
 	| [sS][cC][oO][rR]
 	| [sS][cC][oO][tT]
 	| [sS][dD]
-	| [sS][eE]
 	| [sS][eE][aA][rR][cC][hH]
 	| [sS][eE][aA][tT]
 	| [sS][eE][cC][uU][rR][eE]
@@ -1130,7 +1024,6 @@ ASCIITLD = "." (
 	| [sS][eE][sS]
 	| [sS][eE][vV][eE][nN]
 	| [sS][eE][wW]
-	| [sS][eE][xX]
 	| [sS][eE][xX][yY]
 	| [sS][fF][rR]
 	| [sS][gG]
@@ -1154,8 +1047,6 @@ ASCIITLD = "." (
 	| [sS][iI][nN][gG][lL][eE][sS]
 	| [sS][iI][tT][eE]
 	| [sS][jJ]
-	| [sS][kK]
-	| [sS][kK][iI]
 	| [sS][kK][iI][nN]
 	| [sS][kK][yY]
 	| [sS][kK][yY][pP][eE]
@@ -1166,7 +1057,6 @@ ASCIITLD = "." (
 	| [sS][mM][iI][lL][eE]
 	| [sS][nN]
 	| [sS][nN][cC][fF]
-	| [sS][oO]
 	| [sS][oO][cC][cC][eE][rR]
 	| [sS][oO][cC][iI][aA][lL]
 	| [sS][oO][fF][tT][bB][aA][nN][kK]
@@ -1179,12 +1069,11 @@ ASCIITLD = "." (
 	| [sS][oO][yY]
 	| [sS][pP][aA][cC][eE]
 	| [sS][pP][iI][eE][gG][eE][lL]
+	| [sS][pP][oO][rR][tT]
 	| [sS][pP][oO][tT]
 	| [sS][pP][rR][eE][aA][dD][bB][eE][tT][tT][iI][nN][gG]
-	| [sS][rR]
 	| [sS][rR][lL]
 	| [sS][rR][tT]
-	| [sS][tT]
 	| [sS][tT][aA][dD][aA]
 	| [sS][tT][aA][pP][lL][eE][sS]
 	| [sS][tT][aA][rR]
@@ -1227,11 +1116,8 @@ ASCIITLD = "." (
 	| [tT][aA][tT][aA][mM][oO][tT][oO][rR][sS]
 	| [tT][aA][tT][aA][rR]
 	| [tT][aA][tT][tT][oO][oO]
-	| [tT][aA][xX]
 	| [tT][aA][xX][iI]
-	| [tT][cC]
 	| [tT][cC][iI]
-	| [tT][dD]
 	| [tT][dD][kK]
 	| [tT][eE][aA][mM]
 	| [tT][eE][cC][hH]
@@ -1244,7 +1130,6 @@ ASCIITLD = "." (
 	| [tT][eE][vV][aA]
 	| [tT][fF]
 	| [tT][gG]
-	| [tT][hH]
 	| [tT][hH][dD]
 	| [tT][hH][eE][aA][tT][eE][rR]
 	| [tT][hH][eE][aA][tT][rR][eE]
@@ -1255,7 +1140,6 @@ ASCIITLD = "." (
 	| [tT][iI][pP][sS]
 	| [tT][iI][rR][eE][sS]
 	| [tT][iI][rR][oO][lL]
-	| [tT][jJ]
 	| [tT][jJ][mM][aA][xX][xX]
 	| [tT][jJ][xX]
 	| [tT][kK]
@@ -1264,7 +1148,6 @@ ASCIITLD = "." (
 	| [tT][mM]
 	| [tT][mM][aA][lL][lL]
 	| [tT][nN]
-	| [tT][oO]
 	| [tT][oO][dD][aA][yY]
 	| [tT][oO][kK][yY][oO]
 	| [tT][oO][oO][lL][sS]
@@ -1276,7 +1159,6 @@ ASCIITLD = "." (
 	| [tT][oO][wW][nN]
 	| [tT][oO][yY][oO][tT][aA]
 	| [tT][oO][yY][sS]
-	| [tT][rR]
 	| [tT][rR][aA][dD][eE]
 	| [tT][rR][aA][dD][iI][nN][gG]
 	| [tT][rR][aA][iI][nN][iI][nN][gG]
@@ -1291,7 +1173,6 @@ ASCIITLD = "." (
 	| [tT][uU][iI]
 	| [tT][uU][nN][eE][sS]
 	| [tT][uU][sS][hH][uU]
-	| [tT][vV]
 	| [tT][vV][sS]
 	| [tT][wW]
 	| [tT][zZ]
@@ -1314,14 +1195,12 @@ ASCIITLD = "." (
 	| [vV][aA][nN][aA]
 	| [vV][aA][nN][gG][uU][aA][rR][dD]
 	| [vV][cC]
-	| [vV][eE]
 	| [vV][eE][gG][aA][sS]
 	| [vV][eE][nN][tT][uU][rR][eE][sS]
 	| [vV][eE][rR][iI][sS][iI][gG][nN]
 	| [vV][eE][rR][sS][iI][cC][hH][eE][rR][uU][nN][gG]
 	| [vV][eE][tT]
 	| [vV][gG]
-	| [vV][iI]
 	| [vV][iI][aA][jJ][eE][sS]
 	| [vV][iI][dD][eE][oO]
 	| [vV][iI][gG]
@@ -1369,14 +1248,12 @@ ASCIITLD = "." (
 	| [wW][iI][eE][nN]
 	| [wW][iI][kK][iI]
 	| [wW][iI][lL][lL][iI][aA][mM][hH][iI][lL][lL]
-	| [wW][iI][nN]
 	| [wW][iI][nN][dD][oO][wW][sS]
 	| [wW][iI][nN][eE]
 	| [wW][iI][nN][nN][eE][rR][sS]
 	| [wW][mM][eE]
 	| [wW][oO][lL][tT][eE][rR][sS][kK][lL][uU][wW][eE][rR]
 	| [wW][oO][oO][dD][sS][iI][dD][eE]
-	| [wW][oO][rR][kK]
 	| [wW][oO][rR][kK][sS]
 	| [wW][oO][rR][lL][dD]
 	| [wW][oO][wW]
@@ -1506,6 +1383,7 @@ ASCIITLD = "." (
 	| [xX][nN]--[nN][yY][qQ][yY]26[aA]
 	| [xX][nN]--[oO]3[cC][wW]4[hH]
 	| [xX][nN]--[oO][gG][bB][pP][fF]8[fF][lL]
+	| [xX][nN]--[oO][tT][uU]796[dD]
 	| [xX][nN]--[pP]1[aA][cC][fF]
 	| [xX][nN]--[pP]1[aA][iI]
 	| [xX][nN]--[pP][bB][tT]977[cC]
@@ -1565,3 +1443,132 @@ ASCIITLD = "." (
 	| [zZ][wW]
 	) "."?   // Accept trailing root (empty) domain
 
+// LUCENE-8278: Each TLD in {ASCIITLDprefix_1CharSuffix} is a prefix of another TLD by 1
character
+ASCIITLDprefix_1CharSuffix = "." (
+	  [aA][cC]
+	| [aA][cC][cC][oO][uU][nN][tT][aA][nN][tT]
+	| [aA][dD]
+	| [aA][eE]
+	| [aA][fF]
+	| [aA][iI][gG]
+	| [aA][oO]
+	| [aA][rR][tT]
+	| [aA][uU][dD][iI]
+	| [aA][uU][tT][oO]
+	| [aA][wW]
+	| [aA][xX]
+	| [bB][aA]
+	| [bB][bB]
+	| [bB][eE]
+	| [bB][iI]
+	| [bB][iI][nN][gG]
+	| [bB][mM]
+	| [bB][nN]
+	| [bB][oO][oO]
+	| [bB][zZ]
+	| [cC][aA][lL]
+	| [cC][aA][mM]
+	| [cC][aA][rR]
+	| [cC][aA][rR][eE][eE][rR]
+	| [cC][fF]
+	| [cC][iI][tT][iI]
+	| [cC][oO]
+	| [cC][oO][uU][pP][oO][nN]
+	| [cC][rR]
+	| [cC][rR][uU][iI][sS][eE]
+	| [dD][eE]
+	| [dD][eE][aA][lL]
+	| [dD][oO]
+	| [eE][cC]
+	| [eE][sS]
+	| [eE][uU]
+	| [fF][aA][nN]
+	| [fF][iI]
+	| [fF][oO][oO]
+	| [fF][rR]
+	| [fF][uU][nN]
+	| [gG][aA]
+	| [gG][aA][mM][eE]
+	| [gG][dD]
+	| [gG][eE]
+	| [gG][iI][fF][tT]
+	| [gG][lL]
+	| [gG][mM]
+	| [gG][oO][oO]
+	| [hH][kK]
+	| [iI][nN]
+	| [iI][sS]
+	| [iI][tT]
+	| [jJ][mM]
+	| [jJ][oO]
+	| [kK][iI]
+	| [kK][pP]
+	| [kK][rR]
+	| [lL][aA]
+	| [lL][oO][aA][nN]
+	| [lL][tT][dD]
+	| [mM][aA]
+	| [mM][aA][rR][kK][eE][tT]
+	| [mM][eE][nN]
+	| [mM][lL]
+	| [mM][mM]
+	| [mM][oO]
+	| [mM][sS]
+	| [mM][tT]
+	| [nN][aA]
+	| [nN][eE][wW]
+	| [nN][fF]
+	| [nN][gG]
+	| [nN][oO]
+	| [nN][rR]
+	| [pP][aA]
+	| [pP][eE]
+	| [pP][hH]
+	| [pP][hH][oO][tT][oO]
+	| [pP][iI][nN]
+	| [pP][nN]
+	| [pP][rR][oO]
+	| [pP][wW]
+	| [rR][eE][iI][sS][eE]
+	| [rR][eE][nN]
+	| [rR][eE][vV][iI][eE][wW]
+	| [rR][uU]
+	| [rR][wW]
+	| [sS][aA]
+	| [sS][bB]
+	| [sS][cC]
+	| [sS][eE][xX]
+	| [sS][kK][iI]
+	| [sS][oO]
+	| [sS][rR]
+	| [sS][tT]
+	| [tT][aA][xX]
+	| [tT][cC]
+	| [tT][dD]
+	| [tT][hH]
+	| [tT][jJ]
+	| [tT][oO]
+	| [tT][rR]
+	| [tT][vV]
+	| [vV][eE]
+	| [vV][iI]
+	| [wW][iI][nN]
+	| [wW][oO][rR][kK]
+	) "."?   // Accept trailing root (empty) domain
+
+// LUCENE-8278: Each TLD in {ASCIITLDprefix_2CharSuffix} is a prefix of another TLD by 2
characters
+ASCIITLDprefix_2CharSuffix = "." (
+	  [aA][iI]
+	| [aA][rR]
+	| [bB][oO]
+	| [cC][aA]
+	| [fF][oO]
+	| [lL][tT]
+	| [mM][eE]
+	| [nN][eE]
+	| [pP][rR]
+	| [rR][eE]
+	| [sS][eE]
+	| [sS][kK]
+	) "."?   // Accept trailing root (empty) domain
+


Mime
View raw message