commit 120e1322c668fa876922f66845120d2f5e183c9b
author Akron <nils@diewald-online.de> Thu Sep 01 14:52:29 2022 +0200
committer Nils Diewald <nils@diewald-online.de> Wed Nov 16 11:33:35 2022 +0100
tree 171dcd9929d7868ac4b96849f3150fb1bb3ef130
parent 33451c212e4e745fa17ac475a4cd8d017a407e44

Improve quantifiers in regex

Change-Id: Ibc078b01673348416bd41a9a9dd2b5afc8cb2a43
Reviewed-on: https://korap.ids-mannheim.de/gerrit/c/KorAP/Koral/+/6024
Reviewed-by: Nils Diewald <nils@diewald-online.de>

Changes

diff --git a/Changes b/Changes
index dc38dac..7523e71 100644
--- a/Changes
+++ b/Changes
@@ -1,8 +1,9 @@
-0.38.2 2022-08-31
+0.38.2 2022-10-05
     - [feature] Improve regex treatment of negative
       character classes (diewald)
+    - [bugfix] Improve quantifiers in Regex (diewald)
 
-0.38.1 2022-10-05
+0.38.1 2022-01-07
     - [security] Updated log4j (diewald)
     - Fixed Annis OR group (resolved #96) 
     - [security] Updated jackson (diewald)

src/main/antlr/poliqarpplus/PoliqarpPlusLexer.g4

diff --git a/src/main/antlr/poliqarpplus/PoliqarpPlusLexer.g4 b/src/main/antlr/poliqarpplus/PoliqarpPlusLexer.g4
index 2189657..8871e98 100644
--- a/src/main/antlr/poliqarpplus/PoliqarpPlusLexer.g4
+++ b/src/main/antlr/poliqarpplus/PoliqarpPlusLexer.g4
@@ -101,13 +101,9 @@
 fragment RE_alter    : ((RE_char | ('(' RE_expr ')') | RE_chgroup) '|' RE_expr )+;
 
 fragment RE_chgroup  : '[' '^'? RE_char+ ']';
-fragment RE_quant	 : (RE_star | RE_plus | RE_occ) QMARK?;
-fragment RE_opt      : (RE_char | RE_chgroup | ( '(' RE_expr ')')) '?';
-fragment RE_star     : (RE_char | RE_chgroup | ( '(' RE_expr ')')) '*';
-fragment RE_plus     : (RE_char | RE_chgroup | ( '(' RE_expr ')')) '+';
-fragment RE_occ      : (RE_char | RE_chgroup | ( '(' RE_expr ')')) FOCC;
+fragment RE_quant    : ('.' | RE_char | RE_chgroup | ( '(' RE_expr ')')) ('?' | '*' | '+' | FOCC) QMARK?;
 fragment RE_group    : '(' RE_expr ')';
-fragment RE_expr     : ('.' | RE_char | RE_alter | RE_chgroup | RE_opt | RE_quant | RE_group | '^' )+;
+fragment RE_expr     : ('.' | RE_char | RE_alter | RE_chgroup | RE_quant | RE_group | '^' )+;
 fragment RE_dquote   : ('"'|'„'|'“')  (RE_expr | '\'' | ':' )* ('"'|'“'|'”');
 
 REGEX             : RE_dquote;

src/test/java/de/ids_mannheim/korap/query/test/poliqarpplus/PoliqarpPlusQueryProcessorTest.java

diff --git a/src/test/java/de/ids_mannheim/korap/query/test/poliqarpplus/PoliqarpPlusQueryProcessorTest.java b/src/test/java/de/ids_mannheim/korap/query/test/poliqarpplus/PoliqarpPlusQueryProcessorTest.java
index 675debc..c13c7f4 100644
--- a/src/test/java/de/ids_mannheim/korap/query/test/poliqarpplus/PoliqarpPlusQueryProcessorTest.java
+++ b/src/test/java/de/ids_mannheim/korap/query/test/poliqarpplus/PoliqarpPlusQueryProcessorTest.java
@@ -273,6 +273,12 @@
         qs.setQuery(query, "poliqarpplus");
         res = mapper.readTree(qs.toJSON());
         assertEquals("302", res.at("/errors/0/0").asText());
+
+        query = "\"?\"";
+        qs.setQuery(query, "poliqarpplus");
+        res = mapper.readTree(qs.toJSON());
+        System.err.println(qs.toJSON());
+        assertEquals(302, res.at("/errors/0/0").asInt());
     }
 
     @Test