Added ability to parse nested character classes while using v flag.

javascript/extractor/src/com/semmle/js/parser/RegExpParser.java

@@ -545,6 +545,7 @@ private RegExpTerm parseAtomEscape(SourceLocation loc, boolean inCharClass) {
   }
 
   private RegExpTerm parseCharacterClass() {
+    if (flags != null && flags.contains("v")) return parseNestedCharacterClass();
     SourceLocation loc = new SourceLocation(pos());
     List<RegExpTerm> elements = new ArrayList<>();
 
@@ -560,6 +561,28 @@ private RegExpTerm parseCharacterClass() {
     return this.finishTerm(new CharacterClass(loc, elements, inverted));
   }
 
+  // ECMA 2024 `v` flag allows nested character classes.
+  private RegExpTerm parseNestedCharacterClass() {
+    SourceLocation loc = new SourceLocation(pos());
+    List<RegExpTerm> elements = new ArrayList<>();
+
+    this.match("[");
+    boolean inverted = this.match("^");
+    while (!this.match("]")) {
+      if (this.atEOS()) {
+        this.error(Error.EXPECTED_RBRACKET);
+        break;
+      }
+      if (lookahead("[")) {
+        elements.add(parseNestedCharacterClass());
+      } 
+      else {
+        elements.add(this.parseCharacterClassElement());
+      }
+    }
+    return this.finishTerm(new CharacterClass(loc, elements, inverted));
+  }
+
   private static final List<String> escapeClasses = Arrays.asList("d", "D", "s", "S", "w", "W");
 
   private RegExpTerm parseCharacterClassElement() {

javascript/extractor/tests/es2024/output/trap/regex_nested_character_class.js.trap

@@ -86,181 +86,126 @@ enclosing_stmt(#20027,#20025)
 expr_containers(#20027,#20001)
 literals("/[[]]/v","/[[]]/v",#20027)
 #20028=*
-regexpterm(#20028,1,#20027,0,"[[]]")
+regexpterm(#20028,23,#20027,0,"[[]]")
 #20029=@"loc,{#10000},1,2,1,5"
 locations_default(#20029,#10000,1,2,1,5)
 hasLocation(#20028,#20029)
 #20030=*
-regexpterm(#20030,23,#20028,0,"[[]")
-#20031=@"loc,{#10000},1,2,1,4"
-locations_default(#20031,#10000,1,2,1,4)
+regexpterm(#20030,23,#20028,0,"[]")
+#20031=@"loc,{#10000},1,3,1,4"
+locations_default(#20031,#10000,1,3,1,4)
 hasLocation(#20030,#20031)
 #20032=*
-regexpterm(#20032,14,#20030,0,"[")
-#20033=@"loc,{#10000},1,3,1,3"
-locations_default(#20033,#10000,1,3,1,3)
-hasLocation(#20032,#20033)
-regexp_const_value(#20032,"[")
+stmts(#20032,2,#20001,1,"/[[a]]/v;")
+hasLocation(#20032,#20007)
+stmt_containers(#20032,#20001)
+#20033=*
+exprs(#20033,5,#20032,0,"/[[a]]/v")
+hasLocation(#20033,#20015)
+enclosing_stmt(#20033,#20032)
+expr_containers(#20033,#20001)
+literals("/[[a]]/v","/[[a]]/v",#20033)
 #20034=*
-regexpterm(#20034,14,#20028,1,"]")
-#20035=@"loc,{#10000},1,5,1,5"
-locations_default(#20035,#10000,1,5,1,5)
+regexpterm(#20034,23,#20033,0,"[[a]]")
+#20035=@"loc,{#10000},2,2,2,6"
+locations_default(#20035,#10000,2,2,2,6)
 hasLocation(#20034,#20035)
-regexp_const_value(#20034,"]")
 #20036=*
-regexp_parse_errors(#20036,#20028,"unexpected character")
-hasLocation(#20036,#20035)
-#20037=*
-stmts(#20037,2,#20001,1,"/[[a]]/v;")
-hasLocation(#20037,#20007)
-stmt_containers(#20037,#20001)
+regexpterm(#20036,23,#20034,0,"[a]")
+#20037=@"loc,{#10000},2,3,2,5"
+locations_default(#20037,#10000,2,3,2,5)
+hasLocation(#20036,#20037)
 #20038=*
-exprs(#20038,5,#20037,0,"/[[a]]/v")
-hasLocation(#20038,#20015)
-enclosing_stmt(#20038,#20037)
-expr_containers(#20038,#20001)
-literals("/[[a]]/v","/[[a]]/v",#20038)
-#20039=*
-regexpterm(#20039,1,#20038,0,"[[a]]")
-#20040=@"loc,{#10000},2,2,2,6"
-locations_default(#20040,#10000,2,2,2,6)
-hasLocation(#20039,#20040)
+regexpterm(#20038,14,#20036,0,"a")
+#20039=@"loc,{#10000},2,4,2,4"
+locations_default(#20039,#10000,2,4,2,4)
+hasLocation(#20038,#20039)
+regexp_const_value(#20038,"a")
+#20040=*
+stmts(#20040,2,#20001,2,"/[ [] [ [] [] ] ]/v;")
+hasLocation(#20040,#20009)
+stmt_containers(#20040,#20001)
 #20041=*
-regexpterm(#20041,23,#20039,0,"[[a]")
-#20042=@"loc,{#10000},2,2,2,5"
-locations_default(#20042,#10000,2,2,2,5)
-hasLocation(#20041,#20042)
-#20043=*
-regexpterm(#20043,14,#20041,0,"[")
-#20044=@"loc,{#10000},2,3,2,3"
-locations_default(#20044,#10000,2,3,2,3)
-hasLocation(#20043,#20044)
-regexp_const_value(#20043,"[")
-#20045=*
-regexpterm(#20045,14,#20041,1,"a")
-#20046=@"loc,{#10000},2,4,2,4"
-locations_default(#20046,#10000,2,4,2,4)
-hasLocation(#20045,#20046)
-regexp_const_value(#20045,"a")
-#20047=*
-regexpterm(#20047,14,#20039,1,"]")
-#20048=@"loc,{#10000},2,6,2,6"
-locations_default(#20048,#10000,2,6,2,6)
-hasLocation(#20047,#20048)
-regexp_const_value(#20047,"]")
-#20049=*
-regexp_parse_errors(#20049,#20039,"unexpected character")
-hasLocation(#20049,#20048)
+exprs(#20041,5,#20040,0,"/[ [] [ [] [] ] ]/v")
+hasLocation(#20041,#20019)
+enclosing_stmt(#20041,#20040)
+expr_containers(#20041,#20001)
+literals("/[ [] [ [] [] ] ]/v","/[ [] [ [] [] ] ]/v",#20041)
+#20042=*
+regexpterm(#20042,23,#20041,0,"[ [] [ [] [] ] ]")
+#20043=@"loc,{#10000},3,2,3,17"
+locations_default(#20043,#10000,3,2,3,17)
+hasLocation(#20042,#20043)
+#20044=*
+regexpterm(#20044,14,#20042,0," ")
+#20045=@"loc,{#10000},3,3,3,3"
+locations_default(#20045,#10000,3,3,3,3)
+hasLocation(#20044,#20045)
+regexp_const_value(#20044," ")
+#20046=*
+regexpterm(#20046,23,#20042,1,"[]")
+#20047=@"loc,{#10000},3,4,3,5"
+locations_default(#20047,#10000,3,4,3,5)
+hasLocation(#20046,#20047)
+#20048=*
+regexpterm(#20048,14,#20042,2," ")
+#20049=@"loc,{#10000},3,6,3,6"
+locations_default(#20049,#10000,3,6,3,6)
+hasLocation(#20048,#20049)
+regexp_const_value(#20048," ")
 #20050=*
-stmts(#20050,2,#20001,2,"/[ [] [ [] [] ] ]/v;")
-hasLocation(#20050,#20009)
-stmt_containers(#20050,#20001)
-#20051=*
-exprs(#20051,5,#20050,0,"/[ [] [ [] [] ] ]/v")
-hasLocation(#20051,#20019)
-enclosing_stmt(#20051,#20050)
-expr_containers(#20051,#20001)
-literals("/[ [] [ [] [] ] ]/v","/[ [] [ [] [] ] ]/v",#20051)
+regexpterm(#20050,23,#20042,3,"[ [] [] ]")
+#20051=@"loc,{#10000},3,7,3,15"
+locations_default(#20051,#10000,3,7,3,15)
+hasLocation(#20050,#20051)
 #20052=*
-regexpterm(#20052,1,#20051,0,"[ [] [ [] [] ] ]")
-#20053=@"loc,{#10000},3,2,3,17"
-locations_default(#20053,#10000,3,2,3,17)
+regexpterm(#20052,14,#20050,0," ")
+#20053=@"loc,{#10000},3,8,3,8"
+locations_default(#20053,#10000,3,8,3,8)
 hasLocation(#20052,#20053)
+regexp_const_value(#20052," ")
 #20054=*
-regexpterm(#20054,23,#20052,0,"[ []")
-#20055=@"loc,{#10000},3,2,3,5"
-locations_default(#20055,#10000,3,2,3,5)
+regexpterm(#20054,23,#20050,1,"[]")
+#20055=@"loc,{#10000},3,9,3,10"
+locations_default(#20055,#10000,3,9,3,10)
 hasLocation(#20054,#20055)
 #20056=*
-regexpterm(#20056,14,#20054,0," ")
-#20057=@"loc,{#10000},3,3,3,3"
-locations_default(#20057,#10000,3,3,3,3)
+regexpterm(#20056,14,#20050,2," ")
+#20057=@"loc,{#10000},3,11,3,11"
+locations_default(#20057,#10000,3,11,3,11)
 hasLocation(#20056,#20057)
 regexp_const_value(#20056," ")
 #20058=*
-regexpterm(#20058,14,#20054,1,"[")
-#20059=@"loc,{#10000},3,4,3,4"
-locations_default(#20059,#10000,3,4,3,4)
+regexpterm(#20058,23,#20050,3,"[]")
+#20059=@"loc,{#10000},3,12,3,13"
+locations_default(#20059,#10000,3,12,3,13)
 hasLocation(#20058,#20059)
-regexp_const_value(#20058,"[")
 #20060=*
-regexpterm(#20060,14,#20052,1," ")
-#20061=@"loc,{#10000},3,6,3,6"
-locations_default(#20061,#10000,3,6,3,6)
+regexpterm(#20060,14,#20050,4," ")
+#20061=@"loc,{#10000},3,14,3,14"
+locations_default(#20061,#10000,3,14,3,14)
 hasLocation(#20060,#20061)
 regexp_const_value(#20060," ")
 #20062=*
-regexpterm(#20062,23,#20052,2,"[ []")
-#20063=@"loc,{#10000},3,7,3,10"
-locations_default(#20063,#10000,3,7,3,10)
+regexpterm(#20062,14,#20042,4," ")
+#20063=@"loc,{#10000},3,16,3,16"
+locations_default(#20063,#10000,3,16,3,16)
 hasLocation(#20062,#20063)
+regexp_const_value(#20062," ")
 #20064=*
-regexpterm(#20064,14,#20062,0," ")
-#20065=@"loc,{#10000},3,8,3,8"
-locations_default(#20065,#10000,3,8,3,8)
+entry_cfg_node(#20064,#20001)
+#20065=@"loc,{#10000},1,1,1,0"
+locations_default(#20065,#10000,1,1,1,0)
 hasLocation(#20064,#20065)
-regexp_const_value(#20064," ")
 #20066=*
-regexpterm(#20066,14,#20062,1,"[")
-#20067=@"loc,{#10000},3,9,3,9"
-locations_default(#20067,#10000,3,9,3,9)
-hasLocation(#20066,#20067)
-regexp_const_value(#20066,"[")
-#20068=*
-regexpterm(#20068,14,#20052,3," ")
-#20069=@"loc,{#10000},3,11,3,11"
-locations_default(#20069,#10000,3,11,3,11)
-hasLocation(#20068,#20069)
-regexp_const_value(#20068," ")
-#20070=*
-regexpterm(#20070,23,#20052,4,"[]")
-#20071=@"loc,{#10000},3,12,3,13"
-locations_default(#20071,#10000,3,12,3,13)
-hasLocation(#20070,#20071)
-#20072=*
-regexpterm(#20072,14,#20052,5," ")
-#20073=@"loc,{#10000},3,14,3,14"
-locations_default(#20073,#10000,3,14,3,14)
-hasLocation(#20072,#20073)
-regexp_const_value(#20072," ")
-#20074=*
-regexpterm(#20074,14,#20052,6,"]")
-#20075=@"loc,{#10000},3,15,3,15"
-locations_default(#20075,#10000,3,15,3,15)
-hasLocation(#20074,#20075)
-regexp_const_value(#20074,"]")
-#20076=*
-regexpterm(#20076,14,#20052,7," ")
-#20077=@"loc,{#10000},3,16,3,16"
-locations_default(#20077,#10000,3,16,3,16)
-hasLocation(#20076,#20077)
-regexp_const_value(#20076," ")
-#20078=*
-regexpterm(#20078,14,#20052,8,"]")
-#20079=@"loc,{#10000},3,17,3,17"
-locations_default(#20079,#10000,3,17,3,17)
-hasLocation(#20078,#20079)
-regexp_const_value(#20078,"]")
-#20080=*
-regexp_parse_errors(#20080,#20052,"unexpected character")
-hasLocation(#20080,#20075)
-#20081=*
-regexp_parse_errors(#20081,#20052,"unexpected character")
-hasLocation(#20081,#20079)
-#20082=*
-entry_cfg_node(#20082,#20001)
-#20083=@"loc,{#10000},1,1,1,0"
-locations_default(#20083,#10000,1,1,1,0)
-hasLocation(#20082,#20083)
-#20084=*
-exit_cfg_node(#20084,#20001)
-hasLocation(#20084,#20023)
-successor(#20050,#20051)
-successor(#20051,#20084)
-successor(#20037,#20038)
-successor(#20038,#20050)
+exit_cfg_node(#20066,#20001)
+hasLocation(#20066,#20023)
+successor(#20040,#20041)
+successor(#20041,#20066)
+successor(#20032,#20033)
+successor(#20033,#20040)
 successor(#20025,#20027)
-successor(#20027,#20037)
-successor(#20082,#20025)
+successor(#20027,#20032)
+successor(#20064,#20025)
 numlines(#10000,3,3,1)
 filetype(#10000,"javascript")