dont extract regular expressions from strings that are leaves in a string concat

Author: erik-krogh

javascript/extractor/src/com/semmle/js/extractor/ASTExtractor.java

@@ -596,7 +596,7 @@ public Label visit(Literal nd, Context c) {
         offsets.set(0, 1); // skip the initial '/'
         SourceMap sourceMap = SourceMap.legacyWithStartPos(SourceMap.fromString(nd.getRaw()).offsetBy(0, offsets), startPos);
         regexpExtractor.extract(source.substring(1, source.lastIndexOf('/')), sourceMap, nd, false);
-      } else if (nd.isStringLiteral() && !c.isInsideType() && nd.getRaw().length() < 1000) {
+      } else if (nd.isStringLiteral() && !c.isInsideType() && nd.getRaw().length() < 1000 && !c.isBinopOperand()) {
         SourceMap sourceMap = SourceMap.legacyWithStartPos(SourceMap.fromString(nd.getRaw()).offsetBy(0, makeStringLiteralOffsets(nd.getRaw())), startPos);
         regexpExtractor.extract(valueString, sourceMap, nd, true);
 

javascript/extractor/tests/es2015/output/trap/properties.js.trap

@@ -287,150 +287,144 @@ enclosing_stmt(#20097,#20083)
 expr_containers(#20097,#20001)
 literals("prop","""prop""",#20097)
 #20098=*
-regexpterm(#20098,14,#20097,0,"prop")
-#20099=@"loc,{#10000},3,11,3,14"
-locations_default(#20099,#10000,3,11,3,14)
+exprs(#20098,13,#20095,1,"Math.random()")
+#20099=@"loc,{#10000},3,19,3,31"
+locations_default(#20099,#10000,3,19,3,31)
 hasLocation(#20098,#20099)
-regexp_const_value(#20098,"prop")
+enclosing_stmt(#20098,#20083)
+expr_containers(#20098,#20001)
 #20100=*
-exprs(#20100,13,#20095,1,"Math.random()")
-#20101=@"loc,{#10000},3,19,3,31"
-locations_default(#20101,#10000,3,19,3,31)
+exprs(#20100,14,#20098,-1,"Math.random")
+#20101=@"loc,{#10000},3,19,3,29"
+locations_default(#20101,#10000,3,19,3,29)
 hasLocation(#20100,#20101)
 enclosing_stmt(#20100,#20083)
 expr_containers(#20100,#20001)
 #20102=*
-exprs(#20102,14,#20100,-1,"Math.random")
-#20103=@"loc,{#10000},3,19,3,29"
-locations_default(#20103,#10000,3,19,3,29)
-hasLocation(#20102,#20103)
+exprs(#20102,79,#20100,0,"Math")
+hasLocation(#20102,#20037)
 enclosing_stmt(#20102,#20083)
 expr_containers(#20102,#20001)
+literals("Math","Math",#20102)
+#20103=@"var;{Math};{#20000}"
+variables(#20103,"Math",#20000)
+bind(#20102,#20103)
 #20104=*
-exprs(#20104,79,#20102,0,"Math")
-hasLocation(#20104,#20037)
+exprs(#20104,0,#20100,1,"random")
+hasLocation(#20104,#20041)
 enclosing_stmt(#20104,#20083)
 expr_containers(#20104,#20001)
-literals("Math","Math",#20104)
-#20105=@"var;{Math};{#20000}"
-variables(#20105,"Math",#20000)
-bind(#20104,#20105)
+literals("random","random",#20104)
+#20105=*
+exprs(#20105,3,#20093,1,"23")
+hasLocation(#20105,#20051)
+enclosing_stmt(#20105,#20083)
+expr_containers(#20105,#20001)
+literals("23","23",#20105)
+is_computed(#20093)
 #20106=*
-exprs(#20106,0,#20102,1,"random")
-hasLocation(#20106,#20041)
-enclosing_stmt(#20106,#20083)
-expr_containers(#20106,#20001)
-literals("random","random",#20106)
+properties(#20106,#20091,1,0,"x")
+hasLocation(#20106,#20055)
 #20107=*
-exprs(#20107,3,#20093,1,"23")
-hasLocation(#20107,#20051)
+exprs(#20107,0,#20106,0,"x")
+hasLocation(#20107,#20055)
 enclosing_stmt(#20107,#20083)
 expr_containers(#20107,#20001)
-literals("23","23",#20107)
-is_computed(#20093)
+literals("x","x",#20107)
 #20108=*
-properties(#20108,#20091,1,0,"x")
+exprs(#20108,79,#20106,1,"x")
 hasLocation(#20108,#20055)
+enclosing_stmt(#20108,#20083)
+expr_containers(#20108,#20001)
+literals("x","x",#20108)
+bind(#20108,#20081)
 #20109=*
-exprs(#20109,0,#20108,0,"x")
-hasLocation(#20109,#20055)
-enclosing_stmt(#20109,#20083)
-expr_containers(#20109,#20001)
-literals("x","x",#20109)
-#20110=*
-exprs(#20110,79,#20108,1,"x")
-hasLocation(#20110,#20055)
-enclosing_stmt(#20110,#20083)
-expr_containers(#20110,#20001)
-literals("x","x",#20110)
-bind(#20110,#20081)
+properties(#20109,#20091,2,0,"m() { return 56; }")
+#20110=@"loc,{#10000},5,9,5,26"
+locations_default(#20110,#10000,5,9,5,26)
+hasLocation(#20109,#20110)
 #20111=*
-properties(#20111,#20091,2,0,"m() { return 56; }")
-#20112=@"loc,{#10000},5,9,5,26"
-locations_default(#20112,#10000,5,9,5,26)
-hasLocation(#20111,#20112)
-#20113=*
-exprs(#20113,0,#20111,0,"m")
-hasLocation(#20113,#20059)
-enclosing_stmt(#20113,#20083)
-expr_containers(#20113,#20001)
-literals("m","m",#20113)
+exprs(#20111,0,#20109,0,"m")
+hasLocation(#20111,#20059)
+enclosing_stmt(#20111,#20083)
+expr_containers(#20111,#20001)
+literals("m","m",#20111)
+#20112=*
+exprs(#20112,9,#20109,1,"() { return 56; }")
+#20113=@"loc,{#10000},5,10,5,26"
+locations_default(#20113,#10000,5,10,5,26)
+hasLocation(#20112,#20113)
+enclosing_stmt(#20112,#20083)
+expr_containers(#20112,#20001)
 #20114=*
-exprs(#20114,9,#20111,1,"() { return 56; }")
-#20115=@"loc,{#10000},5,10,5,26"
-locations_default(#20115,#10000,5,10,5,26)
-hasLocation(#20114,#20115)
-enclosing_stmt(#20114,#20083)
-expr_containers(#20114,#20001)
+scopes(#20114,1)
+scopenodes(#20112,#20114)
+scopenesting(#20114,#20000)
+#20115=@"var;{arguments};{#20114}"
+variables(#20115,"arguments",#20114)
+is_arguments_object(#20115)
 #20116=*
-scopes(#20116,1)
-scopenodes(#20114,#20116)
-scopenesting(#20116,#20000)
-#20117=@"var;{arguments};{#20116}"
-variables(#20117,"arguments",#20116)
-is_arguments_object(#20117)
+stmts(#20116,1,#20112,-2,"{ return 56; }")
+#20117=@"loc,{#10000},5,13,5,26"
+locations_default(#20117,#10000,5,13,5,26)
+hasLocation(#20116,#20117)
+stmt_containers(#20116,#20112)
 #20118=*
-stmts(#20118,1,#20114,-2,"{ return 56; }")
-#20119=@"loc,{#10000},5,13,5,26"
-locations_default(#20119,#10000,5,13,5,26)
+stmts(#20118,9,#20116,0,"return 56;")
+#20119=@"loc,{#10000},5,15,5,24"
+locations_default(#20119,#10000,5,15,5,24)
 hasLocation(#20118,#20119)
-stmt_containers(#20118,#20114)
+stmt_containers(#20118,#20112)
 #20120=*
-stmts(#20120,9,#20118,0,"return 56;")
-#20121=@"loc,{#10000},5,15,5,24"
-locations_default(#20121,#10000,5,15,5,24)
-hasLocation(#20120,#20121)
-stmt_containers(#20120,#20114)
-#20122=*
-exprs(#20122,3,#20120,0,"56")
-hasLocation(#20122,#20069)
-enclosing_stmt(#20122,#20120)
-expr_containers(#20122,#20114)
-literals("56","56",#20122)
-is_method(#20111)
+exprs(#20120,3,#20118,0,"56")
+hasLocation(#20120,#20069)
+enclosing_stmt(#20120,#20118)
+expr_containers(#20120,#20112)
+literals("56","56",#20120)
+is_method(#20109)
+#20121=*
+entry_cfg_node(#20121,#20001)
+#20122=@"loc,{#10000},1,1,1,0"
+locations_default(#20122,#10000,1,1,1,0)
+hasLocation(#20121,#20122)
 #20123=*
-entry_cfg_node(#20123,#20001)
-#20124=@"loc,{#10000},1,1,1,0"
-locations_default(#20124,#10000,1,1,1,0)
-hasLocation(#20123,#20124)
-#20125=*
-exit_cfg_node(#20125,#20001)
-hasLocation(#20125,#20079)
+exit_cfg_node(#20123,#20001)
+hasLocation(#20123,#20079)
 successor(#20083,#20086)
 successor(#20091,#20097)
-successor(#20114,#20111)
+successor(#20112,#20109)
+#20124=*
+entry_cfg_node(#20124,#20112)
+#20125=@"loc,{#10000},5,10,5,9"
+locations_default(#20125,#10000,5,10,5,9)
+hasLocation(#20124,#20125)
 #20126=*
-entry_cfg_node(#20126,#20114)
-#20127=@"loc,{#10000},5,10,5,9"
-locations_default(#20127,#10000,5,10,5,9)
+exit_cfg_node(#20126,#20112)
+#20127=@"loc,{#10000},5,27,5,26"
+locations_default(#20127,#10000,5,27,5,26)
 hasLocation(#20126,#20127)
-#20128=*
-exit_cfg_node(#20128,#20114)
-#20129=@"loc,{#10000},5,27,5,26"
-locations_default(#20129,#10000,5,27,5,26)
-hasLocation(#20128,#20129)
-successor(#20118,#20122)
-successor(#20122,#20120)
-successor(#20120,#20128)
-successor(#20126,#20118)
-successor(#20113,#20114)
-successor(#20111,#20088)
-successor(#20110,#20108)
-successor(#20109,#20110)
-successor(#20108,#20113)
-successor(#20107,#20093)
-successor(#20106,#20102)
-successor(#20104,#20106)
-successor(#20102,#20100)
-successor(#20100,#20095)
-successor(#20097,#20104)
-successor(#20095,#20107)
-successor(#20093,#20109)
+successor(#20116,#20120)
+successor(#20120,#20118)
+successor(#20118,#20126)
+successor(#20124,#20116)
+successor(#20111,#20112)
+successor(#20109,#20088)
+successor(#20108,#20106)
+successor(#20107,#20108)
+successor(#20106,#20111)
+successor(#20105,#20093)
+successor(#20104,#20100)
+successor(#20102,#20104)
+successor(#20100,#20098)
+successor(#20098,#20095)
+successor(#20097,#20102)
+successor(#20095,#20105)
+successor(#20093,#20107)
 successor(#20090,#20091)
-successor(#20088,#20125)
+successor(#20088,#20123)
 successor(#20087,#20084)
 successor(#20086,#20087)
 successor(#20084,#20090)
-successor(#20123,#20083)
+successor(#20121,#20083)
 numlines(#10000,6,6,0)
 filetype(#10000,"javascript")