[6.40] [df] Consider quoted strings in collection size short syntax

tree/dataframe/src/RDFInterfaceUtils.cxx

@@ -171,10 +171,21 @@ ParsedExpression ParseRDFExpression(std::string_view expr, const ROOT::Internal:
 {
    // transform `#var` into `R_rdf_sizeof_var`
    TString preProcessedExpr(expr);
-   // match #varname at beginning of the sentence or after not-a-word, but exclude preprocessor directives like #ifdef
+   // regex explanation:
+   // #
+   //     - matches the literal character #
+   // (?!(ifdef|ifndef|if|else|elif|endif|pragma|define|undef|include|line))
+   //     - negative lookahead (i.e. don't match these  occurrences) to exclude preprocessor directives e.g. #ifdef
+   // ([a-zA-Z_][a-zA-Z0-9_]*)
+   //     - match characters that make up the variable name after the #
+   // (?=(?:[^"]*"[^"]*")*[^"]*$)
+   //     - positive lookahead (i.e. match these occurrences) checking that there is always an even number of quotes
+   //       *after* the matched character sequence, using the non-capturing group (?:[^"]*"[^"]*"). We leverage the
+   //       fact that for any character sequence to be within quotes there must be an odd number of quotes after the
+   //       sequence and before the end of the line.
    TPRegexp colSizeReplacer(
-      "(^|\\W)#(?!(ifdef|ifndef|if|else|elif|endif|pragma|define|undef|include|line))([a-zA-Z_][a-zA-Z0-9_]*)");
-   colSizeReplacer.Substitute(preProcessedExpr, "$1R_rdf_sizeof_$3", "g");
+      R"REGEX(#(?!(ifdef|ifndef|if|else|elif|endif|pragma|define|undef|include|line))([a-zA-Z_][a-zA-Z0-9_]*)(?=(?:[^"]*"[^"]*")*[^"]*$))REGEX");
+   colSizeReplacer.Substitute(preProcessedExpr, "R_rdf_sizeof_$2", "g");
 
    ColumnNames_t usedCols;
    ColumnNames_t usedAliases;

tree/dataframe/test/dataframe_interface.cxx

@@ -845,6 +845,39 @@ TEST(RDataFrameInterface, StressShortSyntaxForCollectionSizes)
    EXPECT_EQ(c, 42ull);
 }
 
+// Regression tests for https://github.com/root-project/root/issues/23790
+TEST(RDataFrameInterface, ShortSyntaxForCollectionSizesMultipleStringsAndVars)
+{
+   // We need to define first four target columns that have the longer name for collection size
+   auto df = ROOT::RDF::RNode(ROOT::RDataFrame(1));
+   for (auto i = 1; i <= 4; i++)
+      df = df.Define("R_rdf_sizeof_var" + std::to_string(i), []() -> int { return 1; });
+   std::string code{R"CODE(
+      auto value1 = #var1; // comment
+      #ifdef NONEXISTENTMACRO
+      // comment with "quoted string"
+      #endif
+      std::string simplestr("#simplestr"); // comment
+      std::string mystr("anything#hello"); // comment
+      auto value2 = #var2; // comment with "another string"
+      std::string anotherstr("We don't want #var3 and #var4 to be replaced");
+      auto value5 = #var1 + #var2; std::string innerRawString{R"INNER(
+       This raw string contains #myvar which should
+       not be changed.)INNER"};
+      return #var1 + #var2 + #var3 + #var4 + value1 + value2 + value5;
+   )CODE"};
+   auto sum = df.Define("target_col", code).Sum<int>("target_col");
+   EXPECT_EQ(*sum, 8);
+}
+
+// Regression test for https://github.com/root-project/root/issues/23790
+TEST(RDataFrameInterface, StringWithPotentialAmbiguityWithSortSyntaxForCollectionSizes)
+{
+   auto df = ROOT::RDataFrame(1).Define("col", "std::string(\"#a\")").Filter("col == \"#a\"");
+   EXPECT_EQ(df.Count().GetValue(), 1ull);
+   EXPECT_EQ(*df.Take<std::string>("col")->begin(), "#a");
+}
+
 TEST(RDataFrameInterface, MutableForeach)
 {
    int i = 0;