[flang-commits] [flang] [flang] Handle substring in data statement constant (PR #120130)

[Peter Klausler via flang-commits]

https://github.com/klausler updated https://github.com/llvm/llvm-project/pull/120130

>From 055e9098cb9264d369d2e6ad29700379b9c9b9c0 Mon Sep 17 00:00:00 2001
From: Peter Klausler <pklausler at nvidia.com>
Date: Mon, 16 Dec 2024 11:29:33 -0800
Subject: [PATCH] [flang] Handle substring in data statement constant

The case of a constant substring wasn't handled in the parser
for data statement constants.

Fixes https://github.com/llvm/llvm-project/issues/119005.
---
 flang/include/flang/Parser/parse-tree.h | 7 ++++---
 flang/lib/Parser/Fortran-parsers.cpp    | 7 +++++--
 flang/lib/Parser/expr-parsers.cpp       | 2 +-
 flang/lib/Parser/type-parsers.h         | 1 +
 flang/test/Parser/lit-substr-data.f90   | 7 +++++++
 5 files changed, 18 insertions(+), 6 deletions(-)
 create mode 100644 flang/test/Parser/lit-substr-data.f90

diff --git a/flang/include/flang/Parser/parse-tree.h b/flang/include/flang/Parser/parse-tree.h
index 1d97126d17dbc4..f87a1cfceb37ba 100644
--- a/flang/include/flang/Parser/parse-tree.h
+++ b/flang/include/flang/Parser/parse-tree.h
@@ -1481,9 +1481,10 @@ struct DataStmtConstant {
   UNION_CLASS_BOILERPLATE(DataStmtConstant);
   CharBlock source;
   mutable TypedExpr typedExpr;
-  std::variant<LiteralConstant, SignedIntLiteralConstant,
-      SignedRealLiteralConstant, SignedComplexLiteralConstant, NullInit,
-      common::Indirection<Designator>, StructureConstructor>
+  std::variant<common::Indirection<CharLiteralConstantSubstring>,
+      LiteralConstant, SignedIntLiteralConstant, SignedRealLiteralConstant,
+      SignedComplexLiteralConstant, NullInit, common::Indirection<Designator>,
+      StructureConstructor>
       u;
 };
 
diff --git a/flang/lib/Parser/Fortran-parsers.cpp b/flang/lib/Parser/Fortran-parsers.cpp
index a3d2c363108073..aa0a2a6db7d588 100644
--- a/flang/lib/Parser/Fortran-parsers.cpp
+++ b/flang/lib/Parser/Fortran-parsers.cpp
@@ -929,8 +929,11 @@ TYPE_PARSER(construct<DataStmtRepeat>(intLiteralConstant) ||
 // components can be ambiguous with a scalar-constant-subobject.
 // So we parse literal constants, designator, null-init, and
 // structure-constructor, so that semantics can figure things out later
-// with the symbol table.
-TYPE_PARSER(sourced(first(construct<DataStmtConstant>(literalConstant),
+// with the symbol table.  A literal constant substring must be attempted
+// first to avoid a partial match with a literal constant.
+TYPE_PARSER(sourced(first(
+    construct<DataStmtConstant>(indirect(charLiteralConstantSubstring)),
+    construct<DataStmtConstant>(literalConstant),
     construct<DataStmtConstant>(signedRealLiteralConstant),
     construct<DataStmtConstant>(signedIntLiteralConstant),
     extension<LanguageFeature::SignedComplexLiteral>(
diff --git a/flang/lib/Parser/expr-parsers.cpp b/flang/lib/Parser/expr-parsers.cpp
index 77a13de7fd02d8..0b6e21e1ba2ff2 100644
--- a/flang/lib/Parser/expr-parsers.cpp
+++ b/flang/lib/Parser/expr-parsers.cpp
@@ -68,7 +68,7 @@ TYPE_PARSER(construct<AcImpliedDoControl>(
 // type-param-inquiry is parsed as a structure component, except for
 // substring%KIND/LEN
 constexpr auto primary{instrumented("primary"_en_US,
-    first(construct<Expr>(indirect(Parser<CharLiteralConstantSubstring>{})),
+    first(construct<Expr>(indirect(charLiteralConstantSubstring)),
         construct<Expr>(literalConstant),
         construct<Expr>(construct<Expr::Parentheses>("(" >>
             expr / !","_tok / recovery(")"_tok, SkipPastNested<'(', ')'>{}))),
diff --git a/flang/lib/Parser/type-parsers.h b/flang/lib/Parser/type-parsers.h
index d7e0cd06c3f444..623437f9d2e1d8 100644
--- a/flang/lib/Parser/type-parsers.h
+++ b/flang/lib/Parser/type-parsers.h
@@ -63,6 +63,7 @@ constexpr Parser<KindParam> kindParam; // R709
 constexpr Parser<RealLiteralConstant> realLiteralConstant; // R714
 constexpr Parser<CharLength> charLength; // R723
 constexpr Parser<CharLiteralConstant> charLiteralConstant; // R724
+constexpr Parser<CharLiteralConstantSubstring> charLiteralConstantSubstring;
 constexpr Parser<Initialization> initialization; // R743 & R805
 constexpr Parser<DerivedTypeSpec> derivedTypeSpec; // R754
 constexpr Parser<TypeDeclarationStmt> typeDeclarationStmt; // R801
diff --git a/flang/test/Parser/lit-substr-data.f90 b/flang/test/Parser/lit-substr-data.f90
new file mode 100644
index 00000000000000..7eed616a1ee2ec
--- /dev/null
+++ b/flang/test/Parser/lit-substr-data.f90
@@ -0,0 +1,7 @@
+!RUN: %flang_fc1 -fdebug-unparse %s 2>&1 | FileCheck %s
+!Regression test for bug #119005
+character*2 :: ary4
+!CHECK: DATA ary4/"cd"/
+data ary4/"abcdef"(3:4)/
+end
+