[flang] Catch (and fix) attempts to create an invalid source range for a Scope

[klausler]

Summary

When Scope::AddSourceRange() is called to extend the scope's source range to
include another snippet from a cooked character stream, add a check to ensure
that the new range is part of the same cooked character stream as the rest
of the scope.

And fix the bug that was causing such invalid source ranges to be created:
a submodule's Scope is a children of its parent's in the Scope tree, but
it may or may not be part of the same source file, and it is certainly
not enclosed in the parent's source range. So don't propagate Scope
source range expansion from a submodule to its parent.

[jeanPerier]

Looks good

flang/include/flang/Semantics/scope.h

void set_instantiationContext(parser::Message::Reference &&mref) {
     instantiationContext_ = std::move(mref);
   }
 
   bool hasSAVE() const { return hasSAVE_; }
   void set_hasSAVE(bool yes = true) { hasSAVE_ = yes; }
 
   // The range of the source of this and nested scopes.
   const parser::CharBlock &sourceRange() const { return sourceRange_; }
-  void AddSourceRange(const parser::CharBlock &);
+  void AddSourceRange(parser::CharBlock);
   // Find the smallest scope under this one that contains source
   const Scope *FindScope(parser::CharBlock) const;
   Scope *FindScope(parser::CharBlock);
 
   // Attempts to find a match for a derived type instance
   const DeclTypeSpec *FindInstantiatedDerivedType(const DerivedTypeSpec &,
       DeclTypeSpec::Category = DeclTypeSpec::TypeDerived) const;
 
public:
   }
 
 private:
   Scope &parent_; // this is enclosing scope, not extended derived type base
   const Kind kind_;
   std::size_t size_{0}; // size in bytes
   std::optional<std::size_t> alignment_; // required alignment in bytes
   parser::CharBlock sourceRange_;
+  const parser::CookedSource *cookedSource_{nullptr};
   Symbol *const symbol_; // if not null, symbol_->scope() == this
   std::list<Scope> children_;
   mapType symbols_;
   mapType commonBlocks_;
   std::list<EquivalenceSet> equivalenceSets_;
   mapType crayPointers_;
   std::map<SourceName, common::Reference<Scope>> submodules_;
   std::list<DeclTypeSpec> declTypeSpecs_;

flang/lib/Semantics/scope.cpp

 //===-- lib/Semantics/scope.cpp -------------------------------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "flang/Semantics/scope.h"
 #include "flang/Parser/characters.h"
+#include "flang/Semantics/semantics.h"
 #include "flang/Semantics/symbol.h"
 #include "flang/Semantics/type.h"
 #include "llvm/Support/raw_ostream.h"
 #include <algorithm>
 #include <memory>
 
 namespace Fortran::semantics {
 

 }
 
 const Scope *Scope::FindScope(parser::CharBlock source) const {
   return const_cast<Scope *>(this)->FindScope(source);
 }
 
 Scope *Scope::FindScope(parser::CharBlock source) {
   bool isContained{sourceRange_.Contains(source)};
-  if (!isContained && !IsTopLevel() && !IsModuleFile()) {
+  if (!isContained && !IsTopLevel() && kind_ != Kind::Module) {
     return nullptr;
   }
   for (auto &child : children_) {
     if (auto *scope{child.FindScope(source)}) {
       return scope;
     }
   }
   return isContained && !IsTopLevel() ? this : nullptr;
 }
 
-void Scope::AddSourceRange(const parser::CharBlock &source) {
-  for (auto *scope{this}; !scope->IsGlobal(); scope = &scope->parent()) {
+void Scope::AddSourceRange(parser::CharBlock source) {
+  if (source.empty()) {
+    return;
+  }
+  const parser::AllCookedSources &allCookedSources{context_.allCookedSources()};
+  const parser::CookedSource *cooked{allCookedSources.Find(source)};
+  if (!cooked) {
+    CHECK(context_.IsTempName(source.ToString()));
+    return;
+  }
+  for (auto *scope{this}; !scope->IsTopLevel(); scope = &scope->parent()) {
+    CHECK(scope->sourceRange_.empty() == (scope->cookedSource_ == nullptr));
+    if (!scope->cookedSource_) {
+      scope->cookedSource_ = cooked;
+      scope->sourceRange_ = source;
+    } else if (scope->cookedSource_ == cooked) {
       scope->sourceRange_.ExtendToCover(source);
+    } else {
+      // There's a bug that will be hard to fix; crash informatively
+      const parser::AllSources &allSources{allCookedSources.allSources()};
+      const auto describe{[&](parser::CharBlock src) {
+        if (auto range{allCookedSources.GetProvenanceRange(src)}) {
+          std::size_t offset;
+          if (const parser::SourceFile *
+              file{allSources.GetSourceFile(range->start(), &offset)}) {
+            return "'"s + file->path() + "' at " + std::to_string(offset) +
+                " for " + std::to_string(range->size());
+          } else {
+            return "(GetSourceFile failed)"s;
+          }
+        } else {
+          return "(GetProvenanceRange failed)"s;
+        }
+      }};
+      std::string scopeDesc{describe(scope->sourceRange_)};
+      std::string newDesc{describe(source)};
+      common::die("AddSourceRange would have combined ranges from distinct "
+                  "source files \"%s\" and \"%s\"",
+          scopeDesc.c_str(), newDesc.c_str());
+    }
+    if (scope->IsSubmodule()) {
+      break; // Submodules are child scopes but not contained ranges
+    }
   }
 }
 
 llvm::raw_ostream &operator<<(llvm::raw_ostream &os, const Scope &scope) {
   os << Scope::EnumToString(scope.kind()) << " scope: ";
   if (auto *symbol{scope.symbol()}) {
     os << *symbol << ' ';
   }