feat: Copy/archive input XML files into the output directory

src/coreComponents/dataRepository/unitTests/testXmlWrapper.cpp

@@ -18,6 +18,9 @@
 #include "dataRepository/xmlWrapper.hpp"
 #include "common/format/EnumStrings.hpp"
 
+#include <filesystem>
+#include <fstream>
+
 using namespace geos;
 
 TEST( testXmlWrapper, array3d_errors )
@@ -501,6 +504,214 @@ TEST( testXmlWrapper, testGroupNamesArrayFormats )
   }
 }
 
+class CollectIncludedTest : public ::testing::Test
+{
+protected:
+  std::filesystem::path m_tempDir;
+
+  void SetUp() override
+  {
+    m_tempDir = std::filesystem::temp_directory_path() / "geos_collectIncluded_test";
+    std::filesystem::create_directories( m_tempDir );
+  }
+
+  void TearDown() override
+  {
+    std::filesystem::remove_all( m_tempDir );
+  }
+
+  string filePath( string const & filename )
+  {
+    return ( m_tempDir / filename ).string();
+  }
+
+  void writeXML( string const & filename, string const & content )
+  {
+    std::ofstream f( filePath( filename ) );
+    f << content;
+  }
+};
+
+
+TEST_F( CollectIncludedTest, collectIncluded_noIncludes )
+{
+  writeXML( "base.xml", "<Problem>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncluded( filePath( "base.xml" ) );
+
+  EXPECT_TRUE( result.empty() );
+}
+
+TEST_F( CollectIncludedTest, collectIncluded_singleInclude )
+{
+  writeXML( "child.xml", "<Problem>"
+                         "</Problem>" );
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"child.xml\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncluded( filePath( "base.xml" ) );
+
+  EXPECT_NE( result.find( filePath( "child.xml" ) ), result.end() );
+}
+
+TEST_F( CollectIncludedTest, collectIncluded_multipleIncludes )
+{
+  writeXML( "child1.xml", "<Problem>"
+                          "</Problem>" );
+  writeXML( "child2.xml", "<Problem>"
+                          "</Problem>" );
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"child1.xml\"/>"
+                        "   <File name=\"child2.xml\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncluded( filePath( "base.xml" ) );
+
+  EXPECT_NE( result.find( filePath( "child1.xml" ) ), result.end() );
+  EXPECT_NE( result.find( filePath( "child2.xml" ) ), result.end() );
+}
+
+TEST_F( CollectIncludedTest, collectIncluded_emptyNameAttribute )
+{
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  std::set< string > result;
+
+  EXPECT_ANY_THROW( xmlWrapper::collectIncluded( filePath( "base.xml" ), result ) );
+
+  EXPECT_TRUE( result.empty() );
+}
+
+TEST_F( CollectIncludedTest, collectIncluded_existingEntriesKept )
+{
+  std::set< string > existingCollection;
+  existingCollection.insert( "/somewhere/thereisanalreadyexistingxmlfile.xml" );
+  writeXML( "base.xml", "<Problem>"
+                        "</Problem>" );
+
+  xmlWrapper::collectIncluded( filePath( "base.xml" ), existingCollection );
+
+  EXPECT_NE( existingCollection.find( "/somewhere/thereisanalreadyexistingxmlfile.xml" ),
+             existingCollection.end() );
+}
+
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_noIncludes )
+{
+  writeXML( "base.xml", "<Problem>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncludedRecursive( filePath( "base.xml" ) );
+
+  EXPECT_EQ( result.size(), 1 );  // size 1 because collectIncludedRecursive collects the base file
+}
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_singleInclude )
+{
+  writeXML( "child.xml", "<Problem>"
+                         "</Problem>" );
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"child.xml\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncludedRecursive( filePath( "base.xml" ) );
+
+  EXPECT_NE( result.find( filePath( "child.xml" ) ), result.end() );
+}
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_multipleIncludes )
+{
+  writeXML( "child1.xml", "<Problem>"
+                          "</Problem>" );
+  writeXML( "child2.xml", "<Problem>"
+                          "</Problem>" );
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"child1.xml\"/>"
+                        "   <File name=\"child2.xml\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncludedRecursive( filePath( "base.xml" ) );
+
+  EXPECT_NE( result.find( filePath( "child1.xml" ) ), result.end() );
+  EXPECT_NE( result.find( filePath( "child2.xml" ) ), result.end() );
+}
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_simpleRecursive )
+{
+  writeXML( "child.xml", "<Problem>"
+                         "</Problem>" );
+
+  writeXML( "middle.xml", "<Problem>"
+                          " <Included>"
+                          "   <File name=\"child.xml\"/>"
+                          " </Included>"
+                          "</Problem>" );
+
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"middle.xml\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncludedRecursive( filePath( "base.xml" ) );
+
+  EXPECT_NE( result.find( filePath( "middle.xml" ) ), result.end() );
+  EXPECT_NE( result.find( filePath( "child.xml" ) ), result.end() );
+}
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_cyclePrevention )
+{
+  writeXML( "cycle.xml", "<Problem>"
+                         " <Included>"
+                         "   <File name=\"cycle.xml\"/>"
+                         " </Included>"
+                         "</Problem>" );
+
+  auto result = xmlWrapper::collectIncludedRecursive( filePath( "cycle.xml" ) );
+
+  EXPECT_NE( result.find( filePath( "cycle.xml" ) ), result.end() );
+}
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_noDuplicates )
+{
+  writeXML( "base.xml", "<Problem>"
+                        " <Included>"
+                        "   <File name=\"base.xml\"/>"
+                        " </Included>"
+                        "</Problem>" );
+
+  auto result = xmlWrapper::collectIncludedRecursive( filePath( "base.xml" ) );
+
+  EXPECT_EQ( result.size(), 1 );  // collectIncludedRecursive collects the base file
+}
+
+TEST_F( CollectIncludedTest, collectIncludedRecursive_existingEntriesKept )
+{
+  std::set< string > existingCollection;
+  existingCollection.insert( "/somewhere/thereisanalreadyexistingxmlfile.xml" );
+  writeXML( "base.xml", "<Problem>"
+                        "</Problem>" );
+
+  xmlWrapper::collectIncludedRecursive( filePath( "base.xml" ), existingCollection );
+
+  EXPECT_NE( existingCollection.find( "/somewhere/thereisanalreadyexistingxmlfile.xml" ),
+             existingCollection.end() );
+}
+
 
 int main( int argc, char * argv[] )
 {

src/coreComponents/dataRepository/xmlWrapper.cpp

@@ -24,6 +24,7 @@
 #include "common/format/StringUtilities.hpp"
 #include "common/MpiWrapper.hpp"
 #include "dataRepository/KeyNames.hpp"
+#include "common/Path.hpp"
 
 namespace geos
 {
@@ -268,6 +269,91 @@ string buildMultipleInputXML( string_array const & inputFileList,
   return inputFileName;
 }
 
+void collectIncluded( string const & filePath,
+                      std::set< string > & collection )
+{
+  xmlDocument doc;
+  xmlResult result = doc.loadFile( filePath );
+  GEOS_THROW_IF( !result,
+                 GEOS_FMT( "Could not load XML file '{}': {}", filePath, result.description() ),
+                 InputError );
+  xmlNode rootNode = doc.getFirstChild();
+
+  string const currentDir = splitPath( filePath ).first;
+
+  for( auto & includedNode : rootNode.children( includedListTag ) )
+  {
+    for( auto & fileNode : includedNode.children( includedFileTag ) )
+    {
+      string const fileName = fileNode.attribute( "name" ).value();
+
+      GEOS_THROW_IF( fileName.empty(),
+                     GEOS_FMT( "An included file entry in '{}' has an empty or missing 'name' attribute.", filePath ),
+                     InputError );
+
+      string absolutePath = isAbsolutePath( fileName )
+                            ? getAbsolutePath( fileName )
+                            : getAbsolutePath( joinPath( currentDir, fileName ) );
+      collection.insert( absolutePath );
+    }
+  }
+}
+
+std::set< string > collectIncluded( string const & filePath )
+{
+  std::set< string > collection;
+  collectIncluded( filePath, collection );
+  return collection;
+}
+
+void collectIncludedRecursive( string const & filePath,
+                               std::set< string > & collection )
+{
+  // We want absolute paths
+  string const absFilePath = getAbsolutePath( filePath );
+
+  if( collection.count( absFilePath ) > 0 )
+  {
+    return;
+  }
+  collection.insert( absFilePath );
+
+  xmlDocument doc;
+  xmlResult result = doc.loadFile( absFilePath );
+  GEOS_THROW_IF( !result,
+                 GEOS_FMT( "Could not load XML file '{}': {}", filePath, result.description() ),
+                 InputError );
+  xmlNode rootNode = doc.getFirstChild();
+
+  string const currentDir = splitPath( filePath ).first;
+
+  for( auto & includedNode : rootNode.children( includedListTag ) )
+  {
+    for( auto & fileNode : includedNode.children( includedFileTag ) )
+    {
+      string const includedFilePath = fileNode.attribute( "name" ).value();
+
+      if( includedFilePath.empty() )
+      {
+        continue;
+      }
+
+      string includedAbsPath = isAbsolutePath( includedFilePath )
+                             ? getAbsolutePath( includedFilePath )
+                             : getAbsolutePath( joinPath( currentDir, includedFilePath ) );
+      collectIncludedRecursive( includedAbsPath,
+                                collection );
+    }
+  }
+}
+
+std::set< string > collectIncludedRecursive( string const & filePath )
+{
+  std::set< string > collection;
+  collectIncludedRecursive( filePath, collection );
+  return collection;
+}
+
 bool isFileMetadataAttribute( string const & name )
 {
   static const std::set< string > fileMetadataAttributes {

src/coreComponents/dataRepository/xmlWrapper.hpp

@@ -301,6 +301,50 @@ constexpr char const includedFileTag[] = "File";
 string buildMultipleInputXML( string_array const & inputFileList,
                               string const & outputDir = {} );
 
+/**
+ * @brief Collect the absolute paths of XML files directly included
+ *        by a given xml file
+ * @param[in] filePath absolute path of the xml file to inspect
+ * @param[inout] collection collection to append with absolute file paths
+ *
+ * Only one level of inclusion is collected (files included by the included
+ * files are not added). See collectIncludedRecursive if you want this behavior.
+ * Duplicate entries are not inserted in @p collection
+ */
+void collectIncluded( string const & filePath,
+                      std::set< string > & collection );
+
+/**
+ * @brief Collect the absolute paths of XML files directly included
+ *        by a given xml file
+ * @param[in] filePath absolute path of the xml file to inspect
+ * @return a collection of absolute paths
+ *
+ * Only one level of inclusion is collected (files included by the included
+ * files are not added). See collectIncludedRecursive if you want this behavior.
+ * Duplicate entries are not inserted in @p collection
+ */
+std::set< string > collectIncluded( string const & filePath );
+
+/**
+ * @brief Recursively collect the absolute paths of an XML file and all XML
+ *        files it includes
+ * @param[in] filePath absolute path of the root XML file
+ * @param[inout] collection collection to append with absolute file paths
+ *                          of every visited file (including @p filePath itself)
+ */
+void collectIncludedRecursive( string const & filePath,
+                               std::set< string > & collection );
+
+/**
+ * @brief Recursively collect the absolute paths of an XML file and all XML
+ *        files it includes
+ * @param[in] filePath absolute path of the root XML file
+ * @return a collection of absolute paths of every visited file (including
+ *         @p filePath itself)
+ */
+std::set< string > collectIncludedRecursive( string const & filePath );
+
 /**
  * @return true if the attribute with the specified name declares metadata relative to the xml
  * @param name the name of an attribute

src/coreComponents/fileIO/CMakeLists.txt

@@ -24,6 +24,7 @@ Contains:
 #
 set( fileIO_headers
      LogLevelsInfo.hpp
+     Outputs/ArchiveInputDeck.hpp
      Outputs/BlueprintOutput.hpp
      Outputs/MemoryStatsOutput.hpp
      Outputs/OutputBase.hpp
@@ -43,6 +44,7 @@ set( fileIO_headers
 # Specify all sources
 #
 set( fileIO_sources
+     Outputs/ArchiveInputDeck.cpp
      Outputs/BlueprintOutput.cpp
      Outputs/MemoryStatsOutput.cpp
      Outputs/OutputBase.cpp