logic: sort custom indexer commands by indexed file size and name

src/lib/data/indexer/TaskExecuteCustomCommands.cpp

@@ -16,6 +16,7 @@
 #include "TextAccess.h"
 #include "utility.h"
 #include "utilityApp.h"
+#include "utilityFile.h"
 #include "utilityString.h"
 
 TaskExecuteCustomCommands::TaskExecuteCustomCommands(
@@ -42,7 +43,7 @@ void TaskExecuteCustomCommands::doEnter(std::shared_ptr<Blackboard> blackboard)
 
 	if (m_indexerCommandProvider)
 	{
-		for (const FilePath& sourceFilePath : m_indexerCommandProvider->getAllSourceFilePaths())
+		for (const FilePath& sourceFilePath : utility::partitionFilePathsBySize(m_indexerCommandProvider->getAllSourceFilePaths(), 2))
 		{
 			if (std::shared_ptr<IndexerCommandCustom> indexerCommand =
 				std::dynamic_pointer_cast<IndexerCommandCustom>(m_indexerCommandProvider->consumeCommandForSourceFilePath(sourceFilePath)))
@@ -67,6 +68,9 @@ void TaskExecuteCustomCommands::doEnter(std::shared_ptr<Blackboard> blackboard)
 				}
 			}
 		}
+		// reverse because we pull elements from the back of these vectors
+		std::reverse(m_parallelCommands.begin(), m_parallelCommands.end()); 
+		std::reverse(m_serialCommands.begin(), m_serialCommands.end());
 	}
 }
 

src/lib/data/indexer/TaskFillIndexerCommandQueue.cpp

@@ -1,9 +1,10 @@
 #include "TaskFillIndexerCommandQueue.h"
 
-#include "IndexerCommandProvider.h"
-#include "FileSystem.h"
-#include "logging.h"
 #include "Blackboard.h"
+#include "FileSystem.h"
+#include "IndexerCommandProvider.h"
+#include "logging.h"
+#include "utilityFile.h"
 
 TaskFillIndexerCommandsQueue::TaskFillIndexerCommandsQueue(
 	const std::string& appUUID,
@@ -18,48 +19,11 @@ TaskFillIndexerCommandsQueue::TaskFillIndexerCommandsQueue(
 
 void TaskFillIndexerCommandsQueue::doEnter(std::shared_ptr<Blackboard> blackboard)
 {
-	typedef std::pair<unsigned long long int, FilePath> PairType;
-	std::vector<PairType> sourceFileSizesToCommands;
-	{
-		std::vector<FilePath> allSourceFilePaths;
-		{
-			std::lock_guard<std::mutex> lock(m_commandsMutex);
-			allSourceFilePaths = m_indexerCommandProvider->getAllSourceFilePaths();
-		}
-
-		for (const FilePath& path : allSourceFilePaths)
-		{
-			if (path.exists())
-			{
-				sourceFileSizesToCommands.push_back(std::make_pair(FileSystem::getFileByteSize(path), path));
-			}
-			else
-			{
-				sourceFileSizesToCommands.push_back(std::make_pair(1, path));
-			}
-		}
-		std::sort(sourceFileSizesToCommands.begin(), sourceFileSizesToCommands.end(), [](const PairType& p, const PairType& q) { return p.first > q.first; });
-
-		if (sourceFileSizesToCommands.size() > 2)
-		{
-			std::sort(
-				sourceFileSizesToCommands.begin(),
-				sourceFileSizesToCommands.begin() + sourceFileSizesToCommands.size() / 2,
-				[](const PairType& p, const PairType& q) { return p.second.wstr() < q.second.wstr(); }
-			);
-			std::sort(
-				sourceFileSizesToCommands.begin() + sourceFileSizesToCommands.size() / 2,
-				sourceFileSizesToCommands.end(),
-				[](const PairType& p, const PairType& q) { return p.second.wstr() < q.second.wstr(); }
-			);
-		}
-	}
-
 	{
 		std::lock_guard<std::mutex> lock(m_commandsMutex);
-		for (const PairType &pair : sourceFileSizesToCommands)
+		for (const FilePath& filePath : utility::partitionFilePathsBySize(m_indexerCommandProvider->getAllSourceFilePaths(), 2))
 		{
-			m_filePathQueue.emplace(pair.second);
+			m_filePathQueue.emplace(filePath);
 		}
 	}
 

src/lib/utility/utilityFile.cpp

@@ -1,9 +1,48 @@
 #include "utilityFile.h"
 
 #include "FilePath.h"
+#include "FileSystem.h"
 
 #include "utility.h"
 
+std::vector<FilePath> utility::partitionFilePathsBySize(std::vector<FilePath> filePaths, int partitionCount)
+{
+	typedef std::pair<unsigned long long int, FilePath> PairType;
+	std::vector<PairType> sourceFileSizesToCommands;
+	for (const FilePath& path : filePaths)
+	{
+		if (path.exists())
+		{
+			sourceFileSizesToCommands.push_back(std::make_pair(FileSystem::getFileByteSize(path), path));
+		}
+		else
+		{
+			sourceFileSizesToCommands.push_back(std::make_pair(1, path));
+		}
+	}
+
+	std::sort(sourceFileSizesToCommands.begin(), sourceFileSizesToCommands.end(), [](const PairType& p, const PairType& q) { return p.first > q.first; });
+
+	if (0 < partitionCount && partitionCount < sourceFileSizesToCommands.size())
+	{
+		for (int i = 0; i < partitionCount; i++)
+		{
+			std::sort(
+				sourceFileSizesToCommands.begin() + sourceFileSizesToCommands.size() * i / partitionCount,
+				sourceFileSizesToCommands.begin() + sourceFileSizesToCommands.size() * (i + 1) / partitionCount,
+				[](const PairType& p, const PairType& q) { return p.second.wstr() < q.second.wstr(); }
+			);
+		}
+	}
+
+	std::vector<FilePath> sortedFilePaths;
+	for (const PairType &pair : sourceFileSizesToCommands)
+	{
+		sortedFilePaths.push_back(pair.second);
+	}
+	return sortedFilePaths;
+}
+
 std::vector<FilePath> utility::getTopLevelPaths(const std::vector<FilePath>& paths)
 {
 	return utility::getTopLevelPaths(utility::toSet(paths));

src/lib/utility/utilityFile.h

@@ -8,6 +8,8 @@ class FilePath;
 
 namespace utility
 {
+	std::vector<FilePath> partitionFilePathsBySize(std::vector<FilePath> filePaths, int partitionCount = 0);
+
 	std::vector<FilePath> getTopLevelPaths(const std::vector<FilePath>& paths);
 	std::vector<FilePath> getTopLevelPaths(const std::set<FilePath>& paths);