initial commit

sgutil/sgu_mode_rand_remove.cpp

@@ -0,0 +1,123 @@
+#include "sgu_stdafx.h"
+#include "sgu_mode_rand_remove.h"
+#include "sgu_utils.h"
+
+bool randRemovePartOfFile(const char* inputFileName, const char* outputFileName, int64_t minSize, int64_t maxSize, int32_t counts)
+{
+	if (!inputFileName || inputFileName[0] == 0 || !outputFileName || outputFileName[0] == 0)
+	{
+		printf("Invalid input or output file name.\n");
+		return false;
+	}
+
+	if (minSize == 0 || maxSize == 0 || minSize > maxSize)
+	{
+		printf("Invalid size parameters.\n");
+		return false;
+	}
+
+	// Check if input and output file names are the same
+	if (compareTwoFileName(inputFileName, outputFileName) == 0)
+	{
+		printf("Input and output file names cannot be the same.\n");
+		return false;
+	}
+
+	// Open the input file in binary mode
+	FILE* inputFile;
+	fopen_s(&inputFile, inputFileName, "rb");
+	if (!inputFile)
+	{
+		printf("Error opening input file:%s\n", inputFileName);
+		return false;
+	}
+
+	// Open the output file in binary mode
+	FILE* outputFile;
+	fopen_s(&outputFile, outputFileName, "wb");
+	if (!outputFile)
+	{
+		printf("Error opening output file:%s", outputFileName);
+		fclose(inputFile);
+		return false;
+	}
+
+	// Get the size of the input file
+	fseek(inputFile, 0, SEEK_END);
+	int64_t inputFileSize = (int64_t)ftell(inputFile);
+	fseek(inputFile, 0, SEEK_SET);
+
+	if (maxSize * counts > inputFileSize)
+	{
+		printf("Error: The specified size range(%d*%lld) and count exceed the input file size[%lld].\n",
+			counts, maxSize, inputFileSize);
+		fclose(inputFile);
+		fclose(outputFile);
+		return false;
+	}
+
+	std::vector<std::pair< int64_t, int64_t>> segmentArray;
+	generateRandomSegment(inputFileSize, minSize, maxSize, counts, segmentArray);
+
+	int64_t totalSize = 0;
+	printf("Input file size: %lld bytes\n", inputFileSize);
+	printf("Remove %d segments, size range: %lld ~ %lld bytes\n", counts, minSize, maxSize);
+	for (auto& p : segmentArray)
+	{
+		totalSize += p.second;
+		printf("Remove segment: offset=%lld~%lld, size=%lld\n", p.first, p.first + p.second, p.second);
+	}
+	printf("Total size to remove: %lld bytes\n", totalSize);
+
+	//Temp memory
+	std::vector<uint8_t> buffer;
+
+	size_t offset = 0;
+
+	//read and modify each segment
+	for (const auto& segment : segmentArray)
+	{
+		size_t segmentOffset = segment.first;
+		size_t segmentSize = segment.second;
+		// Read file content before segment
+		buffer.resize(segmentOffset - offset);
+		fseek(inputFile, (long)offset, SEEK_SET);
+		size_t bytesRead = fread(buffer.data(), 1, segmentOffset - offset, inputFile);
+		if (bytesRead != segmentOffset - offset)
+		{
+			printf("Error reading segment before offset %zu, expected %zu bytes, got %zu bytes.\n",
+				offset, segmentOffset - offset, bytesRead);
+			fclose(inputFile);
+			fclose(outputFile);
+			return false;
+		}
+
+		//write to output file
+		fwrite(buffer.data(), 1, segmentOffset - offset, outputFile);
+
+		offset = segmentOffset + segmentSize;
+	}
+
+	//read and write the remaining part of the file
+	buffer.resize(inputFileSize - offset);
+	fseek(inputFile, (long)offset, SEEK_SET);
+	size_t bytesRead = fread(buffer.data(), 1, inputFileSize - offset, inputFile);
+	if (bytesRead != inputFileSize - offset)
+	{
+		printf("Error reading segment before offset %zu, expected %zu bytes, got %zu bytes.\n",
+			offset, inputFileSize - offset, bytesRead);
+		fclose(inputFile);
+		fclose(outputFile);
+		return false;
+	}
+
+	//write to output file
+	fwrite(buffer.data(), 1, inputFileSize - offset, outputFile);
+
+	// Close both files
+	fclose(inputFile);
+	fclose(outputFile);
+
+	return true;
+}
+