proper cmake

src/commands/CMakeLists.txt

@@ -0,0 +1,7 @@
+cmake_minimum_required(VERSION 3.22)
+
+add_library(commands srcs/Command.cpp srcs/CommandDiff.cpp srcs/CommandList.cpp srcs/CommandListFiles.cpp srcs/CommandRestore.cpp srcs/CommandRun.cpp srcs/CommandsCommon.cpp srcs/Diff.cpp)
+
+target_include_directories(commands PUBLIC includes)
+
+target_link_libraries(commands crypto repo chunkers utils change_detectors)

src/commands/includes/Command.h

@@ -5,7 +5,7 @@
 #ifndef SEMBACKUP_COMMAND_H
 #define SEMBACKUP_COMMAND_H
 
-#include "../Context.h"
+#include "Context.h"
 
 /// Abstract base class for some process running with some Context
 class Command {

src/commands/includes/CommandRestore.h

@@ -7,7 +7,7 @@
 
 #include "Command.h"
 
-#include "../repo/objects/File.h"
+#include "../../repo/includes/objects/File.h"
 
 #include "CommandsCommon.h"
 

src/commands/includes/Context.h

@@ -0,0 +1,18 @@
+//
+// Created by Stepan Usatiuk on 23.05.2023.
+//
+
+#ifndef SEMBACKUP_CONTEXT_H
+#define SEMBACKUP_CONTEXT_H
+
+#include "../../repo/includes/Repository.h"
+#include "../../utils/includes/Config.h"
+#include "../../utils/includes/Logger.h"
+
+struct Context {
+    Logger *logger;
+    Repository *repo;
+};
+
+
+#endif//SEMBACKUP_CONTEXT_H

src/commands/includes/Diff.h

@@ -0,0 +1,38 @@
+//
+// Created by Stepan Usatiuk on 06.05.2023.
+//
+
+#ifndef SEMBACKUP_DIFF_H
+#define SEMBACKUP_DIFF_H
+
+#include <sstream>
+#include <string>
+
+#include "../../change_detectors/includes/ComparableFile.h"
+
+/// Utility class to compute difference between two ComparableFile%s
+class Diff {
+public:
+    /// Compute the difference between two ComparableFile%s
+    /// If the file is binary, calls diffPercent, which outputs the difference between files in bytes
+    /// Otherwise prints linewise difference
+    /// \param c1 Constant reference to the first ComparableFile
+    /// \param c2 Constant reference to the second ComparableFile
+    /// \returns  Difference message
+    static std::string diff(const ComparableFile &c1, const ComparableFile &c2);
+
+    /// Calculates the difference between \p c1 amd \p c2 in bytes
+    /// \param c1 Constant reference to the first ComparableFile
+    /// \param c2 Constant reference to the second ComparableFile
+    /// \returns  Difference message
+    static std::string diffPercent(const ComparableFile &c1, const ComparableFile &c2);
+
+    /// Checks if a file is binary
+    /// A file is considered binary if its first 2048 bytes contain a null byte
+    /// \param c1 Constant reference to the checked ComparableFile
+    /// \return True if the file is considered binary, false otherwise
+    static bool isBinary(const ComparableFile &c1);
+};
+
+
+#endif//SEMBACKUP_DIFF_H

src/commands/srcs/Command.cpp

@@ -2,7 +2,7 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "Command.h"
+#include "../includes/Command.h"
 
 Command::Command(std::string name) : name(std::move(name)) {}
 

src/commands/srcs/CommandDiff.cpp

@@ -2,20 +2,20 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "CommandDiff.h"
+#include "../includes/CommandDiff.h"
 
-#include "../BytesFormatter.h"
-#include "../Diff.h"
-#include "../Exception.h"
-#include "../Progress.h"
-#include "../RunningDiffAverage.h"
-#include "../Signals.h"
-#include "../ThreadPool.h"
-#include "../change_detectors/ChangeDetectorFactory.h"
-#include "../chunkers/ChunkerFactory.h"
-#include "../repo/Serialize.h"
-#include "../repo/objects/Archive.h"
-#include "../repo/objects/Chunk.h"
+#include "../../change_detectors/includes/ChangeDetectorFactory.h"
+#include "../../chunkers/includes/ChunkerFactory.h"
+#include "../../repo/includes/Serialize.h"
+#include "../../repo/includes/objects/Archive.h"
+#include "../../repo/includes/objects/Chunk.h"
+#include "../../utils/includes/BytesFormatter.h"
+#include "../../utils/includes/Exception.h"
+#include "../../utils/includes/Progress.h"
+#include "../../utils/includes/RunningDiffAverage.h"
+#include "../../utils/includes/Signals.h"
+#include "../../utils/includes/ThreadPool.h"
+#include "../includes/Diff.h"
 
 using namespace CommandsCommon;
 

src/commands/srcs/CommandList.cpp

@@ -2,7 +2,7 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "CommandList.h"
+#include "../includes/CommandList.h"
 
 CommandList::CommandList() : Command("list") {
 }

src/commands/srcs/CommandListFiles.cpp

@@ -2,13 +2,13 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "CommandListFiles.h"
+#include "../includes/CommandListFiles.h"
 
-#include "../BytesFormatter.h"
-#include "../repo/Serialize.h"
-#include "../repo/objects/Archive.h"
-#include "../repo/objects/Chunk.h"
-#include "../repo/objects/File.h"
+#include "../../repo/includes/Serialize.h"
+#include "../../repo/includes/objects/Archive.h"
+#include "../../repo/includes/objects/Chunk.h"
+#include "../../repo/includes/objects/File.h"
+#include "../../utils/includes/BytesFormatter.h"
 
 CommandListFiles::CommandListFiles() : Command("list-files") {
 }

src/commands/srcs/CommandRestore.cpp

@@ -2,21 +2,21 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "CommandRestore.h"
+#include "../includes/CommandRestore.h"
 
 #include <fstream>
 #include <sstream>
 
-#include "../BytesFormatter.h"
-#include "../Exception.h"
-#include "../Progress.h"
-#include "../RunningDiffAverage.h"
-#include "../Signals.h"
-#include "../ThreadPool.h"
-#include "../chunkers/ChunkerFactory.h"
-#include "../repo/Serialize.h"
-#include "../repo/objects/Archive.h"
-#include "../repo/objects/Chunk.h"
+#include "../../chunkers/includes/ChunkerFactory.h"
+#include "../../repo/includes/Serialize.h"
+#include "../../repo/includes/objects/Archive.h"
+#include "../../repo/includes/objects/Chunk.h"
+#include "../../utils/includes/BytesFormatter.h"
+#include "../../utils/includes/Exception.h"
+#include "../../utils/includes/Progress.h"
+#include "../../utils/includes/RunningDiffAverage.h"
+#include "../../utils/includes/Signals.h"
+#include "../../utils/includes/ThreadPool.h"
 
 using namespace CommandsCommon;
 

src/commands/srcs/CommandRun.cpp

@@ -2,27 +2,27 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "CommandRun.h"
+#include "../includes/CommandRun.h"
 
 #include <fstream>
 #include <iomanip>
 #include <sstream>
 
-#include "../BytesFormatter.h"
-#include "../Exception.h"
-#include "../Progress.h"
-#include "../RunningDiffAverage.h"
-#include "../Signals.h"
-#include "../ThreadPool.h"
-#include "../change_detectors/ChangeDetectorFactory.h"
-#include "../chunkers/ChunkerFactory.h"
-#include "../crypto/MD5.h"
-#include "../repo/Serialize.h"
-#include "../repo/objects/Archive.h"
-#include "../repo/objects/Chunk.h"
-#include "../repo/objects/File.h"
+#include "../../change_detectors/includes/ChangeDetectorFactory.h"
+#include "../../chunkers/includes/ChunkerFactory.h"
+#include "../../crypto/includes/MD5.h"
+#include "../../repo/includes/Serialize.h"
+#include "../../repo/includes/objects/Archive.h"
+#include "../../repo/includes/objects/Chunk.h"
+#include "../../repo/includes/objects/File.h"
+#include "../../utils/includes/BytesFormatter.h"
+#include "../../utils/includes/Exception.h"
+#include "../../utils/includes/Progress.h"
+#include "../../utils/includes/RunningDiffAverage.h"
+#include "../../utils/includes/Signals.h"
+#include "../../utils/includes/ThreadPool.h"
 
-#include "CommandsCommon.h"
+#include "../includes/CommandsCommon.h"
 
 using namespace CommandsCommon;
 

src/commands/srcs/CommandsCommon.cpp

@@ -2,13 +2,13 @@
 // Created by Stepan Usatiuk on 23.05.2023.
 //
 
-#include "CommandsCommon.h"
+#include "../includes/CommandsCommon.h"
 
 #include <fstream>
 #include <regex>
 
-#include "../Exception.h"
-#include "../Signals.h"
+#include "../../utils/includes/Exception.h"
+#include "../../utils/includes/Signals.h"
 
 void CommandsCommon::workerCallback(unsigned long long int bytesWritten, unsigned long long int bytesSkipped, unsigned long long int filesWritten, WorkerStats &to) {
     to.bytesWritten += bytesWritten;

src/commands/srcs/Diff.cpp

@@ -0,0 +1,103 @@
+//
+// Created by Stepan Usatiuk on 06.05.2023.
+//
+
+#include "../includes/Diff.h"
+
+#include "../../chunkers/includes/BuzhashChunker.h"
+#include "../../utils/includes/BytesFormatter.h"
+#include "../../utils/includes/Exception.h"
+#include "../../utils/includes/Signals.h"
+
+bool Diff::isBinary(const ComparableFile &c) {
+    auto b = c.contents();
+    for (unsigned int i = 0; i < std::min(c.bytes, 2048ULL); i++) {
+        auto e = b->sbumpc();
+        if (std::streambuf::traits_type::to_char_type(e) == '\0') return true;
+        if (e == std::streambuf::traits_type::eof()) return false;
+    }
+    return false;
+}
+
+std::string Diff::diff(const ComparableFile &c1, const ComparableFile &c2) {
+    if (isBinary(c1) || isBinary(c2)) {
+        if (!(isBinary(c1) && isBinary(c2))) return "One of the files is binary, the other is not";
+        return diffPercent(c1, c2);
+    }
+
+    std::stringstream out;
+    auto b1 = c1.contents();
+    auto b2 = c2.contents();
+    std::multimap<std::string, unsigned long> f1lines;
+    std::multimap<std::string, unsigned long> f2diff;
+    std::string line;
+    std::istream is1(b1.get());
+    std::istream is2(b2.get());
+
+    int i = 0;
+    while (std::getline(is1, line)) {
+        /// Exit when asked to
+        if (Signals::shouldQuit) throw Exception("Quitting");
+        f1lines.emplace(line, ++i);
+    }
+
+    i = 0;
+    while (std::getline(is2, line)) {
+        /// Exit when asked to
+        if (Signals::shouldQuit) throw Exception("Quitting");
+        if (f1lines.count(line) > 0) f1lines.erase(f1lines.find(line));
+        else
+            f2diff.emplace(line, ++i);
+    }
+
+    out << "\nLines only in first file: " << std::endl;
+    for (const auto &s: f1lines) {
+        out << s.second << "<" << s.first << std::endl;
+    }
+    out << "Lines only in second file: " << std::endl;
+    for (const auto &s: f2diff) {
+        out << s.second << ">" << s.first << std::endl;
+    }
+    out << "^^^\n";
+    return out.str();
+}
+
+std::string Diff::diffPercent(const ComparableFile &c1, const ComparableFile &c2) {
+    auto b1 = c1.contents();
+    auto b2 = c2.contents();
+    BuzhashChunker ch1(b1.get(), 512 * 1024, 1024 * 1024, 19, 31);
+    BuzhashChunker ch2(b2.get(), 512 * 1024, 1024 * 1024, 19, 31);
+    std::multiset<std::string> ch1hashes;
+    std::multiset<std::string> ch2diff;
+    std::unordered_map<std::string, unsigned long long> hashsize;
+    for (auto chunkp: ch1) {
+        /// Exit when asked to
+        if (Signals::shouldQuit) throw Exception("Quitting");
+        if (chunkp.second.empty()) continue;
+        std::string md5(chunkp.first.begin(), chunkp.first.end());
+        ch1hashes.emplace(md5);
+        hashsize[md5] = chunkp.second.size();
+    }
+
+    for (auto chunkp: ch2) {
+        /// Exit when asked to
+        if (Signals::shouldQuit) throw Exception("Quitting");
+        if (chunkp.second.empty()) continue;
+        std::string md5(chunkp.first.begin(), chunkp.first.end());
+        hashsize[md5] = chunkp.second.size();
+        if (ch1hashes.count(md5) > 0) ch1hashes.erase(md5);
+        else if (ch1hashes.count(md5) == 0)
+            ch2diff.emplace(md5);
+    }
+
+    unsigned long long diff = 0;
+
+    for (const auto &c: ch1hashes) {
+        diff += hashsize[c];
+    }
+    for (const auto &c: ch2diff) {
+        diff += hashsize[c];
+    }
+
+    return "at most " + BytesFormatter::formatStr(diff);
+}