From 09d285d94d844311e2d691ddb2f9a4b512eec72b Mon Sep 17 00:00:00 2001
From: mrbesen <y.g.2@gmx.de>
Date: Tue, 8 Nov 2022 21:51:29 +0100
Subject: [PATCH] more options

---
 inc/dedup.h         |  2 ++
 inc/fileindexer.h   |  4 ++++
 src/dedup.cpp       | 18 ++++++++++++------
 src/fileindexer.cpp | 22 ++++++++++++++++------
 4 files changed, 34 insertions(+), 12 deletions(-)

diff --git a/inc/dedup.h b/inc/dedup.h
index 184a831..ae6a9c1 100644
--- a/inc/dedup.h
+++ b/inc/dedup.h
@@ -35,6 +35,8 @@ private:
 	bool hardlink = true;
 	bool ignoredotfiles = false;
 	bool dryrun = true;
+	bool deleteDuplicates = true;
+	bool ignoreInodeID = true;
 	uint64_t deduplicatedBytes = 0; // amount of bytes deduplicated
 	uint64_t deduplicatedFiles = 0;
 };
diff --git a/inc/fileindexer.h b/inc/fileindexer.h
index 0ac040e..6ccf03c 100644
--- a/inc/fileindexer.h
+++ b/inc/fileindexer.h
@@ -16,10 +16,14 @@ public:
 
 	using map_t = std::multimap<uint64_t, std::shared_ptr<File>>;
 	map_t index(const SearchFolder& sf);
+
+	void setIgnoreDotFiles(bool b = true);
+	void setIgnoreInodeIDs(bool b = true);
 private:
 	void handleFolderContent(dirent* dircont, const SearchFolder& sf, std::list<SearchFolder>& newfolders, map_t& newfiles);
 	void handleNewFile(ino_t inode, const std::string& path, map_t& newfiles);
 
 	std::set<uint64_t> knownInodes; // file inodes, that are known and should not be indexed again
 	bool ignoredotfiles = false;
+	bool ignoreInodeID = false;
 };
diff --git a/src/dedup.cpp b/src/dedup.cpp
index 653051c..dfcae59 100644
--- a/src/dedup.cpp
+++ b/src/dedup.cpp
@@ -38,6 +38,8 @@ Files& getFiles(uint64_t fs, std::map<uint64_t, Files>& m) {
 
 void Dedup::start() {
 	FileIndexer::map_t foundfiles;
+	indexer.setIgnoreDotFiles(ignoredotfiles);
+	indexer.setIgnoreInodeIDs(ignoreInodeID);
 	for(const SearchFolder& sf : folders) {
 		foundfiles.merge(indexer.index(sf));
 	}
@@ -192,7 +194,9 @@ bool Dedup::relinkFile(const std::string& linkbase, const std::string& replacedf
 
 	if(dryrun || !hardlink) {
 		Log::note << "delete " << replacedfile;
-		Log::note << "link " << linkbase << " -> " << replacedfile;
+		if(!deleteDuplicates) {
+			Log::note << "link " << linkbase << " -> " << replacedfile;
+		}
 	} else {
 		int res = ::unlink(replacedfile.c_str());
 		if(res != 0) {
@@ -200,11 +204,13 @@ bool Dedup::relinkFile(const std::string& linkbase, const std::string& replacedf
 			return false;
 		}
 
-		res = ::link(linkbase.c_str(), replacedfile.c_str());
-		if(res != 0) {
-			Log::error << "link(" << linkbase << ", " << replacedfile << ") failed: " << strerror(errno) << " (" << errno << ')';
-			// TODO try to symlink?
-			return false;
+		if(!deleteDuplicates) {
+			res = ::link(linkbase.c_str(), replacedfile.c_str());
+			if(res != 0) {
+				Log::error << "link(" << linkbase << ", " << replacedfile << ") failed: " << strerror(errno) << " (" << errno << ')';
+				// TODO try to symlink?
+				return false;
+			}
 		}
 	}
 	return true;
diff --git a/src/fileindexer.cpp b/src/fileindexer.cpp
index cce6092..a740662 100644
--- a/src/fileindexer.cpp
+++ b/src/fileindexer.cpp
@@ -35,6 +35,13 @@ std::multimap<uint64_t, std::shared_ptr<File>> FileIndexer::index(const SearchFo
 	return out;
 }
 
+void FileIndexer::setIgnoreDotFiles(bool b) {
+	ignoredotfiles = b;
+}
+
+void FileIndexer::setIgnoreInodeIDs(bool b) {
+	ignoreInodeID = b;
+}
 
 void FileIndexer::handleFolderContent(dirent* dircont, const SearchFolder& sf, std::list<SearchFolder>& newfolders, map_t& newfiles) {
 	std::string name(dircont->d_name);
@@ -62,10 +69,12 @@ void FileIndexer::handleFolderContent(dirent* dircont, const SearchFolder& sf, s
 
 void FileIndexer::handleNewFile(ino_t inode, const std::string& path, map_t& newfiles) {
 	// check for already scanned inodes
-	auto it = knownInodes.find(inode);
-	if(it != knownInodes.end()) {
-		Log::note << "found already detected file: " << inode << ' ' << path;
-		return;
+	if(!ignoreInodeID) {
+		auto it = knownInodes.find(inode);
+		if(it != knownInodes.end()) {
+			Log::note << "found already detected file: " << inode << ' ' << path;
+			return;
+		}
 	}
 
 	struct stat statbuf;
@@ -78,8 +87,9 @@ void FileIndexer::handleNewFile(ino_t inode, const std::string& path, map_t& new
 	loff_t fileSize = statbuf.st_size;
 	nlink_t linkCount = statbuf.st_nlink;
 
-
-	knownInodes.insert(inode);
+	if(!ignoreInodeID) {
+		knownInodes.insert(inode);
+	}
 
 	newfiles.insert({fileSize, std::make_shared<File>(fileSize, inode, linkCount, path)});
 }