From 3ee81f48e96031f02bc5caab533b0dd1994f9717 Mon Sep 17 00:00:00 2001
From: Enver Haase <enver.haase@infraleap.com>
Date: Mon, 4 May 2020 11:49:02 +0200
Subject: [PATCH] Introduce -protectsametree option useful e.g. if you want to
 delete duplicates from deviated backup copy from your original tree (where
 you may want to keep duplicates) in order to let only the deviated / added
 files survive.

---
 Rdutil.cc | 21 ++++++++++++---------
 Rdutil.hh |  5 +++--
 rdfind.cc |  7 ++++++-
 3 files changed, 21 insertions(+), 12 deletions(-)

diff --git a/Rdutil.cc b/Rdutil.cc
index f098d2c..6c552f0 100644
--- a/Rdutil.cc
+++ b/Rdutil.cc
@@ -61,7 +61,7 @@ Rdutil::printtofile(const std::string& filename) const
 // returns how many times the function was invoked.
 template<typename Function>
 std::size_t
-applyactiononfile(std::vector<Fileinfo>& m_list, Function f)
+applyactiononfile(std::vector<Fileinfo>& m_list, bool m_protectSameTree, Function f)
 {
 
   const auto first = m_list.begin();
@@ -79,9 +79,12 @@ applyactiononfile(std::vector<Fileinfo>& m_list, Function f)
                "original file should have positive identity");
       } break;
 
-      case Fileinfo::duptype::DUPTYPE_OUTSIDE_TREE:
+      case Fileinfo::duptype::DUPTYPE_WITHIN_SAME_TREE:
+        if (m_protectSameTree) {
+          break;
+        }
         // intentional fallthrough
-      case Fileinfo::duptype::DUPTYPE_WITHIN_SAME_TREE: {
+      case Fileinfo::duptype::DUPTYPE_OUTSIDE_TREE: {
         assert(original != last);
         // double check that "it" shall be ~linked to "src"
         assert(it->getidentity() == -original->getidentity() &&
@@ -140,11 +143,11 @@ Rdutil::deleteduplicates(bool dryrun) const
   if (dryrun) {
     const bool outputBname = false;
     dryrun_helper<outputBname> obj("delete ");
-    auto ret = applyactiononfile(m_list, obj);
+    auto ret = applyactiononfile(m_list, m_protectSameTree, obj);
     std::cout.flush();
     return ret;
   } else {
-    return applyactiononfile(m_list, &Fileinfo::static_deletefile);
+    return applyactiononfile(m_list, m_protectSameTree, &Fileinfo::static_deletefile);
   }
 }
 
@@ -154,11 +157,11 @@ Rdutil::makesymlinks(bool dryrun) const
   if (dryrun) {
     const bool outputBname = true;
     dryrun_helper<outputBname> obj("symlink ", " to ");
-    auto ret = applyactiononfile(m_list, obj);
+    auto ret = applyactiononfile(m_list, m_protectSameTree, obj);
     std::cout.flush();
     return ret;
   } else {
-    return applyactiononfile(m_list, &Fileinfo::static_makesymlink);
+    return applyactiononfile(m_list, m_protectSameTree, &Fileinfo::static_makesymlink);
   }
 }
 
@@ -168,11 +171,11 @@ Rdutil::makehardlinks(bool dryrun) const
   if (dryrun) {
     const bool outputBname = true;
     dryrun_helper<outputBname> obj("hardlink ", " to ");
-    const auto ret = applyactiononfile(m_list, obj);
+    const auto ret = applyactiononfile(m_list, m_protectSameTree, obj);
     std::cout.flush();
     return ret;
   } else
-    return applyactiononfile(m_list, &Fileinfo::static_makehardlink);
+    return applyactiononfile(m_list, m_protectSameTree, &Fileinfo::static_makehardlink);
 }
 
 // mark files with a unique number
diff --git a/Rdutil.hh b/Rdutil.hh
index b39e2e9..d4967b6 100644
--- a/Rdutil.hh
+++ b/Rdutil.hh
@@ -16,8 +16,8 @@
 class Rdutil
 {
 public:
-  explicit Rdutil(std::vector<Fileinfo>& list)
-    : m_list(list){};
+  explicit Rdutil(std::vector<Fileinfo>& list, bool protectSameTree)
+    : m_list(list), m_protectSameTree(protectSameTree) {};
 
   /**
    * print file names to a file, with extra information.
@@ -121,6 +121,7 @@ public:
 
 private:
   std::vector<Fileinfo>& m_list;
+  bool m_protectSameTree;
 };
 
 #endif
diff --git a/rdfind.cc b/rdfind.cc
index facdda7..4c66589 100644
--- a/rdfind.cc
+++ b/rdfind.cc
@@ -72,6 +72,8 @@ usage()
     << " -outputname  name  sets the results file name to \"name\" "
        "(default results.txt)\n"
     << " -deleteduplicates  true |(false) delete duplicate files\n"
+    << " -protectsametree   true |(false) do not touch duplicates found in "
+       "the same tree as the original\n"
     << " -sleep              Xms          sleep for X milliseconds between "
        "file reads.\n"
     << "                                  Default is 0. Only a few values\n"
@@ -99,6 +101,7 @@ struct Options
   Fileinfo::filesizetype maximumfilesize =
     0; // if nonzero, files this size or larger are ignored
   bool deleteduplicates = false;      // delete duplicate files
+  bool protectsametree = false;       // do not touch duplicates in same tree
   bool followsymlinks = false;        // follow symlinks
   bool dryrun = false;                // only dryrun, dont destroy anything
   bool remove_identical_inode = true; // remove files with identical inodes
@@ -156,6 +159,8 @@ parseOptions(Parser& parser)
       o.maximumfilesize = maxsize;
     } else if (parser.try_parse_bool("-deleteduplicates")) {
       o.deleteduplicates = parser.get_parsed_bool();
+    } else if (parser.try_parse_bool("-protectsametree")) {
+      o.protectsametree = parser.get_parsed_bool();
     } else if (parser.try_parse_bool("-followsymlinks")) {
       o.followsymlinks = parser.get_parsed_bool();
     } else if (parser.try_parse_bool("-dryrun")) {
@@ -286,7 +291,7 @@ main(int narg, const char* argv[])
   const std::string dryruntext(o.dryrun ? "(DRYRUN MODE) " : "");
 
   // an object to do sorting and duplicate finding
-  Rdutil gswd(filelist);
+  Rdutil gswd(filelist, o.protectsametree);
 
   // an object to traverse the directory structure
   Dirlist dirlist(o.followsymlinks);