operate on all detected hardlinks

Fileinfo.cc

@@ -166,11 +166,19 @@ Fileinfo::Fileinfostat::Fileinfostat()
 int
 Fileinfo::deletefile()
 {
-  const int ret = unlink(name().c_str());
-  if (ret) {
-    std::cerr << "Failed deleting file " << name() << '\n';
+  std::vector<std::string> allfiles;
+  allfiles.push_back(m_filename);
+  allfiles.insert(allfiles.end(), m_aliases.begin(), m_aliases.end());
+
+  for (const std::string& n : allfiles) {
+    const int ret = unlink(n.c_str());
+
+    if (ret) {
+      std::cerr << "Failed deleting file " << n << '\n';
+      return ret;
+    }
   }
-  return ret;
+  return 0;
 }
 
 namespace {
@@ -267,39 +275,56 @@ transactional_operation(const std::string& filename, const Func& f)
 int
 Fileinfo::makesymlink(const Fileinfo& A)
 {
-  const int retval =
-    transactional_operation(name(), [&](const std::string& filename) {
-      // The tricky thing is that the path must be correct, as seen from
-      // the directory where *this is. Making the path absolute solves this
-      // problem. Doing string manipulations to find how to make the path
-      // relative is prone to error because directories can be symlinks.
-      std::string target = A.name();
-      makeAbsolute(target);
-      // clean up the path, so it does not contain sequences "/./" or "//"
-      simplifyPath(target);
-      return symlink(target.c_str(), filename.c_str());
-    });
-
-  if (retval) {
-    std::cerr << "Failed to make symlink " << name() << " to " << A.name()
-              << '\n';
+  std::vector<std::string> allfiles;
+  allfiles.push_back(m_filename);
+  allfiles.insert(allfiles.end(), m_aliases.begin(), m_aliases.end());
+
+  // The tricky thing is that the path must be correct, as seen from
+  // the directory where *this is. Making the path absolute solves this
+  // problem. Doing string manipulations to find how to make the path
+  // relative is prone to error because directories can be symlinks.
+  std::string target = A.name();
+  makeAbsolute(target);
+  // clean up the path, so it does not contain sequences "/./" or "//"
+  simplifyPath(target);
+
+  for (const std::string& n : allfiles) {
+    const int retval =
+      transactional_operation(n, [&target](const std::string& filename) {
+        return symlink(target.c_str(), filename.c_str());
+      });
+
+    if (retval) {
+      std::cerr << "Failed to make symlink " << n << " to " << A.name()
+                << '\n';
+      return retval;
+    }
   }
-  return retval;
+  return 0;
 }
 
 // makes a hard link that points to A
 int
 Fileinfo::makehardlink(const Fileinfo& A)
 {
-  return transactional_operation(name(), [&](const std::string& filename) {
-    // make a hardlink.
-    const int retval = link(A.name().c_str(), filename.c_str());
+  std::vector<std::string> allfiles;
+  allfiles.push_back(m_filename);
+  allfiles.insert(allfiles.end(), m_aliases.begin(), m_aliases.end());
+
+  for (const std::string& n : allfiles) {
+    const int retval =
+      transactional_operation(n, [&A](const std::string& filename) {
+        // make a hardlink.
+        return link(A.name().c_str(), filename.c_str());
+      });
+
     if (retval) {
-      std::cerr << "Failed to make hardlink " << filename << " to " << A.name()
+      std::cerr << "Failed to make hardlink " << n << " to " << A.name()
                 << '\n';
+      return retval;
     }
-    return retval;
-  });
+  }
+  return 0;
 }
 
 int

Fileinfo.hh

@@ -10,6 +10,7 @@
 #include <array>
 #include <cstdint>
 #include <string>
+#include <vector>
 
 // os specific headers
 #include <sys/types.h> //for off_t and others.
@@ -127,6 +128,12 @@ public:
   // gets the filename
   const std::string& name() const { return m_filename; }
 
+  // gets the filename
+  const std::vector<std::string>& aliases() const { return m_aliases; }
+
+  // adds a filename alias
+  void add_alias(std::string n) { m_aliases.push_back(n); }
+
   // gets the command line index this item was found at
   int get_cmdline_index() const { return m_cmdline_index; }
 
@@ -174,6 +181,9 @@ private:
   // to be deleted or not
   bool m_delete;
 
+  // list of hardlinks
+  std::vector<std::string> m_aliases;
+
   duptype m_duptype;
 
   // If two files are found to be identical, the one with highest ranking is

Rdutil.cc

@@ -42,14 +42,19 @@ Rdutil::printtofile(const std::string& filename) const
   // This uses "priority" instead of "cmdlineindex". Change this the day
   // a change in output format is allowed (for backwards compatibility).
   output << "# Automatically generated\n";
-  output << "# duptype id depth size device inode priority name\n";
+  output << "# duptype id depth size device inode priority name(s)\n";
 
   std::vector<Fileinfo>::iterator it;
   for (it = m_list.begin(); it != m_list.end(); ++it) {
     output << Fileinfo::getduptypestring(*it) << " " << it->getidentity() << " "
            << it->depth() << " " << it->size() << " " << it->device() << " "
-           << it->inode() << " " << it->get_cmdline_index() << " " << it->name()
-           << '\n';
+           << it->inode() << " " << it->get_cmdline_index() << " " << it->name();
+    if (!it->aliases().empty()) {
+      output << " aliases:";
+      for (const std::string& n : it->aliases())
+        output << " " << n;
+    }
+    output << '\n';
   }
   output << "# end of file\n";
   f1.close();
@@ -313,9 +318,9 @@ Rdutil::removeIdenticalInodes()
       // let the highest-ranking element not be deleted. do this in order, to be
       // cache friendly.
       auto best = std::min_element(first, last, cmpRank);
-      std::for_each(first, best, [](Fileinfo& f) { f.setdeleteflag(true); });
+      std::for_each(first, last, [](Fileinfo& f) { f.setdeleteflag(true); });
       best->setdeleteflag(false);
-      std::for_each(best + 1, last, [](Fileinfo& f) { f.setdeleteflag(true); });
+      std::for_each(first, last, [&best](Fileinfo& f) { if (f.deleteflag()) best->add_alias(f.name()); });
     });
   return cleanup();
 }