From df7b136f245cf6beee7572eeca2f9f5cdc82ad7c Mon Sep 17 00:00:00 2001
From: Danny Robson <danny@nerdcruft.net>
Date: Fri, 26 Apr 2019 09:25:41 +1000
Subject: [PATCH] tools/compare: record the source of the matched blocks

---
 tools/compare.cpp | 59 ++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 46 insertions(+), 13 deletions(-)

diff --git a/tools/compare.cpp b/tools/compare.cpp
index 66a2c44..a9ab558 100644
--- a/tools/compare.cpp
+++ b/tools/compare.cpp
@@ -32,7 +32,7 @@ namespace emory {
         static_hash::digest_t digest;
 
         struct digest_equality {
-            bool operator() (chunk const &a, chunk const &b)
+            bool operator() (chunk const &a, chunk const &b) const
             {
                 return a.digest == b.digest;
             }
@@ -40,7 +40,7 @@ namespace emory {
 
 
         struct digest_ordering {
-            bool operator() (chunk const &a, chunk const &b)
+            bool operator() (chunk const &a, chunk const &b) const
             {
                 return a.digest < b.digest;
             }
@@ -50,23 +50,35 @@ namespace emory {
 
     template <typename IdT>
     struct match {
-        struct {
+        struct side {
             IdT id;
             chunk data;
         } src, dst;
 
+
         struct src_equality {
-            bool operator() (match const &a, match const &b)
+            bool operator() (match const &a, match const &b) const
             {
                 return a.src == b.src;
             }
         };
 
 
-        struct src_ordering {
-            bool operator() (match const &a, match const &b)
+        struct dst_equality {
+            bool operator() (match const &a, match const &b) const
             {
-                return a.src < b.src;
+                return chunk::digest_equality {} (
+                    a.dst.data,
+                    b.dst.data
+                );
+            }
+        };
+
+
+        struct src_ordering {
+            bool operator() (match const &a, match const &b) const
+            {
+                return chunk::digest_ordering {} (a.src.data, b.src.data);
             }
         };
     };
@@ -189,7 +201,7 @@ int main (int argc, char const **argv)
     emory::map target (cruft::mapped_file (argv[ARGS_TARGET]), p);
     std::sort (target.elements.begin (), target.elements.end (), emory::chunk::digest_ordering {});
 
-    std::vector<emory::chunk> found;
+    std::vector<emory::match<int>> found;
 
     for (int i = ARGS_SOURCE; i != argc; ++i) {
         auto const path = argv[i];
@@ -205,21 +217,39 @@ int main (int argc, char const **argv)
             std::begin (source_matches),
             std::end   (source_matches),
             std::back_inserter (found),
-            [] (auto const &i) { return i.src.data; }
+            [&] (auto const &j) -> emory::match<int>
+        {
+            return {
+                .src = {           i, j.src.data },
+                .dst = { ARGS_TARGET, j.dst.data },
+            };
+        });
+
+        std::sort (
+            std::begin (found),
+            std::end   (found),
+            emory::match<int>::src_ordering {}
         );
 
-        std::sort (std::begin (found), std::end (found), emory::chunk::digest_ordering {});
         found.erase (
-            std::unique (std::begin (found), std::end (found), emory::chunk::digest_equality {}),
+            std::unique (
+                std::begin (found),
+                std::end   (found),
+                emory::match<int>::dst_equality {}
+            ),
             found.end ()
         );
     }
 
+
     std::clog << "Finalising\n";
     std::size_t matching = 0;
+    std::vector<std::size_t> source_bytes (argc - ARGS_SOURCE, 0);
     for (auto const &i: found) {
-        std::cout << i << '\n';
-        matching += i.offset.second - i.offset.first;
+        std::cout << i.dst.data << '\n';
+        auto const size = i.dst.data.offset.second - i.dst.data.offset.first;
+        matching += size;
+        source_bytes[i.src.id - ARGS_SOURCE] += size;
     }
 
     std::size_t const total = std::accumulate (
@@ -230,4 +260,7 @@ int main (int argc, char const **argv)
     );
 
     std::cout << "Found " << found.size () << " chunks. " << matching << "/" << total << " bytes for a factor of " << float (matching) / total << "\n";
+
+    for (int i = ARGS_SOURCE; i != argc; ++i)
+        std::cerr << argv[i] << " contributed: " << source_bytes[i - ARGS_SOURCE] << '\n';
 }