Trying to get rid of duplicated audio on sf64

3 files changed, 41 insertions, 2 deletions

diff --git a/src/factories/naudio/v1/AudioContext.cpp b/src/factories/naudio/v1/AudioContext.cpp
index 82d034a..45beef5 100644
--- a/src/factories/naudio/v1/AudioContext.cpp
+++ b/src/factories/naudio/v1/AudioContext.cpp
@@ -4,6 +4,8 @@
 
 std::unordered_map<AudioTableType, TableEntry> AudioContext::tables;
 NAudioDrivers AudioContext::driver = NAudioDrivers::UNKNOWN;
+std::unordered_map<uint64_t, std::string> AudioContext::sampleDedup;
+std::unordered_map<uint32_t, std::string> AudioContext::sampleAddrRemap;
 
 std::optional<std::shared_ptr<IParsedData>> AudioContextFactory::parse(std::vector<uint8_t>& buffer, YAML::Node& node) {
     auto driver = GetSafeNode<std::string>(node, "driver");
@@ -83,6 +85,14 @@ uint64_t AudioContext::GetPathByAddr(uint32_t addr) {
         return 0;
     }
 
+    // If this ROM address belongs to a duplicate SampleData struct, return the
+    // canonical's hash so instruments/drums reference the canonical entry directly.
+    auto remapIt = sampleAddrRemap.find(addr);
+    if (remapIt != sampleAddrRemap.end()) {
+        SPDLOG_INFO("Remapping duplicate 0x{:X} to canonical {}", addr, remapIt->second);
+        return CRC64(remapIt->second.c_str());
+    }
+
     auto dec = Companion::Instance->GetNodeByAddr(addr);
     if (dec.has_value()) {
         std::string path = std::get<0>(dec.value());
diff --git a/src/factories/naudio/v1/AudioContext.h b/src/factories/naudio/v1/AudioContext.h
index 415d1f1..05e5516 100644
--- a/src/factories/naudio/v1/AudioContext.h
+++ b/src/factories/naudio/v1/AudioContext.h
@@ -3,6 +3,8 @@
 #include "AudioTableFactory.h"
 #include "factories/naudio/v0/AudioHeaderFactory.h"
 #include "factories/BaseFactory.h"
+#include <unordered_map>
+#include <string>
 
 struct TunedSample {
     uint32_t sample;
@@ -25,6 +27,13 @@ class AudioContext {
 public:
     static std::unordered_map<AudioTableType, TableEntry> tables;
     static NAudioDrivers driver;
+    // Key: (sampleBankId << 32) | sampleAddr → canonical archive path.
+    // Populated during parse so all dedup decisions are made before any export runs.
+    static std::unordered_map<uint64_t, std::string> sampleDedup;
+    // Key: ROM offset of a duplicate SampleData struct → canonical archive path.
+    // Used by GetPathByAddr() so instrument/drum export writes the canonical hash
+    // instead of the duplicate's hash, making redirect entries permanently unreferenced.
+    static std::unordered_map<uint32_t, std::string> sampleAddrRemap;
 
     static LUS::BinaryReader MakeReader(AudioTableType type, uint32_t offset);
     static TunedSample LoadTunedSample(LUS::BinaryReader& reader, uint32_t parent, uint32_t sampleBankId);
diff --git a/src/factories/naudio/v1/SampleFactory.cpp b/src/factories/naudio/v1/SampleFactory.cpp
index 2faafe2..4616d3f 100644
--- a/src/factories/naudio/v1/SampleFactory.cpp
+++ b/src/factories/naudio/v1/SampleFactory.cpp
@@ -28,9 +28,10 @@ ExportResult NSampleCodeExporter::Export(std::ostream& write, std::shared_ptr<IP
 
 ExportResult NSampleBinaryExporter::Export(std::ostream& write, std::shared_ptr<IParsedData> raw,
                                            std::string& entryName, YAML::Node& node, std::string* replacement) {
-    auto writer = LUS::BinaryWriter();
+    // parse() returns nullopt for duplicates, so Export() is only reached for
+    // canonical samples — no redirect check needed here.
     auto data = std::static_pointer_cast<NSampleData>(raw);
-
+    auto writer = LUS::BinaryWriter();
     WriteHeader(writer, Torch::ResourceType::Sample, 1);
     writer.Write((uint8_t)data->codec);
     writer.Write((uint8_t)data->medium);
@@ -190,5 +191,24 @@ std::optional<std::shared_ptr<IParsedData>> NSampleFactory::parse(std::vector<ui
     sample->sampleBankId = sampleBankId;
     sample->sampleRate = sampleRate;
 
+    // Build dedup maps at parse time so GetPathByAddr() has full information
+    // before any instrument/drum export runs.
+    if (addr != 0) {
+        uint64_t key = ((uint64_t)sampleBankId << 32) | (uint64_t)addr;
+        auto it = AudioContext::sampleDedup.find(key);
+        if (it == AudioContext::sampleDedup.end()) {
+            // First time we see this audio data: this struct is the canonical.
+            auto pathDec = Companion::Instance->GetNodeByAddr(offset);
+            if (pathDec.has_value()) {
+                AudioContext::sampleDedup[key] = std::get<0>(pathDec.value());
+            }
+        } else {
+            // Duplicate: map this ROM offset → canonical path for GetPathByAddr(),
+            // then return nullopt so the Companion produces no archive entry for it.
+            AudioContext::sampleAddrRemap[offset] = it->second;
+            return std::nullopt;
+        }
+    }
+
     return sample;
 }