Implement metadata handling in pcmaudio categorizer

src/dwarfs/categorizer/pcmaudio_categorizer.cpp

@@ -30,6 +30,8 @@
 
 #include <fmt/format.h>
 
+#include <folly/Synchronized.h>
+
 #include "dwarfs/categorizer.h"
 #include "dwarfs/error.h"
 #include "dwarfs/logger.h"
@@ -40,8 +42,76 @@ namespace po = boost::program_options;
 
 namespace {
 
-constexpr std::string_view const AIFF_CATEGORY{"aiff"};
+constexpr std::string_view const PCMAUDIO_CATEGORY{"pcmaudio"};
-constexpr std::string_view const WAV_CATEGORY{"wav"};
+
+enum class endianness : uint8_t {
+  BIG,
+  LITTLE,
+};
+enum class signedness : uint8_t {
+  SIGNED,
+  UNSIGNED,
+};
+
+char const* endianness_string(endianness e) {
+  switch (e) {
+  case endianness::BIG:
+    return "big";
+  case endianness::LITTLE:
+    return "little";
+  }
+}
+
+char const* signedness_string(signedness s) {
+  switch (s) {
+  case signedness::SIGNED:
+    return "signed";
+  case signedness::UNSIGNED:
+    return "unsigned";
+  }
+}
+
+struct pcmaudio_metadata {
+  endianness sample_endianness;
+  signedness sample_signedness;
+  uint8_t bits_per_sample;
+  uint16_t number_of_channels;
+
+  //// Sample rate should be irrelevant
+  // uint32_t samples_per_second;
+
+  auto operator<=>(pcmaudio_metadata const&) const = default;
+};
+
+class pcmaudio_metadata_store {
+ public:
+  pcmaudio_metadata_store() = default;
+
+  size_t add(pcmaudio_metadata const& m) {
+    auto it = reverse_index_.find(m);
+    if (it == reverse_index_.end()) {
+      auto r = reverse_index_.emplace(m, forward_index_.size());
+      assert(r.second);
+      forward_index_.emplace_back(m);
+      it = r.first;
+    }
+    return it->second;
+  }
+
+  folly::dynamic lookup(size_t ix) const {
+    auto const& m = DWARFS_NOTHROW(forward_index_.at(ix));
+    folly::dynamic obj = folly::dynamic::object;
+    obj.insert("endianness", endianness_string(m.sample_endianness));
+    obj.insert("signedness", signedness_string(m.sample_signedness));
+    obj.insert("bits_per_sample", m.bits_per_sample);
+    obj.insert("number_of_channels", m.number_of_channels);
+    return obj;
+  }
+
+ private:
+  std::vector<pcmaudio_metadata> forward_index_;
+  std::map<pcmaudio_metadata, size_t> reverse_index_;
+};
 
 class pcmaudio_categorizer_base : public random_access_categorizer {
  public:
@@ -60,20 +130,20 @@ class pcmaudio_categorizer_ final : public pcmaudio_categorizer_base {
 
   bool is_single_fragment() const override { return false; }
 
-  folly::dynamic category_metadata(fragment_category) const override {
+  folly::dynamic category_metadata(fragment_category c) const override {
-    // TODO
+    DWARFS_CHECK(c.has_subcategory(), "expected pcmaudio to have subcategory");
-    return folly::dynamic();
+    return meta_.rlock()->lookup(c.subcategory());
   }
 
  private:
   LOG_PROXY_DECL(LoggerPolicy);
+  folly::Synchronized<pcmaudio_metadata_store> mutable meta_;
 };
 
 std::span<std::string_view const>
 pcmaudio_categorizer_base::categories() const {
   static constexpr std::array const s_categories{
-      AIFF_CATEGORY,
+      PCMAUDIO_CATEGORY,
-      WAV_CATEGORY,
   };
   return s_categories;
 }