From 82aeadc79d1fb4c61bd49af855cef0dbb050ee09 Mon Sep 17 00:00:00 2001
From: Marcus Holland-Moritz <github@mhxnet.de>
Date: Tue, 26 Aug 2025 23:20:42 +0200
Subject: [PATCH] feat: expose block category metadata in `filesystem_v2`

---
 include/dwarfs/reader/filesystem_v2.h        |  7 +++++++
 include/dwarfs/reader/internal/metadata_v2.h |  8 ++++++++
 src/reader/filesystem_v2.cpp                 | 16 ++++++++++++++--
 src/reader/internal/metadata_v2.cpp          | 18 ++++++++++++++++++
 4 files changed, 47 insertions(+), 2 deletions(-)

diff --git a/include/dwarfs/reader/filesystem_v2.h b/include/dwarfs/reader/filesystem_v2.h
index 42b09ef2..541ca2c4 100644
--- a/include/dwarfs/reader/filesystem_v2.h
+++ b/include/dwarfs/reader/filesystem_v2.h
@@ -324,6 +324,11 @@ class filesystem_v2_lite {
     return lite_->get_block_category(block_number);
   }
 
+  std::optional<nlohmann::json>
+  get_block_category_metadata(size_t block_number) const {
+    return lite_->get_block_category_metadata(block_number);
+  }
+
   void cache_blocks_by_category(std::string_view category) const {
     lite_->cache_blocks_by_category(category);
   }
@@ -420,6 +425,8 @@ class filesystem_v2_lite {
     virtual std::vector<file_stat::gid_type> get_all_gids() const = 0;
     virtual std::optional<std::string>
     get_block_category(size_t block_number) const = 0;
+    virtual std::optional<nlohmann::json>
+    get_block_category_metadata(size_t block_number) const = 0;
     virtual void cache_blocks_by_category(std::string_view category) const = 0;
     virtual void cache_all_blocks() const = 0;
     virtual std::shared_ptr<internal::filesystem_parser> get_parser() const = 0;
diff --git a/include/dwarfs/reader/internal/metadata_v2.h b/include/dwarfs/reader/internal/metadata_v2.h
index 5d656983..8c0305f6 100644
--- a/include/dwarfs/reader/internal/metadata_v2.h
+++ b/include/dwarfs/reader/internal/metadata_v2.h
@@ -157,6 +157,11 @@ class metadata_v2 {
     return impl_->get_block_category(block_number);
   }
 
+  std::optional<nlohmann::json>
+  get_block_category_metadata(size_t block_number) const {
+    return impl_->get_block_category_metadata(block_number);
+  }
+
   std::vector<std::string> get_all_block_categories() const {
     return impl_->get_all_block_categories();
   }
@@ -232,6 +237,9 @@ class metadata_v2 {
     virtual std::optional<std::string>
     get_block_category(size_t block_number) const = 0;
 
+    virtual std::optional<nlohmann::json>
+    get_block_category_metadata(size_t block_number) const = 0;
+
     virtual std::vector<std::string> get_all_block_categories() const = 0;
 
     virtual std::vector<file_stat::uid_type> get_all_uids() const = 0;
diff --git a/src/reader/filesystem_v2.cpp b/src/reader/filesystem_v2.cpp
index 6eaaa89e..2a8b3ff3 100644
--- a/src/reader/filesystem_v2.cpp
+++ b/src/reader/filesystem_v2.cpp
@@ -307,6 +307,10 @@ class filesystem_ final {
   std::optional<std::string> get_block_category(size_t block_no) const {
     return meta_.get_block_category(block_no);
   }
+  std::optional<nlohmann::json>
+  get_block_category_metadata(size_t block_no) const {
+    return meta_.get_block_category_metadata(block_no);
+  }
 
   void cache_blocks_by_category(std::string_view category) const {
     auto const max_blocks = get_max_cache_blocks();
@@ -689,7 +693,9 @@ void filesystem_<LoggerPolicy>::dump(std::ostream& os,
 
       std::string metadata;
 
-      if (bd) {
+      if (auto m = meta_.get_block_category_metadata(block_no)) {
+        metadata = fmt::format(", metadata={}", m->dump());
+      } else if (bd) {
         if (auto m = bd->metadata()) {
           metadata = fmt::format(", metadata={}", *m);
         }
@@ -769,7 +775,9 @@ filesystem_<LoggerPolicy>::info_as_json(fsinfo_options const& opts,
         section_info["size"] = uncompressed_size;
         section_info["ratio"] = float(s.length()) / uncompressed_size;
 
-        if (auto m = bd->metadata()) {
+        if (auto m = meta_.get_block_category_metadata(block_no)) {
+          section_info["metadata"] = *m;
+        } else if (auto m = bd->metadata()) {
           section_info["metadata"] = nlohmann::json::parse(*m);
         }
       }
@@ -1328,6 +1336,10 @@ class filesystem_common_ : public Base {
   get_block_category(size_t block_no) const override {
     return fs_.get_block_category(block_no);
   }
+  std::optional<nlohmann::json>
+  get_block_category_metadata(size_t block_no) const override {
+    return fs_.get_block_category_metadata(block_no);
+  }
 
   void cache_blocks_by_category(std::string_view category) const override {
     fs_.cache_blocks_by_category(category);
diff --git a/src/reader/internal/metadata_v2.cpp b/src/reader/internal/metadata_v2.cpp
index 85d97d18..f7a8755b 100644
--- a/src/reader/internal/metadata_v2.cpp
+++ b/src/reader/internal/metadata_v2.cpp
@@ -331,6 +331,19 @@ class metadata_v2_data {
     return std::nullopt;
   }
 
+  std::optional<nlohmann::json>
+  get_block_category_metadata(size_t block_number) const {
+    if (auto meta_json = meta_.category_metadata_json()) {
+      if (auto block_cat_meta = meta_.block_category_metadata()) {
+        if (auto it = block_cat_meta->find(block_number);
+            it != block_cat_meta->end()) {
+          return nlohmann::json::parse(meta_json.value()[it->second()]);
+        }
+      }
+    }
+    return std::nullopt;
+  }
+
   std::vector<std::string> get_all_block_categories() const {
     std::vector<std::string> rv;
 
@@ -2192,6 +2205,11 @@ class metadata_ final : public metadata_v2::impl {
     return data_.get_block_category(block_number);
   }
 
+  std::optional<nlohmann::json>
+  get_block_category_metadata(size_t block_number) const override {
+    return data_.get_block_category_metadata(block_number);
+  }
+
   std::vector<std::string> get_all_block_categories() const override {
     return data_.get_all_block_categories();
   }