Add metadata export to dwarfsck

CMakeLists.txt

@@ -332,6 +332,7 @@ add_library(
   ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp2/protocol/BinaryProtocol.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp2/protocol/DebugProtocol.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp2/protocol/JSONProtocolCommon.cpp
+  ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp2/protocol/JSONProtocol.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp/protocol/TProtocolException.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp/util/VarintUtils.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/fbthrift/thrift/lib/cpp2/gen/module_types_cpp.cpp

include/dwarfs/filesystem_v2.h

@@ -74,6 +74,10 @@ class filesystem_v2 {
     return impl_->metadata_as_dynamic();
   }
 
+  std::string serialize_metadata_as_json(bool simple) const {
+    return impl_->serialize_metadata_as_json(simple);
+  }
+
   void walk(std::function<void(entry_view)> const& func) const {
     impl_->walk(func);
   }
@@ -134,6 +138,7 @@ class filesystem_v2 {
 
     virtual void dump(std::ostream& os, int detail_level) const = 0;
     virtual folly::dynamic metadata_as_dynamic() const = 0;
+    virtual std::string serialize_metadata_as_json(bool simple) const = 0;
     virtual void walk(std::function<void(entry_view)> const& func) const = 0;
     virtual std::optional<entry_view> find(const char* path) const = 0;
     virtual std::optional<entry_view> find(int inode) const = 0;

include/dwarfs/metadata_v2.h

@@ -71,6 +71,10 @@ class metadata_v2 {
 
   folly::dynamic as_dynamic() const { return impl_->as_dynamic(); }
 
+  std::string serialize_as_json(bool simple) const {
+    return impl_->serialize_as_json(simple);
+  }
+
   static void get_stat_defaults(struct ::stat* defaults);
 
   // TODO: check if this is needed
@@ -142,6 +146,7 @@ class metadata_v2 {
         std::function<void(const std::string&, uint32_t)> const& icb) const = 0;
 
     virtual folly::dynamic as_dynamic() const = 0;
+    virtual std::string serialize_as_json(bool simple) const = 0;
 
     virtual size_t size() const = 0;
     virtual bool empty() const = 0;

src/dwarfs/filesystem_v2.cpp

@@ -195,6 +195,7 @@ class filesystem_ : public filesystem_v2::impl {
 
   void dump(std::ostream& os, int detail_level) const override;
   folly::dynamic metadata_as_dynamic() const override;
+  std::string serialize_metadata_as_json(bool simple) const override;
   void walk(std::function<void(entry_view)> const& func) const override;
   std::optional<entry_view> find(const char* path) const override;
   std::optional<entry_view> find(int inode) const override;
@@ -278,6 +279,12 @@ folly::dynamic filesystem_<LoggerPolicy>::metadata_as_dynamic() const {
   return meta_.as_dynamic();
 }
 
+template <typename LoggerPolicy>
+std::string
+filesystem_<LoggerPolicy>::serialize_metadata_as_json(bool simple) const {
+  return meta_.serialize_as_json(simple);
+}
+
 template <typename LoggerPolicy>
 void filesystem_<LoggerPolicy>::walk(
     std::function<void(entry_view)> const& func) const {

src/dwarfs/metadata_v2.cpp

@@ -34,6 +34,7 @@
 
 #include <thrift/lib/cpp2/frozen/FrozenUtil.h>
 #include <thrift/lib/cpp2/protocol/DebugProtocol.h>
+#include <thrift/lib/cpp2/protocol/Serializer.h>
 
 #include <fmt/format.h>
 
@@ -170,6 +171,7 @@ class metadata_ : public metadata_v2::impl {
       const override;
 
   folly::dynamic as_dynamic() const override;
+  std::string serialize_as_json(bool simple) const override;
 
   size_t size() const override { return data_.size(); }
 
@@ -525,6 +527,19 @@ folly::dynamic metadata_<LoggerPolicy>::as_dynamic() const {
   return obj;
 }
 
+template <typename LoggerPolicy>
+std::string metadata_<LoggerPolicy>::serialize_as_json(bool simple) const {
+  std::string json;
+  if (simple) {
+    apache::thrift::SimpleJSONSerializer serializer;
+    serializer.serialize(meta_.thaw(), &json);
+  } else {
+    apache::thrift::JSONSerializer serializer;
+    serializer.serialize(meta_.thaw(), &json);
+  }
+  return json;
+}
+
 template <typename LoggerPolicy>
 std::string metadata_<LoggerPolicy>::modestring(uint16_t mode) const {
   std::ostringstream oss;

src/dwarfsck.cpp

@@ -24,6 +24,8 @@
 
 #include <boost/program_options.hpp>
 
+#include <folly/File.h>
+#include <folly/FileUtil.h>
 #include <folly/String.h>
 #include <folly/json.h>
 
@@ -38,7 +40,7 @@ namespace dwarfs {
 namespace po = boost::program_options;
 
 int dwarfsck(int argc, char** argv) {
-  std::string log_level, input;
+  std::string log_level, input, export_metadata;
   int detail;
   bool json = false;
 
@@ -54,6 +56,9 @@ int dwarfsck(int argc, char** argv) {
     ("json",
         po::value<bool>(&json)->zero_tokens(),
         "print metadata in JSON format")
+    ("export-metadata",
+        po::value<std::string>(&export_metadata),
+        "export raw metadata as JSON to file")
     ("log-level",
         po::value<std::string>(&log_level)->default_value("info"),
         "log level (error, warn, info, debug, trace)")
@@ -87,7 +92,15 @@ int dwarfsck(int argc, char** argv) {
   try {
     auto mm = std::make_shared<mmap>(input);
 
-    if (json) {
+    if (!export_metadata.empty()) {
+      auto of = folly::File(export_metadata, O_RDWR | O_CREAT);
+      filesystem_v2 fs(lgr, mm);
+      auto json = fs.serialize_metadata_as_json(true);
+      if (folly::writeFull(of.fd(), json.data(), json.size()) < 0) {
+        LOG_ERROR << "failed to export metadata";
+      }
+      of.close();
+    } else if (json) {
       filesystem_v2 fs(lgr, mm);
       std::cout << folly::toPrettyJson(fs.metadata_as_dynamic()) << std::endl;
     } else {
@@ -99,6 +112,9 @@ int dwarfsck(int argc, char** argv) {
   } catch (runtime_error const& e) {
     LOG_ERROR << folly::exceptionStr(e);
     return 1;
+  } catch (std::system_error const& e) {
+    LOG_ERROR << folly::exceptionStr(e);
+    return 1;
   }
 
   return 0;