3 files changed, 87 insertions, 20 deletions

diff --git a/src/database/database.cpp b/src/database/database.cpp
index 9b978b8c..e826f576 100644
--- a/src/database/database.cpp
+++ b/src/database/database.cpp
@@ -28,6 +28,7 @@
 #include "leveldb/iterator.h"
 #include "leveldb/options.h"
 #include "leveldb/slice.h"
+#include "leveldb/status.h"
 #include "leveldb/write_batch.h"
 
 #include "db_events.hpp"
@@ -51,12 +52,13 @@ static const char kDbPath[] = "/.tangara-db";
 
 static const char kKeyDbVersion[] = "schema_version";
 static const uint8_t kCurrentDbVersion = 3;
-
+static const char kKeyCollator[] = "collator";
 static const char kKeyTrackId[] = "next_track_id";
 
 static std::atomic<bool> sIsDbOpen(false);
 
-static auto CreateNewDatabase(leveldb::Options& options) -> leveldb::DB* {
+static auto CreateNewDatabase(leveldb::Options& options, locale::ICollator& col)
+    -> leveldb::DB* {
   Database::Destroy();
   leveldb::DB* db;
   options.create_if_missing = true;
@@ -71,9 +73,48 @@ static auto CreateNewDatabase(leveldb::Options& options) -> leveldb::DB* {
     delete db;
     return nullptr;
   }
+  ESP_LOGI(kTag, "opening db with collator %s",
+           col.Describe().value_or("NULL").c_str());
+  status = db->Put(leveldb::WriteOptions{}, kKeyCollator,
+                   col.Describe().value_or(""));
+  if (!status.ok()) {
+    delete db;
+    return nullptr;
+  }
   return db;
 }
 
+static auto CheckDatabase(leveldb::DB& db, locale::ICollator& col) -> bool {
+  leveldb::Status status;
+
+  std::string raw_version;
+  std::optional<uint8_t> version{};
+  status = db.Get(leveldb::ReadOptions{}, kKeyDbVersion, &raw_version);
+  if (status.ok()) {
+    version = std::stoi(raw_version);
+  }
+  if (!version || *version != kCurrentDbVersion) {
+    ESP_LOGW(kTag, "db version missing or incorrect");
+    return false;
+  }
+
+  std::string collator;
+  status = db.Get(leveldb::ReadOptions{}, kKeyCollator, &collator);
+  if (!status.ok()) {
+    ESP_LOGW(kTag, "db collator is unknown");
+    return false;
+  }
+  auto needed = col.Describe();
+
+  if ((needed && needed.value() != collator) ||
+      (!needed && !collator.empty())) {
+    ESP_LOGW(kTag, "db collator is mismatched");
+    return false;
+  }
+
+  return true;
+}
+
 auto Database::Open(IFileGatherer& gatherer,
                     ITagParser& parser,
                     locale::ICollator& collator)
@@ -107,23 +148,16 @@ auto Database::Open(IFileGatherer& gatherer,
             auto status = leveldb::DB::Open(options, kDbPath, &db);
             if (!status.ok()) {
               ESP_LOGI(kTag, "opening db failed. recreating.");
-              db = CreateNewDatabase(options);
+              db = CreateNewDatabase(options, collator);
               if (db == nullptr) {
                 return cpp::fail(FAILED_TO_OPEN);
               }
             }
 
-            std::string raw_version;
-            std::optional<uint8_t> version{};
-            status =
-                db->Get(leveldb::ReadOptions{}, kKeyDbVersion, &raw_version);
-            if (status.ok()) {
-              version = std::stoi(raw_version);
-            }
-            if (!version || *version != kCurrentDbVersion) {
-              ESP_LOGI(kTag, "db version missing or incorrect. recreating.");
+            if (!CheckDatabase(*db, collator)) {
+              ESP_LOGI(kTag, "db incompatible. recreating.");
               delete db;
-              db = CreateNewDatabase(options);
+              db = CreateNewDatabase(options, collator);
               if (db == nullptr) {
                 return cpp::fail(FAILED_TO_OPEN);
               }
diff --git a/src/locale/collation.cpp b/src/locale/collation.cpp
index f5e8272a..2d787b5a 100644
--- a/src/locale/collation.cpp
+++ b/src/locale/collation.cpp
@@ -58,19 +58,26 @@ auto GLibCollator::create() -> GLibCollator* {
     return nullptr;
   }
 
+  // We reserve the first 8 bytes of the partition for an identifier / name.
+  // Copy it out, then crop the rest of the region so that the LC_COLLATE parser
+  // doesn't see it.
+  std::string name{static_cast<const char*>(region)};
+  region = static_cast<const std::byte*>(region) + 8;
+
   auto data = std::make_unique<locale_data_t>();
-  if (!parse_locale_data(region, partition->size, data.get())) {
+  if (!parse_locale_data(region, partition->size - 8, data.get())) {
     ESP_LOGE(kTag, "parsing locale data failed");
     esp_partition_munmap(handle);
     return nullptr;
   }
 
-  return new GLibCollator(handle, std::move(data));
+  return new GLibCollator(name, handle, std::move(data));
 }
 
-GLibCollator::GLibCollator(const esp_partition_mmap_handle_t handle,
+GLibCollator::GLibCollator(const std::string& name,
+                           const esp_partition_mmap_handle_t handle,
                            std::unique_ptr<locale_data_t> locale)
-    : handle_(handle), locale_data_(std::move(locale)) {}
+    : name_(name), handle_(handle), locale_data_(std::move(locale)) {}
 
 GLibCollator::~GLibCollator() {
   esp_partition_munmap(handle_);
diff --git a/src/locale/include/collation.hpp b/src/locale/include/collation.hpp
index e8b6fa17..b666860d 100644
--- a/src/locale/include/collation.hpp
+++ b/src/locale/include/collation.hpp
@@ -8,6 +8,7 @@
 
 #include <cstddef>
 #include <memory>
+#include <optional>
 #include <string>
 
 #include "esp_partition.h"
@@ -17,31 +18,56 @@
 
 namespace locale {
 
+/*
+ * Interface for sorting database entries.
+ * For performance, our database exclusively orders entries via byte
+ * comparisons of each key. Our collators therefore work by transforming keys
+ * such that a byte-order comparison results in a natural ordering.
+ */
 class ICollator {
  public:
   virtual ~ICollator() {}
 
+  /*
+   * Returns an identify that uniquely describes this collator. Does not need
+   * to be human readable.
+   */
+  virtual auto Describe() -> std::optional<std::string> = 0;
   virtual auto Transform(const std::string&) -> std::string = 0;
 };
 
+/* Creates and returns the best available collator. */
+auto CreateCollator() -> std::unique_ptr<ICollator>;
+
+/*
+ * Collator that doesn't do anything. Used only when there is no available
+ * locale data.
+ */
 class NoopCollator : public ICollator {
  public:
+  auto Describe() -> std::optional<std::string> override { return {}; }
   auto Transform(const std::string& in) -> std::string override { return in; }
 };
 
-auto CreateCollator() -> std::unique_ptr<ICollator>;
-
+/*
+ * Collator that uses glibc's `strxfrm` to transform keys. Relies on an
+ * LC_COLLATE file (+ 8 byte name header) flashed on a partition in internal
+ * flash.
+ */
 class GLibCollator : public ICollator {
  public:
   static auto create() -> GLibCollator*;
   ~GLibCollator();
 
+  auto Describe() -> std::optional<std::string> override { return name_; }
   auto Transform(const std::string& in) -> std::string override;
 
  private:
-  GLibCollator(const esp_partition_mmap_handle_t,
+  GLibCollator(const std::string& name,
+               const esp_partition_mmap_handle_t,
                std::unique_ptr<locale_data_t>);
 
+  const std::string name_;
   const esp_partition_mmap_handle_t handle_;
   std::unique_ptr<locale_data_t> locale_data_;
 };