add locale-aware colation to db indexes

2 files changed, 84 insertions, 0 deletions

diff --git a/src/locale/include/collation.hpp b/src/locale/include/collation.hpp
new file mode 100644
index 00000000..e8b6fa17
--- /dev/null
+++ b/src/locale/include/collation.hpp
@@ -0,0 +1,49 @@
+/*
+ * Copyright 2023 jacqueline <me@jacqueline.id.au>
+ *
+ * SPDX-License-Identifier: GPL-3.0-only
+ */
+
+#pragma once
+
+#include <cstddef>
+#include <memory>
+#include <string>
+
+#include "esp_partition.h"
+#include "span.hpp"
+
+#include "strxfrm.h"
+
+namespace locale {
+
+class ICollator {
+ public:
+  virtual ~ICollator() {}
+
+  virtual auto Transform(const std::string&) -> std::string = 0;
+};
+
+class NoopCollator : public ICollator {
+ public:
+  auto Transform(const std::string& in) -> std::string override { return in; }
+};
+
+auto CreateCollator() -> std::unique_ptr<ICollator>;
+
+class GLibCollator : public ICollator {
+ public:
+  static auto create() -> GLibCollator*;
+  ~GLibCollator();
+
+  auto Transform(const std::string& in) -> std::string override;
+
+ private:
+  GLibCollator(const esp_partition_mmap_handle_t,
+               std::unique_ptr<locale_data_t>);
+
+  const esp_partition_mmap_handle_t handle_;
+  std::unique_ptr<locale_data_t> locale_data_;
+};
+
+}  // namespace locale
diff --git a/src/locale/include/strxfrm.h b/src/locale/include/strxfrm.h
new file mode 100644
index 00000000..a40f7596
--- /dev/null
+++ b/src/locale/include/strxfrm.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2023 jacqueline <me@jacqueline.id.au>
+ *
+ * SPDX-License-Identifier: GPL-3.0-only
+ */
+
+#pragma once
+
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct {
+  uint_fast32_t nrules;
+  unsigned char* rulesets;
+  unsigned char* weights;
+  int32_t* table;
+  unsigned char* extra;
+  int32_t* indirect;
+} locale_data_t;
+
+bool parse_locale_data(const void* raw_data, size_t size, locale_data_t* out);
+
+size_t glib_strxfrm(char* dest,
+                    const char* src,
+                    size_t n,
+                    locale_data_t* locale);
+
+#ifdef __cplusplus
+}
+#endif