Sort cinemas and DKDM recipients correctly using the collator (#2950).

6 files changed, 37 insertions, 8 deletions

diff --git a/src/lib/cinema_list.cc b/src/lib/cinema_list.cc
index 4ee538efe..8673f9c8b 100644
--- a/src/lib/cinema_list.cc
+++ b/src/lib/cinema_list.cc
@@ -244,7 +244,7 @@ cinemas_from_result(SQLiteStatement& statement)
 vector<pair<CinemaID, Cinema>>
 CinemaList::cinemas() const
 {
-	SQLiteStatement statement(_db, _cinemas.select("ORDER BY name ASC"));
+	SQLiteStatement statement(_db, _cinemas.select("ORDER BY name COLLATE unicode ASC"));
 	return cinemas_from_result(statement);
 }
 
diff --git a/src/lib/collator.cc b/src/lib/collator.cc
index 8de1857ab..21e89fd78 100644
--- a/src/lib/collator.cc
+++ b/src/lib/collator.cc
@@ -27,6 +27,7 @@
 #include <unicode/utypes.h>
 #include <unicode/usearch.h>
 #include <unicode/ustring.h>
+#include <fmt/format.h>
 #include <boost/scoped_array.hpp>
 #include <cstring>
 #include <vector>
@@ -36,10 +37,14 @@ using std::string;
 using std::vector;
 
 
-Collator::Collator(char const* locale)
+Collator::Collator()
 {
 	UErrorCode status = U_ZERO_ERROR;
-	_collator = ucol_open(locale, &status);
+#ifdef DCPOMATIC_POSIX
+	_collator = ucol_open("POSIX", &status);
+#else
+	_collator = ucol_open(nullptr, &status);
+#endif
 	if (_collator) {
 		ucol_setAttribute(_collator, UCOL_NORMALIZATION_MODE, UCOL_ON, &status);
 		/* Ignore case and character encoding (and probably some other things) */
@@ -72,9 +77,12 @@ int
 Collator::compare (string const& utf8_a, string const& utf8_b) const
 {
 	if (_collator) {
-		auto utf16_a = utf8_to_utf16(utf8_a);
-		auto utf16_b = utf8_to_utf16(utf8_b);
-		return ucol_strcoll(_collator, utf16_a.data(), -1, utf16_b.data(), -1);
+		UErrorCode error = U_ZERO_ERROR;
+		auto const result = ucol_strcollUTF8(_collator, utf8_a.data(), -1, utf8_b.data(), -1, &error);
+		if (error != U_ZERO_ERROR) {
+			throw std::runtime_error(fmt::format("Failed to compare strings ({})", static_cast<int>(error)));
+		}
+		return result;
 	} else {
 		return strcoll(utf8_a.c_str(), utf8_b.c_str());
 	}
diff --git a/src/lib/collator.h b/src/lib/collator.h
index 3cbb40748..1768c43c0 100644
--- a/src/lib/collator.h
+++ b/src/lib/collator.h
@@ -32,7 +32,7 @@ struct UCollator;
 class Collator
 {
 public:
-	Collator(char const* locale = nullptr);
+	Collator();
 	~Collator();
 
 	Collator(Collator const &) = delete;
diff --git a/src/lib/dkdm_recipient_list.cc b/src/lib/dkdm_recipient_list.cc
index c57ade437..5b8a373f6 100644
--- a/src/lib/dkdm_recipient_list.cc
+++ b/src/lib/dkdm_recipient_list.cc
@@ -179,7 +179,7 @@ dkdm_recipients_from_result(SQLiteStatement& statement)
 vector<std::pair<DKDMRecipientID, DKDMRecipient>>
 DKDMRecipientList::dkdm_recipients() const
 {
-	SQLiteStatement statement(_db, _dkdm_recipients.select("ORDER BY name ASC"));
+	SQLiteStatement statement(_db, _dkdm_recipients.select("ORDER BY name COLLATE unicode ASC"));
 	return dkdm_recipients_from_result(statement);
 }
 
diff --git a/src/lib/sqlite_database.cc b/src/lib/sqlite_database.cc
index f0270129b..e1ad15143 100644
--- a/src/lib/sqlite_database.cc
+++ b/src/lib/sqlite_database.cc
@@ -24,6 +24,20 @@
 #include <sqlite3.h>
 
 
+using std::string;
+
+
+static
+int collator_compare(void* context, int length_a, const void* string_a, int length_b, const void* string_b)
+{
+	auto collator = reinterpret_cast<Collator*>(context);
+	return collator->compare(
+		string(reinterpret_cast<char const*>(string_a), length_a),
+		string(reinterpret_cast<char const*>(string_b), length_b)
+	);
+}
+
+
 SQLiteDatabase::SQLiteDatabase(boost::filesystem::path path)
 {
 #ifdef DCPOMATIC_WINDOWS
@@ -36,6 +50,11 @@ SQLiteDatabase::SQLiteDatabase(boost::filesystem::path path)
 	}
 
 	sqlite3_busy_timeout(_db, 500);
+
+	rc = sqlite3_create_collation(_db, "unicode", SQLITE_UTF8, &_collator, collator_compare);
+	if (rc != SQLITE_OK) {
+		throw std::runtime_error("Could not set SQLite database collation");
+	}
 }
 
 
diff --git a/src/lib/sqlite_database.h b/src/lib/sqlite_database.h
index 7858cbf5c..1f459091f 100644
--- a/src/lib/sqlite_database.h
+++ b/src/lib/sqlite_database.h
@@ -23,6 +23,7 @@
 #define DCPOMATIC_SQLITE_DATABASE_H
 
 
+#include "collator.h"
 #include <boost/filesystem.hpp>
 
 struct sqlite3;
@@ -46,6 +47,7 @@ public:
 
 private:
 	sqlite3* _db = nullptr;
+	Collator _collator;
 };