SpellingsDB reuses an existing database

Also the underlying Xapian database is now in a single-file format.

include/spelling_correction.h

@@ -20,6 +20,7 @@
 #ifndef KIWIX_SPELLING_CORRECTION_H
 #define KIWIX_SPELLING_CORRECTION_H
 
+#include <filesystem>
 #include <memory>
 #include <string>
 #include <vector>
@@ -40,7 +41,7 @@ namespace kiwix
 class SpellingsDB
 {
 public: // functions
-  explicit SpellingsDB(const zim::Archive& archive, std::string path);
+  SpellingsDB(const zim::Archive& archive, std::filesystem::path path);
   ~SpellingsDB();
 
   SpellingsDB(const SpellingsDB& ) = delete;

src/spelling_correction.cpp

@@ -39,19 +39,37 @@ std::vector<std::string> getAllTitles(const zim::Archive& a)
   return result;
 }
 
+void createXapianDB(std::string path, const zim::Archive& archive)
+{
+  const int flags = Xapian::DB_BACKEND_GLASS|Xapian::DB_CREATE;
+  const auto tmpDbPath = path + ".tmp";
+  Xapian::WritableDatabase db(tmpDbPath, flags);
+  for (const auto& t : getAllTitles(archive)) {
+    db.add_spelling(t);
+  }
+  db.commit();
+  db.compact(path, Xapian::DBCOMPACT_SINGLE_FILE);
+  db.close();
+  std::filesystem::remove_all(tmpDbPath);
+}
+
 std::unique_ptr<Xapian::Database> openOrCreateXapianDB(std::string path, const zim::Archive& archive)
 {
-  auto db(std::make_unique<Xapian::WritableDatabase>(path, Xapian::DB_BACKEND_GLASS));
-  for (const auto& t : getAllTitles(archive)) {
-    db->add_spelling(t);
+  try
+  {
+    return std::make_unique<Xapian::Database>(path);
+  }
+  catch (const Xapian::DatabaseOpeningError& )
+  {
+    createXapianDB(path, archive);
+    return std::make_unique<Xapian::Database>(path);
   }
-  return std::move(db);
 }
 
 } // unnamed namespace
 
-SpellingsDB::SpellingsDB(const zim::Archive& archive, std::string path)
-  : impl_(openOrCreateXapianDB(path, archive))
+SpellingsDB::SpellingsDB(const zim::Archive& archive, std::filesystem::path path)
+  : impl_(openOrCreateXapianDB(path.string(), archive))
 {
 }
 

test/spelling_correction.cpp

@@ -19,27 +19,45 @@
 
 #include "gtest/gtest.h"
 #include "../include/spelling_correction.h"
+#include "../src/tools/pathTools.h"
 #include "zim/archive.h"
 
 #include <filesystem>
 
+#include <xapian.h>
+
 const std::string TEST_DB_PATH = "./spellings.db";
 
 class SpellingCorrectionTest : public ::testing::Test
 {
-  void removeDb()
-  {
-    std::filesystem::remove_all(TEST_DB_PATH);
-  }
-
 protected:
   void SetUp() override {
-    removeDb();
+    tmpDirPath = makeTmpDirectory();
+    archive = std::make_unique<zim::Archive>("./test/spelling_correction_test.zim");
   }
 
   void TearDown() override {
-    removeDb();
+    std::filesystem::permissions(
+        tmpDirPath,
+        std::filesystem::perms::owner_write,
+        std::filesystem::perm_options::add
+    );
+    std::filesystem::remove_all(tmpDirPath);
   }
+
+  void makeTmpDirReadOnly() {
+    using std::filesystem::perms;
+
+    std::filesystem::permissions(
+        tmpDirPath,
+        perms::owner_write | perms::group_write | perms::others_write,
+        std::filesystem::perm_options::remove
+    );
+  }
+
+protected:
+  std::filesystem::path tmpDirPath;
+  std::unique_ptr<zim::Archive> archive;
 };
 
 void testSpellingCorrections(const kiwix::SpellingsDB& spellingsDB)
@@ -169,9 +187,29 @@ void testSpellingCorrections(const kiwix::SpellingsDB& spellingsDB)
   EXPECT_THROW(spellingsDB.getSpellingCorrections("Kung", 2), std::runtime_error);
 }
 
+TEST_F(SpellingCorrectionTest, SpellingsDBCannotBeCreatedInAReadOnlyDirectory)
+{
+  makeTmpDirReadOnly();
+
+  EXPECT_THROW(
+    const kiwix::SpellingsDB spellingsDB(*archive, tmpDirPath / "spellings.db"),
+    Xapian::DatabaseCreateError
+  );
+}
+
 TEST_F(SpellingCorrectionTest, allInOne)
 {
-  const auto archive = zim::Archive("./test/spelling_correction_test.zim");
-  kiwix::SpellingsDB spellingsDB(archive, TEST_DB_PATH);
-  testSpellingCorrections(spellingsDB);
+  const auto spellingsDbPath = tmpDirPath / "spellings.db";
+
+  {
+    const kiwix::SpellingsDB spellingsDB(*archive, spellingsDbPath);
+    testSpellingCorrections(spellingsDB);
+  }
+
+  makeTmpDirReadOnly();
+
+  {
+    const kiwix::SpellingsDB spellingsDB(*archive, spellingsDbPath);
+    testSpellingCorrections(spellingsDB);
+  }
 }