Completely rework provider Nuspell, require Nuspell v5.1.0.

The new provider is much smaller in code size, more correct and uses some
new APIs introduced in v5.1.0 for enumerating the dictionaries.

3 files changed, 88 insertions, 336 deletions

diff --git a/.appveyor.yml b/.appveyor.yml
index 82cc38b..c413347 100644
--- a/.appveyor.yml
+++ b/.appveyor.yml
@@ -26,7 +26,7 @@ for:
   init:
     - sudo apt-get -y update && sudo apt-get -y upgrade
     - sudo apt-get -y install libglib2.0-dev libaspell-dev hspell libhunspell-dev libvoikko-dev voikko-fi aspell-en libunittest++-dev hunspell-fr libicu-dev ninja-build
-    - wget https://github.com/nuspell/nuspell/archive/refs/tags/v5.0.0.tar.gz -O - | tar -xz
+    - wget https://github.com/nuspell/nuspell/archive/refs/tags/v5.1.0.tar.gz -O - | tar -xz
     - cmake -S nuspell-* -B nuspell-build -G Ninja -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTING=0
     - cmake --build nuspell-build
     - sudo cmake --install nuspell-build
diff --git a/configure.ac b/configure.ac
index 35e3b80..78df0ce 100644
--- a/configure.ac
+++ b/configure.ac
@@ -209,7 +209,7 @@ build_providers=
 
 dnl Standard providers
 ENCHANT_CHECK_PKG_CONFIG_PROVIDER([hunspell], [HUNSPELL])
-ENCHANT_CHECK_PKG_CONFIG_PROVIDER([nuspell], [NUSPELL], [nuspell >= 4.1.0])
+ENCHANT_CHECK_PKG_CONFIG_PROVIDER([nuspell], [NUSPELL], [nuspell >= 5.1.0])
 ENCHANT_CHECK_LIB_PROVIDER([aspell], [ASPELL], [get_aspell_dict_info_list])
 ENCHANT_CHECK_LIB_PROVIDER([hspell], [HSPELL], [hspell_get_dictionary_path],, [-lz])
 ENCHANT_CHECK_PKG_CONFIG_PROVIDER([voikko], [VOIKKO], [libvoikko])
diff --git a/providers/enchant_nuspell.cpp b/providers/enchant_nuspell.cpp
index 01e3edd..91d9502 100644
--- a/providers/enchant_nuspell.cpp
+++ b/providers/enchant_nuspell.cpp
@@ -1,4 +1,5 @@
 /* enchant
+ * Copyright (C) 2022 Dimitrij Mijoski
  * Copyright (C) 2020 Sander van Geloven
  *
  * This library is free software; you can redistribute it and/or
@@ -35,12 +36,7 @@
 
 #include "config.h"
 
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-
-#include <string>
-#include <vector>
+#include <memory>
 
 #include "enchant-provider.h"
 #include "unused-parameter.h"
@@ -51,372 +47,130 @@
 #include <glib.h>
 
 using namespace std;
-using namespace nuspell;
-
-/***************************************************************************/
 
-class NuspellChecker
+// EnchantDict functions
+static int nuspell_dict_check(EnchantDict* me, const char* const word,
+                              size_t len)
 {
-public:
-	bool checkWord (const char *word, size_t len);
-	char **suggestWord (const char* const word, size_t len, size_t *out_n_suggs);
-
-	bool requestDictionary (const char * szLang);
-
-private:
-	Dictionary nuspell;
-};
+	auto dict = static_cast<nuspell::Dictionary*>(me->user_data);
 
-/***************************************************************************/
-
-bool
-NuspellChecker::checkWord(const char *utf8Word, size_t len)
-{
-	// the 8-bit encodings use precomposed forms
-	char *normalizedWord = g_utf8_normalize (utf8Word, len, G_NORMALIZE_NFC);
-	auto ret = nuspell.spell(normalizedWord);
-	g_free(normalizedWord);
-	return ret;
+	using UniquePtr = unique_ptr<char[], decltype(&g_free)>;
+	auto normalized_word =
+	    UniquePtr(g_utf8_normalize(word, len, G_NORMALIZE_NFC), g_free);
+	return !dict->spell(normalized_word.get());
 }
 
-char**
-NuspellChecker::suggestWord(const char* const utf8Word, size_t len, size_t *nsug)
+static char** nuspell_dict_suggest(EnchantDict* me, const char* const word,
+                                   size_t len, size_t* out_n_suggs)
 {
+	auto dict = static_cast<nuspell::Dictionary*>(me->user_data);
+
+	using UniquePtr = unique_ptr<char[], decltype(&g_free)>;
 	// the 8-bit encodings use precomposed forms
-	char *normalizedWord = g_utf8_normalize (utf8Word, len, G_NORMALIZE_NFC);
+	auto normalized_word =
+	    UniquePtr(g_utf8_normalize(word, len, G_NORMALIZE_NFC), g_free);
 	auto suggestions = vector<string>();
-	nuspell.suggest(normalizedWord, suggestions);
-	g_free(normalizedWord);
-	if (suggestions.empty())
+	dict->suggest(normalized_word.get(), suggestions);
+	if (empty(suggestions)) {
+		*out_n_suggs = 0;
 		return nullptr;
-	*nsug = suggestions.size();
-	char **sug = g_new0 (char *, *nsug + 1);
-	size_t i = 0;
-	for (auto& suggest : suggestions) {
-		char *word = g_new0(char, suggest.size() + 1);
-		strcpy(word, suggest.c_str());
-		sug[i] = word;
-		i++;
-	}
-	return sug;
-}
-
-static void
-s_buildDictionaryDirs (vector<string> & dirs)
-{
-	dirs.clear ();
-
-	/* 1. personal overrides for Enchant
-	 *    ~/.config/enchant/nuspell
-	 */
-	gchar * tmp;
-	char * config_dir = enchant_get_user_config_dir ();
-	tmp = g_build_filename (config_dir, "nuspell", nullptr);
-	dirs.push_back (tmp);
-	free (config_dir);
-	g_free(tmp);
-
-	/* Dynamically retrieved from Nuspell dictionary finder:
-	 * 2. personal overrides for Hunspell
-	 *    $XDG_DATA_HOME/hunspell
-	 *    $XDG_DATA_HOME by default is $HOME/.local/share/
-	 * 3. system installed for Hunspell
-	 *    $XDG_DATA_DIRS/hunspell
-	 *    $XDG_DATA_DIRS/myspell (needed for Fedora)
-	 *    $XDG_DATA_DIRS by default are /usr/local/share and /usr/share
-	 */
-	nuspell::append_default_dir_paths(dirs);
-
-	/* 5. system installations by Enchant
-	 *    /usr/local/share/enchant/nuspell
-	 *    /usr/share/enchant/nuspell
-	 */
-	char * enchant_prefix = enchant_get_prefix_dir();
-	if (enchant_prefix) {
-		tmp = g_build_filename(enchant_prefix, "share", "enchant", "nuspell", nullptr);
-		dirs.push_back (tmp);
-		g_free(enchant_prefix);
-		g_free(tmp);
-	}
-
-	/* Hunspell paths are used, therefore ENCHANT_NUSPELL_DICT_DIR is
-	 * irrelevant. Hence, the following paths are not to be considered:
-	 * /usr/local/share/nuspell and /usr/share/nuspell
-	 */
-}
-
-static void
-s_buildHashNames (vector<string> & names, const char * dict)
-{
-	names.clear ();
-
-	vector<string> dirs;
-	s_buildDictionaryDirs (dirs);
-
-	char *dict_dic = g_strconcat(dict, ".dic", nullptr);
-	for (size_t i = 0; i < dirs.size(); i++) {
-		char *tmp = g_build_filename (dirs[i].c_str(), dict_dic, nullptr);
-		names.push_back (tmp);
-		g_free (tmp);
-	}
-
-	g_free(dict_dic);
-}
-
-static const string
-s_correspondingAffFile(const string & dicFile)
-{
-	string aff = dicFile;
-	if (aff.size() >= 4 && aff.compare(aff.size() - 4, 4, ".dic") == 0) {
-		aff.erase(aff.size() - 3);
-		aff += "aff";
 	}
-	return aff;
+	char** sug_list = g_new0(char*, size(suggestions) + 1);
+	transform(begin(suggestions), end(suggestions), sug_list,
+	          [](const string& sug) { return g_strdup(sug.c_str()); });
+	*out_n_suggs = size(suggestions);
+	return sug_list;
 }
+// End EnchantDict functions
 
-static bool
-s_fileExists(const string & file)
-{
-	return g_file_test(file.c_str(), G_FILE_TEST_EXISTS) != 0;
-}
+// EnchantProvider functions
+static void nuspell_provider_dispose(EnchantProvider* me) { g_free(me); }
 
-static bool is_plausible_dict_for_tag(const char *dir_entry, const char *tag)
+static EnchantDict*
+nuspell_provider_request_dict([[maybe_unused]] EnchantProvider* me,
+                              const char* const tag)
 {
-	const char *dic_suffix = ".dic";
-	size_t dic_suffix_len = strlen(dic_suffix);
-	size_t dir_entry_len = strlen(dir_entry);
-	size_t tag_len = strlen(tag);
-
-	if (dir_entry_len - dic_suffix_len < tag_len)
-		return false;
-	if (strcmp(dir_entry + dir_entry_len - dic_suffix_len, dic_suffix) != 0)
-		return false;
-	if (strncmp (dir_entry, tag, tag_len) != 0)
-		return false;
-	//e.g. requested dict for "fi",
-	//reject "fil_PH.dic"
-	//allow "fi-FOO.dic", "fi_FOO.dic", "fi.dic", etc.
-	if (!ispunct(dir_entry[tag_len]))
-		return false;
-	return true;
-}
-
-static char *
-nuspell_request_dictionary (const char * tag)
-{
-	vector<string> names;
-
-	s_buildHashNames (names, tag);
-
-	for (size_t i = 0; i < names.size (); i++) {
-		if (g_file_test(names[i].c_str(), G_FILE_TEST_EXISTS) &&
-		    s_fileExists(s_correspondingAffFile(names[i]))) {
-			return strdup (names[i].c_str());
-		}
-	}
-
-	vector<string> dirs;
-	s_buildDictionaryDirs (dirs);
-
-	for (size_t i = 0; i < dirs.size(); i++) {
-		GDir *dir = g_dir_open (dirs[i].c_str(), 0, nullptr);
-		if (dir) {
-			const char *dir_entry;
-			while ((dir_entry = g_dir_read_name (dir)) != NULL) {
-				if (is_plausible_dict_for_tag(dir_entry, tag)) {
-					char *dict = g_build_filename (dirs[i].c_str(),
-								       dir_entry, nullptr);
-					if(s_fileExists(s_correspondingAffFile(dict))) {
-						g_dir_close (dir);
-						return dict;
-					}
-					g_free(dict);
-				}
-			}
-
-			g_dir_close (dir);
-		}
-	}
-
-	return NULL;
-}
+	auto dirs = vector<filesystem::path>();
+	nuspell::append_default_dir_paths(dirs);
+	auto dic_path = nuspell::search_dirs_for_one_dict(dirs, tag);
+	if (empty(dic_path))
+		return nullptr;
 
-bool
-NuspellChecker::requestDictionary(const char *szLang)
-{
-	char *dic = nuspell_request_dictionary (szLang);
-	if (!dic)
-		return false;
-	string aff(s_correspondingAffFile(dic));
-	if (!s_fileExists(aff))
-		return false;
-	auto path = string(dic);
-	free(dic);
-	if (path.size() >= 4 && path.compare(path.size() - 4, 4, ".dic") == 0)
-		path.erase(path.size() - 4);
-	else
-		return false;
+	auto dict_cpp = make_unique<nuspell::Dictionary>();
 	try {
-		nuspell = nuspell::Dictionary::load_from_path(path);
-	} catch (const std::runtime_error& e) {
-		return false;
+		dict_cpp->load_aff_dic(dic_path);
 	}
-
-	return true;
-}
-
-/*
- * Enchant
- */
-
-static char **
-nuspell_dict_suggest (EnchantDict * me, const char *const word,
-		      size_t len, size_t * out_n_suggs)
-{
-	NuspellChecker * checker = static_cast<NuspellChecker *>(me->user_data);
-	return checker->suggestWord (word, len, out_n_suggs);
-}
-
-static int
-nuspell_dict_check (EnchantDict * me, const char *const word, size_t len)
-{
-	NuspellChecker * checker = static_cast<NuspellChecker *>(me->user_data);
-
-	return !(checker->checkWord(word, len));
-}
-
-static int
-nuspell_dict_is_word_character (EnchantDict * me _GL_UNUSED_PARAMETER,
-				uint32_t uc, size_t n _GL_UNUSED_PARAMETER)
-{
-	return g_unichar_isalpha(uc);
-}
-
-static void
-nuspell_provider_enum_dicts (const char * const directory,
-			     vector<string> & out_dicts)
-{
-	GDir * dir = g_dir_open (directory, 0, nullptr);
-	if (dir) {
-		const char * entry;
-		while ((entry = g_dir_read_name (dir)) != NULL) {
-			char * utf8_entry = g_filename_to_utf8 (entry, -1, nullptr, nullptr, nullptr);
-			if (utf8_entry) {
-				string dir_entry (utf8_entry);
-				g_free (utf8_entry);
-
-				int hit = dir_entry.rfind (".dic");
-				// don't include hyphenation dictionaries
-				if (hit != -1) {
-					// require .aff file to be present
-					if(dir_entry.compare (0, 5, "hyph_") != 0) {
-						char * dic = g_build_filename(directory, dir_entry.c_str(), nullptr);
-						if (s_fileExists(s_correspondingAffFile(dic))) {
-							out_dicts.push_back (dir_entry.substr (0, hit));
-						}
-						g_free(dic);
-					}
-				}
-			}
-		}
-
-		g_dir_close (dir);
-	}
-}
-
-extern "C" {
-
-static char **
-nuspell_provider_list_dicts (EnchantProvider * me _GL_UNUSED_PARAMETER,
-			     size_t * out_n_dicts)
-{
-	vector<string> dict_dirs, dicts;
-	char ** dictionary_list = NULL;
-
-	s_buildDictionaryDirs (dict_dirs);
-
-	for (size_t i = 0; i < dict_dirs.size(); i++) {
-		nuspell_provider_enum_dicts (dict_dirs[i].c_str(), dicts);
-	}
-
-	if (dicts.size () > 0) {
-		dictionary_list = g_new0 (char *, dicts.size() + 1);
-
-		for (size_t i = 0; i < dicts.size(); i++)
-			dictionary_list[i] = g_strdup (dicts[i].c_str());
-	}
-
-	*out_n_dicts = dicts.size ();
-	return dictionary_list;
-}
-
-static EnchantDict *
-nuspell_provider_request_dict(EnchantProvider * me _GL_UNUSED_PARAMETER, const char *const tag)
-{
-	NuspellChecker * checker = new NuspellChecker();
-
-	if (!checker)
-		return NULL;
-
-	if (!checker->requestDictionary(tag)) {
-		delete checker;
-		return NULL;
+	catch (const nuspell::Dictionary_Loading_Error&) {
+		return nullptr;
 	}
 
-	EnchantDict *dict = g_new0(EnchantDict, 1);
-	dict->user_data = (void *) checker;
+	EnchantDict* dict = g_new0(EnchantDict, 1);
+	dict->user_data = static_cast<void*>(dict_cpp.release());
 	dict->check = nuspell_dict_check;
 	dict->suggest = nuspell_dict_suggest;
-	// don't implement personal, session
-	dict->is_word_character = nuspell_dict_is_word_character;
-
 	return dict;
 }
 
-static void
-nuspell_provider_dispose_dict (EnchantProvider * me _GL_UNUSED_PARAMETER, EnchantDict * dict)
+static void nuspell_provider_dispose_dict([[maybe_unused]] EnchantProvider* me,
+                                          EnchantDict* dict)
 {
-	NuspellChecker *checker = (NuspellChecker *) dict->user_data;
-	delete checker;
-
-	g_free (dict);
+	auto dict_cpp = static_cast<nuspell::Dictionary*>(dict->user_data);
+	delete dict_cpp;
+	g_free(dict);
 }
 
 static int
-nuspell_provider_dictionary_exists (struct str_enchant_provider * me _GL_UNUSED_PARAMETER,
-				    const char *const tag)
+nuspell_provider_dictionary_exists([[maybe_unused]] EnchantProvider* me,
+                                   const char* const tag)
 {
-	vector <string> names;
-	s_buildHashNames (names, tag);
-	for (size_t i = 0; i < names.size(); i++) {
-		if (g_file_test (names[i].c_str(), G_FILE_TEST_EXISTS) &&
-		    s_fileExists(s_correspondingAffFile(names[i]))) {
-			return 1;
-		}
-	}
-
-	return 0;
+	auto dirs = vector<filesystem::path>();
+	nuspell::append_default_dir_paths(dirs);
+	auto dic_path = nuspell::search_dirs_for_one_dict(dirs, tag);
+	return !empty(dic_path);
 }
 
-static void
-nuspell_provider_dispose (EnchantProvider * me)
+static const char*
+nuspell_provider_identify([[maybe_unused]] EnchantProvider* me)
 {
-	g_free (me);
+	return "nuspell";
 }
 
-static const char *
-nuspell_provider_identify (EnchantProvider * me _GL_UNUSED_PARAMETER)
+static const char*
+nuspell_provider_describe([[maybe_unused]] EnchantProvider* me)
 {
-	return "nuspell";
+	return "Nuspell Provider";
 }
 
-static const char *
-nuspell_provider_describe (EnchantProvider * me _GL_UNUSED_PARAMETER)
+static char**
+nuspell_provider_list_dicts(EnchantProvider* me _GL_UNUSED_PARAMETER,
+                            size_t* out_n_dicts)
 {
-	return "Nuspell Provider";
+	auto dicts = nuspell::search_default_dirs_for_dicts();
+	if (empty(dicts)) {
+		*out_n_dicts = 0;
+		return nullptr;
+	}
+	for (auto& d : dicts)
+		d = d.stem();
+	sort(begin(dicts), end(dicts));
+	auto it = unique(begin(dicts), end(dicts));
+	it = remove_if(begin(dicts), it, [](const filesystem::path& p) {
+		auto& n = p.native();
+		return any_of(begin(n), end(n),
+		              [](auto c) { return c < 0 || c > 127; });
+	});
+	dicts.erase(it, end(dicts));
+
+	char** dictionary_list = g_new0(char*, size(dicts) + 1);
+	transform(begin(dicts), end(dicts), dictionary_list,
+	          [](const filesystem::path& p) {
+		          return g_strdup(p.string().c_str());
+	          });
+	*out_n_dicts = size(dicts);
+	return dictionary_list;
 }
 
-EnchantProvider *init_enchant_provider (void);
+extern "C" EnchantProvider* init_enchant_provider(void);
 
 EnchantProvider *
 init_enchant_provider (void)
@@ -432,5 +186,3 @@ init_enchant_provider (void)
 
 	return provider;
 }
-
-} // extern C linkage