19596a2c1Sopenharmony_ci/*
29596a2c1Sopenharmony_ci * Copyright (c) 2023 Huawei Device Co., Ltd.
39596a2c1Sopenharmony_ci * Licensed under the Apache License, Version 2.0 (the "License");
49596a2c1Sopenharmony_ci * you may not use this file except in compliance with the License.
59596a2c1Sopenharmony_ci * You may obtain a copy of the License at
69596a2c1Sopenharmony_ci *
79596a2c1Sopenharmony_ci *     http://www.apache.org/licenses/LICENSE-2.0
89596a2c1Sopenharmony_ci *
99596a2c1Sopenharmony_ci * Unless required by applicable law or agreed to in writing, software
109596a2c1Sopenharmony_ci * distributed under the License is distributed on an "AS IS" BASIS,
119596a2c1Sopenharmony_ci * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
129596a2c1Sopenharmony_ci * See the License for the specific language governing permissions and
139596a2c1Sopenharmony_ci * limitations under the License.
149596a2c1Sopenharmony_ci */
159596a2c1Sopenharmony_ci
169596a2c1Sopenharmony_ci#include <cstring>
179596a2c1Sopenharmony_ci#include <filesystem>
189596a2c1Sopenharmony_ci#include <sys/stat.h>
199596a2c1Sopenharmony_ci#include "i18n_hilog.h"
209596a2c1Sopenharmony_ci#include "libxml/globals.h"
219596a2c1Sopenharmony_ci#include "libxml/tree.h"
229596a2c1Sopenharmony_ci#include "libxml/xmlstring.h"
239596a2c1Sopenharmony_ci#include "locale_compare.h"
249596a2c1Sopenharmony_ci#include "taboo.h"
259596a2c1Sopenharmony_ci
269596a2c1Sopenharmony_cinamespace OHOS {
279596a2c1Sopenharmony_cinamespace Global {
289596a2c1Sopenharmony_cinamespace I18n {
299596a2c1Sopenharmony_ciconst char* Taboo::ROOT_TAG = "taboo";
309596a2c1Sopenharmony_ciconst char* Taboo::ITEM_TAG = "item";
319596a2c1Sopenharmony_ciconst char* Taboo::NAME_TAG = "name";
329596a2c1Sopenharmony_ciconst char* Taboo::VALUE_TAG = "value";
339596a2c1Sopenharmony_cistd::set<std::string> Taboo::supportedRegions;
349596a2c1Sopenharmony_cistd::set<std::string> Taboo::supportedLanguages;
359596a2c1Sopenharmony_cistd::map<std::string, std::string> Taboo::RESOURCES = {};
369596a2c1Sopenharmony_cistd::mutex Taboo::TABOO_MUTEX;
379596a2c1Sopenharmony_ci
389596a2c1Sopenharmony_ciTaboo::Taboo()
399596a2c1Sopenharmony_ci{
409596a2c1Sopenharmony_ci}
419596a2c1Sopenharmony_ci
429596a2c1Sopenharmony_ciTaboo::Taboo(const std::string& path) : tabooDataPath(path)
439596a2c1Sopenharmony_ci{
449596a2c1Sopenharmony_ci    using std::filesystem::directory_iterator;
459596a2c1Sopenharmony_ci
469596a2c1Sopenharmony_ci    std::string tabooConfigFilePath = tabooDataPath + tabooConfigFileName;
479596a2c1Sopenharmony_ci    struct stat s;
489596a2c1Sopenharmony_ci    isTabooDataExist = stat(tabooConfigFilePath.c_str(), &s) == 0;
499596a2c1Sopenharmony_ci    if (isTabooDataExist) {
509596a2c1Sopenharmony_ci        if (RESOURCES.size() == 0) {
519596a2c1Sopenharmony_ci            std::lock_guard<std::mutex> tabooLock(TABOO_MUTEX);
529596a2c1Sopenharmony_ci            if (RESOURCES.size() == 0) {
539596a2c1Sopenharmony_ci                // parse taboo-config.xml to obtain supported regions and languages for name replacement.
549596a2c1Sopenharmony_ci                ParseTabooData(tabooConfigFilePath, DataFileType::CONFIG_FILE);
559596a2c1Sopenharmony_ci                ReadResourceList();
569596a2c1Sopenharmony_ci            }
579596a2c1Sopenharmony_ci        }
589596a2c1Sopenharmony_ci    }
599596a2c1Sopenharmony_ci}
609596a2c1Sopenharmony_ci
619596a2c1Sopenharmony_ciTaboo::~Taboo()
629596a2c1Sopenharmony_ci{
639596a2c1Sopenharmony_ci}
649596a2c1Sopenharmony_ci
659596a2c1Sopenharmony_cistd::string Taboo::ReplaceCountryName(const std::string& region, const std::string& displayLanguage,
669596a2c1Sopenharmony_ci    const std::string& name)
679596a2c1Sopenharmony_ci{
689596a2c1Sopenharmony_ci    if (!isTabooDataExist) {
699596a2c1Sopenharmony_ci        HILOG_INFO_I18N("Taboo::ReplaceCountryName Taboo data not exist.");
709596a2c1Sopenharmony_ci        return name;
719596a2c1Sopenharmony_ci    }
729596a2c1Sopenharmony_ci    if (supportedRegions.find(region) == supportedRegions.end()) {
739596a2c1Sopenharmony_ci        HILOG_INFO_I18N("Taboo::ReplaceCountryName taboo data don't support region %{public}s", region.c_str());
749596a2c1Sopenharmony_ci        return name;
759596a2c1Sopenharmony_ci    }
769596a2c1Sopenharmony_ci    std::string key = regionKey + region;
779596a2c1Sopenharmony_ci    std::vector<std::string> fallbackRegionKeys = QueryKeyFallBack(key);
789596a2c1Sopenharmony_ci    std::string fallbackLanguage;
799596a2c1Sopenharmony_ci    std::string fileName;
809596a2c1Sopenharmony_ci    std::tie(fallbackLanguage, fileName) = LanguageFallBack(displayLanguage);
819596a2c1Sopenharmony_ci    if (fallbackLanguage.length() == 0) {
829596a2c1Sopenharmony_ci        HILOG_INFO_I18N("Taboo::ReplaceCountryName language %{public}s fallback failed", displayLanguage.c_str());
839596a2c1Sopenharmony_ci        return name;
849596a2c1Sopenharmony_ci    }
859596a2c1Sopenharmony_ci    if (localeTabooData.find(fallbackLanguage) == localeTabooData.end()) {
869596a2c1Sopenharmony_ci        localeTabooData[fallbackLanguage] = {};
879596a2c1Sopenharmony_ci        std::string localeTabooDataFilePath = tabooDataPath + fileName + filePathSplitor + tabooLocaleDataFileName;
889596a2c1Sopenharmony_ci        ParseTabooData(localeTabooDataFilePath, DataFileType::DATA_FILE, fallbackLanguage);
899596a2c1Sopenharmony_ci    }
909596a2c1Sopenharmony_ci    auto tabooData = localeTabooData[fallbackLanguage];
919596a2c1Sopenharmony_ci    for (auto it = fallbackRegionKeys.begin(); it != fallbackRegionKeys.end(); ++it) {
929596a2c1Sopenharmony_ci        if (tabooData.find(*it) != tabooData.end()) {
939596a2c1Sopenharmony_ci            return tabooData[*it];
949596a2c1Sopenharmony_ci        }
959596a2c1Sopenharmony_ci    }
969596a2c1Sopenharmony_ci    HILOG_INFO_I18N("Taboo::ReplaceCountryName not find taboo data correspond to region %{public}s",
979596a2c1Sopenharmony_ci        region.c_str());
989596a2c1Sopenharmony_ci    return name;
999596a2c1Sopenharmony_ci}
1009596a2c1Sopenharmony_ci
1019596a2c1Sopenharmony_cistd::string Taboo::ReplaceLanguageName(const std::string& language, const std::string& displayLanguage,
1029596a2c1Sopenharmony_ci    const std::string& name)
1039596a2c1Sopenharmony_ci{
1049596a2c1Sopenharmony_ci    if (!isTabooDataExist) {
1059596a2c1Sopenharmony_ci        HILOG_INFO_I18N("Taboo::ReplaceLanguageName Taboo data not exist.");
1069596a2c1Sopenharmony_ci        return name;
1079596a2c1Sopenharmony_ci    }
1089596a2c1Sopenharmony_ci    if (supportedLanguages.find(language) == supportedLanguages.end()) {
1099596a2c1Sopenharmony_ci        HILOG_ERROR_I18N("Taboo::ReplaceLanguageName taboo data don't support language %{public}s",
1109596a2c1Sopenharmony_ci            language.c_str());
1119596a2c1Sopenharmony_ci        return name;
1129596a2c1Sopenharmony_ci    }
1139596a2c1Sopenharmony_ci    std::string key = languageKey + language;
1149596a2c1Sopenharmony_ci    std::vector<std::string> fallbackLanguageKeys = QueryKeyFallBack(key);
1159596a2c1Sopenharmony_ci    std::string fallbackLanguage;
1169596a2c1Sopenharmony_ci    std::string fileName;
1179596a2c1Sopenharmony_ci    std::tie(fallbackLanguage, fileName) = LanguageFallBack(displayLanguage);
1189596a2c1Sopenharmony_ci    if (fallbackLanguage.size() == 0) {
1199596a2c1Sopenharmony_ci        HILOG_ERROR_I18N("Taboo::ReplaceLanguageName language %{public}s fallback failed", displayLanguage.c_str());
1209596a2c1Sopenharmony_ci        return name;
1219596a2c1Sopenharmony_ci    }
1229596a2c1Sopenharmony_ci    if (localeTabooData.find(fallbackLanguage) == localeTabooData.end()) {
1239596a2c1Sopenharmony_ci        localeTabooData[fallbackLanguage] = {};
1249596a2c1Sopenharmony_ci        std::string localeTabooDataFilePath = tabooDataPath + fileName + filePathSplitor + tabooLocaleDataFileName;
1259596a2c1Sopenharmony_ci        ParseTabooData(localeTabooDataFilePath, DataFileType::DATA_FILE, fallbackLanguage);
1269596a2c1Sopenharmony_ci    }
1279596a2c1Sopenharmony_ci    auto tabooData = localeTabooData[fallbackLanguage];
1289596a2c1Sopenharmony_ci    for (auto it = fallbackLanguageKeys.begin(); it != fallbackLanguageKeys.end(); ++it) {
1299596a2c1Sopenharmony_ci        if (tabooData.find(*it) != tabooData.end()) {
1309596a2c1Sopenharmony_ci            return tabooData[*it];
1319596a2c1Sopenharmony_ci        }
1329596a2c1Sopenharmony_ci    }
1339596a2c1Sopenharmony_ci    HILOG_ERROR_I18N("Taboo::ReplaceLanguageName not find taboo data correspond to language %{public}s",
1349596a2c1Sopenharmony_ci        language.c_str());
1359596a2c1Sopenharmony_ci    return name;
1369596a2c1Sopenharmony_ci}
1379596a2c1Sopenharmony_ci
1389596a2c1Sopenharmony_civoid Taboo::ParseTabooData(const std::string& path, DataFileType fileType, const std::string& locale)
1399596a2c1Sopenharmony_ci{
1409596a2c1Sopenharmony_ci    xmlKeepBlanksDefault(0);
1419596a2c1Sopenharmony_ci    xmlDocPtr doc = xmlParseFile(path.c_str());
1429596a2c1Sopenharmony_ci    if (doc == nullptr) {
1439596a2c1Sopenharmony_ci        HILOG_ERROR_I18N("Taboo parse taboo data file failed: %{public}s", path.c_str());
1449596a2c1Sopenharmony_ci        return;
1459596a2c1Sopenharmony_ci    }
1469596a2c1Sopenharmony_ci    xmlNodePtr cur = xmlDocGetRootElement(doc);
1479596a2c1Sopenharmony_ci    if (cur == nullptr || xmlStrcmp(cur->name, reinterpret_cast<const xmlChar*>(ROOT_TAG)) != 0) {
1489596a2c1Sopenharmony_ci        xmlFreeDoc(doc);
1499596a2c1Sopenharmony_ci        HILOG_ERROR_I18N("Taboo get root tag from taboo data file failed: %{public}s", path.c_str());
1509596a2c1Sopenharmony_ci        return;
1519596a2c1Sopenharmony_ci    }
1529596a2c1Sopenharmony_ci    cur = cur->xmlChildrenNode;
1539596a2c1Sopenharmony_ci    const xmlChar* nameTag = reinterpret_cast<const xmlChar*>(NAME_TAG);
1549596a2c1Sopenharmony_ci    const xmlChar* valueTag = reinterpret_cast<const xmlChar*>(VALUE_TAG);
1559596a2c1Sopenharmony_ci    while (cur != nullptr && xmlStrcmp(cur->name, reinterpret_cast<const xmlChar*>(ITEM_TAG)) == 0) {
1569596a2c1Sopenharmony_ci        xmlChar* name = xmlGetProp(cur, nameTag);
1579596a2c1Sopenharmony_ci        xmlChar* value = xmlGetProp(cur, valueTag);
1589596a2c1Sopenharmony_ci        if (name == nullptr || value == nullptr) {
1599596a2c1Sopenharmony_ci            HILOG_ERROR_I18N("Taboo get name and value property failed: %{public}s", path.c_str());
1609596a2c1Sopenharmony_ci            cur = cur->next;
1619596a2c1Sopenharmony_ci            continue;
1629596a2c1Sopenharmony_ci        }
1639596a2c1Sopenharmony_ci        std::string nameStr = reinterpret_cast<const char*>(name);
1649596a2c1Sopenharmony_ci        std::string valueStr = reinterpret_cast<const char*>(value);
1659596a2c1Sopenharmony_ci        switch (fileType) {
1669596a2c1Sopenharmony_ci            case DataFileType::CONFIG_FILE:
1679596a2c1Sopenharmony_ci                ProcessTabooConfigData(nameStr, valueStr);
1689596a2c1Sopenharmony_ci                break;
1699596a2c1Sopenharmony_ci            case DataFileType::DATA_FILE:
1709596a2c1Sopenharmony_ci                ProcessTabooLocaleData(locale, nameStr, valueStr);
1719596a2c1Sopenharmony_ci                break;
1729596a2c1Sopenharmony_ci        }
1739596a2c1Sopenharmony_ci        xmlFree(name);
1749596a2c1Sopenharmony_ci        xmlFree(value);
1759596a2c1Sopenharmony_ci        cur = cur->next;
1769596a2c1Sopenharmony_ci    }
1779596a2c1Sopenharmony_ci    xmlFreeDoc(doc);
1789596a2c1Sopenharmony_ci}
1799596a2c1Sopenharmony_ci
1809596a2c1Sopenharmony_civoid Taboo::ProcessTabooConfigData(const std::string& name, const std::string& value)
1819596a2c1Sopenharmony_ci{
1829596a2c1Sopenharmony_ci    if (name.compare(supportedRegionsTag) == 0) {
1839596a2c1Sopenharmony_ci        SplitValue(value, supportedRegions);
1849596a2c1Sopenharmony_ci    } else if (name.compare(supportedLanguagesTag) == 0) {
1859596a2c1Sopenharmony_ci        SplitValue(value, supportedLanguages);
1869596a2c1Sopenharmony_ci    }
1879596a2c1Sopenharmony_ci}
1889596a2c1Sopenharmony_ci
1899596a2c1Sopenharmony_civoid Taboo::ProcessTabooLocaleData(const std::string& locale, const std::string& name, const std::string& value)
1909596a2c1Sopenharmony_ci{
1919596a2c1Sopenharmony_ci    if (localeTabooData.find(locale) != localeTabooData.end()) {
1929596a2c1Sopenharmony_ci        localeTabooData[locale][name] = value;
1939596a2c1Sopenharmony_ci    } else {
1949596a2c1Sopenharmony_ci        std::map<std::string, std::string> data;
1959596a2c1Sopenharmony_ci        data[name] = value;
1969596a2c1Sopenharmony_ci        localeTabooData[locale] = data;
1979596a2c1Sopenharmony_ci    }
1989596a2c1Sopenharmony_ci}
1999596a2c1Sopenharmony_ci
2009596a2c1Sopenharmony_civoid Taboo::SplitValue(const std::string& value, std::set<std::string>& collation)
2019596a2c1Sopenharmony_ci{
2029596a2c1Sopenharmony_ci    size_t startPos = 0;
2039596a2c1Sopenharmony_ci    while (startPos < value.length()) {
2049596a2c1Sopenharmony_ci        size_t endPos = value.find(tabooDataSplitor, startPos);
2059596a2c1Sopenharmony_ci        if (endPos == std::string::npos) {
2069596a2c1Sopenharmony_ci            collation.insert(value.substr(startPos));
2079596a2c1Sopenharmony_ci            endPos = value.length();
2089596a2c1Sopenharmony_ci        } else {
2099596a2c1Sopenharmony_ci            collation.insert(value.substr(startPos, endPos - startPos));
2109596a2c1Sopenharmony_ci        }
2119596a2c1Sopenharmony_ci        startPos = endPos + 1;
2129596a2c1Sopenharmony_ci    }
2139596a2c1Sopenharmony_ci}
2149596a2c1Sopenharmony_ci
2159596a2c1Sopenharmony_cistd::vector<std::string> Taboo::QueryKeyFallBack(const std::string& key)
2169596a2c1Sopenharmony_ci{
2179596a2c1Sopenharmony_ci    std::vector<std::string> fallback;
2189596a2c1Sopenharmony_ci    fallback.push_back(key + "_r_all");
2199596a2c1Sopenharmony_ci    return fallback;
2209596a2c1Sopenharmony_ci}
2219596a2c1Sopenharmony_ci
2229596a2c1Sopenharmony_cistd::tuple<std::string, std::string> Taboo::LanguageFallBack(const std::string& language)
2239596a2c1Sopenharmony_ci{
2249596a2c1Sopenharmony_ci    std::string bestMatch;
2259596a2c1Sopenharmony_ci    std::string fileName;
2269596a2c1Sopenharmony_ci    int32_t bestScore = -1;
2279596a2c1Sopenharmony_ci
2289596a2c1Sopenharmony_ci    for (auto it = RESOURCES.begin(); it != RESOURCES.end(); ++it) {
2299596a2c1Sopenharmony_ci        std::string resLanguage = it->first;
2309596a2c1Sopenharmony_ci        int32_t score = LocaleCompare::Compare(language, resLanguage);
2319596a2c1Sopenharmony_ci        if (score > bestScore) {
2329596a2c1Sopenharmony_ci            bestMatch = resLanguage;
2339596a2c1Sopenharmony_ci            fileName = it->second;
2349596a2c1Sopenharmony_ci            bestScore = score;
2359596a2c1Sopenharmony_ci        }
2369596a2c1Sopenharmony_ci    }
2379596a2c1Sopenharmony_ci    if (bestScore < 0) {
2389596a2c1Sopenharmony_ci        return std::make_tuple("", "");
2399596a2c1Sopenharmony_ci    }
2409596a2c1Sopenharmony_ci    return std::make_tuple(bestMatch, fileName);
2419596a2c1Sopenharmony_ci}
2429596a2c1Sopenharmony_ci
2439596a2c1Sopenharmony_civoid Taboo::ReadResourceList()
2449596a2c1Sopenharmony_ci{
2459596a2c1Sopenharmony_ci    using std::filesystem::directory_iterator;
2469596a2c1Sopenharmony_ci    struct stat s;
2479596a2c1Sopenharmony_ci    for (const auto &dirEntry : directory_iterator{tabooDataPath}) {
2489596a2c1Sopenharmony_ci        std::string path = dirEntry.path();
2499596a2c1Sopenharmony_ci        if (stat(path.c_str(), &s) != 0) {
2509596a2c1Sopenharmony_ci            HILOG_ERROR_I18N("get path status failed");
2519596a2c1Sopenharmony_ci            continue;
2529596a2c1Sopenharmony_ci        }
2539596a2c1Sopenharmony_ci        if (s.st_mode & S_IFDIR) {
2549596a2c1Sopenharmony_ci            std::string fileName = path.substr(tabooDataPath.length());
2559596a2c1Sopenharmony_ci            std::string language = GetLanguageFromFileName(fileName);
2569596a2c1Sopenharmony_ci            RESOURCES[language] = fileName;
2579596a2c1Sopenharmony_ci        }
2589596a2c1Sopenharmony_ci    }
2599596a2c1Sopenharmony_ci}
2609596a2c1Sopenharmony_ci
2619596a2c1Sopenharmony_cistd::string Taboo::GetLanguageFromFileName(const std::string& fileName)
2629596a2c1Sopenharmony_ci{
2639596a2c1Sopenharmony_ci    if (fileName.length() == 3) { // 3 is the length of file 'xml'
2649596a2c1Sopenharmony_ci        return "en";
2659596a2c1Sopenharmony_ci    }
2669596a2c1Sopenharmony_ci    std::string language = fileName.substr(4);
2679596a2c1Sopenharmony_ci    if (language[0] == 'b' && language[1] == '+') {
2689596a2c1Sopenharmony_ci        language = language.substr(2); // 2 is the length of "b+"
2699596a2c1Sopenharmony_ci    }
2709596a2c1Sopenharmony_ci    size_t pos = language.find("+");
2719596a2c1Sopenharmony_ci    if (pos != std::string::npos) {
2729596a2c1Sopenharmony_ci        language = language.replace(pos, 1, "-");
2739596a2c1Sopenharmony_ci    }
2749596a2c1Sopenharmony_ci    pos = language.find("-r");
2759596a2c1Sopenharmony_ci    if (pos != std::string::npos) {
2769596a2c1Sopenharmony_ci        language = language.replace(pos, 2, "-"); // 2 is the length of "-r"
2779596a2c1Sopenharmony_ci    }
2789596a2c1Sopenharmony_ci    return language;
2799596a2c1Sopenharmony_ci}
2809596a2c1Sopenharmony_ci} // namespace I18n
2819596a2c1Sopenharmony_ci} // namespace Global
2829596a2c1Sopenharmony_ci} // namespace OHOS