diff --git a/tools/asset_extract/extractor.cpp b/tools/asset_extract/extractor.cpp index 3c61bef3..7a847806 100644 --- a/tools/asset_extract/extractor.cpp +++ b/tools/asset_extract/extractor.cpp @@ -14,7 +14,6 @@ #include #include #include -#include #include #include #include @@ -32,12 +31,6 @@ namespace tools { namespace fs = std::filesystem; using wowee::pipeline::DBCFile; -// Archive descriptor for priority-based loading -struct ArchiveDesc { - std::string path; - int priority; -}; - static std::string toLowerStr(std::string s) { std::transform(s.begin(), s.end(), s.begin(), [](unsigned char c) { return static_cast(std::tolower(c)); }); @@ -394,163 +387,150 @@ std::string Extractor::detectExpansion(const std::string& mpqDir) { return ""; } -std::string Extractor::detectLocale(const std::string& mpqDir) { - for (const auto& loc : kKnownLocales) { - if (fs::is_directory(mpqDir + "/" + loc)) - return loc; +static std::string findCaseInsensitiveDirectory(const std::string& parentDir, + const std::string& directoryName) { + if (!fs::exists(parentDir) || !fs::is_directory(parentDir)) return ""; + std::string lowerDirectoryName = toLowerStr(directoryName); + for (const auto& entry : fs::directory_iterator(parentDir)) { + if (!entry.is_directory()) continue; + std::string name = entry.path().filename().string(); + if (toLowerStr(name) == lowerDirectoryName) { + return name; + } } return ""; } +std::string Extractor::detectLocale(const std::string& mpqDir) { + if (!fs::exists(mpqDir) || !fs::is_directory(mpqDir)) return ""; + for (const auto& entry : fs::directory_iterator(mpqDir)) { + if (!entry.is_directory()) continue; + std::string name = entry.path().filename().string(); + std::string lower = toLowerStr(name); + for (const auto& loc : kKnownLocales) { + if (toLowerStr(loc) == lower) { + return name; + } + } + } + return ""; +} + +static std::unordered_map buildCaseMap(const std::string& dir) { + std::unordered_map map; + if (!fs::exists(dir) || !fs::is_directory(dir)) return map; + for (const auto& entry : fs::directory_iterator(dir)) { + if (entry.is_regular_file()) { + std::string filename = entry.path().filename().string(); + if (filename.rfind("._", 0) == 0) { + continue; + } + std::string ext = toLowerStr(entry.path().extension().string()); + if (ext == ".mpq") { + std::string lower = toLowerStr(filename); + map[lower] = filename; + } + } + } + return map; +} + // Discover archive files with expansion-specific and locale-aware loading -static std::vector discoverArchives(const std::string& mpqDir, +static std::vector discoverArchives(const std::string& mpqDir, const std::string& expansion, const std::string& locale) { - std::vector result; + std::vector result; - auto tryAdd = [&](const std::string& name, int prio) { - std::string fullPath = mpqDir + "/" + name; - if (fs::exists(fullPath)) { - result.push_back({fullPath, prio}); + auto caseMap = buildCaseMap(mpqDir); + std::string lowerLocale = toLowerStr(locale); + if (!locale.empty()) { + std::string actualLocaleDir = findCaseInsensitiveDirectory(mpqDir, locale); + if (actualLocaleDir.empty()) { + actualLocaleDir = locale; + } + std::string localeDir = mpqDir + "/" + actualLocaleDir; + std::cout << "Locale directory: " << localeDir << "\n"; + auto localeMap = buildCaseMap(localeDir); + for (auto& [name, realName] : localeMap) { + caseMap[lowerLocale + "/" + name] = actualLocaleDir + "/" + realName; + } + } + + std::vector baseSequence; + std::vector localeSequence; + + if (expansion == "classic" || expansion == "turtle") { + baseSequence = { + "base.mpq", "backup.mpq", "dbc.mpq", "fonts.mpq", + "interface.mpq", "misc.mpq", "model.mpq", "sound.mpq", + "speech.mpq", "terrain.mpq", "texture.mpq", "wmo.mpq" + }; + } else if (expansion == "tbc") { + baseSequence = { "common.mpq", "expansion.mpq" }; + if (!locale.empty()) { + localeSequence = { + lowerLocale + "/backup-" + lowerLocale + ".mpq", + lowerLocale + "/base-" + lowerLocale + ".mpq", + lowerLocale + "/locale-" + lowerLocale + ".mpq", + lowerLocale + "/speech-" + lowerLocale + ".mpq", + lowerLocale + "/expansion-locale-" + lowerLocale + ".mpq", + lowerLocale + "/expansion-speech-" + lowerLocale + ".mpq", + }; + } + } else { + baseSequence = { "common.mpq", "common-2.mpq", "expansion.mpq", "lichking.mpq" }; + if (!locale.empty()) { + localeSequence = { + lowerLocale + "/backup-" + lowerLocale + ".mpq", + lowerLocale + "/base-" + lowerLocale + ".mpq", + lowerLocale + "/locale-" + lowerLocale + ".mpq", + lowerLocale + "/speech-" + lowerLocale + ".mpq", + lowerLocale + "/expansion-locale-" + lowerLocale + ".mpq", + lowerLocale + "/expansion-speech-" + lowerLocale + ".mpq", + lowerLocale + "/lichking-locale-" + lowerLocale + ".mpq", + lowerLocale + "/lichking-speech-" + lowerLocale + ".mpq", + }; + } + } + + std::vector sequence; + for (const auto& name : baseSequence) { + sequence.push_back(name); + } + for (const auto& name : localeSequence) { + sequence.push_back(name); + } + + // Interleave patches: base patch then locale patch for each tier + std::vector patchSuffixes = {""}; + for (int i = 2; i <= 9; ++i) { + patchSuffixes.push_back(std::string("-") + std::to_string(i)); + } + for (char c = 'a'; c <= 'z'; ++c) { + patchSuffixes.push_back(std::string("-") + c); + } + + for (const auto& suffix : patchSuffixes) { + sequence.push_back("patch" + suffix + ".mpq"); + if (!locale.empty()) { + sequence.push_back(lowerLocale + "/patch-" + lowerLocale + suffix + ".mpq"); + } + } + + auto addIfPresent = [&](const std::string& expected) { + auto it = caseMap.find(toLowerStr(expected)); + if (it != caseMap.end()) { + result.push_back(mpqDir + "/" + it->second); } }; - if (expansion == "classic" || expansion == "turtle") { - // Vanilla-era base archives (also used by Turtle WoW clients) - tryAdd("base.MPQ", 90); - tryAdd("base.mpq", 90); - tryAdd("backup.MPQ", 95); - tryAdd("backup.mpq", 95); - tryAdd("dbc.MPQ", 100); - tryAdd("dbc.mpq", 100); - tryAdd("fonts.MPQ", 100); - tryAdd("fonts.mpq", 100); - tryAdd("interface.MPQ", 100); - tryAdd("interface.mpq", 100); - tryAdd("misc.MPQ", 100); - tryAdd("misc.mpq", 100); - tryAdd("model.MPQ", 100); - tryAdd("model.mpq", 100); - tryAdd("sound.MPQ", 100); - tryAdd("sound.mpq", 100); - tryAdd("speech.MPQ", 100); - tryAdd("speech.mpq", 100); - tryAdd("terrain.MPQ", 100); - tryAdd("terrain.mpq", 100); - tryAdd("texture.MPQ", 100); - tryAdd("texture.mpq", 100); - tryAdd("wmo.MPQ", 100); - tryAdd("wmo.mpq", 100); - - // Patches - tryAdd("patch.MPQ", 150); - tryAdd("patch.mpq", 150); - for (int i = 1; i <= 9; ++i) { - tryAdd("patch-" + std::to_string(i) + ".MPQ", 160 + (i * 10)); - tryAdd("patch-" + std::to_string(i) + ".mpq", 160 + (i * 10)); - } - // Turtle WoW uses letter patch MPQs (patch-a.mpq ... patch-z.mpq). - for (char c = 'a'; c <= 'z'; ++c) { - tryAdd(std::string("patch-") + c + ".mpq", 800 + (c - 'a')); - tryAdd(std::string("Patch-") + static_cast(std::toupper(c)) + ".mpq", 900 + (c - 'a')); - } - - // Locale - if (!locale.empty()) { - tryAdd(locale + "/base-" + locale + ".MPQ", 230); - tryAdd(locale + "/speech-" + locale + ".MPQ", 240); - tryAdd(locale + "/locale-" + locale + ".MPQ", 250); - tryAdd(locale + "/patch-" + locale + ".MPQ", 450); - } - } else if (expansion == "tbc") { - // TBC 2.4.x base archives - tryAdd("common.MPQ", 100); - tryAdd("common-2.MPQ", 100); - tryAdd("expansion.MPQ", 100); - - // Patches - tryAdd("patch.MPQ", 150); - tryAdd("patch-2.MPQ", 200); - tryAdd("patch-3.MPQ", 300); - tryAdd("patch-4.MPQ", 400); - tryAdd("patch-5.MPQ", 500); - - // Letter patches - for (char c = 'a'; c <= 'z'; ++c) { - tryAdd(std::string("patch-") + c + ".mpq", 800 + (c - 'a')); - tryAdd(std::string("Patch-") + static_cast(std::toupper(c)) + ".mpq", 900 + (c - 'a')); - } - - // Locale - if (!locale.empty()) { - tryAdd(locale + "/backup-" + locale + ".MPQ", 225); - tryAdd(locale + "/base-" + locale + ".MPQ", 230); - tryAdd(locale + "/speech-" + locale + ".MPQ", 240); - tryAdd(locale + "/expansion-speech-" + locale + ".MPQ", 245); - tryAdd(locale + "/expansion-locale-" + locale + ".MPQ", 246); - tryAdd(locale + "/locale-" + locale + ".MPQ", 250); - tryAdd(locale + "/patch-" + locale + ".MPQ", 450); - tryAdd(locale + "/patch-" + locale + "-2.MPQ", 460); - tryAdd(locale + "/patch-" + locale + "-3.MPQ", 470); - } - } else { - // WotLK 3.3.5a (default) - tryAdd("common.MPQ", 100); - tryAdd("common-2.MPQ", 100); - tryAdd("expansion.MPQ", 100); - tryAdd("lichking.MPQ", 100); - - // Patches - tryAdd("patch.MPQ", 150); - tryAdd("patch-2.MPQ", 200); - tryAdd("patch-3.MPQ", 300); - tryAdd("patch-4.MPQ", 400); - tryAdd("patch-5.MPQ", 500); - - // Letter patches - for (char c = 'a'; c <= 'z'; ++c) { - tryAdd(std::string("patch-") + c + ".mpq", 800 + (c - 'a')); - tryAdd(std::string("Patch-") + static_cast(std::toupper(c)) + ".mpq", 900 + (c - 'a')); - } - - // Locale - if (!locale.empty()) { - tryAdd(locale + "/backup-" + locale + ".MPQ", 225); - tryAdd(locale + "/base-" + locale + ".MPQ", 230); - tryAdd(locale + "/speech-" + locale + ".MPQ", 240); - tryAdd(locale + "/expansion-speech-" + locale + ".MPQ", 245); - tryAdd(locale + "/expansion-locale-" + locale + ".MPQ", 246); - tryAdd(locale + "/lichking-speech-" + locale + ".MPQ", 248); - tryAdd(locale + "/lichking-locale-" + locale + ".MPQ", 249); - tryAdd(locale + "/locale-" + locale + ".MPQ", 250); - tryAdd(locale + "/patch-" + locale + ".MPQ", 450); - tryAdd(locale + "/patch-" + locale + "-2.MPQ", 460); - tryAdd(locale + "/patch-" + locale + "-3.MPQ", 470); - } + for (const auto& entry : sequence) { + addIfPresent(entry); } - // Sort by priority so highest-priority archives are last - // (we'll iterate highest-prio first when extracting) - std::sort(result.begin(), result.end(), - [](const ArchiveDesc& a, const ArchiveDesc& b) { return a.priority < b.priority; }); - return result; } -// Read a text file into a vector of lines (for external listfile loading) -static std::vector readLines(const std::string& path) { - std::vector lines; - std::ifstream f(path); - if (!f) return lines; - std::string line; - while (std::getline(f, line)) { - // Trim trailing \r - if (!line.empty() && line.back() == '\r') line.pop_back(); - if (!line.empty()) lines.push_back(std::move(line)); - } - return lines; -} - // Extract the (listfile) from an MPQ archive into a set of filenames static void extractInternalListfile(HANDLE hMpq, std::set& out) { HANDLE hFile = nullptr; @@ -595,14 +575,9 @@ bool Extractor::enumerateFiles(const Options& opts, std::cout << "Found " << archives.size() << " MPQ archives\n"; - // Load external listfile into memory once (avoids repeated file I/O) - std::vector externalEntries; - std::vector externalPtrs; - if (!opts.listFile.empty()) { - externalEntries = readLines(opts.listFile); - externalPtrs.reserve(externalEntries.size()); - for (const auto& e : externalEntries) externalPtrs.push_back(e.c_str()); - std::cout << " Loaded external listfile: " << externalEntries.size() << " entries\n"; + const bool haveExternalListFile = !opts.listFile.empty(); + if (haveExternalListFile) { + std::cout << " Using external listfile: " << opts.listFile << "\n"; } const auto wantedDbcs = buildWantedDbcSet(opts); @@ -611,21 +586,20 @@ bool Extractor::enumerateFiles(const Options& opts, // Enumerate from highest priority first so first-seen files win for (auto it = archives.rbegin(); it != archives.rend(); ++it) { HANDLE hMpq = nullptr; - if (!SFileOpenArchive(it->path.c_str(), 0, 0, &hMpq)) { - std::cerr << " Failed to open: " << it->path << "\n"; + if (!SFileOpenArchive(it->c_str(), 0, 0, &hMpq)) { + std::cerr << " Failed to open: " << *it << "\n"; continue; } - // Inject external listfile entries into archive's in-memory name table. - // SFileAddListFileEntries is fast — it only hashes the names against the - // archive's hash table, no file I/O involved. - if (!externalPtrs.empty()) { - SFileAddListFileEntries(hMpq, externalPtrs.data(), - static_cast(externalPtrs.size())); + // Inject external listfile into archive's in-memory name table. + // SFileAddListFile reads the file and hashes names against the + // archive's hash table. + if (haveExternalListFile) { + SFileAddListFile(hMpq, opts.listFile.c_str()); } if (opts.verbose) { - std::cout << " Scanning: " << it->path << " (priority " << it->priority << ")\n"; + std::cout << " Scanning: " << *it << "\n"; } SFILE_FIND_DATA findData; @@ -721,19 +695,18 @@ bool Extractor::run(const Options& opts) { // thread-safe even with separate handles, so we serialize all MPQ reads. struct SharedArchive { HANDLE handle; - int priority; std::string path; }; std::vector sharedHandles; - for (const auto& ad : archives) { + for (const auto& path : archives) { HANDLE h = nullptr; - if (SFileOpenArchive(ad.path.c_str(), 0, 0, &h)) { + if (SFileOpenArchive(path.c_str(), 0, 0, &h)) { if (!opts.listFile.empty()) { SFileAddListFile(h, opts.listFile.c_str()); } - sharedHandles.push_back({h, ad.priority, ad.path}); + sharedHandles.push_back({h, path}); } else { - std::cerr << " Failed to open archive: " << ad.path << "\n"; + std::cerr << " Failed to open archive: " << path << "\n"; } } if (sharedHandles.empty()) {