Locale support - extract

README.md

@@ -310,6 +310,15 @@ Extract a single file using the `-f` option. If the target file in the MPQ archi
 mpqcli extract -f "Documentation\Layout\Greeting.html" "World of Warcraft_1.12.1.5875/Data/base.MPQ"
 ```
 
+### Extract one specific file with locale
+
+Use the `--locale` argument to specify the locale of the file to extract. If there is no file with the requested name and locale, the default locale will be used instead.
+
+```
+mpqcli extract -f "rez\gluBNRes.res" Patch_rt.mpq --locale deDE
+```
+
+
 ### Read a specific file from an MPQ archive
 
 Read the `patch.cmd` file from an MPQ archive and print the file contents to stdout. Even though the subcommand always outputs bytes, plaintext files will be human-readable.

src/main.cpp

@@ -137,6 +137,7 @@ int main(int argc, char **argv) {
     extract->add_flag("-k,--keep", extractKeepFolderStructure, "Keep folder structure (default false)");
     extract->add_option("-l,--listfile", baseListfileName, "File listing content of an MPQ archive")
         ->check(CLI::ExistingFile);
+    extract->add_option("--locale", baseLocale, "Preferred locale for extracted file");
 
     // Subcommand: Read
     CLI::App* read = app.add_subcommand("read", "Read a file from an MPQ archive");
@@ -296,10 +297,15 @@ int main(int argc, char **argv) {
             return 1;
         }
 
+        LCID locale = LangToLocale(baseLocale);
+        if (baseLocale != "default" && locale == defaultLocale) {
+            std::cout << "[!] Warning: The locale '" << baseLocale << "' is unknown. Will use default locale instead." << std::endl;
+        }
+
         if (baseFile != "default") {
-            ExtractFile(hArchive, baseOutput, baseFile, extractKeepFolderStructure);
+            ExtractFile(hArchive, baseOutput, baseFile, extractKeepFolderStructure, locale);
         } else {
-            ExtractFiles(hArchive, baseOutput, baseListfileName);
+            ExtractFiles(hArchive, baseOutput, baseListfileName, locale);
         }
     }
 

src/mpq.cpp

@@ -38,7 +38,8 @@ int SignMpqArchive(HANDLE hArchive) {
     return 1;
 }
 
-int ExtractFiles(HANDLE hArchive, const std::string& output, const std::string& listfileName) {
+int ExtractFiles(HANDLE hArchive, const std::string& output, const std::string& listfileName, LCID preferredLocale) {
+    SFileSetLocale(preferredLocale);
     // Check if the user provided a listfile input
     const char *listfile = (listfileName == "default") ? NULL : listfileName.c_str();
 
@@ -55,7 +56,8 @@ int ExtractFiles(HANDLE hArchive, const std::string& output, const std::string&
             hArchive,
             output,
             findData.cFileName,
-            true  // Keep folder structure
+            true,  // Keep folder structure
+            preferredLocale
         );
         if (result != 0) {
             return result;
@@ -67,7 +69,8 @@ int ExtractFiles(HANDLE hArchive, const std::string& output, const std::string&
     return 0;
 }
 
-int ExtractFile(HANDLE hArchive, const std::string& output, const std::string& fileName, bool keepFolderStructure) {
+int ExtractFile(HANDLE hArchive, const std::string& output, const std::string& fileName, bool keepFolderStructure, LCID preferredLocale) {
+    SFileSetLocale(preferredLocale);
     const char *szFileName = fileName.c_str();
     if (!SFileHasFile(hArchive, szFileName)) {
         std::cerr << "[!] Failed: File doesn't exist: " << szFileName << std::endl;

src/mpq.h

@@ -11,8 +11,8 @@ namespace fs = std::filesystem;
 int OpenMpqArchive(const std::string &filename, HANDLE *hArchive, int32_t flags);
 int CloseMpqArchive(HANDLE hArchive);
 int SignMpqArchive(HANDLE hArchive);
-int ExtractFiles(HANDLE hArchive, const std::string& output, const std::string &listfileName);
-int ExtractFile(HANDLE hArchive, const std::string& output, const std::string& fileName, bool keepFolderStructure);
+int ExtractFiles(HANDLE hArchive, const std::string& output, const std::string &listfileName, LCID preferredLocale);
+int ExtractFile(HANDLE hArchive, const std::string& output, const std::string& fileName, bool keepFolderStructure, LCID preferredLocale);
 HANDLE CreateMpqArchive(std::string outputArchiveName, int32_t fileCount, int32_t mpqVersion);
 int AddFiles(HANDLE hArchive, const std::string& inputPath, LCID locale);
 int AddFile(HANDLE hArchive, const fs::path& localFile, const std::string& archiveFilePath, LCID locale);

test/test_extract.py

@@ -127,6 +127,8 @@ def test_extract_file_from_mpq_output_directory_specified(binary_path, generate_
     script_dir = Path(__file__).parent
     test_file = script_dir / "data" / "mpq_with_output_v1.mpq"
     output_dir = script_dir / "data" / "extracted_file"
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
 
 
     expected_output = {
@@ -155,3 +157,234 @@ def test_extract_file_from_mpq_output_directory_specified(binary_path, generate_
     assert output_lines == expected_lines, f"Unexpected output: {output_lines}"
     assert output_file.exists(), "Output directory was not created"
     assert output_files == expected_output, f"Unexpected files: {output_files}"
+
+
+def test_extract_file_from_mpq_with_locale(binary_path, generate_locales_mpq_test_files):
+    """
+    Test MPQ archive file extraction with a specified locale.
+
+    This test checks:
+    - If the file with the given locale is extracted correctly.
+    - If the output files match the expected files.
+    """
+    _ = generate_locales_mpq_test_files
+    script_dir = Path(__file__).parent
+    test_file = script_dir / "data" / "mpq_with_many_locales.mpq"
+    output_dir = script_dir / "data" / "extracted_file"
+    file_to_extract = "cats.txt"
+    locale = "esES"
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
+
+    result = subprocess.run(
+        [str(binary_path), "extract", "-o", str(output_dir), "-f", file_to_extract, str(test_file), "--locale", locale],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        text=True
+    )
+
+    expected_stdout = {
+        "[*] Extracted: " + file_to_extract
+    }
+    output_file = output_dir / file_to_extract
+    expected_content = "Este es un archivo sobre gatos."
+
+    output_lines = set(result.stdout.splitlines())
+    assert result.returncode == 0, f"mpqcli failed with error: {result.stderr}"
+    assert output_lines == expected_stdout, f"Unexpected output: {output_lines}"
+    assert output_file.exists(), "Output directory was not created"
+    assert output_file.read_text(encoding="utf-8") == expected_content, "Unexpected file content"
+
+
+def test_extract_file_from_mpq_with_default_locale(binary_path, generate_locales_mpq_test_files):
+    """
+    Test MPQ archive file extraction with no specified locale but many matching file names.
+
+    This test checks:
+    - If the file with the default locale is extracted correctly.
+    - If the output files match the expected files.
+    """
+    _ = generate_locales_mpq_test_files
+    script_dir = Path(__file__).parent
+    test_file = script_dir / "data" / "mpq_with_many_locales.mpq"
+    output_dir = script_dir / "data" / "extracted_file"
+    file_to_extract = "cats.txt"
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
+
+
+    result = subprocess.run(
+        [str(binary_path), "extract", "-o", str(output_dir), "-f", file_to_extract, str(test_file)],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        text=True
+    )
+
+    expected_stdout = {
+        "[*] Extracted: " + file_to_extract
+    }
+    output_file = output_dir / file_to_extract
+    expected_content = "This is a file about cats."
+
+    output_lines = set(result.stdout.splitlines())
+    assert result.returncode == 0, f"mpqcli failed with error: {result.stderr}"
+    assert output_lines == expected_stdout, f"Unexpected output: {output_lines}"
+    assert output_file.exists(), "Output directory was not created"
+    assert output_file.read_text(encoding="utf-8") == expected_content, "Unexpected file content"
+
+
+def test_extract_file_from_mpq_with_illegal_locale(binary_path, generate_locales_mpq_test_files):
+    """
+    Test MPQ archive file extraction with an illegal locale.
+
+    This test checks:
+    - When a locale is given that does not exist in the file, the file with the default locale is extracted.
+    - If the output files match the expected files.
+    """
+    _ = generate_locales_mpq_test_files
+    script_dir = Path(__file__).parent
+    test_file = script_dir / "data" / "mpq_with_many_locales.mpq"
+    output_dir = script_dir / "data" / "extracted_file"
+    file_to_extract = "cats.txt"
+    locale = "nosuchlocale"
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
+
+
+    result = subprocess.run(
+        [str(binary_path), "extract", "-o", str(output_dir), "-f", file_to_extract, str(test_file), "--locale", locale],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        text=True
+    )
+
+    expected_stdout = {
+        "[!] Warning: The locale 'nosuchlocale' is unknown. Will use default locale instead.",
+        "[*] Extracted: " + file_to_extract
+    }
+    output_file = output_dir / file_to_extract
+    expected_content = "This is a file about cats."
+
+    output_lines = set(result.stdout.splitlines())
+    assert result.returncode == 0, f"mpqcli failed with error: {result.stderr}"
+    assert output_lines == expected_stdout, f"Unexpected output: {output_lines}"
+    assert output_file.exists(), "Output directory was not created"
+    assert output_file.read_text(encoding="utf-8") == expected_content, "Unexpected file content"
+
+
+def test_extract_file_from_mpq_with_locale_not_in_file(binary_path, generate_locales_mpq_test_files):
+    """
+    Test MPQ archive file extraction with a specified locale that is not in the file.
+
+    This test checks:
+    - When a locale is given that does not exist in the file, the file with the default locale is extracted.
+    - If the output files match the expected files.
+    """
+    _ = generate_locales_mpq_test_files
+    script_dir = Path(__file__).parent
+    test_file = script_dir / "data" / "mpq_with_many_locales.mpq"
+    output_dir = script_dir / "data" / "extracted_file"
+    file_to_extract = "cats.txt"
+    locale = "ptPT"  # There is no file for this locale
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
+
+    result = subprocess.run(
+        [str(binary_path), "extract", "-o", str(output_dir), "-f", file_to_extract, str(test_file), "--locale", locale],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        text=True
+    )
+
+    expected_stdout = {
+        "[*] Extracted: " + file_to_extract
+    }
+    output_file = output_dir / file_to_extract
+    expected_content = "This is a file about cats."
+
+    output_lines = set(result.stdout.splitlines())
+    assert result.returncode == 0, f"mpqcli failed with error: {result.stderr}"
+    assert output_lines == expected_stdout, f"Unexpected output: {output_lines}"
+    assert output_file.exists(), "Output directory was not created"
+    assert output_file.read_text(encoding="utf-8") == expected_content, "Unexpected file content"
+
+
+def test_extract_file_from_mpq_with_no_locale_argument_and_no_default_locale(binary_path, generate_locales_mpq_test_files):
+    """
+    Test MPQ archive file extraction without a specified locale, and no file with the default locale.
+
+    This test checks:
+    - When no locale is given, and no file by that name exists for the default locale,
+    but one does for a different one, no file is extracted.
+    """
+    _ = generate_locales_mpq_test_files
+    script_dir = Path(__file__).parent
+    test_file = script_dir / "data" / "mpq_with_one_locale.mpq"
+    output_dir = script_dir / "data" / "extracted_file"
+    file_to_extract = "cats.txt"  # There is a file by this name, but for locale esES
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
+
+
+    result = subprocess.run(
+        [str(binary_path), "extract", "-o", str(output_dir), "-f", file_to_extract, str(test_file)],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        text=True
+    )
+
+    expected_stdout_output = set()
+    expected_stderr_output = {
+        "[!] Failed: File doesn't exist: " + file_to_extract,
+    }
+
+    stdout_output_lines = set(result.stdout.splitlines())
+    stderr_output_lines = set(result.stderr.splitlines())
+
+    output_file = output_dir / file_to_extract
+
+    assert result.returncode == 0, f"mpqcli failed with error: {result.stderr}"
+    assert stdout_output_lines == expected_stdout_output, f"Unexpected output: {stdout_output_lines}"
+    assert stderr_output_lines == expected_stderr_output, f"Unexpected output: {stderr_output_lines}"
+    assert not output_file.exists(), "Output directory was not created"
+
+
+def test_extract_file_from_mpq_with_wrong_locale_argument_and_no_default_locale(binary_path, generate_locales_mpq_test_files):
+    """
+    Test MPQ archive file extraction with a specified locale, and no file with the default locale.
+
+    This test checks:
+    - When no locale is given, and no file by that name exists for the default locale,
+    but one does for a different one, no file is extracted.
+    """
+    _ = generate_locales_mpq_test_files
+    script_dir = Path(__file__).parent
+    test_file = script_dir / "data" / "mpq_with_one_locale.mpq"
+    output_dir = script_dir / "data" / "extracted_file"
+    file_to_extract = "cats.txt"  # There is a file by this name, but for locale esES
+    locale = "deDE"
+    for fi in output_dir.glob("*"):
+        fi.unlink(missing_ok=True)
+
+
+    result = subprocess.run(
+        [str(binary_path), "extract", "-o", str(output_dir), "-f", file_to_extract, str(test_file), "--locale", locale],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+        text=True
+    )
+
+    expected_stdout_output = set()
+    expected_stderr_output = {
+        "[!] Failed: File doesn't exist: " + file_to_extract,
+    }
+
+    stdout_output_lines = set(result.stdout.splitlines())
+    stderr_output_lines = set(result.stderr.splitlines())
+
+    output_file = output_dir / file_to_extract
+
+    assert result.returncode == 0, f"mpqcli failed with error: {result.stderr}"
+    assert stdout_output_lines == expected_stdout_output, f"Unexpected output: {stdout_output_lines}"
+    assert stderr_output_lines == expected_stderr_output, f"Unexpected output: {stderr_output_lines}"
+    assert not output_file.exists(), "Output directory was not created"