Refactor: Move file extension checks from Tokenizer to Path class. This has also functional change as now also file.JAVA is considered a Java file.

aggro80 · aggro80 · commit e91d239ea5ca · 2012-01-06T21:56:28.000+02:00
diff --git a/lib/path.cpp b/lib/path.cpp
@@ -108,6 +108,13 @@ bool Path::sameFileName(const std::string &fname1, const std::string &fname2)
 #endif
 }
 
+// This wrapper exists because Sun's CC does not allow a static_cast
+// from extern "C" int(*)(int) to int(*)(int).
+static int tolowerWrapper(int c)
+{
+    return std::tolower(c);
+}
+
 std::string Path::removeQuotationMarks(std::string path)
 {
     path.erase(std::remove(path.begin(), path.end(), '\"'), path.end());
@@ -124,26 +131,29 @@ std::string Path::getFilenameExtension(const std::string &path)
     return extension;
 }
 
-
-// This wrapper exists because Sun's CC does not allow a static_cast
-// from extern "C" int(*)(int) to int(*)(int).
-static int tolowerWrapper(int c)
+std::string Path::getFilenameExtensionInLowerCase(const std::string &path)
 {
-    return std::tolower(c);
+    std::string extension = getFilenameExtension(path);
+    std::transform(extension.begin(), extension.end(), extension.begin(), tolowerWrapper);
+    return extension;
 }
 
-
-bool Path::acceptFile(const std::string &filename)
+bool Path::isC(const std::string &path)
 {
-    std::string extension = Path::getFilenameExtension(filename);
-    if (extension == "")
-        return false;
-    std::transform(extension.begin(), extension.end(), extension.begin(), tolowerWrapper);
+    const std::string extension = getFilenameExtensionInLowerCase(path);
+    if (extension == ".c") {
+        return true;
+    }
 
+    return false;
+}
+
+bool Path::isCPP(const std::string &path)
+{
+    const std::string extension = getFilenameExtensionInLowerCase(path);
     if (extension == ".cpp" ||
         extension == ".cxx" ||
         extension == ".cc" ||
-        extension == ".c" ||
         extension == ".c++" ||
         extension == ".tpp" ||
         extension == ".txx") {
@@ -153,3 +163,33 @@ bool Path::acceptFile(const std::string &filename)
     return false;
 }
 
+bool Path::isJava(const std::string &path)
+{
+    const std::string extension = getFilenameExtensionInLowerCase(path);
+    if (extension == ".java") {
+        return true;
+    }
+
+    return false;
+}
+
+bool Path::isCSharp(const std::string &path)
+{
+    const std::string extension = getFilenameExtensionInLowerCase(path);
+    if (extension == ".cs") {
+        return true;
+    }
+
+    return false;
+}
+
+bool Path::acceptFile(const std::string &filename)
+{
+    if (Path::isCPP(filename) ||
+        Path::isC(filename)) {
+        return true;
+    }
+
+    return false;
+}
+
diff --git a/lib/path.h b/lib/path.h
@@ -73,10 +73,17 @@ class Path {
     /**
       * @brief Get an extension of the filename.
       * @param path Path containing filename.
-      * @return Filename extension (containing the dot, e.g. ".h").
+      * @return Filename extension (containing the dot, e.g. ".h" or ".CPP").
       */
     static std::string getFilenameExtension(const std::string &path);
 
+    /**
+      * @brief Get an extension of the filename in lower case.
+      * @param path Path containing filename.
+      * @return Filename extension (containing the dot, e.g. ".h").
+      */
+    static std::string getFilenameExtensionInLowerCase(const std::string &path);
+
     /**
      * @brief Check if the file extension indicates that it's a C/C++ source file.
      * Check if the file has source file extension: *.c;*.cpp;*.cxx;*.c++;*.cc;*.txx
@@ -85,6 +92,33 @@ class Path {
      */
     static bool acceptFile(const std::string &filename);
 
+    /**
+     * @brief Identify language based on file extension.
+     * @param extensionInLowerCase e.g. ".c"
+     * @return true if extension is meant for C files
+     */
+    static bool isC(const std::string &extensionInLowerCase);
+
+    /**
+     * @brief Identify language based on file extension.
+     * @param extensionInLowerCase e.g. ".cpp"
+     * @return true if extension is meant for C++ files
+     */
+    static bool isCPP(const std::string &extensionInLowerCase);
+
+    /**
+     * @brief Identify language based on file extension.
+     * @param extensionInLowerCase e.g. ".java"
+     * @return true if extension is meant for Java files
+     */
+    static bool isJava(const std::string &extensionInLowerCase);
+
+    /**
+     * @brief Identify language based on file extension.
+     * @param extensionInLowerCase e.g. ".cs"
+     * @return true if extension is meant for C# files
+     */
+    static bool isCSharp(const std::string &extensionInLowerCase);
 };
 
 /// @}
diff --git a/lib/tokenize.cpp b/lib/tokenize.cpp
@@ -9503,3 +9503,29 @@ void Tokenizer::printUnknownTypes()
             _errorLogger->reportOut(ss.str());
     }
 }
+
+std::string Tokenizer::getSourceFilePath() const {
+    if (_files.empty())
+        return std::string("");
+    return _files[0];
+}
+
+bool Tokenizer::isJava() const {
+    return Path::isJava(getSourceFilePath());
+}
+
+bool Tokenizer::isCSharp() const {
+    return Path::isCSharp(getSourceFilePath());
+}
+
+bool Tokenizer::isJavaOrCSharp() const {
+    return isJava() || isCSharp();
+}
+
+bool Tokenizer::isC() const {
+    return Path::isC(getSourceFilePath());
+}
+
+bool Tokenizer::isCPP() const {
+    return Path::isCPP(getSourceFilePath());
+}
diff --git a/lib/tokenize.h b/lib/tokenize.h
@@ -49,38 +49,23 @@ class Tokenizer {
     Tokenizer(const Settings * settings, ErrorLogger *errorLogger);
     virtual ~Tokenizer();
 
-    /** The file extension. Used by isC() etc. */
-    std::string fileExtension() const {
-        if (_files.empty())
-            return std::string("");
-        return Path::getFilenameExtension(_files[0]);
-    }
+    /** Returns the source file path. e.g. "file.cpp" */
+    std::string getSourceFilePath() const;
 
     /** Is the code JAVA. Used for bailouts */
-    bool isJava() const {
-        return fileExtension() == ".java";
-    }
+    bool isJava() const;
 
     /** Is the code C#. Used for bailouts */
-    bool isCSharp() const {
-        return fileExtension() == ".cs";
-    }
+    bool isCSharp() const;
 
     /** Is the code JAVA/C#. Used for bailouts */
-    bool isJavaOrCSharp() const {
-        return isJava() || isCSharp();
-    }
+    bool isJavaOrCSharp() const;
 
     /** Is the code C. Used for bailouts */
-    bool isC() const {
-        const std::string ext = fileExtension();
-        return (ext == ".c" || ext == ".C");
-    }
+    bool isC() const;
 
     /** Is the code CPP. Used for bailouts */
-    bool isCPP() const {
-        return !isC() && (_files.size() && Path::acceptFile(_files[0]));
-    }
+    bool isCPP() const;
 
     /**
      * Check if inner scope ends with a call to a noreturn function
diff --git a/test/testpath.cpp b/test/testpath.cpp
@@ -29,6 +29,11 @@ class TestPath : public TestFixture {
 
     void run() {
         TEST_CASE(simplify_path);
+        TEST_CASE(accept_file);
+        TEST_CASE(is_c);
+        TEST_CASE(is_cpp);
+        TEST_CASE(is_java);
+        TEST_CASE(is_csharp);
     }
 
     void simplify_path() {
@@ -58,6 +63,48 @@ class TestPath : public TestFixture {
         ASSERT_EQUALS("the/path to/index.cpp", Path::removeQuotationMarks("the/\"path to\"/index.cpp"));
         ASSERT_EQUALS("the/path to/index.cpp", Path::removeQuotationMarks("\"the/path to/index.cpp\""));
     }
+
+    void accept_file() {
+        ASSERT(Path::acceptFile("index.cpp"));
+        ASSERT(Path::acceptFile("index.invalid.cpp"));
+        ASSERT(Path::acceptFile("index.invalid.Cpp"));
+        ASSERT(Path::acceptFile("index.invalid.C"));
+        ASSERT(Path::acceptFile("index.invalid.C++"));
+        ASSERT(Path::acceptFile("index.")==false);
+        ASSERT(Path::acceptFile("index")==false);
+        ASSERT(Path::acceptFile("")==false);
+        ASSERT(Path::acceptFile("C")==false);
+    }
+
+    void is_c() {
+        ASSERT(Path::isC("index.cpp")==false);
+        ASSERT(Path::isC("")==false);
+        ASSERT(Path::isC("c")==false);
+        ASSERT(Path::isC("index.c"));
+        ASSERT(Path::isC("C:\\foo\\index.c"));
+        ASSERT(Path::isC("C:\\foo\\index.C"));
+    }
+
+    void is_cpp() {
+        ASSERT(Path::isCPP("index.c")==false);
+        ASSERT(Path::isCPP("index.cpp"));
+        ASSERT(Path::isCPP("C:\\foo\\index.cpp"));
+        ASSERT(Path::isCPP("C:\\foo\\index.Cpp"));
+    }
+
+    void is_java() {
+        ASSERT(Path::isJava("index.cpp")==false);
+        ASSERT(Path::isJava("index.java"));
+        ASSERT(Path::isJava("C:\\foo\\index.java"));
+        ASSERT(Path::isJava("C:\\foo\\index.Java"));
+    }
+
+    void is_csharp() {
+        ASSERT(Path::isCSharp("index.cpp")==false);
+        ASSERT(Path::isCSharp("index.cs"));
+        ASSERT(Path::isCSharp("C:\\foo\\index.cs"));
+        ASSERT(Path::isCSharp("C:\\foo\\index.Cs"));
+    }
 };
 
 REGISTER_TEST(TestPath)
