pcre2

Author: firewave

cli/cmdlineparser.cpp

@@ -1356,6 +1356,9 @@ CmdLineParser::Result CmdLineParser::parseFromArgs(int argc, const char* const a
                             if (std::strcmp(engine, "pcre") == 0) {
                                 rule.engine = Regex::Engine::Pcre;
                             }
+                            else if (std::strcmp(engine, "pcre2") == 0) {
+                                rule.engine = Regex::Engine::Pcre2;
+                            }
                             else {
                                 mLogger.printError(std::string("unknown regex engine '") + engine + "'.");
                                 return Result::Fail;

cmake/findDependencies.cmake

@@ -32,7 +32,13 @@ if(HAVE_RULES)
     find_path(PCRE_INCLUDE pcre.h)
     find_library(PCRE_LIBRARY NAMES pcre pcred)
     if(NOT PCRE_LIBRARY OR NOT PCRE_INCLUDE)
-        message(FATAL_ERROR "pcre dependency for RULES has not been found")
+        message(FATAL_ERROR "PCRE dependency for RULES has not been found")
+    endif()
+
+    find_path(PCRE2_INCLUDE pcre2.h)
+    find_library(PCRE2_LIBRARY NAMES pcre2-8) # TODO: is this the proper library?
+    if(NOT PCRE2_LIBRARY OR NOT PCRE2_INCLUDE)
+        message(FATAL_ERROR "PCRE2 dependency for RULES has not been found")
     endif()
 else()
     set(PCRE_LIBRARY "")

cmake/printInfo.cmake

@@ -80,6 +80,8 @@ message(STATUS)
 message(STATUS "HAVE_RULES =            ${HAVE_RULES}")
 if(HAVE_RULES)
     message(STATUS "PCRE_LIBRARY =          ${PCRE_LIBRARY}")
+    message(STATUS "PCRE2_INCLUDE =         ${PCRE2_INCLUDE}")
+    message(STATUS "PCRE2_LIBRARY =         ${PCRE2_LIBRARY}")
 endif()
 message(STATUS)
 message(STATUS "DISALLOW_THREAD_EXECUTOR = ${DISALLOW_THREAD_EXECUTOR}")

gui/test/resultstree/CMakeLists.txt

@@ -24,8 +24,8 @@ add_executable(test-resultstree
 target_include_directories(test-resultstree PRIVATE ${CMAKE_SOURCE_DIR}/gui)
 target_link_libraries(test-resultstree cppcheck-core simplecpp tinyxml2)
 if (HAVE_RULES)
-    target_link_libraries(test-resultstree ${PCRE_LIBRARY})
-    target_include_directories(test-resultstree SYSTEM PRIVATE ${PCRE_INCLUDE})
+    target_link_libraries(test-resultstree ${PCRE_LIBRARY} ${PCRE2_LIBRARY})
+    target_include_directories(test-resultstree SYSTEM PRIVATE ${PCRE_INCLUDE} ${PCRE2_INCLUDE})
 endif()
 target_compile_definitions(test-resultstree PRIVATE SRCDIR="${CMAKE_CURRENT_SOURCE_DIR}")
 target_link_libraries(test-resultstree ${QT_CORE_LIB} ${QT_GUI_LIB} ${QT_WIDGETS_LIB} ${QT_TEST_LIB})

lib/CMakeLists.txt

@@ -53,10 +53,10 @@ endif()
 target_dll_compile_definitions(cppcheck-core EXPORT CPPCHECKLIB_EXPORT IMPORT CPPCHECKLIB_IMPORT)
 
 target_include_directories(cppcheck-core PUBLIC .)
-target_link_libraries(cppcheck-core PRIVATE tinyxml2 simplecpp picojson ${PCRE_LIBRARY})
+target_link_libraries(cppcheck-core PRIVATE tinyxml2 simplecpp picojson ${PCRE_LIBRARY} ${PCRE2_LIBRARY})
 
 if (HAVE_RULES)
-    target_include_directories(cppcheck-core SYSTEM PRIVATE ${PCRE_INCLUDE})
+    target_include_directories(cppcheck-core SYSTEM PRIVATE ${PCRE_INCLUDE} ${PCRE2_INCLUDE})
 endif()
 if (Boost_FOUND)
     target_include_directories(cppcheck-core SYSTEM PUBLIC ${Boost_INCLUDE_DIRS})

lib/regex.cpp

@@ -27,6 +27,9 @@
 #endif
 #include <pcre.h>
 
+#define PCRE2_CODE_UNIT_WIDTH 8
+#include <pcre2.h>
+
 namespace {
     class PcreRegex : public Regex
     {
@@ -248,6 +251,85 @@ namespace {
     }
 }
 
+namespace {
+    class Pcre2Regex : public Regex
+    {
+    public:
+        explicit Pcre2Regex(std::string pattern)
+            : mPattern(std::move(pattern))
+        {}
+
+        ~Pcre2Regex() override
+        {
+            if (mRe) {
+                pcre_free(mRe);
+                mRe = nullptr;
+            }
+        }
+
+        std::string compile();
+        std::string match(const std::string& str, const MatchFn& match) const override;
+
+    private:
+        std::string mPattern;
+        pcre2_code* mRe{};
+    };
+
+    std::string Pcre2Regex::compile()
+    {
+        if (mRe)
+            return "regular expression has already been compiled";
+
+        int errnumber = 0;
+        size_t erroffset = 0;
+        pcre2_code * const re = pcre2_compile(reinterpret_cast<PCRE2_SPTR8>(mPattern.c_str()), PCRE2_ZERO_TERMINATED, 0, &errnumber, &erroffset, nullptr);
+        if (!re) {
+            PCRE2_UCHAR buffer[256];
+            pcre2_get_error_message(errnumber, buffer, sizeof(buffer));
+            return reinterpret_cast<char*>(buffer);
+        }
+
+        mRe = re;
+
+        return "";
+    }
+
+    std::string Pcre2Regex::match(const std::string& str, const MatchFn& match) const
+    {
+        if (!mRe)
+            return "regular expression has not been compiled yet";
+
+        pcre2_match_data* match_data = pcre2_match_data_create_from_pattern(mRe, NULL);
+
+        int pos = 0;
+        while (pos < static_cast<int>(str.size())) {
+            const int pcreExecRet = pcre2_match(mRe, reinterpret_cast<PCRE2_SPTR8>(str.c_str()), static_cast<int>(str.size()), pos, 0, match_data, nullptr);
+            if (pcreExecRet == PCRE2_ERROR_NOMATCH)
+                return "";
+            if (pcreExecRet < 0) {
+                PCRE2_UCHAR errorMessageBuf[120];
+                const int res = pcre2_get_error_message(pcreExecRet, errorMessageBuf, sizeof(errorMessageBuf));
+                if (res < 0)
+                    return std::string("failed to get error message ") + std::to_string(res) + " (pos: " + std::to_string(pos) + ")";
+                return std::string(reinterpret_cast<char*>(errorMessageBuf)) + " (pos: " + std::to_string(pos) + ")";
+            }
+            PCRE2_SIZE* ovector = pcre2_get_ovector_pointer(match_data);
+            const uint32_t ovcount = pcre2_get_ovector_count(match_data);
+            if (ovcount != 1)
+                return "invalid ovector count";
+            const auto pos1 = static_cast<unsigned int>(ovector[0]);
+            const auto pos2 = static_cast<unsigned int>(ovector[1]);
+
+            match(pos1, pos2);
+
+            // jump to the end of the match for the next pcre_exec
+            pos = static_cast<int>(pos2);
+        }
+
+        return "";
+    }
+}
+
 template<typename T>
 static T* createAndCompileRegex(std::string pattern, std::string& err)
 {
@@ -261,6 +343,8 @@ std::shared_ptr<Regex> Regex::create(std::string pattern, Engine engine, std::st
     Regex* regex = nullptr;
     if (engine == Engine::Pcre)
         regex = createAndCompileRegex<PcreRegex>(std::move(pattern), err);
+    else if (engine == Engine::Pcre2)
+        regex = createAndCompileRegex<Pcre2Regex>(std::move(pattern), err);
     else {
         err = "unknown regular expression engine";
     }

lib/regex.h

@@ -41,7 +41,8 @@ class CPPCHECKLIB Regex
     enum class Engine : std::uint8_t
     {
         Unknown = 0,
-        Pcre = 1
+        Pcre = 1,
+        Pcre2 = 2
     };
 
     static std::shared_ptr<Regex> create(std::string pattern, Engine engine, std::string& err);

releasenotes.txt

@@ -21,3 +21,4 @@ Infrastructure & dependencies:
 
 Other:
 - Make it possible to specify the regular expression engine using the `engine` element in a rule XML.
+- Added support for PCRE2 (`pcre`) as the regular expression engine for rules.

test/testcmdlineparser.cpp

@@ -2682,10 +2682,20 @@ class TestCmdlineParser : public TestFixture {
                         "<summary>ruleSummary2</summary>\n"
                         "</message>\n"
                         "</rule>\n"
+                        "<rule>\n"
+                        "<engine>pcre2</engine>\n"
+                        "<tokenlist>raw</tokenlist>\n"
+                        "<pattern>[A-Z]</pattern>\n"
+                        "<message>\n"
+                        "<severity>style</severity>\n"
+                        "<id>ruleId3</id>\n"
+                        "<summary>ruleSummary3</summary>\n"
+                        "</message>\n"
+                        "</rule>\n"
                         "</rules>");
         const char * const argv[] = {"cppcheck", "--rule-file=rule.xml", "file.cpp"};
         ASSERT_EQUALS_ENUM(CmdLineParser::Result::Success, parseFromArgs(argv));
-        ASSERT_EQUALS(2, settings->rules.size());
+        ASSERT_EQUALS(3, settings->rules.size());
         auto it = settings->rules.cbegin();
         ASSERT_EQUALS_ENUM(Regex::Engine::Pcre, it->engine);
         ASSERT_EQUALS("raw", it->tokenlist);
@@ -2700,6 +2710,13 @@ class TestCmdlineParser : public TestFixture {
         ASSERT_EQUALS_ENUM(Severity::warning, it->severity);
         ASSERT_EQUALS("ruleId2", it->id);
         ASSERT_EQUALS("ruleSummary2", it->summary);
+        ++it;
+        ASSERT_EQUALS_ENUM(Regex::Engine::Pcre2, it->engine);
+        ASSERT_EQUALS("raw", it->tokenlist);
+        ASSERT_EQUALS("[A-Z]", it->pattern);
+        ASSERT_EQUALS_ENUM(Severity::style, it->severity);
+        ASSERT_EQUALS("ruleId3", it->id);
+        ASSERT_EQUALS("ruleSummary3", it->summary);
     }
 
     void ruleFileSingle() {

test/testregex.cpp

@@ -85,6 +85,8 @@ class TestRegExBase : public TestFixture {
         std::string exp;
         if (mEngine == Regex::Engine::Pcre)
             exp = "missing terminating ] for character class";
+        else if (mEngine == Regex::Engine::Pcre2)
+            exp = "missing terminating ] for character class";
 
         (void)assertRegex("[", exp);
     }
@@ -201,6 +203,13 @@ class TestRegExPcre : public TestRegExBase {
     TestRegExPcre() : TestRegExBase("TestRegExPcre", Regex::Engine::Pcre) {}
 };
 
+class TestRegExPcre2 : public TestRegExBase {
+public:
+    TestRegExPcre2() : TestRegExBase("TestRegExPcre2", Regex::Engine::Pcre2) {}
+};
+
+
 REGISTER_TEST(TestRegExPcre)
+REGISTER_TEST(TestRegExPcre2)
 
 #endif // HAVE_RULES