reimplement javascriptcore regexp via std::regex

Signed-off-by: Ivailo Monev <xakepa10@gmail.com>

.github/workflows/codeql-analysis.yml

@@ -29,7 +29,7 @@ jobs:
     - name: Install dependencies
       run: |
         sudo apt-get update -qq
-        sudo apt-get install -qq libpcre3-dev libdeflate-dev libjansson-dev libc6-dev libpng-dev libcups2-dev libfreetype6-dev libfontconfig1-dev libdbus-1-dev libicu-dev xorg-dev
+        sudo apt-get install -qq libdeflate-dev libjansson-dev libc6-dev libpng-dev libcups2-dev libfreetype6-dev libfontconfig1-dev libdbus-1-dev libicu-dev xorg-dev
 
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL

CMakeLists.txt

@@ -186,15 +186,6 @@ set_package_properties(Jansson PROPERTIES
     TYPE REQUIRED
 )
 
-# v7.7+ required for JavaScript compatibility
-find_package(PCRE 7.7)
-set_package_properties(PCRE PROPERTIES
-    PURPOSE "Required for regexp support in script component"
-    DESCRIPTION "Perl Compatible Regular Expressions"
-    URL "https://www.pcre.org/"
-    TYPE REQUIRED
-)
-
 find_package(X11)
 set_package_properties(X11 PROPERTIES
     PURPOSE "Required for X11/X.Org integration support"

README

@@ -12,7 +12,7 @@ There are several things you should be aware before considering Katie:
 
  - some things have changed:
   - QMake build system replaced with CMake
-  - Jansson, PCRE, Freetype, ICU, X11 and libpng are required for building
+  - Jansson, Freetype, ICU, X11 and libpng are required for building
   - D-Bus and CUPS must be linked to during build
   - QtUiTools is build as shared library by default
   - moc, uic, rcc, etc. are linked to components
@@ -69,7 +69,7 @@ There are several things you should be aware before considering Katie:
   - improved namespaces support and introduced QT_NAMESPACE_COMPAT
   - improved cross-compilation support
   - improved POSIX, XSI, SUSv2, etc. requirements build checks
-  - script component relies on PCRE instead of internal copy
+  - script component uses std::regex instead of internal PCRE copy
   - tests and benchmarks can be run without prior installation
 
 Bugs fixed in Katie, some of which in Qt5 too:

appveyor.yml

@@ -12,7 +12,7 @@ build_script:
 
     sudo apt-get update -qq
 
-    sudo apt-get install -qq libpcre3-dev libdeflate-dev libc6-dev \
+    sudo apt-get install -qq libdeflate-dev libc6-dev \
         libpng-dev libcups2-dev libfreetype6-dev libfontconfig1-dev \
         libdbus-1-dev libicu-dev xorg-dev dbus-x11 libjansson-dev ccache \
         fonts-freefont-ttf

cmake/modules/FindPCRE.cmake

@@ -1,32 +0,0 @@
-# - Try to find the PCRE
-# Once done this will define
-#
-#  PCRE_FOUND - system has PCRE
-#  PCRE_INCLUDES - the PCRE include directory
-#  PCRE_LIBRARIES - the libraries needed to use PCRE
-#
-# Copyright (C) 2016, Ivailo Monev, <xakepa10@gmail.com>
-#
-# Redistribution and use is allowed according to the terms of the BSD license.
-
-include(FindPkgConfig)
-include(FindPackageHandleStandardArgs)
-
-pkg_check_modules(PC_PCRE QUIET libpcre)
-
-find_path(PCRE_INCLUDES
-    NAMES pcre.h
-    HINTS $ENV{PCREDIR}/include ${PC_PCRE_INCLUDEDIR}
-)
-
-find_library(PCRE_LIBRARIES
-    NAMES pcre
-    HINTS $ENV{PCREDIR}/lib ${PC_PCRE_LIBDIR}
-)
-
-find_package_handle_standard_args(PCRE
-    VERSION_VAR PC_PCRE_VERSION
-    REQUIRED_VARS PCRE_LIBRARIES PCRE_INCLUDES
-)
-
-mark_as_advanced(PCRE_INCLUDES PCRE_LIBRARIES)

package/debian/control

@@ -7,8 +7,8 @@ Vcs-browser: https://github.com/fluxer/katie
 Standards-Version: 4.12.0
 Build-Depends: debhelper (>= 9~), libdeflate-dev,
  libc6-dev, libjansson-dev, libpng-dev, libcups2-dev, libfreetype6-dev,
- libfontconfig1-dev, libpcre3-dev, libdbus-1-dev, libicu-dev, cmake,
- git, xserver-xorg-dev, libxinerama-dev, libxrandr-dev, libxrender-dev,
+ libfontconfig1-dev, libdbus-1-dev, libicu-dev, cmake, git,
+ xserver-xorg-dev, libxinerama-dev, libxrandr-dev, libxrender-dev,
  libxcursor-dev, libsm-dev, unifdef | dpkg
 
 Package: katie-runtime

package/fedora/katie.spec

@@ -7,7 +7,7 @@ Summary: C++ toolkit derived from the Qt 4.8 framework
 License: BSD and LGPLv2+
 URL: https://github.com/fluxer/katie
 
-BuildRequires: gcc-c++ cmake libicu-devel libdeflate-devel jansson-devel libpng-devel freetype-devel pcre-devel libX11-devel libXinerama-devel libXrandr-devel libXrender-devel libXfixes-devel libXcursor-devel libSM-devel libICE-devel dbus-devel fontconfig-devel cups-devel unifdef
+BuildRequires: gcc-c++ cmake libicu-devel libdeflate-devel jansson-devel libpng-devel freetype-devel libX11-devel libXinerama-devel libXrandr-devel libXrender-devel libXfixes-devel libXcursor-devel libSM-devel libICE-devel dbus-devel fontconfig-devel cups-devel unifdef
 Requires: xdg-utils gnu-free-fonts
 Requires(post): /sbin/ldconfig
 Requires(postun): /sbin/ldconfig

package/freebsd/Makefile

@@ -20,9 +20,9 @@ RUN_DEPENDS = xdg-open:devel/xdg-utils \
 	${LOCALBASE}/share/fonts/freefont-ttf/FreeSans.ttf:x11-fonts/freefont-ttf
 LIB_DEPENDS = libdeflate.so:archivers/libdeflate libicuuc.so:devel/icu \
 	libicui18n.so:devel/icu libjansson.so:devel/jansson \
-	libpcre.so:devel/pcre libpng.so:graphics/png \
-	libfreetype.so:print/freetype2 libfontconfig.so:x11-fonts/fontconfig \
-	libdbus-1.so:devel/dbus libcups.so:print/cups
+	libpng.so:graphics/png libfreetype.so:print/freetype2 \
+	libfontconfig.so:x11-fonts/fontconfig libdbus-1.so:devel/dbus \
+	libcups.so:print/cups
 CMAKE_ARGS = -DKATIE_TOOLS_SUFFIX="-katie" -Wno-dev
 
 OPTIONS_DEFINE = NLS

package/netbsd/Makefile

@@ -28,7 +28,6 @@ BUILD_DEPENDS = unifdef-[0-9]*:../../devel/unifdef
 .include "../../sysutils/desktop-file-utils/desktopdb.mk"
 .include "../../textproc/jansson/buildlink3.mk"
 .include "../../textproc/icu/buildlink3.mk"
-.include "../../devel/pcre/buildlink3.mk"
 .include "../../x11/libXinerama/buildlink3.mk"
 .include "../../x11/libXcursor/buildlink3.mk"
 .include "../../x11/libXext/buildlink3.mk"

package/openbsd/Makefile

@@ -21,10 +21,10 @@ COMPILER = base-clang ports-gcc
 MODULES = devel/cmake
 BUILD_DEPENDS = devel/gettext,-tools
 RUN_DEPENDS = devel/desktop-file-utils devel/xdg-utils fonts/freefont-ttf
-LIB_DEPENDS = textproc/icu4c devel/jansson devel/pcre \
-	graphics/png x11/dbus print/cups,-libs devel/gettext,-runtime
+LIB_DEPENDS = textproc/icu4c devel/jansson graphics/png x11/dbus \
+	print/cups,-libs devel/gettext,-runtime
 WANTLIB = ${COMPILER_LIBCXX} ICE SM X11 Xcursor Xext Xfixes Xinerama Xrandr \
-	Xrender fontconfig freetype icui18n icuuc pcre png \
+	Xrender fontconfig freetype icui18n icuuc png \
 	dbus-1 cups intl z c m
 SEPARATE_BUILD = Yes
 CONFIGURE_ARGS = -DKATIE_TOOLS_SUFFIX="-katie" -Wno-dev

src/3rdparty/javascriptcore/runtime/RegExp.cpp

@@ -27,14 +27,11 @@
 #include <string.h>
 #include <wtf/Assertions.h>
 
-#include <pcre.h>
-
 namespace JSC {
 
 inline RegExp::RegExp(const UString& pattern)
     : m_pattern(pattern)
     , m_flagBits(0)
-    , m_constructionError(0)
     , m_numSubpatterns(0)
 {
     compile();
@@ -43,7 +40,6 @@ inline RegExp::RegExp(const UString& pattern)
 inline RegExp::RegExp(const UString& pattern, const UString& flags)
     : m_pattern(pattern)
     , m_flagBits(0)
-    , m_constructionError(0)
     , m_numSubpatterns(0)
 {
     // NOTE: The global flag is handled on a case-by-case basis by functions like
@@ -63,7 +59,6 @@ inline RegExp::RegExp(const UString& pattern, const UString& flags)
             break;
         default:
             m_constructionError = flagError;
-            m_regExp = 0;
             return;
         }
     }
@@ -73,7 +68,6 @@ inline RegExp::RegExp(const UString& pattern, const UString& flags)
 
 RegExp::~RegExp()
 {
-    pcre_free(m_regExp);
 }
 
 PassRefPtr<RegExp> RegExp::create(const UString& pattern)
@@ -88,17 +82,38 @@ PassRefPtr<RegExp> RegExp::create(const UString& pattern, const UString& flags)
 
 void RegExp::compile()
 {
-    m_regExp = nullptr;
+    m_constructionError = std::string();
+    m_regExp = std::regex();
+    m_numSubpatterns = 0;
 
-    int regexOptions = PCRE_JAVASCRIPT_COMPAT | PCRE_NO_UTF8_CHECK;
+    int regexOptions = std::regex_constants::ECMAScript;
     if (ignoreCase())
-        regexOptions |= PCRE_CASELESS;
+        regexOptions |= std::regex_constants::icase;
     if (multiline())
-        regexOptions |= PCRE_MULTILINE;
-    int errorOffset;
-    m_regExp = pcre_compile(m_pattern.ascii(), regexOptions, &m_constructionError, &errorOffset, nullptr);
+#if __cplusplus >= 201703L
+        regexOptions |= std::regex_constants::multiline;
+#endif
 
-    pcre_fullinfo(m_regExp, nullptr, PCRE_INFO_CAPTURECOUNT, &m_numSubpatterns);
+#ifndef QT_NO_EXCEPTIONS
+    try {
+        const std::string regexpattern = m_pattern.ascii();
+        m_regExp = std::regex(regexpattern.c_str(), regexOptions);
+        std::sregex_iterator matchbegin = std::sregex_iterator(regexpattern.begin(), regexpattern.end(), m_regExp);
+        std::sregex_iterator matchend = std::sregex_iterator();
+        m_numSubpatterns = std::distance(matchbegin, matchend);
+    } catch (const std::regex_error &err) {
+        m_constructionError = err.what();
+    } catch (...) {
+        m_constructionError = "Exception caught during regex compilation";
+    }
+#else
+    // no exceptions, no way to find out if error occured
+    const std::string regexpattern = m_pattern.ascii();
+    m_regExp = std::regex(regexpattern.c_str(), regexOptions);
+    std::sregex_iterator matchbegin = std::sregex_iterator(regexpattern.begin(), regexpattern.end(), m_regExp);
+    std::sregex_iterator matchend = std::sregex_iterator();
+    m_numSubpatterns = std::distance(matchbegin, matchend);
+#endif
 }
 
 int RegExp::match(const UString& s, int startOffset, Vector<int, 32>* ovector)
@@ -111,9 +126,9 @@ int RegExp::match(const UString& s, int startOffset, Vector<int, 32>* ovector)
     if (startOffset > s.size() || s.isNull())
         return -1;
 
-    if (m_regExp) {
+    if (isValid()) {
         // Set up the offset vector for the result.
-        // First 2/3 used for result, the last third used by PCRE.
+        // First 2/3 used for result, the last third unsed but there for compatibility.
         int* offsetVector;
         int offsetVectorSize;
         int fixedSizeOffsetVector[3];
@@ -126,18 +141,40 @@ int RegExp::match(const UString& s, int startOffset, Vector<int, 32>* ovector)
             offsetVector = ovector->data();
         }
 
-        const int numMatches = pcre_exec(m_regExp, nullptr, s.ascii(), s.size(), startOffset, 0, offsetVector, offsetVectorSize);
+#ifndef QT_NO_EXCEPTIONS
+        bool didmatch = false;
+        try {
+            didmatch = std::regex_match(s.ascii() + startOffset, m_regExp);
+        } catch (const std::regex_error &err) {
+            m_constructionError = err.what();
+        } catch (...) {
+            m_constructionError = "Exception caught during regex matching";
+        }
+#else
+        const bool didmatch = std::regex_match(s.ascii() + startOffset, m_regExp);
+#endif
 
-        if (numMatches < 0) {
+        if (!didmatch) {
 #ifndef QT_NO_DEBUG
-            if (numMatches != PCRE_ERROR_NOMATCH)
-                fprintf(stderr, "jsRegExpExecute failed with result %d\n", numMatches);
+            fprintf(stderr, "jsRegExpExecute failed with result\n");
 #endif
             if (ovector)
                 ovector->clear();
             return -1;
         }
 
+        const std::string regexpattern = m_pattern.ascii();
+        std::sregex_iterator matchbegin = std::sregex_iterator(regexpattern.begin(), regexpattern.end(), m_regExp);
+        std::sregex_iterator matchend = std::sregex_iterator();
+        size_t nummatches = 0;
+        for (std::sregex_iterator iter = matchbegin; iter != matchend; iter++) {
+            const std::smatch itermatch = *iter;
+            offsetVector[nummatches] = itermatch.position();
+            offsetVector[nummatches + 1] = itermatch.length();
+            offsetVector[nummatches + 2] = 0;
+            nummatches++;
+        }
+
         return offsetVector[0];
     }
 

src/3rdparty/javascriptcore/runtime/RegExp.h

@@ -26,7 +26,7 @@
 #include <wtf/Forward.h>
 #include <wtf/RefCounted.h>
 
-#include <pcre.h>
+#include <regex>
 
 namespace JSC {
 
@@ -42,8 +42,8 @@ namespace JSC {
 
         const UString& pattern() const { return m_pattern; }
 
-        bool isValid() const { return !m_constructionError; }
-        const char* errorMessage() const { return m_constructionError; }
+        bool isValid() const { return m_constructionError.empty(); }
+        const char* errorMessage() const { return m_constructionError.c_str(); }
 
         int match(const UString&, int startOffset, Vector<int, 32>* ovector = 0);
         unsigned numSubpatterns() const { return m_numSubpatterns; }
@@ -58,10 +58,10 @@ namespace JSC {
 
         UString m_pattern; // FIXME: Just decompile m_regExp instead of storing this.
         int m_flagBits;
-        const char* m_constructionError;
+        std::string m_constructionError;
         unsigned m_numSubpatterns;
 
-        pcre* m_regExp;
+        std::regex m_regExp;
     };
 
 } // namespace JSC

src/script/CMakeLists.txt

@@ -8,7 +8,6 @@ add_definitions(
 )
 set(EXTRA_SCRIPT_LIBS
     KtCore
-    ${PCRE_LIBRARIES}
     ${CMAKE_THREAD_LIBS_INIT}
 )
 
@@ -47,7 +46,6 @@ include_directories(
     ${CMAKE_SOURCE_DIR}/src/3rdparty/javascriptcore/API
     ${CMAKE_SOURCE_DIR}/src/3rdparty/javascriptcore/bytecode
     ${CMAKE_SOURCE_DIR}/src/3rdparty/javascriptcore/generated
-    ${PCRE_INCLUDES}
 )
 
 set(SCRIPT_HEADERS

tests/benchmarks/script/sunspider/tst_sunspider.cpp

@@ -20,6 +20,7 @@
 ****************************************************************************/
 
 #include <qtest.h>
+#include <QtCore/qdebug.h>
 #include <QtCore/qdir.h>
 #include <QtCore/qfile.h>
 #include <QtCore/qtextstream.h>
@@ -95,6 +96,9 @@ void tst_SunSpider::benchmark()
     QBENCHMARK {
         engine.evaluate(testContents);
     }
+    if (engine.hasUncaughtException()) {
+        qWarning() << engine.uncaughtException().toString();
+    }
     QVERIFY(!engine.hasUncaughtException());
 }