Add support for PCRE2 (to replace deprecated PCRE) (fixes #4822)

.github/workflows/ubuntu_20.04/build-deps.sh

@@ -15,7 +15,7 @@ DEBIAN_FRONTEND=noninteractive apt-get install -y --fix-missing --no-install-rec
     libnetcdf-dev libpoppler-dev libpoppler-private-dev \
     libspatialite-dev swig libhdf4-alt-dev libhdf5-serial-dev \
     libfreexl-dev unixodbc-dev libwebp-dev libepsilon-dev \
-    liblcms2-2 libpcre3-dev libcrypto++-dev libdap-dev libfyba-dev \
+    liblcms2-2 libpcre2-dev libcrypto++-dev libdap-dev libfyba-dev \
     libkml-dev libmysqlclient-dev libogdi-dev \
     libcfitsio-dev openjdk-8-jdk libzstd-dev \
     libpq-dev libssl-dev libboost-dev \

GDALmake.opt.in

@@ -223,6 +223,7 @@ SPATIALITE_SONAME = @SPATIALITE_SONAME@
 SPATIALITE_INC = @SPATIALITE_INC@
 SPATIALITE_412_OR_LATER = @SPATIALITE_412_OR_LATER@
 HAVE_PCRE = @HAVE_PCRE@
+HAVE_PCRE2 = @HAVE_PCRE2@
 HAVE_RASTERLITE2 = @HAVE_RASTERLITE2@
 RASTERLITE2_CFLAGS = @RASTERLITE2_CFLAGS@
 

configure.ac

@@ -4430,6 +4430,43 @@ fi
 AC_SUBST([HAVE_RASTERLITE2], $HAVE_RASTERLITE2)
 AC_SUBST([RASTERLITE2_CFLAGS], $RASTERLITE2_CFLAGS)
 
+dnl ---------------------------------------------------------------------------
+dnl Check for PCRE2.
+dnl ---------------------------------------------------------------------------
+
+HAVE_PCRE2=no
+
+AC_ARG_WITH(pcre2,
+    [  --with-pcre2             Include libpcre2 support (REGEXP support for SQLite)],
+    ,,)
+
+if test "$with_pcre2" = "no"; then
+    AC_MSG_RESULT(disabled)
+    HAVE_PCRE2=no
+else
+    save_CPPFLAGS="$CPPFLAGS"
+    CPPFLAGS="$CPPFLAGS -DPCRE2_CODE_UNIT_WIDTH=8"
+    AC_CHECK_HEADERS([pcre2.h])
+    CPPFLAGS="$save_CPPFLAGS"
+    if test "$ac_cv_header_pcre2_h" = "no" ; then
+        if test "$with_pcre2" = "yes"; then
+            AC_MSG_ERROR([cannot find pcre2.h])
+        fi
+    else
+        AC_CHECK_LIB(pcre2-8,pcre2_compile_8,HAVE_PCRE2=yes,HAVE_PCRE2=no,)
+        if test "$HAVE_PCRE2" = "yes" ; then
+            HAVE_PCRE2=yes
+            LIBS="$LIBS -lpcre2-8"
+        else
+            if test "$with_pcre2" = "yes"; then
+                AC_MSG_ERROR([cannot find libpcre2])
+            fi
+        fi
+    fi
+fi
+
+AC_SUBST(HAVE_PCRE2)
+
 dnl ---------------------------------------------------------------------------
 dnl Check for PCRE.
 dnl ---------------------------------------------------------------------------
@@ -4443,7 +4480,7 @@ AC_ARG_WITH(pcre,
 if test "$with_pcre" = "no"; then
     AC_MSG_RESULT(disabled)
     HAVE_PCRE=no
-else
+elif test "$HAVE_PCRE2" = "no"; then
     AC_CHECK_HEADERS([pcre.h])
     if test "$ac_cv_header_pcre_h" = "no" ; then
         if test "$with_pcre" = "yes"; then
@@ -6251,6 +6288,7 @@ LOC_MSG([  OpenJPEG support:          ${HAVE_OPENJPEG}])
 LOC_MSG([  PCIDSK support:            ${PCIDSK_SETTING}])
 LOC_MSG([  PCRaster support:          ${PCRASTER_SETTING}])
 LOC_MSG([  PCRE support:              ${HAVE_PCRE}])
+LOC_MSG([  PCRE2 support:             ${HAVE_PCRE2}])
 LOC_MSG([  PDFium support:            ${HAVE_PDFIUM}])
 LOC_MSG([  Podofo support:            ${HAVE_PODOFO}])
 LOC_MSG([  Poppler support:           ${HAVE_POPPLER}])

nmake.opt

@@ -664,7 +664,11 @@ INGRES_LIB = "$(INGRES_HOME)\lib\iilibapi.lib" \
 # Uncomment following line if spatialite is 4.1.2 or later
 #SPATIALITE_412_OR_LATER=yes
 
-# PCRE Library (REGEXP support for SQLite) for example from http://sourceforge.net/projects/gnuwin32/files/pcre/7.0/pcre-7.0.exe/download
+# PCRE2 Library (REGEXP support for SQLite)
+#PCRE2_INC=-I"C:\dev\install-pcre2\include" -DHAVE_PCRE2
+#PCRE2_LIB="C:\dev\install-pcre2\lib\pcre2-8.lib"
+
+# Deprecated: PCRE Library (REGEXP support for SQLite) for example from http://sourceforge.net/projects/gnuwin32/files/pcre/7.0/pcre-7.0.exe/download
 #PCRE_INC=-I"C:\Program Files\GNUWin32\include" -DHAVE_PCRE
 #PCRE_LIB="C:\Program Files\GNUWin32\lib\pcre.lib"
 
@@ -1180,6 +1184,6 @@ EXTERNAL_LIBS =	$(OGDILIB) $(XERCES_LIB) $(EXPAT_LIB) $(OCI_LIB) $(PG_LIB) \
 	$(ODBCLIB) $(JASPER_LIB) $(PNG_LIB) $(ZLIB_LIB) $(LIBDEFLATE_LIB) $(ADD_LIBS) $(OPENJPEG_LIB) \
 	$(MRSID_LIDAR_LIB) $(LIBKML_LIBS) $(SOSI_LIBS) $(PDF_LIB_LINK) $(LZMA_LIBS) $(ZSTD_LIBS) $(BLOSC_LIBS) $(LZ4_LIBS) \
 	$(LIBICONV_LIBRARY) $(WEBP_LIBS) $(TILEDB_LIBS) $(FGDB_LIB_LINK) $(FREEXL_LIBS) $(GTA_LIBS) \
-	$(INGRES_LIB) $(LIBXML2_LIB) $(PCRE_LIB) $(MONGODB_LIB_LINK) $(MONGODBV3_LIB_LINK) $(CRYPTOPP_LIB) $(OPENSSL_LIB) $(CHARLS_LIB) ws2_32.lib \
+	$(INGRES_LIB) $(LIBXML2_LIB) $(PCRE_LIB) $(PCRE2_LIB) $(MONGODB_LIB_LINK) $(MONGODBV3_LIB_LINK) $(CRYPTOPP_LIB) $(OPENSSL_LIB) $(CHARLS_LIB) ws2_32.lib \
 	$(RDB_LIB) $(CRUNCH_LIB) $(OPENEXR_LIB) $(HEIF_LIB) $(LERC_LIB) $(JXL_LIB) $(BROTLI_LIB) $(BRUNSLI_LIB)\
     kernel32.lib psapi.lib wbemuuid.lib

ogr/ogrsf_frmts/sqlite/GNUmakefile

@@ -23,6 +23,10 @@ ifeq ($(HAVE_PCRE),yes)
 CPPFLAGS +=  -DHAVE_PCRE
 endif
 
+ifeq ($(HAVE_PCRE2),yes)
+CPPFLAGS +=  -DHAVE_PCRE2
+endif
+
 ifeq ($(SQLITE_HAS_COLUMN_METADATA),yes)
 CPPFLAGS +=  -DSQLITE_HAS_COLUMN_METADATA
 endif

ogr/ogrsf_frmts/sqlite/makefile.vc

@@ -11,7 +11,9 @@ GDAL_ROOT	=	..\..\..
 
 EXTRAFLAGS = -I.. -I..\.. $(PROJ_FLAGS) $(PROJ_INCLUDE) $(SQLITE_INC) $(PCRE_EXTRAFLAGS) $(SQLITE_HAS_COLUMN_METADATA_EXTRAFLAGS) $(SPATIALITE_412_OR_LATER_EXTRAFLAGS) -DSQLITE_HAS_NON_DEPRECATED_AUTO_EXTENSION
 
-!IFDEF PCRE_INC
+!IFDEF PCRE2_INC
+PCRE_EXTRAFLAGS = $(PCRE2_INC)
+!ELSEIFDEF PCRE_INC
 PCRE_EXTRAFLAGS = $(PCRE_INC)
 !ENDIF
 

ogr/ogrsf_frmts/sqlite/ogrsqliteregexp.cpp

@@ -45,10 +45,137 @@
  * commercial or non-commercial, and by any means.
  */
 
+// The pcre2 variant has been ported from https://github.com/pfmoore/sqlite-pcre2/blob/main/src/pcre.c
+// which has the same license as above.
+
 #include "ogrsqliteregexp.h"
 #include "sqlite3.h"
 
-#ifdef HAVE_PCRE
+#ifdef HAVE_PCRE2
+
+#define PCRE2_CODE_UNIT_WIDTH 8
+#include <pcre2.h>
+
+typedef struct {
+    char *s;
+    pcre2_code *p;
+} cache_entry;
+
+constexpr int CACHE_SIZE = 16;
+
+static
+pcre2_code *re_compile_with_cache(sqlite3_context *ctx, const char *re)
+{
+    cache_entry *cache = static_cast<cache_entry*>(sqlite3_user_data(ctx));
+
+    CPLAssert(cache);
+
+    bool found = false;
+    int i;
+    for (i = 0; i < CACHE_SIZE && cache[i].s; i++)
+        if (strcmp(re, cache[i].s) == 0) {
+            found = true;
+            break;
+        }
+
+    if (found) {
+        if (i > 0) {
+            /* Get the found entry */
+            cache_entry c = cache[i];
+            /* Move 0..i-1 up one - args are (dest, src, size) */
+            memmove(cache + 1, cache, i * sizeof(cache_entry));
+            /* Put the found entry at the start */
+            cache[0] = c;
+        }
+    }
+    else {
+        /* Create a new entry */
+        int errorcode = 0;
+        PCRE2_SIZE pos = 0;
+        uint32_t has_jit = 0;
+        PCRE2_UCHAR8 err_buff[256];
+
+        pcre2_code* pat = pcre2_compile(reinterpret_cast<const PCRE2_UCHAR8*>(re),
+                            PCRE2_ZERO_TERMINATED, 0, &errorcode, &pos, nullptr);
+        if (!pat) {
+            pcre2_get_error_message(errorcode, err_buff, sizeof(err_buff));
+            char *e2 = sqlite3_mprintf("%s: %s (offset %d)", re, err_buff, pos);
+            sqlite3_result_error(ctx, e2, -1);
+            sqlite3_free(e2);
+            return nullptr;
+        }
+        pcre2_config(PCRE2_CONFIG_JIT, &has_jit);
+        if (has_jit) {
+            errorcode = pcre2_jit_compile(pat, 0);
+            if (errorcode) {
+                pcre2_get_error_message(errorcode, err_buff, sizeof(err_buff));
+                char *e2 = sqlite3_mprintf("%s: %s", re, err_buff);
+                sqlite3_result_error(ctx, e2, -1);
+                sqlite3_free(e2);
+                pcre2_code_free(pat);
+                return nullptr;
+            }
+        }
+        /* Free the last cache entry if necessary */
+        i = CACHE_SIZE - 1;
+        if (cache[i].s) {
+            VSIFree(cache[i].s);
+            CPLAssert(cache[i].p);
+            pcre2_code_free(cache[i].p);
+        }
+        /* Move everything up to make space */
+        memmove(cache + 1, cache, i * sizeof(cache_entry));
+        cache[0].s = VSIStrdup(re);
+        cache[0].p = pat;
+    }
+
+    return cache[0].p;
+}
+
+/************************************************************************/
+/*                         OGRSQLiteREGEXPFunction()                    */
+/************************************************************************/
+
+static
+void OGRSQLiteREGEXPFunction( sqlite3_context *ctx,
+                              CPL_UNUSED int argc, sqlite3_value **argv )
+{
+    CPLAssert(argc == 2);
+
+    const char *re = (const char *) sqlite3_value_text(argv[0]);
+    if (!re) {
+        sqlite3_result_error(ctx, "no regexp", -1);
+        return;
+    }
+
+    if( sqlite3_value_type(argv[1]) == SQLITE_NULL )
+    {
+        sqlite3_result_int(ctx, 0);
+        return;
+    }
+
+    const char *str = (const char *) sqlite3_value_text(argv[1]);
+    if (!str) {
+        sqlite3_result_error(ctx, "no string", -1);
+        return;
+    }
+
+    pcre2_code* p = re_compile_with_cache(ctx, re);
+    if (!p)
+        return;
+
+    pcre2_match_data *md = pcre2_match_data_create_from_pattern(p, nullptr);
+    if (!str) {
+        sqlite3_result_error(ctx, "could not create match data block", -1);
+        return;
+    }
+    int rc = pcre2_match(p, reinterpret_cast<const PCRE2_UCHAR8*>(str),
+                         PCRE2_ZERO_TERMINATED, 0, 0, md, nullptr);
+    sqlite3_result_int(ctx, rc >= 0);
+}
+
+
+#elif defined(HAVE_PCRE)
 
 #include <pcre.h>
 
@@ -160,12 +287,12 @@ void OGRSQLiteREGEXPFunction( sqlite3_context *ctx,
 
 static
 void* OGRSQLiteRegisterRegExpFunction(sqlite3*
-#ifdef HAVE_PCRE
+#if defined(HAVE_PCRE) || defined(HAVE_PCRE2)
                                        hDB
 #endif
                                       )
 {
-#ifdef HAVE_PCRE
+#if defined(HAVE_PCRE) || defined(HAVE_PCRE2)
 
     /* For debugging purposes mostly */
     if( !CPLTestBool(CPLGetConfigOption("OGR_SQLITE_REGEXP", "YES")) )
@@ -198,12 +325,24 @@ void* OGRSQLiteRegisterRegExpFunction(sqlite3*
 
 static
 void OGRSQLiteFreeRegExpCache(void*
-#ifdef HAVE_PCRE
+#if defined(HAVE_PCRE) || defined(HAVE_PCRE2)
                               hRegExpCache
 #endif
                               )
 {
-#ifdef HAVE_PCRE
+#ifdef HAVE_PCRE2
+    if( hRegExpCache == nullptr )
+        return;
+
+    cache_entry *cache = (cache_entry*) hRegExpCache;
+    for( int i = 0; i < CACHE_SIZE && cache[i].s; i++ )
+    {
+        CPLFree(cache[i].s);
+        CPLAssert(cache[i].p);
+        pcre2_code_free(cache[i].p);
+    }
+    CPLFree(cache);
+#elif defined(HAVE_PCRE)
     if( hRegExpCache == nullptr )
         return;