More fine granular regex config options

CMakeLists.txt

@@ -17,7 +17,9 @@ list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 
 option(HAVE_OVERPASS_XML "Build with Overpass XML dialect" ON)
 option(HAVE_FASTCGI   "Build with FastCGI library" ON)
+option(HAVE_POSIX     "Use C POSIX regular expresions" ON)
 option(HAVE_PCRE      "Build with PCRE library" ON)
+option(HAVE_PCREJIT   "Build with PCRE library (JIT)" ON)
 option(HAVE_ICU       "BUild with ICU library" OFF)
 option(HAVE_TCMALLOC  "Build with tcmalloc library" ON)
 option(HAVE_LZ4       "Build with lz4 library" ON)
@@ -107,7 +109,7 @@ if (HAVE_FASTCGI)
   find_package(FCGI++ REQUIRED)
 endif()
 
-if (HAVE_PCRE)
+if (HAVE_PCRE OR HAVE_PCREJIT)
   find_package(PCRE2 REQUIRED)
 endif()
 
@@ -160,7 +162,7 @@ if (HAVE_TCMALLOC)
   list(APPEND LIBS ${Tcmalloc_LIBRARY})
 endif()
 
-if (HAVE_PCRE)
+if (HAVE_PCRE OR HAVE_PCREJIT)
   list(APPEND LIBS ${PCRE2_LIBRARIES})
 endif()
 

cmake/config.h.in

@@ -1,5 +1,7 @@
 #cmakedefine HAVE_FASTCGI 1
+#cmakedefine HAVE_POSIX 1
 #cmakedefine HAVE_PCRE 1
+#cmakedefine HAVE_PCREJIT 1
 #cmakedefine HAVE_TCMALLOC 1
 #cmakedefine HAVE_LZ4 1
 #cmakedefine HAVE_OPENMP 1

src/configure.ac

@@ -151,8 +151,10 @@ AC_LANG_POP([C++])
 
 
 AC_DEFINE(HAVE_LIBOSMIUM, 1, [Define if you have libosmium library])
+AC_DEFINE(HAVE_POSIX, 1, [Define if you have C POSIX regex library])
 AC_DEFINE(HAVE_ICU, 1, [Define if you have ICU library])
 AC_DEFINE(HAVE_PCRE, 1, [Define if you have PCRE library])
+AC_DEFINE(HAVE_PCREJIT, 1, [Define if you have PCRE library with JIT])
 AC_DEFINE(HAVE_OPENMP, 1, [Define if you have OpenMP])
 
 # Some unused output formats are no longer compiled in by default

src/overpass_api/data/regular_expression.h

@@ -26,7 +26,9 @@
 
 #include "sys/types.h"
 #include "locale.h"
+#ifdef HAVE_POSIX
 #include "regex.h"
+#endif
 
 #include <iostream>
 #include <string>
@@ -40,7 +42,7 @@ using icu::UnicodeString;
 using icu::RegexMatcher;
 #endif
 
-#ifdef HAVE_PCRE
+#if defined(HAVE_PCRE) || defined(HAVE_PCREJIT)
 #define PCRE2_CODE_UNIT_WIDTH 8
 #include <pcre2.h>
 #endif
@@ -95,8 +97,8 @@ class Regular_Expression
     Strategy strategy;
 };
 
-
-class Regular_Expression_POSIX : public Regular_Expression
+#ifdef HAVE_POSIX
+class Regular_Expression_POSIX final : public Regular_Expression
 {
   public:
 
@@ -153,13 +155,13 @@ class Regular_Expression_POSIX : public Regular_Expression
   private:
     regex_t preg;
 };
-
+#endif
 
 
 
 #ifdef HAVE_ICU
 
-class Regular_Expression_ICU : public Regular_Expression
+class Regular_Expression_ICU final : public Regular_Expression
 {
   public:
 
@@ -265,9 +267,9 @@ class Regular_Expression_ICU : public Regular_Expression
 
 #endif
 
-#ifdef HAVE_PCRE
+#if defined(HAVE_PCRE) || defined(HAVE_PCREJIT)
 
-class Regular_Expression_PCRE : public Regular_Expression
+class Regular_Expression_PCRE final : public Regular_Expression
 {
   public:
 
@@ -301,6 +303,7 @@ class Regular_Expression_PCRE : public Regular_Expression
           throw Regular_Expression_Error(std::string(reinterpret_cast<const char*>(buffer), size));
         }
 
+#ifdef HAVE_PCREJIT
         if (enable_jit) {
           pcre2_config(PCRE2_CONFIG_JIT, &pcre2_jit_on);
         }
@@ -323,6 +326,7 @@ class Regular_Expression_PCRE : public Regular_Expression
 
           pcre2_jit_stack_assign(mcontext, NULL, jit_stack);
         }
+#endif
 
         match_data = pcre2_match_data_create_from_pattern(re, NULL);
         if (match_data == nullptr) {
@@ -367,9 +371,17 @@ class Regular_Expression_PCRE : public Regular_Expression
 
       uint32_t options = 0;
 
-      int rc;
+      int rc = 0;
 
+#if defined(HAVE_PCRE) && defined(HAVE_PCREJIT)
       if (pcre2_jit_on) {
+#elif defined(HAVE_PCREJIT)
+      if (true) {
+#else
+      if (false) {
+#endif
+
+#ifdef HAVE_PCREJIT
         rc = pcre2_jit_match(
           re,                                            /* the compiled pattern */
           reinterpret_cast<PCRE2_SPTR>(line.data()),     /* the subject string */
@@ -378,8 +390,10 @@ class Regular_Expression_PCRE : public Regular_Expression
           options,                                       /* options */
           match_data,                                    /* block for storing the result */
           NULL);                                         /* use default match context */
+#endif
       }
       else {
+#ifdef HAVE_PCRE
         rc = pcre2_match(
           re,                                            /* the compiled pattern */
           reinterpret_cast<PCRE2_SPTR>(line.data()),     /* the subject string */
@@ -388,6 +402,7 @@ class Regular_Expression_PCRE : public Regular_Expression
           options,                                       /* options */
           match_data,                                    /* block for storing the result */
           NULL);                                         /* use default match context */
+#endif
       }
 
       if (rc < 0)  {
@@ -426,9 +441,17 @@ class Regular_Expression_PCRE : public Regular_Expression
 
       uint32_t options = 0;
 
-      int rc;
+      int rc = 0;
 
+#if defined(HAVE_PCRE) && defined(HAVE_PCREJIT)
       if (pcre2_jit_on) {
+#elif defined(HAVE_PCREJIT)
+      if (true) {
+#else
+      if (false) {
+#endif
+
+#ifdef HAVE_PCREJIT
         rc = pcre2_jit_match(
           re,                                            /* the compiled pattern */
           reinterpret_cast<PCRE2_SPTR>(line.data()),     /* the subject string */
@@ -437,8 +460,10 @@ class Regular_Expression_PCRE : public Regular_Expression
           options,                                       /* options */
           match_data,                                    /* block for storing the result */
           NULL);                                         /* use default match context */
+#endif
       }
       else {
+#ifdef HAVE_PCRE
         rc = pcre2_match(
           re,                                            /* the compiled pattern */
           reinterpret_cast<PCRE2_SPTR>(line.data()),     /* the subject string */
@@ -447,6 +472,7 @@ class Regular_Expression_PCRE : public Regular_Expression
           options,                                       /* options */
           match_data,                                    /* block for storing the result */
           NULL);                                         /* use default match context */
+#endif
       }
 
       if (rc < 0)  {
@@ -488,30 +514,26 @@ class Regular_Expression_Factory
 
 public:
 
-  static Regular_Expression* get_regexp_engine(const std::string& engine, const std::string& regex, bool case_sensitive )
+  static Regular_Expression* get_regexp(const std::string& engine, const std::string& regex, bool case_sensitive )
   {
-    if (engine == "ICU") {
+
 #ifdef HAVE_ICU
-      return new Regular_Expression_ICU(regex, case_sensitive);
-#else
-      throw std::runtime_error("ICU support not available");
+    if (engine == "ICU") return new Regular_Expression_ICU(regex, case_sensitive);
 #endif
-    } else if (engine == "PCRE") {
+
 #ifdef HAVE_PCRE
-      return new Regular_Expression_PCRE(regex, case_sensitive, false);
-#else
-      throw std::runtime_error("PCRE support not available");
+    if (engine == "PCRE") return new Regular_Expression_PCRE(regex, case_sensitive, false);
 #endif
-   } else if (engine == "PCREJIT") {
-#ifdef HAVE_PCRE
-      return new Regular_Expression_PCRE(regex, case_sensitive, true);
-#else
-      throw std::runtime_error("PCRE support not available");
+
+#ifdef HAVE_PCREJIT
+   if (engine == "PCREJIT") return new Regular_Expression_PCRE(regex, case_sensitive, true);
+#endif
+
+#ifdef HAVE_POSIX
+    if (engine == "POSIX") return new Regular_Expression_POSIX(regex, case_sensitive);
 #endif
-    }
 
-    // always fall back to POSIX
-    return new Regular_Expression_POSIX(regex, case_sensitive);
+    throw std::runtime_error("No suitable regular expression engine found.");
   }
 };
 

src/overpass_api/statements/osm_script.cc

@@ -19,6 +19,8 @@
 #include <fstream>
 #include <iostream>
 #include <sstream>
+#include <string>
+#include <vector>
 
 #include "../core/settings.h"
 #include "../frontend/output_handler_parser.h"
@@ -27,6 +29,7 @@
 #include "print.h"
 
 
+
 Generic_Statement_Maker< Osm_Script_Statement > Osm_Script_Statement::statement_maker("osm-script");
 
 
@@ -63,6 +66,7 @@ Osm_Script_Statement::Osm_Script_Statement
        desired_timestamp(NOW), comparison_timestamp(0), add_deletion_information(false),
        max_allowed_time(0), max_allowed_space(0)
 {
+
   std::map< std::string, std::string > attributes;
 
   attributes["bbox"] = "";
@@ -121,14 +125,17 @@ Osm_Script_Statement::Osm_Script_Statement
     }
   }
 
-  if (attributes["regexp"] == "POSIX" ||
-      attributes["regexp"] == "ICU" ||
-      attributes["regexp"] == "PCRE" ||
-      attributes["regexp"] == "PCREJIT")
+  if (std::find(supported_engines.begin(), supported_engines.end(), attributes["regexp"]) != supported_engines.end())
     global_settings.set_regexp_engine(attributes["regexp"]);
   else
   {
-    add_static_error("For the attribute \"regexp\" of the element \"osm-script\" the only allowed values are \"POSIX\", \"PCRE\", \"PCREJIT\" and \"ICU\".");
+    if (!supported_engines.empty()) {
+      std::ostringstream oss;
+      std::copy(supported_engines.begin(), supported_engines.end(), std::ostream_iterator<std::string>(oss, ", "));
+      std::string tmp = oss.str();
+      std::string res(tmp.data(), tmp.size() - 2);
+      add_static_error("For the attribute \"regexp\" of the element \"osm-script\" the only allowed values are: " + res + ".");
+    }
   }
 
   if (!attributes["bbox"].empty())

src/overpass_api/statements/osm_script.h

@@ -25,6 +25,11 @@
 #include "bbox_query.h"
 #include "statement.h"
 
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#undef VERSION
+#endif
+
 
 class Output_Handle;
 
@@ -54,6 +59,24 @@ class Osm_Script_Statement : public Statement
     uint32 max_allowed_time;
     uint64 max_allowed_space;
     Statement::Factory* factory = nullptr;
+
+    inline static const std::vector<std::string> supported_engines
+    {
+    #ifdef HAVE_POSIX
+      "POSIX",
+    #endif
+    #ifdef HAVE_ICU
+      "ICU",
+    #endif
+    #ifdef HAVE_PCRE
+      "PCRE",
+    #endif
+    #ifdef HAVE_PCREJIT
+      "PCREJIT",
+    #endif
+     };
+
 };
 
+
 #endif

src/overpass_api/statements/query.cc

@@ -158,7 +158,7 @@ void Query_Statement::add_statement(Statement* statement, std::string text)
 
 
 
-struct Trivial_Regex
+struct Trivial_Regex final
 {
 public:
   bool matches(const std::string&, bool use_buffer = true) const { return true; }
@@ -2382,7 +2382,7 @@ Has_Kv_Statement::Has_Kv_Statement
 
     try
     {
-      key_regex = Regular_Expression_Factory::get_regexp_engine(global_settings.get_regexp_engine(), attributes["regk"], case_sensitive);
+      key_regex = Regular_Expression_Factory::get_regexp(global_settings.get_regexp_engine(), attributes["regk"], case_sensitive);
       key = attributes["regk"];
     }
     catch (Regular_Expression_Error& e)
@@ -2400,7 +2400,7 @@ Has_Kv_Statement::Has_Kv_Statement
 
     try
     {
-      regex = Regular_Expression_Factory::get_regexp_engine(global_settings.get_regexp_engine(), attributes["regv"], case_sensitive);
+      regex = Regular_Expression_Factory::get_regexp(global_settings.get_regexp_engine(), attributes["regv"], case_sensitive);
       value = attributes["regv"];
     }
     catch (Regular_Expression_Error& e)