Use PCRE2 instead of PCRE (#153)

PCRE is now at end of life and is no longer actively maintained.
Lift the dependency to the next major version, i.e. PCRE2.

Implementation notes:
- Removed the pcre study option since:
  "The new API ... was simplified by abolishing the separate "study" optimizing
  function; in PCRE2, patterns are automatically optimized where possible."
- If asprintf() fails the content of the 'strp' variable is undefined.
  Lets check the return value and return NULL upon error.
- Pattern and subject can straightforwardly be cast to PCRE2_SPTR since we
  only work with 8-bit code units.

.travis-ci/install.sh

@@ -10,7 +10,7 @@ apt-get install -qq \
     cmake \
     graphviz-dev \
     libjemalloc-dev \
-    libpcre3-dev \
+    libpcre2-dev \
     libtool \
     ninja-build \
     pkg-config

CMakeLists.txt

@@ -5,7 +5,7 @@ list(APPEND CMAKE_MODULE_PATH ${PROJECT_SOURCE_DIR}/cmake/Modules)
 set(CMAKE_C_STANDARD 99)
 
 find_package(Check)
-find_package(PCRE REQUIRED)
+find_package(PCRE2 REQUIRED)
 
 include(CheckSymbolExists)
 include(CheckIncludeFile)

README.md

@@ -25,7 +25,7 @@ Requirement
 
 ### Runtime Requirement
 
-* pcre
+* pcre2
 * (optional) graphviz version 2.38.0 (20140413.2041)
 * (optional) libjson-c-dev
 
@@ -187,13 +187,13 @@ Optimization
 Simple regular expressions are optimized through a regexp pattern to opcode
 translator, which translates simple patterns into small & fast scanners.
 
-By using this method, r3 reduces the matching overhead of pcre library.
+By using this method, r3 reduces the matching overhead of pcre2 library.
 
 Optimized patterns are: `[a-z]+`, `[0-9]+`, `\d+`, `\w+`, `[^/]+`, `[^-]+` or `.*`.
 
 Slugs without specified regular expression will be compiled into the `[^/]+` pattern. therefore, it's optimized too.
 
-Complex regular expressions will still use libpcre to match URL (partially).
+Complex regular expressions will still use libpcre2 to match URL (partially).
 
 
 Performance
@@ -356,15 +356,15 @@ if ( $error ) {
 Install
 ----------------------
 
-    sudo apt-get install check libpcre3 libpcre3-dev libjemalloc-dev libjemalloc1 build-essential libtool automake autoconf pkg-config
+    sudo apt-get install check libpcre2 libpcre2-dev libjemalloc-dev libjemalloc1 build-essential libtool automake autoconf pkg-config
     sudo apt-get install graphviz-dev graphviz  # if you want graphviz
     ./autogen.sh
     ./configure && make
     sudo make install
 
 And we support debian-based distro now!
 
-    sudo apt-get install build-essential autoconf automake libpcre3-dev pkg-config debhelper libtool check
+    sudo apt-get install build-essential autoconf automake libpcre2-dev pkg-config debhelper libtool check
     mv dist-debian debian
     dpkg-buildpackage -b -us -uc
     sudo gdebi ../libr3*.deb

cmake/Modules/FindPCRE2.cmake

@@ -0,0 +1,37 @@
+# Copyright (C) 2007-2009 LuaDist.
+# Created by Peter Kapec <kapecp@gmail.com>
+# Redistribution and use of this file is allowed according to the terms of the MIT license.
+# For details see the COPYRIGHT file distributed with LuaDist.
+#	Note:
+#		Searching headers and libraries is very simple and is NOT as powerful as scripts
+#		distributed with CMake, because LuaDist defines directories to search for.
+#		Everyone is encouraged to contact the author with improvements. Maybe this file
+#		becomes part of CMake distribution sometimes.
+
+# - Find pcre2
+# Find the native PCRE2 headers and libraries.
+#
+# PCRE2_INCLUDE_DIRS	- where to find pcre2.h, etc.
+# PCRE2_LIBRARIES	- List of libraries when using pcre2.
+# PCRE2_FOUND		- True if pcre2 found.
+
+# Look for the header file.
+FIND_PATH(PCRE2_INCLUDE_DIR NAMES pcre2.h)
+
+# Look for the library.
+FIND_LIBRARY(PCRE2_LIBRARY NAMES pcre2-8)
+
+# Handle the QUIETLY and REQUIRED arguments and set PCRE2_FOUND to TRUE if all listed variables are TRUE.
+INCLUDE(FindPackageHandleStandardArgs)
+FIND_PACKAGE_HANDLE_STANDARD_ARGS(PCRE2 DEFAULT_MSG PCRE2_LIBRARY PCRE2_INCLUDE_DIR)
+
+# Copy the results to the output variables.
+IF(PCRE2_FOUND)
+	SET(PCRE2_LIBRARIES ${PCRE2_LIBRARY})
+	SET(PCRE2_INCLUDE_DIRS ${PCRE2_INCLUDE_DIR})
+ELSE(PCRE2_FOUND)
+	SET(PCRE2_LIBRARIES)
+	SET(PCRE2_INCLUDE_DIRS)
+ENDIF(PCRE2_FOUND)
+
+MARK_AS_ADVANCED(PCRE2_INCLUDE_DIRS PCRE2_LIBRARIES)

configure.ac

@@ -73,7 +73,7 @@ AM_CONDITIONAL(USE_JEMALLOC, test "x$have_jemalloc" = "xyes")
 # AC_DEFINE(USE_JEMALLOC, test "x$found_jemalloc" = "xyes" , "use jemalloc")
 
 
-PKG_CHECK_MODULES(DEPS, [libpcre])
+PKG_CHECK_MODULES(DEPS, [libpcre2-8])
 AC_SUBST(DEPS_CFLAGS)
 AC_SUBST(DEPS_LIBS)
 

dist-debian/control

@@ -2,7 +2,7 @@ Source: libr3
 Priority: optional
 Maintainer: Ronmi Ren <ronmi.ren@gmail.com>
 Build-Depends: debhelper (>= 8.0.0), automake, autotools-dev, autoconf,
-               libtool, libpcre3-dev, pkg-config, check
+               libtool, libpcre2-dev, pkg-config, check
 Standards-Version: 3.9.4
 Section: libs
 Homepage: https://github.com/c9s/r3

include/r3.h

@@ -10,7 +10,8 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <pcre.h>
+#define PCRE2_CODE_UNIT_WIDTH 8
+#include <pcre2.h>
 
 #if __STDC_VERSION__ <= 201710L
 #ifdef HAVE_STDBOOL_H
@@ -43,13 +44,12 @@ struct _node  {
     R3_VECTOR(R3Edge) edges;
     R3_VECTOR(R3Route) routes;
     char * combined_pattern;
-    pcre * pcre_pattern;
-    pcre_extra * pcre_extra;
+    pcre2_code * pcre_pattern;
+    pcre2_match_data * match_data;
 
     // edges are mostly less than 255
     unsigned int compare_type; // compare_type: pcre, opcode, string
     unsigned int endpoint; // endpoint, should be zero for non-endpoint nodes
-    unsigned int ov_cnt; // capture vector array size for pcre
 
     // the pointer of R3Route data
     void * data;

r3.pc.in

@@ -6,6 +6,6 @@ libdir=@libdir@
 Name: r3
 Description: High-performance URL router library
 Version: @PACKAGE_VERSION@
-Requires: libpcre
+Requires: libpcre2-8
 Libs: -L${libdir} -lr3
 CFlags: -I${includedir}

src/CMakeLists.txt

@@ -19,7 +19,7 @@ target_include_directories(r3
 
 target_link_libraries(r3
   PUBLIC
-    ${PCRE_LIBRARIES})
+    ${PCRE2_LIBRARIES})
 
 install(
   TARGETS r3

src/edge.c

@@ -13,8 +13,6 @@
 // Jemalloc memory management
 // #include <jemalloc/jemalloc.h>
 
-// PCRE
-#include <pcre.h>
 #include "r3.h"
 #include "r3_slug.h"
 #include "slug.h"

src/match_entry.c

@@ -8,7 +8,6 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <pcre.h>
 #include <assert.h>
 
 #include "r3.h"

src/node.c

@@ -7,9 +7,6 @@
 #include <netinet/in.h>
 #include <arpa/inet.h>
 
-// PCRE
-#include <pcre.h>
-
 #include "r3.h"
 #include "r3_slug.h"
 #include "slug.h"
@@ -75,13 +72,11 @@ void r3_tree_free(R3Node * tree) {
     }
     free(tree->routes.entries);
     if (tree->pcre_pattern) {
-        pcre_free(tree->pcre_pattern);
+        pcre2_code_free(tree->pcre_pattern);
     }
-#ifdef PCRE_STUDY_JIT_COMPILE
-    if (tree->pcre_extra) {
-        pcre_free_study(tree->pcre_extra);
+    if (tree->match_data) {
+        pcre2_match_data_free(tree->match_data);
     }
-#endif
     free(tree->combined_pattern);
     free(tree);
     tree = NULL;
@@ -223,41 +218,44 @@ int r3_tree_compile_patterns(R3Node * n, char **errstr) {
     free(n->combined_pattern);
     n->combined_pattern = cpat;
 
-    const char *pcre_error = NULL;
-    int pcre_erroffset = 0;
+    int pcre_errorcode = 0;
+    PCRE2_SIZE pcre_erroffset = 0;
     unsigned int option_bits = 0;
 
-    n->ov_cnt = (1 + n->edges.size) * 3;
-
     if (n->pcre_pattern) {
-        pcre_free(n->pcre_pattern);
+        pcre2_code_free(n->pcre_pattern);
     }
-    n->pcre_pattern = pcre_compile(
-            n->combined_pattern,              /* the pattern */
+    n->pcre_pattern = pcre2_compile(
+            (PCRE2_SPTR)n->combined_pattern,  /* the pattern, 8-bit code units */
+            PCRE2_ZERO_TERMINATED,
             option_bits,                                /* default options */
-            &pcre_error,               /* for error message */
+            &pcre_errorcode,           /* for error code */
             &pcre_erroffset,           /* for error offset */
-            NULL);                /* use default character tables */
+            NULL);                     /* compile context */
     if (n->pcre_pattern == NULL) {
         if (errstr) {
-            int r = asprintf(errstr, "PCRE compilation failed at offset %d: %s, pattern: %s", pcre_erroffset, pcre_error, n->combined_pattern);
-            if (r) {};
+            PCRE2_UCHAR buf[128];
+            pcre2_get_error_message(pcre_errorcode, buf, sizeof(buf));
+            int r = asprintf(errstr, "PCRE compilation failed at offset %ld: %s, pattern: %s", pcre_erroffset, buf, n->combined_pattern);
+            if (r < 0) {
+                *errstr = NULL; /* the content of errstr is undefined when asprintf() fails */
+            }
         }
         return -1;
     }
-#ifdef PCRE_STUDY_JIT_COMPILE
-    if (n->pcre_extra) {
-        pcre_free_study(n->pcre_extra);
+    if (n->match_data) {
+        pcre2_match_data_free(n->match_data);
     }
-    n->pcre_extra = pcre_study(n->pcre_pattern, 0, &pcre_error);
-    if (!n->pcre_extra && pcre_error) {
+    n->match_data = pcre2_match_data_create_from_pattern(n->pcre_pattern, NULL);
+    if (n->match_data == NULL) {
         if (errstr) {
-            int r = asprintf(errstr, "PCRE study failed at offset %s, pattern: %s", pcre_error, n->combined_pattern);
-            if (r) {};
+            int r = asprintf(errstr, "Failed to allocate match data block");
+            if (r < 0) {
+                *errstr = NULL; /* the content of errstr is undefined when asprintf() fails */
+            }
         }
         return -1;
     }
-#endif
     return 0;
 }
 
@@ -339,28 +337,26 @@ static R3Node * r3_tree_matchl_base(const R3Node * n, const char * path,
         info("COMPARE PCRE_PATTERN\n");
         const char *substring_start = 0;
         int   substring_length = 0;
-        int   ov[ n->ov_cnt ];
         int   rc;
 
         info("pcre matching %s on [%s]\n", n->combined_pattern, path);
 
-        rc = pcre_exec(
+        rc = pcre2_match(
                 n->pcre_pattern, /* the compiled pattern */
-                n->pcre_extra,
-                path,         /* the subject string */
+                (PCRE2_SPTR)path,/* the subject string, 8-bit code units */
                 path_len,     /* the length of the subject */
                 0,            /* start at offset 0 in the subject */
                 0,            /* default options */
-                ov,           /* output vector for substring information */
-                n->ov_cnt);      /* number of elements in the output vector */
+                n->match_data,/* match data results */
+                NULL);        /* match context */
 
         // does not match all edges, return NULL;
         if (rc < 0) {
 #ifdef DEBUG
             printf("pcre rc: %d\n", rc );
             switch(rc)
             {
-                case PCRE_ERROR_NOMATCH:
+                case PCRE2_ERROR_NOMATCH:
                     printf("pcre: no match '%s' on pattern '%s'\n", path, n->combined_pattern);
                     break;
 
@@ -373,23 +369,22 @@ static R3Node * r3_tree_matchl_base(const R3Node * n, const char * path,
             return NULL;
         }
 
-
+        PCRE2_SIZE *ov = pcre2_get_ovector_pointer(n->match_data);
 
         restlen = path_len - ov[1]; // if it's fully matched to the end (rest string length)
-        int *inv = ov + 2;
+
         if (!restlen) {
             // Check the substring to decide we should go deeper on which edge
             for (i = 1; i < rc; i++)
             {
-                substring_length = *(inv+1) - *inv;
+                substring_length = ov[2*i+1] - ov[2*i];
 
                 // if it's not matched for this edge, just skip them quickly
                 if (!is_end && !substring_length) {
-                    inv += 2;
                     continue;
                 }
 
-                substring_start = path + *inv;
+                substring_start = path + ov[2*i];
                 e = n->edges.entries + i - 1;
 
                 if (entry && e->has_slug) {
@@ -404,18 +399,16 @@ static R3Node * r3_tree_matchl_base(const R3Node * n, const char * path,
 
 
         // Check the substring to decide we should go deeper on which edge
-        inv = ov + 2;
         for (i = 1; i < rc; i++)
         {
-            substring_length = *(inv+1) - *inv;
+            substring_length = ov[2*i+1] - ov[2*i];
 
             // if it's not matched for this edge, just skip them quickly
             if (!is_end && !substring_length) {
-                inv += 2;
                 continue;
             }
 
-            substring_start = path + *inv;
+            substring_start = path + ov[2*i];
             e = n->edges.entries + i - 1;
 
             if (entry && e->has_slug) {
@@ -520,7 +513,6 @@ inline R3Edge * r3_node_find_edge_str(const R3Node * n, const char * str, int st
 //     n->endpoint = 0;
 //     n->combined_pattern = NULL;
 //     n->pcre_pattern = NULL;
-//     n->pcre_extra = NULL;
 //     n->data = NULL;
 //     return n;
 // }