Control: tags -1 + patch sid trixie Please find attached a patch which I admit I could not test properly.
This package is fairly specialized and I'm absolutely unfamiliar with it. It requires some extra knowledge and right configuration in order to expose the functionality that has to be tested. Also, some of the affected plugins fail to load due to #1020989. The testsuite passes, and the server runs and loads some of the pcre-based global plugins. I also tested loading of the remap plugins with a hack to workaround #1020989 (I'll follow-up there with more details).
Description: Port to PCRE2. Bug-Debian: https://bugs.debian.org/bug=999938 Bug: https://github.com/apache/trafficserver/issues/8780 Author: Yavor Doganov <ya...@gnu.org> Forwarded: no Last-Update: 2023-12-29 --- --- trafficserver-9.2.3+ds.orig/build/pcre.m4 +++ trafficserver-9.2.3+ds/build/pcre.m4 @@ -45,12 +45,12 @@ fi ], [ - AC_CHECK_PROG(PCRE_CONFIG, pcre-config, pcre-config) + AC_CHECK_PROG(PCRE_CONFIG, pcre2-config, pcre2-config) if test "x$PCRE_CONFIG" != "x"; then enable_pcre=yes pcre_base_dir="`$PCRE_CONFIG --prefix`" pcre_include="`$PCRE_CONFIG --cflags | sed -es/-I//`" - pcre_ldflags="`$PCRE_CONFIG --libs | sed -es/-lpcre// -es/-L//`" + pcre_ldflags="`$PCRE_CONFIG --libs8 | sed -es/-lpcre2-8// -es/-L//`" fi ]) @@ -58,7 +58,7 @@ AC_MSG_CHECKING([for pcre location]) AC_CACHE_VAL(ats_cv_pcre_dir,[ for dir in /usr/local /usr ; do - if test -d $dir && ( test -f $dir/include/pcre.h || test -f $dir/include/pcre/pcre.h ); then + if test -d $dir && ( test -f $dir/include/pcre2.h ); then ats_cv_pcre_dir=$dir break fi @@ -76,7 +76,7 @@ fi else AC_MSG_CHECKING(for pcre headers in $pcre_include) - if test -d $pcre_include && test -d $pcre_ldflags && ( test -f $pcre_include/pcre.h || test -f $pcre_include/pcre/pcre.h ); then + if test -d $pcre_include && test -d $pcre_ldflags && ( test -f $pcre_include/pcre2.h ); then AC_MSG_RESULT([ok]) else AC_MSG_RESULT([not found]) @@ -93,14 +93,15 @@ TS_ADDTO(LDFLAGS, [-L${pcre_ldflags}]) TS_ADDTO_RPATH(${pcre_ldflags}) fi - AC_CHECK_LIB([pcre], [pcre_exec], [pcre_have_libs=1]) + AC_CHECK_LIB([pcre2-8], [pcre2_match_8], [pcre_have_libs=1]) if test "$pcre_have_libs" != "0"; then - AC_CHECK_HEADERS(pcre.h, [pcre_have_headers=1]) - AC_CHECK_HEADERS(pcre/pcre.h, [pcre_have_headers=1]) + AC_CHECK_HEADERS([pcre2.h], [pcre_have_headers=1], [], [[ +#define PCRE2_CODE_UNIT_WIDTH 8 +]]) fi if test "$pcre_have_headers" != "0"; then AC_DEFINE(HAVE_LIBPCRE,1,[Compiling with pcre support]) - AC_SUBST(LIBPCRE, [-lpcre]) + AC_SUBST(LIBPCRE, [-lpcre2-8]) else enable_pcre=no CPPFLAGS=$saved_cppflags --- trafficserver-9.2.3+ds.orig/include/tscore/Regex.h +++ trafficserver-9.2.3+ds/include/tscore/Regex.h @@ -30,11 +30,8 @@ #include "tscore/ink_config.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> /// Match flags for regular expression evaluation. enum REFlags { @@ -94,8 +91,7 @@ int get_capture_count(); private: - pcre *regex = nullptr; - pcre_extra *regex_extra = nullptr; + pcre2_code *regex = nullptr; }; /** Deterministic Finite state Automata container. --- trafficserver-9.2.3+ds.orig/include/tscore/ink_memory.h +++ trafficserver-9.2.3+ds/include/tscore/ink_memory.h @@ -22,6 +22,7 @@ */ #pragma once +#ifdef __cplusplus #include <cctype> #include <cstring> #include <strings.h> @@ -29,6 +30,7 @@ #include <limits> #include <string> #include <string_view> +#endif #include "tscore/ink_config.h" @@ -56,6 +58,9 @@ #include <malloc.h> #endif // ! TS_HAS_JEMALLOC +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> + #ifndef MADV_NORMAL #define MADV_NORMAL 0 #endif @@ -80,18 +85,9 @@ extern "C" { #endif /* __cplusplus */ -struct IOVec : public iovec { - IOVec() - { - iov_base = nullptr; - iov_len = 0; - } - IOVec(void *base, size_t len) - { - iov_base = base; - iov_len = len; - } -}; +extern pcre2_general_context *gen_ctxt; +extern pcre2_compile_context *comp_ctxt; +extern pcre2_match_context *match_ctxt; void *ats_malloc(size_t size); void *ats_calloc(size_t nelem, size_t elsize); @@ -100,6 +96,10 @@ void ats_free(void *ptr); void *ats_free_null(void *ptr); +void pcre_init(void); +void *pcre_malloc(PCRE2_SIZE size, void *data); +void pcre_free(void *ptr, void *data); + int ats_msync(caddr_t addr, size_t len, caddr_t end, int flags); int ats_madvise(caddr_t addr, size_t len, int flags); int ats_mlock(caddr_t addr, size_t len); @@ -142,6 +142,19 @@ #include <memory> +struct IOVec : public iovec { + IOVec() + { + iov_base = nullptr; + iov_len = 0; + } + IOVec(void *base, size_t len) + { + iov_base = base; + iov_len = len; + } +}; + // this is to help with migration to a std::string issue with older code that // expects char* being copied. As more code moves to std::string, this can be // removed to avoid these extra copies. --- trafficserver-9.2.3+ds.orig/src/tscore/Regex.cc +++ trafficserver-9.2.3+ds/src/tscore/Regex.cc @@ -28,67 +28,55 @@ #include "tscore/ink_memory.h" #include "tscore/Regex.h" -#ifdef PCRE_CONFIG_JIT -static pcre_jit_stack * +static pcre2_jit_stack * get_jit_stack(void *data ATS_UNUSED) { thread_local struct JitStack { JitStack() { - jit_stack = pcre_jit_stack_alloc(ats_pagesize(), 1024 * 1024); // 1 page min and 1MB max + jit_stack = pcre2_jit_stack_create(ats_pagesize(), 1024 * 1024, gen_ctxt); // 1 page min and 1MB max } - ~JitStack() { pcre_jit_stack_free(jit_stack); } + ~JitStack() { pcre2_jit_stack_free(jit_stack); } - pcre_jit_stack *jit_stack = nullptr; + pcre2_jit_stack *jit_stack = nullptr; } stack; return stack.jit_stack; } -#endif -Regex::Regex(Regex &&that) noexcept : regex(that.regex), regex_extra(that.regex_extra) +Regex::Regex(Regex &&that) noexcept : regex(that.regex) { that.regex = nullptr; - that.regex_extra = nullptr; } bool Regex::compile(const char *pattern, const unsigned flags) { - const char *error; - int erroffset; - int options = 0; - int study_opts = 0; + int error; + PCRE2_SIZE erroffset; + uint32_t options = 0; if (regex) { return false; } if (flags & RE_CASE_INSENSITIVE) { - options |= PCRE_CASELESS; + options |= PCRE2_CASELESS; } if (flags & RE_ANCHORED) { - options |= PCRE_ANCHORED; + options |= PCRE2_ANCHORED; } - regex = pcre_compile(pattern, options, &error, &erroffset, nullptr); - if (error) { + pcre_init(); + regex = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, options, &error, &erroffset, comp_ctxt); + if (0 != erroffset) { regex = nullptr; return false; } -#ifdef PCRE_CONFIG_JIT - study_opts |= PCRE_STUDY_JIT_COMPILE; -#endif - - regex_extra = pcre_study(regex, study_opts, &error); - -#ifdef PCRE_CONFIG_JIT - if (regex_extra) { - pcre_assign_jit_stack(regex_extra, &get_jit_stack, nullptr); - } -#endif + pcre2_jit_compile(regex, PCRE2_JIT_COMPLETE); + pcre2_jit_stack_assign(match_ctxt, &get_jit_stack, nullptr); return true; } @@ -97,7 +85,7 @@ Regex::get_capture_count() { int captures = -1; - if (pcre_fullinfo(regex, regex_extra, PCRE_INFO_CAPTURECOUNT, &captures) != 0) { + if (pcre2_pattern_info(regex, PCRE2_INFO_CAPTURECOUNT, &captures) != 0) { return -1; } @@ -114,23 +102,19 @@ bool Regex::exec(std::string_view const &str, int *ovector, int ovecsize) const { + pcre2_match_data *md; int rv; - rv = pcre_exec(regex, regex_extra, str.data(), int(str.size()), 0, 0, ovector, ovecsize); + md = pcre2_match_data_create_from_pattern(regex, gen_ctxt); + rv = pcre2_match(regex, (PCRE2_SPTR)str.data(), str.size(), 0, 0, md, match_ctxt); + pcre2_match_data_free(md); return rv > 0; } Regex::~Regex() { - if (regex_extra) { -#ifdef PCRE_CONFIG_JIT - pcre_free_study(regex_extra); -#else - pcre_free(regex_extra); -#endif - } if (regex) { - pcre_free(regex); + pcre2_code_free(regex); } } --- trafficserver-9.2.3+ds.orig/src/tscore/ink_memory.cc +++ trafficserver-9.2.3+ds/src/tscore/ink_memory.cc @@ -46,6 +46,33 @@ #include <cstdlib> #include <cstring> +pcre2_general_context *gen_ctxt = nullptr; +pcre2_compile_context *comp_ctxt = nullptr; +pcre2_match_context *match_ctxt = nullptr; + +void +pcre_init(void) +{ + if (!gen_ctxt) + gen_ctxt = pcre2_general_context_create(pcre_malloc, pcre_free, nullptr); + if (!comp_ctxt) + comp_ctxt = pcre2_compile_context_create(gen_ctxt); + if (!match_ctxt) + match_ctxt = pcre2_match_context_create(gen_ctxt); +} + +void * +pcre_malloc(PCRE2_SIZE size, void *data) +{ + return ats_malloc(size); +} + +void +pcre_free(void *ptr, void *data) +{ + ats_free(ptr); +} + void * ats_malloc(size_t size) { --- trafficserver-9.2.3+ds.orig/proxy/ControlMatcher.cc +++ trafficserver-9.2.3+ds/proxy/ControlMatcher.cc @@ -375,7 +375,7 @@ template <class Data, class MatchResult> RegexMatcher<Data, MatchResult>::~RegexMatcher() { for (int i = 0; i < num_el; i++) { - pcre_free(re_array[i]); + pcre2_code_free(re_array[i]); ats_free(re_str[i]); } delete[] re_str; @@ -408,8 +408,8 @@ // Should not have been allocated before ink_assert(array_len == -1); - re_array = static_cast<pcre **>(ats_malloc(sizeof(pcre *) * num_entries)); - memset(re_array, 0, sizeof(pcre *) * num_entries); + re_array = static_cast<pcre2_code **>(ats_malloc(sizeof(pcre2_code *) * num_entries)); + memset(re_array, 0, sizeof(pcre2_code *) * num_entries); data_array = new Data[num_entries]; @@ -429,8 +429,9 @@ { Data *cur_d; char *pattern; - const char *errptr; - int erroffset; + PCRE2_UCHAR errptr[120]; + PCRE2_SIZE erroffset; + int err; Result error = Result::ok(); // Make sure space has been allocated @@ -446,9 +447,10 @@ ink_assert(pattern != nullptr); // Create the compiled regular expression - re_array[num_el] = pcre_compile(pattern, 0, &errptr, &erroffset, nullptr); + re_array[num_el] = pcre2_compile((PCRE2_SPTR)pattern, strlen(pattern), 0, &err, &erroffset, comp_ctxt); if (!re_array[num_el]) { - return Result::failure("%s regular expression error at line %d position %d : %s", matcher_name, line_info->line_num, erroffset, + pcre2_get_error_message(err, errptr, sizeof(errptr)); + return Result::failure("%s regular expression error at line %d position %zu : %s", matcher_name, line_info->line_num, erroffset, errptr); } re_str[num_el] = ats_strdup(pattern); @@ -465,7 +467,7 @@ // There was a problem so undo the effects this function ats_free(re_str[num_el]); re_str[num_el] = nullptr; - pcre_free(re_array[num_el]); + pcre2_code_free(re_array[num_el]); re_array[num_el] = nullptr; } else { num_el++; @@ -484,6 +486,7 @@ void RegexMatcher<Data, MatchResult>::Match(RequestData *rdata, MatchResult *result) const { + pcre2_match_data *md; char *url_str; int r; @@ -506,7 +509,9 @@ // HttpRequestData::get_string(); therefore, no need to call again here. for (int i = 0; i < num_el; i++) { - r = pcre_exec(re_array[i], nullptr, url_str, strlen(url_str), 0, 0, nullptr, 0); + md = pcre2_match_data_create_from_pattern(re_array[i], gen_ctxt); + r = pcre2_match(re_array[i], (PCRE2_SPTR)url_str, strlen(url_str), 0, 0, md, match_ctxt); + pcre2_match_data_free(md); if (r > -1) { Debug("matcher", "%s Matched %s with regex at line %d", matcher_name, url_str, data_array[i].line_num); data_array[i].UpdateMatch(result, rdata); @@ -537,6 +542,7 @@ void HostRegexMatcher<Data, MatchResult>::Match(RequestData *rdata, MatchResult *result) const { + pcre2_match_data *md; const char *url_str; int r; @@ -554,7 +560,9 @@ url_str = ""; } for (int i = 0; i < this->num_el; i++) { - r = pcre_exec(this->re_array[i], nullptr, url_str, strlen(url_str), 0, 0, nullptr, 0); + md = pcre2_match_data_create_from_pattern(this->re_array[i], gen_ctxt); + r = pcre2_match(this->re_array[i], (PCRE2_SPTR)url_str, strlen(url_str), 0, 0, md, match_ctxt); + pcre2_match_data_free(md); if (r != -1) { Debug("matcher", "%s Matched %s with regex at line %d", const_cast<char *>(this->matcher_name), url_str, this->data_array[i].line_num); --- trafficserver-9.2.3+ds.orig/proxy/ControlMatcher.h +++ trafficserver-9.2.3+ds/proxy/ControlMatcher.h @@ -223,7 +223,7 @@ using super::array_len; protected: - pcre **re_array = nullptr; // array of compiled regexs + pcre2_code **re_array = nullptr; // array of compiled regexs char **re_str = nullptr; // array of uncompiled regex strings }; --- trafficserver-9.2.3+ds.orig/iocore/net/P_SSLSNI.h +++ trafficserver-9.2.3+ds/iocore/net/P_SSLSNI.h @@ -53,7 +53,7 @@ void operator()(void *p) { - pcre_free(p); + pcre2_code_free((pcre2_code *)p); } }; @@ -66,7 +66,7 @@ void set_glob_name(std::string name); void set_regex_name(const std::string ®ex_name); - std::unique_ptr<pcre, PcreFreer> match; + std::unique_ptr<pcre2_code, PcreFreer> match; }; struct ActionElement : public NamedElement { --- trafficserver-9.2.3+ds.orig/iocore/net/SSLSNIConfig.cc +++ trafficserver-9.2.3+ds/iocore/net/SSLSNIConfig.cc @@ -42,9 +42,10 @@ #include <sstream> #include <utility> -#include <pcre.h> +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> -static constexpr int OVECSIZE{30}; +static constexpr uint32_t OVECSIZE = 30; //// // NamedElement @@ -82,10 +83,10 @@ void NamedElement::set_regex_name(const std::string ®ex_name) { - const char *err_ptr; - int err_offset = 0; + int err; + PCRE2_SIZE err_offset; if (!regex_name.empty()) { - match.reset(pcre_compile(regex_name.c_str(), PCRE_ANCHORED | PCRE_CASELESS, &err_ptr, &err_offset, nullptr)); + match.reset(pcre2_compile((PCRE2_SPTR)regex_name.c_str(), PCRE2_ZERO_TERMINATED, PCRE2_ANCHORED | PCRE2_CASELESS, &err, &err_offset, comp_ctxt)); } } @@ -95,14 +96,17 @@ const NextHopProperty * SNIConfigParams::get_property_config(const std::string &servername) const { + pcre2_match_data *md; const NextHopProperty *nps = nullptr; + md = pcre2_match_data_create(OVECSIZE, gen_ctxt); for (auto &&item : next_hop_list) { - if (pcre_exec(item.match.get(), nullptr, servername.c_str(), servername.length(), 0, 0, nullptr, 0) >= 0) { + if (pcre2_match(item.match.get(), (PCRE2_SPTR)servername.c_str(), servername.length(), 0, 0, md, match_ctxt) >= 0) { // Found a match nps = &item.prop; break; } } + pcre2_match_data_free(md); return nps; } @@ -183,18 +187,23 @@ std::pair<const ActionVector *, ActionItem::Context> SNIConfigParams::get(std::string_view servername) const { - int ovector[OVECSIZE]; + pcre2_match_data *md; + PCRE2_SIZE *ovector = nullptr; + md = pcre2_match_data_create(OVECSIZE, gen_ctxt); for (const auto &retval : sni_action_list) { - int length = servername.length(); + size_t length = servername.length(); if (retval.match == nullptr && length == 0) { + pcre2_match_data_free(md); return {&retval.actions, {}}; - } else if (auto offset = pcre_exec(retval.match.get(), nullptr, servername.data(), length, 0, 0, ovector, OVECSIZE); + } else if (auto offset = pcre2_match(retval.match.get(), (PCRE2_SPTR)servername.data(), length, 0, 0, md, match_ctxt); offset >= 0) { if (offset == 1) { // first pair identify the portion of the subject string matched by the entire pattern + ovector = pcre2_get_ovector_pointer(md); if (ovector[0] == 0 && ovector[1] == length) { // full match + pcre2_match_data_free(md); return {&retval.actions, {}}; } else { continue; @@ -214,9 +223,11 @@ groups.emplace_back(servername.data() + start, length); } + pcre2_match_data_free(md); return {&retval.actions, {std::move(groups)}}; } } + pcre2_match_data_free(md); return {nullptr, {}}; } --- trafficserver-9.2.3+ds.orig/proxy/http/remap/UrlMapping.cc +++ trafficserver-9.2.3+ds/proxy/http/remap/UrlMapping.cc @@ -153,8 +153,9 @@ referer_info::referer_info(char *_ref, bool *error_flag, char *errmsgbuf, int errmsgbuf_size) : next(nullptr), referer(nullptr), referer_size(0), any(false), negative(false), regx_valid(false) { - const char *error; - int erroffset; + int err; + const char error[120] = {}; + PCRE2_SIZE erroffset; if (error_flag) { *error_flag = false; @@ -171,9 +172,10 @@ if (!strcmp(referer, "*")) { any = true; } else { - regx = pcre_compile(referer, PCRE_CASELESS, &error, &erroffset, nullptr); + regx = pcre2_compile((PCRE2_SPTR)referer, PCRE2_ZERO_TERMINATED, PCRE2_CASELESS, &err, &erroffset, comp_ctxt); if (!regx) { if (errmsgbuf && (errmsgbuf_size - 1) > 0) { + pcre2_get_error_message(err, (PCRE2_UCHAR *)error, sizeof(error)); ink_strlcpy(errmsgbuf, error, errmsgbuf_size); } if (error_flag) { @@ -197,7 +199,7 @@ referer_size = 0; if (regx_valid) { - pcre_free(regx); + pcre2_code_free(regx); regx = nullptr; regx_valid = false; } --- trafficserver-9.2.3+ds.orig/proxy/http/remap/UrlMapping.h +++ trafficserver-9.2.3+ds/proxy/http/remap/UrlMapping.h @@ -51,7 +51,7 @@ bool any; /* any flag '*' */ bool negative; /* negative referer '~' */ bool regx_valid; - pcre *regx; + pcre2_code *regx; }; /** --- trafficserver-9.2.3+ds.orig/iocore/cache/CachePages.cc +++ trafficserver-9.2.3+ds/iocore/cache/CachePages.cc @@ -613,16 +613,19 @@ int res = CACHE_SCAN_RESULT_CONTINUE; for (unsigned s = 0; show_cache_urlstrs[s][0] != '\0'; s++) { - const char *error; - int erroffset; - pcre *preq = pcre_compile(show_cache_urlstrs[s], 0, &error, &erroffset, nullptr); + pcre2_match_data *md; + int error; + PCRE2_SIZE erroffset; + pcre2_code *preq = pcre2_compile((PCRE2_SPTR)show_cache_urlstrs[s], PCRE2_ZERO_TERMINATED, 0, &error, &erroffset, comp_ctxt); Debug("cache_inspector", "matching url '%s' '%s' with regex '%s'", m, xx, show_cache_urlstrs[s]); if (preq) { - int r = pcre_exec(preq, nullptr, xx, ib, 0, 0, nullptr, 0); + md = pcre2_match_data_create_from_pattern(preq, gen_ctxt); + int r = pcre2_match(preq, (PCRE2_SPTR)xx, ib, 0, 0, md, match_ctxt); - pcre_free(preq); + pcre2_code_free(preq); + pcre2_match_data_free(md); if (r != -1) { linecount++; if ((linecount % 5) == 0) { --- trafficserver-9.2.3+ds.orig/proxy/http/remap/RemapProcessor.cc +++ trafficserver-9.2.3+ds/proxy/http/remap/RemapProcessor.cc @@ -155,6 +155,7 @@ // Check referer filtering rules if ((s->filter_mask & URL_REMAP_FILTER_REFERER) != 0 && (ri = map->referer_list) != nullptr) { + pcre2_match_data *md = nullptr; const char *referer_hdr = nullptr; int referer_len = 0; bool enabled_flag = map->optional_referer ? true : false; @@ -167,16 +168,19 @@ memcpy(tmp_referer_buf, referer_hdr, referer_len); tmp_referer_buf[referer_len] = 0; for (enabled_flag = false; ri; ri = ri->next) { + pcre2_match_data_free(md); + md = pcre2_match_data_create_from_pattern(ri->regx, gen_ctxt); if (ri->any) { enabled_flag = true; if (!map->negative_referer) { break; } - } else if (ri->regx_valid && (pcre_exec(ri->regx, nullptr, tmp_referer_buf, referer_len, 0, 0, nullptr, 0) != -1)) { + } else if (ri->regx_valid && (pcre2_match(ri->regx, (PCRE2_SPTR)tmp_referer_buf, referer_len, 0, 0, md, match_ctxt) != -1)) { enabled_flag = ri->negative ? false : true; break; } } + pcre2_match_data_free(md); } if (!enabled_flag) { --- trafficserver-9.2.3+ds.orig/mgmt/WebMgmtUtils.cc +++ trafficserver-9.2.3+ds/mgmt/WebMgmtUtils.cc @@ -1020,17 +1020,20 @@ bool recordRegexCheck(const char *pattern, const char *value) { - pcre *regex; - const char *error; - int erroffset; + pcre2_code *regex; + pcre2_match_data *md; + int error; + PCRE2_SIZE erroffset; - regex = pcre_compile(pattern, 0, &error, &erroffset, nullptr); + regex = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, 0, &error, &erroffset, comp_ctxt); if (!regex) { return false; } else { - int r = pcre_exec(regex, nullptr, value, strlen(value), 0, 0, nullptr, 0); + md = pcre2_match_data_create_from_pattern(regex, gen_ctxt); + int r = pcre2_match(regex, (PCRE2_SPTR)value, strlen(value), 0, 0, md, match_ctxt); - pcre_free(regex); + pcre2_match_data_free(md); + pcre2_code_free(regex); return (r != -1) ? true : false; } --- trafficserver-9.2.3+ds.orig/src/traffic_server/traffic_server.cc +++ trafficserver-9.2.3+ds/src/traffic_server/traffic_server.cc @@ -1740,8 +1740,7 @@ mcheck_pedantic(NULL); #endif - pcre_malloc = ats_malloc; - pcre_free = ats_free; + pcre_init(); // Define the version info appVersionInfo.setup(PACKAGE_NAME, "traffic_server", PACKAGE_VERSION, __DATE__, __TIME__, BUILD_MACHINE, BUILD_PERSON, ""); --- trafficserver-9.2.3+ds.orig/src/traffic_via/traffic_via.cc +++ trafficserver-9.2.3+ds/src/traffic_via/traffic_via.cc @@ -23,6 +23,7 @@ #include "tscore/ink_platform.h" #include "tscore/ink_args.h" +#include "tscore/ink_memory.h" #include "tscore/I_Version.h" #include "mgmtapi.h" #include <cstdio> @@ -32,11 +33,8 @@ #include "tscore/Regex.h" /// XXX Use DFA or Regex wrappers? -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #define SUBSTRING_VECTOR_COUNT 30 // Should be multiple of 3 @@ -257,11 +255,12 @@ static TSMgmtError filterViaHeader() { - const pcre *compiledReg; - const pcre_extra *extraReg = nullptr; - int subStringVector[SUBSTRING_VECTOR_COUNT]; - const char *err; - int errOffset; + const pcre2_code *compiledReg; + pcre2_match_data *md; + PCRE2_SIZE *subStringVector; + PCRE2_UCHAR errbuf[120]; + int err; + PCRE2_SIZE errOffset; int pcreExecCode; int i; const char *viaPattern = @@ -269,17 +268,19 @@ std::string line; // Compile PCRE via header pattern - compiledReg = pcre_compile(viaPattern, 0, &err, &errOffset, nullptr); + compiledReg = pcre2_compile((PCRE2_SPTR)viaPattern, PCRE2_ZERO_TERMINATED, 0, &err, &errOffset, comp_ctxt); if (compiledReg == nullptr) { - printf("PCRE regex compilation failed with error %s at offset %d\n", err, errOffset); + pcre2_get_error_message(err, errbuf, sizeof(errbuf)); + printf("PCRE regex compilation failed with error %s at offset %zu\n", errbuf, errOffset); return TS_ERR_FAIL; } + md = pcre2_match_data_create(SUBSTRING_VECTOR_COUNT, gen_ctxt); // Read all lines from stdin while (std::getline(std::cin, line)) { // Match for via header pattern - pcreExecCode = pcre_exec(compiledReg, extraReg, line.data(), line.size(), 0, 0, subStringVector, SUBSTRING_VECTOR_COUNT); + pcreExecCode = pcre2_match(compiledReg, (PCRE2_SPTR)line.data(), line.size(), 0, 0, md, match_ctxt); // Match failed, don't worry. Continue to next line. if (pcreExecCode < 0) { @@ -293,12 +294,14 @@ } // Loop based on number of matches found + subStringVector = pcre2_get_ovector_pointer(md); for (i = 1; i < pcreExecCode; i++) { std::string_view match{line.data() + subStringVector[2 * i], size_t(subStringVector[2 * i + 1] - subStringVector[2 * i])}; // Decode matched substring decodeViaHeader(match); } } + pcre2_match_data_free(md); return TS_ERR_OKAY; } @@ -316,6 +319,7 @@ HELP_ARGUMENT_DESCRIPTION(), }; + pcre_init(); process_args(&appVersionInfo, argument_descriptions, countof(argument_descriptions), argv); for (unsigned i = 0; i < n_file_arguments; ++i) { --- trafficserver-9.2.3+ds.orig/src/traffic_layout/info.cc +++ trafficserver-9.2.3+ds/src/traffic_layout/info.cc @@ -189,7 +189,7 @@ print_var("openssl", LBW().print("{:#x}", OPENSSL_VERSION_NUMBER).view(), json); print_var("openssl_str", LBW().print(OPENSSL_VERSION_TEXT).view(), json); - print_var("pcre", LBW().print("{}.{}", PCRE_MAJOR, PCRE_MINOR).view(), json); + print_var("pcre", LBW().print("{}.{}", PCRE2_MAJOR, PCRE2_MINOR).view(), json); // These are optional, for now at least. #if TS_USE_HWLOC print_var("hwloc", LBW().print("{:#x}", HWLOC_API_VERSION).view(), json); --- trafficserver-9.2.3+ds.orig/plugins/header_rewrite/regex_helper.cc +++ trafficserver-9.2.3+ds/plugins/header_rewrite/regex_helper.cc @@ -20,21 +20,17 @@ bool regexHelper::setRegexMatch(const std::string &s) { - const char *errorComp = nullptr; - const char *errorStudy = nullptr; - int erroffset; + int errorComp; + PCRE2_SIZE erroffset; regexString = s; - regex = pcre_compile(regexString.c_str(), 0, &errorComp, &erroffset, nullptr); + regex = pcre2_compile((PCRE2_SPTR)regexString.c_str(), PCRE2_ZERO_TERMINATED, 0, &errorComp, &erroffset, comp_ctxt); if (regex == nullptr) { return false; } - regexExtra = pcre_study(regex, 0, &errorStudy); - if ((regexExtra == nullptr) && (errorStudy != nullptr)) { - return false; - } - if (pcre_fullinfo(regex, regexExtra, PCRE_INFO_CAPTURECOUNT, ®exCcount) != 0) { + + if (pcre2_pattern_info(regex, PCRE2_INFO_CAPTURECOUNT, ®exCcount) != 0) { return false; } return true; @@ -43,12 +39,17 @@ int regexHelper::regexMatch(const char *str, int len, int ovector[]) const { - return pcre_exec(regex, // the compiled pattern - regexExtra, // Extra data from study (maybe) - str, // the subject std::string + pcre2_match_data *md; + int rc; + + md = pcre2_match_data_create(OVECCOUNT, gen_ctxt); + rc = pcre2_match(regex, // the compiled pattern + (PCRE2_SPTR)str, // the subject std::string len, // the length of the subject 0, // start at offset 0 in the subject 0, // default options - ovector, // output vector for substring information - OVECCOUNT); // number of elements in the output vector + md, // match data + match_ctxt); // match context + pcre2_match_data_free(md); + return rc; }; --- trafficserver-9.2.3+ds.orig/plugins/header_rewrite/regex_helper.h +++ trafficserver-9.2.3+ds/plugins/header_rewrite/regex_helper.h @@ -18,12 +18,10 @@ #pragma once #include "tscore/ink_defs.h" +#include "tscore/ink_memory.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include <string> @@ -34,8 +32,7 @@ public: ~regexHelper() { - pcre_free(regex); - pcre_free(regexExtra); + pcre2_code_free(regex); } bool setRegexMatch(const std::string &s); @@ -43,7 +40,6 @@ private: std::string regexString; - pcre *regex = nullptr; - pcre_extra *regexExtra = nullptr; + pcre2_code *regex = nullptr; int regexCcount = 0; }; --- trafficserver-9.2.3+ds.orig/plugins/cachekey/pattern.cc +++ trafficserver-9.2.3+ds/plugins/cachekey/pattern.cc @@ -140,14 +140,9 @@ Pattern::pcreFree() { if (_re) { - pcre_free(_re); + pcre2_code_free(_re); _re = nullptr; } - - if (_extra) { - pcre_free(_extra); - _extra = nullptr; - } } /** @@ -205,6 +200,7 @@ bool Pattern::match(const String &subject) { + pcre2_match_data *md; int matchCount; CacheKeyDebug("matching '%s' to '%s'", _pattern.c_str(), subject.c_str()); @@ -212,9 +208,11 @@ return false; } - matchCount = pcre_exec(_re, _extra, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, nullptr, 0); + md = pcre2_match_data_create_from_pattern(_re, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); + pcre2_match_data_free(md); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { CacheKeyError("matching error %d", matchCount); } return false; @@ -231,8 +229,9 @@ bool Pattern::capture(const String &subject, StringVector &result) { + pcre2_match_data *md; int matchCount; - int ovector[OVECOUNT]; + PCRE2_SIZE *ovector; CacheKeyDebug("capturing '%s' from '%s'", _pattern.c_str(), subject.c_str()); @@ -241,23 +240,27 @@ return false; } - matchCount = pcre_exec(_re, nullptr, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, ovector, OVECOUNT); + md = pcre2_match_data_create(OVECOUNT, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { CacheKeyError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } + ovector = pcre2_get_ovector_pointer(md); for (int i = 0; i < matchCount; i++) { - int start = ovector[2 * i]; - int length = ovector[2 * i + 1] - ovector[2 * i]; + int start = (int)ovector[2 * i]; + int length = (int)(ovector[2 * i + 1] - ovector[2 * i]); String dst(subject, start, length); - CacheKeyDebug("capturing '%s' %d[%d,%d]", dst.c_str(), i, ovector[2 * i], ovector[2 * i + 1]); + CacheKeyDebug("capturing '%s' %d[%zu,%zu]", dst.c_str(), i, ovector[2 * i], ovector[2 * i + 1]); result.push_back(dst); } + pcre2_match_data_free(md); return true; } @@ -271,8 +274,9 @@ bool Pattern::replace(const String &subject, String &result) { + pcre2_match_data *md; int matchCount; - int ovector[OVECOUNT]; + PCRE2_SIZE *ovector; CacheKeyDebug("replacing:'%s' in pattern:'%s', subject:'%s'", _replacement.c_str(), _pattern.c_str(), subject.c_str()); @@ -281,11 +285,13 @@ return false; } - matchCount = pcre_exec(_re, nullptr, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, ovector, OVECOUNT); + md = pcre2_match_data_create(OVECOUNT, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { CacheKeyError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } @@ -293,15 +299,17 @@ for (int i = 0; i < _tokenCount; i++) { if (_tokens[i] >= matchCount) { CacheKeyError("invalid reference in replacement string: $%d", _tokens[i]); + pcre2_match_data_free(md); return false; } } + ovector = pcre2_get_ovector_pointer(md); int previous = 0; for (int i = 0; i < _tokenCount; i++) { int replIndex = _tokens[i]; - int start = ovector[2 * replIndex]; - int length = ovector[2 * replIndex + 1] - ovector[2 * replIndex]; + int start = (int)ovector[2 * replIndex]; + int length = (int)(ovector[2 * replIndex + 1] - ovector[2 * replIndex]); /* Handle the case when no match / a group capture result in an empty string */ if (start < 0) { @@ -323,6 +331,7 @@ result.append(_replacement, previous, _replacement.length() - previous); CacheKeyDebug("replacing '%s' resulted in '%s'", _replacement.c_str(), result.c_str()); + pcre2_match_data_free(md); return true; } @@ -334,31 +343,24 @@ bool Pattern::compile() { - const char *errPtr; /* PCRE error */ - int errOffset; /* PCRE error offset */ + int errCode; /* PCRE error code */ + PCRE2_UCHAR errPtr[120]; /* PCRE error */ + PCRE2_SIZE errOffset; /* PCRE error offset */ CacheKeyDebug("compiling pattern:'%s', replace: %s, replacement:'%s'", _pattern.c_str(), _replace ? "true" : "false", _replacement.c_str()); - _re = pcre_compile(_pattern.c_str(), /* the pattern */ - 0, /* options */ - &errPtr, /* for error message */ - &errOffset, /* for error offset */ - nullptr); /* use default character tables */ + _re = pcre2_compile((PCRE2_SPTR)_pattern.c_str(), /* the pattern */ + PCRE2_ZERO_TERMINATED, + 0, /* options */ + &errCode, /* for error code */ + &errOffset, /* for error offset */ + comp_ctxt); /* compile context */ if (nullptr == _re) { - CacheKeyError("compile of regex '%s' at char %d: %s", _pattern.c_str(), errOffset, errPtr); - - return false; - } - - _extra = pcre_study(_re, 0, &errPtr); + pcre2_get_error_message(errCode, errPtr, sizeof(errPtr)); + CacheKeyError("compile of regex '%s' at char %zu: %s", _pattern.c_str(), errOffset, errPtr); - if ((nullptr == _extra) && (nullptr != errPtr) && (0 != *errPtr)) { - CacheKeyError("failed to study regex '%s': %s", _pattern.c_str(), errPtr); - - pcre_free(_re); - _re = nullptr; return false; } --- trafficserver-9.2.3+ds.orig/plugins/cachekey/pattern.h +++ trafficserver-9.2.3+ds/plugins/cachekey/pattern.h @@ -24,12 +24,10 @@ #pragma once #include "tscore/ink_defs.h" +#include "tscore/ink_memory.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include "common.h" @@ -57,8 +55,7 @@ bool compile(); void pcreFree(); - pcre *_re = nullptr; /**< @brief PCRE compiled info structure, computed during initialization */ - pcre_extra *_extra = nullptr; /**< @brief PCRE study data block, computed during initialization */ + pcre2_code *_re = nullptr; /**< @brief PCRE compiled info structure, computed during initialization */ String _pattern; /**< @brief PCRE pattern string, containing PCRE patterns and capturing groups. */ String _replacement; /**< @brief PCRE replacement string, containing $0..$9 to be replaced with content of the capturing groups */ --- trafficserver-9.2.3+ds.orig/plugins/prefetch/pattern.cc +++ trafficserver-9.2.3+ds/plugins/prefetch/pattern.cc @@ -139,14 +139,9 @@ Pattern::pcreFree() { if (_re) { - pcre_free(_re); + pcre2_code_free(_re); _re = nullptr; } - - if (_extra) { - pcre_free(_extra); - _extra = nullptr; - } } /** @@ -204,6 +199,7 @@ bool Pattern::match(const String &subject) { + pcre2_match_data *md; int matchCount; PrefetchDebug("matching '%s' to '%s'", _pattern.c_str(), subject.c_str()); @@ -211,13 +207,16 @@ return false; } - matchCount = pcre_exec(_re, _extra, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, nullptr, 0); + md = pcre2_match_data_create_from_pattern(_re, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { PrefetchError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } + pcre2_match_data_free(md); return true; } @@ -230,8 +229,9 @@ bool Pattern::capture(const String &subject, StringVector &result) { + pcre2_match_data *md; int matchCount; - int ovector[OVECOUNT]; + PCRE2_SIZE *ovector; PrefetchDebug("matching '%s' to '%s'", _pattern.c_str(), subject.c_str()); @@ -239,23 +239,27 @@ return false; } - matchCount = pcre_exec(_re, nullptr, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, ovector, OVECOUNT); + md = pcre2_match_data_create(OVECOUNT, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { PrefetchError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } + ovector = pcre2_get_ovector_pointer(md); for (int i = 0; i < matchCount; i++) { - int start = ovector[2 * i]; - int length = ovector[2 * i + 1] - ovector[2 * i]; + int start = (int)ovector[2 * i]; + int length = (int)(ovector[2 * i + 1] - ovector[2 * i]); String dst(subject, start, length); - PrefetchDebug("capturing '%s' %d[%d,%d]", dst.c_str(), i, ovector[2 * i], ovector[2 * i + 1]); + PrefetchDebug("capturing '%s' %d[%zu,%zu]", dst.c_str(), i, ovector[2 * i], ovector[2 * i + 1]); result.push_back(dst); } + pcre2_match_data_free(md); return true; } @@ -269,8 +273,9 @@ bool Pattern::replace(const String &subject, String &result) { + pcre2_match_data *md; int matchCount; - int ovector[OVECOUNT]; + PCRE2_SIZE *ovector; PrefetchDebug("matching '%s' to '%s'", _pattern.c_str(), subject.c_str()); @@ -278,11 +283,13 @@ return false; } - matchCount = pcre_exec(_re, nullptr, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, ovector, OVECOUNT); + md = pcre2_match_data_create(OVECOUNT, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { PrefetchError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } @@ -294,11 +301,12 @@ } } + ovector = pcre2_get_ovector_pointer(md); int previous = 0; for (int i = 0; i < _tokenCount; i++) { int replIndex = _tokens[i]; - int start = ovector[2 * replIndex]; - int length = ovector[2 * replIndex + 1] - ovector[2 * replIndex]; + int start = (int)ovector[2 * replIndex]; + int length = (int)(ovector[2 * replIndex + 1] - ovector[2 * replIndex]); String src(_replacement, _tokenOffset[i], 2); String dst(subject, start, length); @@ -314,6 +322,7 @@ result.append(_replacement, previous, _replacement.length() - previous); PrefetchDebug("replacing '%s' resulted in '%s'", _replacement.c_str(), result.c_str()); + pcre2_match_data_free(md); return true; } @@ -325,30 +334,23 @@ bool Pattern::compile() { - const char *errPtr; /* PCRE error */ - int errOffset; /* PCRE error offset */ + int errCode; /* PCRE error code */ + PCRE2_UCHAR errPtr[120]; /* PCRE error */ + PCRE2_SIZE errOffset; /* PCRE error offset */ PrefetchDebug("compiling pattern:'%s', replacement:'%s'", _pattern.c_str(), _replacement.c_str()); - _re = pcre_compile(_pattern.c_str(), /* the pattern */ - 0, /* options */ - &errPtr, /* for error message */ - &errOffset, /* for error offset */ - nullptr); /* use default character tables */ + _re = pcre2_compile((PCRE2_SPTR)_pattern.c_str(), /* the pattern */ + PCRE2_ZERO_TERMINATED, + 0, /* options */ + &errCode, /* for error code */ + &errOffset, /* for error offset */ + comp_ctxt); /* compile context */ if (nullptr == _re) { - PrefetchError("compile of regex '%s' at char %d: %s", _pattern.c_str(), errOffset, errPtr); - - return false; - } - - _extra = pcre_study(_re, 0, &errPtr); + pcre2_get_error_message(errCode, errPtr, sizeof(errPtr)); + PrefetchError("compile of regex '%s' at char %zu: %s", _pattern.c_str(), errOffset, errPtr); - if ((nullptr == _extra) && (nullptr != errPtr) && (0 != *errPtr)) { - PrefetchError("failed to study regex '%s': %s", _pattern.c_str(), errPtr); - - pcre_free(_re); - _re = nullptr; return false; } --- trafficserver-9.2.3+ds.orig/plugins/prefetch/pattern.h +++ trafficserver-9.2.3+ds/plugins/prefetch/pattern.h @@ -23,11 +23,10 @@ #pragma once -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#include "tscore/ink_memory.h" + +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include "common.h" @@ -56,8 +55,7 @@ bool failed(const String &subject) const; void pcreFree(); - pcre *_re = nullptr; /**< @brief PCRE compiled info structure, computed during initialization */ - pcre_extra *_extra = nullptr; /**< @brief PCRE study data block, computed during initialization */ + pcre2_code *_re = nullptr; /**< @brief PCRE compiled info structure, computed during initialization */ String _pattern; /**< @brief PCRE pattern string, containing PCRE patterns and capturing groups. */ String _replacement; /**< @brief PCRE replacement string, containing $0..$9 to be replaced with content of the capturing groups */ --- trafficserver-9.2.3+ds.orig/plugins/regex_remap/regex_remap.cc +++ trafficserver-9.2.3+ds/plugins/regex_remap/regex_remap.cc @@ -44,11 +44,8 @@ #include "tscore/ink_time.h" #include "tscore/ink_inet.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> static const char *PLUGIN_NAME = "regex_remap"; @@ -114,16 +111,17 @@ public: ~RemapRegex() { + uint32_t depth; + TSDebug(PLUGIN_NAME, "Calling destructor"); TSfree(_rex_string); TSfree(_subst); if (_rex) { - pcre_free(_rex); - } - if (_extra) { - pcre_free(_extra); + pcre2_code_free(_rex); } + pcre2_config(PCRE2_CONFIG_DEPTHLIMIT, &depth); + pcre2_set_depth_limit(match_ctxt, depth); } bool initialize(const std::string ®, const std::string &sub, const std::string &opt); @@ -140,25 +138,31 @@ fprintf(stderr, "[%s]: Regex %d ( %s ): %.2f%%\n", now, ix, _rex_string, 100.0 * _hits / max); } - int compile(const char *&error, int &erroffset); + int compile(char *&error, PCRE2_SIZE &erroffset); // Perform the regular expression matching against a string. int - match(const char *str, int len, int ovector[]) + match(const char *str, int len, PCRE2_SIZE ovector[] ATS_UNUSED) { - return pcre_exec(_rex, // the compiled pattern - _extra, // Extra data from study (maybe) - str, // the subject string + pcre2_match_data *md; + int rc; + + md = pcre2_match_data_create(OVECCOUNT, gen_ctxt); + rc = pcre2_match(_rex, // the compiled pattern + (PCRE2_SPTR)str, // the subject string len, // the length of the subject 0, // start at offset 0 in the subject 0, // default options - ovector, // output vector for substring information - OVECCOUNT); // number of elements in the output vector + md, // match data + match_ctxt); // match context + ovector = pcre2_get_ovector_pointer(md); + pcre2_match_data_free(md); + return rc; } // Substitutions - int get_lengths(const int ovector[], int lengths[], TSRemapRequestInfo *rri, UrlComponents *req_url); - int substitute(char dest[], const char *src, const int ovector[], const int lengths[], TSHttpTxn txnp, TSRemapRequestInfo *rri, + int get_lengths(const PCRE2_SIZE ovector[], int lengths[], TSRemapRequestInfo *rri, UrlComponents *req_url); + int substitute(char dest[], const char *src, const PCRE2_SIZE ovector[], const int lengths[], TSHttpTxn txnp, TSRemapRequestInfo *rri, UrlComponents *req_url, bool lowercase_substitutions); // setter / getters for members the linked list. @@ -248,13 +252,12 @@ int _subst_len = 0; int _num_subs = -1; int _hits = 0; - int _options = 0; + uint32_t _options = 0; int _order = -1; bool _lowercase_substitutions = false; - pcre *_rex = nullptr; - pcre_extra *_extra = nullptr; + pcre2_code *_rex = nullptr; RemapRegex *_next = nullptr; TSHttpStatus _status = static_cast<TSHttpStatus>(0); @@ -306,7 +309,7 @@ // These take an option 0|1 value, without value it implies 1 if (opt.compare(start, 8, "caseless") == 0) { - _options |= PCRE_CASELESS; + _options |= PCRE2_CASELESS; } else if (opt.compare(start, 23, "lowercase_substitutions") == 0) { _lowercase_substitutions = true; } else if (opt_val.size() <= 0) { @@ -370,36 +373,33 @@ // Compile and study the regular expression. int -RemapRegex::compile(const char *&error, int &erroffset) +RemapRegex::compile(char *&error, PCRE2_SIZE &erroffset) { + PCRE2_UCHAR buf[120]; char *str; int ccount; + int errcode; + uint32_t match_limit_recursion; - // Initialize these in case they are not set. - error = "unknown error"; - erroffset = -1; - - _rex = pcre_compile(_rex_string, // the pattern - _options, // options - &error, // for error message - &erroffset, // for error offset - nullptr); // use default character tables + _rex = pcre2_compile((PCRE2_SPTR)_rex_string, // the pattern + PCRE2_ZERO_TERMINATED, + _options, // options + &errcode, // for error message + &erroffset, // for error offset + comp_ctxt); // use default character tables + pcre2_get_error_message(errcode, buf, sizeof(buf)); + error = strdup((const char *)buf); if (nullptr == _rex) { return -1; } - _extra = pcre_study(_rex, PCRE_STUDY_EXTRA_NEEDED, &error); - if (error != nullptr) { - return -1; - } - // POOMA - also dependent on actual stack size. Crashes with previous value of 2047, - _extra->match_limit_recursion = 1750; - _extra->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION; + match_limit_recursion = 1750; + pcre2_set_depth_limit(match_ctxt, match_limit_recursion); - if (pcre_fullinfo(_rex, _extra, PCRE_INFO_CAPTURECOUNT, &ccount) != 0) { - error = "call to pcre_fullinfo() failed"; + if (pcre2_pattern_info(_rex, PCRE2_INFO_CAPTURECOUNT, &ccount) != 0) { + error = strdup("call to pcre2_pattern_info() failed"); return -1; } @@ -452,7 +452,7 @@ if (ix > -1) { if ((ix < 10) && (ix > ccount)) { - error = "using unavailable captured substring ($n) in substitution"; + error = strdup("using unavailable captured substring ($n) in substitution"); return -1; } @@ -474,7 +474,7 @@ // We also calculate a total length for the new string, which is the max length the // substituted string can have (use it to allocate a buffer before calling substitute() ). int -RemapRegex::get_lengths(const int ovector[], int lengths[], TSRemapRequestInfo *rri, UrlComponents *req_url) +RemapRegex::get_lengths(const PCRE2_SIZE ovector[], int lengths[], TSRemapRequestInfo *rri, UrlComponents *req_url) { int len = _subst_len + 1; // Bigger then necessary @@ -482,7 +482,7 @@ int ix = _sub_ix[i]; if (ix < 10) { - lengths[ix] = ovector[2 * ix + 1] - ovector[2 * ix]; // -1 - -1 == 0 + lengths[ix] = (int)(ovector[2 * ix + 1] - ovector[2 * ix]); // -1 - -1 == 0 len += lengths[ix]; } else { int tmp_len; @@ -531,7 +531,7 @@ // regex that was matches, while $1 - $9 are the corresponding groups. Return the final // length of the string as written to dest (not including the trailing '0'). int -RemapRegex::substitute(char dest[], const char *src, const int ovector[], const int lengths[], TSHttpTxn txnp, +RemapRegex::substitute(char dest[], const char *src, const PCRE2_SIZE ovector[], const int lengths[], TSHttpTxn txnp, TSRemapRequestInfo *rri, UrlComponents *req_url, bool lowercase_substitutions) { if (_num_subs > 0) { @@ -792,8 +792,8 @@ continue; } - const char *error; - int erroffset; + char *error; + PCRE2_SIZE erroffset; if (cur->compile(error, erroffset) < 0) { std::ostringstream oss; oss << '[' << PLUGIN_NAME << "] PCRE failed in " << (ri->filename).c_str() << " (line " << lineno << ')'; @@ -818,6 +818,7 @@ } ri->last = tmp; } + free(error); } // Make sure we got something... @@ -925,7 +926,7 @@ UrlComponents req_url; req_url.populate(src_url.bufp, src_url.loc); - int ovector[OVECCOUNT]; + PCRE2_SIZE *ovector = nullptr; int lengths[OVECCOUNT / 2 + 1]; int dest_len; TSRemapStatus retval = TSREMAP_DID_REMAP; --- trafficserver-9.2.3+ds.orig/plugins/regex_revalidate/regex_revalidate.c +++ trafficserver-9.2.3+ds/plugins/regex_revalidate/regex_revalidate.c @@ -20,6 +20,7 @@ */ #include <ts/ts.h> +#include <tscore/ink_memory.h> #include <stdlib.h> #include <stdio.h> @@ -32,11 +33,8 @@ #include <sys/stat.h> #include <unistd.h> -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #define CONFIG_TMOUT 60000 #define FREE_TMOUT 300000 @@ -113,8 +111,7 @@ typedef struct invalidate_t { const char *regex_text; - pcre *regex; - pcre_extra *regex_extra; + pcre2_code *regex; time_t epoch; time_t expiry; TSCacheLookupResult new_result; @@ -134,7 +131,6 @@ { i->regex_text = NULL; i->regex = NULL; - i->regex_extra = NULL; i->epoch = 0; i->expiry = 0; i->new_result = TS_CACHE_LOOKUP_HIT_STALE; @@ -145,18 +141,11 @@ static void free_invalidate_t(invalidate_t *i) { - if (i->regex_extra) { -#ifndef PCRE_STUDY_JIT_COMPILE - pcre_free(i->regex_extra); -#else - pcre_free_study(i->regex_extra); -#endif - } if (i->regex) { - pcre_free(i->regex); + pcre2_code_free(i->regex); } if (i->regex_text) { - pcre_free_substring(i->regex_text); + pcre2_substring_free((PCRE2_UCHAR *)i->regex_text); } TSfree(i); } @@ -204,13 +193,10 @@ copy_invalidate_t(invalidate_t *i) { invalidate_t *iptr; - const char *errptr; - int erroffset; iptr = (invalidate_t *)TSmalloc(sizeof(invalidate_t)); iptr->regex_text = TSstrdup(i->regex_text); - iptr->regex = pcre_compile(iptr->regex_text, 0, &errptr, &erroffset, NULL); // There is no pcre_copy :-( - iptr->regex_extra = pcre_study(iptr->regex, 0, &errptr); // Assuming no errors since this worked before :-/ + iptr->regex = pcre2_code_copy(i->regex); iptr->epoch = i->epoch; iptr->expiry = i->expiry; iptr->new_result = i->new_result; @@ -280,9 +266,10 @@ struct stat s; char line[LINE_MAX]; time_t now; - pcre *config_re = NULL; - const char *errptr; - int erroffset, ovector[OVECTOR_SIZE], rc; + pcre2_code *config_re = NULL; + pcre2_match_data *md; + int err, rc; + PCRE2_SIZE erroffset, *ovector; int ln = 0; if (!*ilist) { @@ -301,19 +288,21 @@ now = time(NULL); - config_re = pcre_compile("^([^#].+?)\\s+(\\d+)\\s+(\\d+)\\s+(\\w+)\\s*$", 0, &errptr, &erroffset, NULL); + config_re = pcre2_compile((PCRE2_SPTR)"^([^#].+?)\\s+(\\d+)\\s+(\\d+)\\s+(\\w+)\\s*$", PCRE2_ZERO_TERMINATED, 0, &err, &erroffset, comp_ctxt); TSAssert(NULL != config_re); + md = pcre2_match_data_create(OVECTOR_SIZE, gen_ctxt); while (fgets(line, LINE_MAX, fs) != NULL) { TSDebug(PLUGIN_NAME, "state: processing: %d %s", ln, line); ++ln; - rc = pcre_exec(config_re, NULL, line, strlen(line), 0, 0, ovector, OVECTOR_SIZE); + rc = pcre2_match(config_re, (PCRE2_SPTR)line, strlen(line), 0, 0, md, match_ctxt); + ovector = pcre2_get_ovector_pointer(md); if (5 == rc) { invalidate_t *const inv = (invalidate_t *)TSmalloc(sizeof(invalidate_t)); init_invalidate_t(inv); - pcre_get_substring(line, ovector, rc, 1, &(inv->regex_text)); + pcre2_substring_get_bynumber(md, 1, (PCRE2_UCHAR **)&(inv->regex_text), &erroffset); inv->epoch = atoi(line + ovector[4]); inv->expiry = atoi(line + ovector[6]); @@ -323,7 +312,7 @@ continue; } - int const len = ovector[9] - ovector[8]; + int const len = (int)(ovector[9] - ovector[8]); char const *const type = line + ovector[8]; if (0 == strncasecmp(type, RESULT_STALE, len)) { @@ -358,7 +347,8 @@ } } - pcre_free(config_re); + pcre2_match_data_free(md); + pcre2_code_free(config_re); fclose(fs); return true; } @@ -372,9 +362,10 @@ char *path; char line[LINE_MAX]; time_t now; - pcre *config_re; - const char *errptr; - int erroffset, ovector[OVECTOR_SIZE], rc; + pcre2_code *config_re; + pcre2_match_data *md = NULL; + int err, rc; + PCRE2_SIZE erroffset, *ovector; int ln = 0; invalidate_t *iptr, *i; @@ -395,25 +386,27 @@ TSDebug(PLUGIN_NAME, "Could not open %s for reading", path); return false; } - config_re = pcre_compile("^([^#].+?)\\s+(\\d+)(\\s+(\\w+))?\\s*$", 0, &errptr, &erroffset, NULL); + config_re = pcre2_compile((PCRE2_SPTR)"^([^#].+?)\\s+(\\d+)(\\s+(\\w+))?\\s*$", PCRE2_ZERO_TERMINATED, 0, &err, &erroffset, comp_ctxt); TSAssert(NULL != config_re); + md = pcre2_match_data_create(OVECTOR_SIZE, gen_ctxt); while (fgets(line, LINE_MAX, fs) != NULL) { TSDebug(PLUGIN_NAME, "Processing: %d %s", ln, line); ++ln; - rc = pcre_exec(config_re, NULL, line, strlen(line), 0, 0, ovector, OVECTOR_SIZE); + rc = pcre2_match(config_re, (PCRE2_SPTR)line, strlen(line), 0, 0, md, match_ctxt); + ovector = pcre2_get_ovector_pointer(md); if (3 <= rc) { i = (invalidate_t *)TSmalloc(sizeof(invalidate_t)); init_invalidate_t(i); - pcre_get_substring(line, ovector, rc, 1, &i->regex_text); + pcre2_substring_get_bynumber(md, 1, (PCRE2_UCHAR **)&i->regex_text, &erroffset); - i->regex = pcre_compile(i->regex_text, 0, &errptr, &erroffset, NULL); + i->regex = pcre2_compile((PCRE2_SPTR)i->regex_text, PCRE2_ZERO_TERMINATED, 0, &err, &erroffset, comp_ctxt); i->epoch = now; i->expiry = atoi(line + ovector[4]); if (5 == rc) { - int const len = ovector[9] - ovector[8]; + int const len = (int)(ovector[9] - ovector[8]); char const *const type = line + ovector[8]; if (0 == strncasecmp(type, RESULT_MISS, len)) { TSDebug(PLUGIN_NAME, "Regex line set to result type %s: '%s'", RESULT_MISS, i->regex_text); @@ -432,7 +425,6 @@ free_invalidate_t(i); i = NULL; } else { - i->regex_extra = pcre_study(i->regex, 0, &errptr); if (!*ilist) { *ilist = i; TSDebug(PLUGIN_NAME, "Created new list and Loaded %s %d %d %s", i->regex_text, (int)i->epoch, (int)i->expiry, @@ -470,13 +462,15 @@ TSDebug(PLUGIN_NAME, "Skipping line %d", ln); } } - pcre_free(config_re); + pcre2_match_data_free(md); + pcre2_code_free(config_re); fclose(fs); pstate->last_load = s.st_mtime; return true; } else { TSDebug(PLUGIN_NAME, "File mod time is not newer: %d >= %d", (int)pstate->last_load, (int)s.st_mtime); } + pcre2_match_data_free(md); return false; } @@ -607,6 +601,7 @@ int status; invalidate_t *iptr; plugin_state_t *pstate; + pcre2_match_data *md = NULL; time_t date = 0, now = 0; char *url = NULL; @@ -618,6 +613,7 @@ if (status == TS_CACHE_LOOKUP_HIT_FRESH) { pstate = (plugin_state_t *)TSContDataGet(cont); iptr = pstate->invalidate_list; + md = pcre2_match_data_create_from_pattern(iptr->regex, gen_ctxt); while (iptr) { if (!date) { date = get_date_from_cached_hdr(txn); @@ -627,7 +623,7 @@ if (!url) { url = TSHttpTxnEffectiveUrlStringGet(txn, &url_len); } - if (pcre_exec(iptr->regex, iptr->regex_extra, url, url_len, 0, 0, NULL, 0) >= 0) { + if (pcre2_match(iptr->regex, (PCRE2_SPTR)url, url_len, 0, 0, md, match_ctxt) >= 0) { TSHttpTxnCacheLookupStatusSet(txn, iptr->new_result); increment_stat(iptr->new_result); TSDebug(PLUGIN_NAME, "Forced revalidate - %.*s %s", url_len, url, strForResult(iptr->new_result)); @@ -649,6 +645,7 @@ } TSHttpTxnReenable(txn, TS_EVENT_HTTP_CONTINUE); + pcre2_match_data_free(md); return 0; } --- trafficserver-9.2.3+ds.orig/plugins/experimental/access_control/pattern.cc +++ trafficserver-9.2.3+ds/plugins/experimental/access_control/pattern.cc @@ -149,14 +149,9 @@ Pattern::pcreFree() { if (_re) { - pcre_free(_re); + pcre2_code_free(_re); _re = nullptr; } - - if (_extra) { - pcre_free(_extra); - _extra = nullptr; - } } /** @@ -214,6 +209,7 @@ bool Pattern::match(const String &subject) { + pcre2_match_data *md; int matchCount; AccessControlDebug("matching '%s' to '%s'", _pattern.c_str(), subject.c_str()); @@ -221,9 +217,11 @@ return false; } - matchCount = pcre_exec(_re, _extra, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, nullptr, 0); + md = pcre2_match_data_create_from_pattern(_re, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); + pcre2_match_data_free(md); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { AccessControlError("matching error %d", matchCount); } return false; @@ -240,8 +238,9 @@ bool Pattern::capture(const String &subject, StringVector &result) { + pcre2_match_data *md; int matchCount; - int ovector[OVECOUNT]; + PCRE2_SIZE *ovector; AccessControlDebug("capturing '%s' from '%s'", _pattern.c_str(), subject.c_str()); @@ -250,23 +249,27 @@ return false; } - matchCount = pcre_exec(_re, nullptr, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, ovector, OVECOUNT); + md = pcre2_match_data_create(OVECOUNT, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { AccessControlError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } + ovector = pcre2_get_ovector_pointer(md); for (int i = 0; i < matchCount; i++) { - int start = ovector[2 * i]; - int length = ovector[2 * i + 1] - ovector[2 * i]; + int start = (int)ovector[2 * i]; + int length = (int)(ovector[2 * i + 1] - ovector[2 * i]); String dst(subject, start, length); - AccessControlDebug("capturing '%s' %d[%d,%d]", dst.c_str(), i, ovector[2 * i], ovector[2 * i + 1]); + AccessControlDebug("capturing '%s' %d[%zu,%zu]", dst.c_str(), i, ovector[2 * i], ovector[2 * i + 1]); result.push_back(dst); } + pcre2_match_data_free(md); return true; } @@ -280,8 +283,9 @@ bool Pattern::replace(const String &subject, String &result) { + pcre2_match_data *md; int matchCount; - int ovector[OVECOUNT]; + PCRE2_SIZE *ovector; AccessControlDebug("replacing:'%s' in pattern:'%s', subject:'%s'", _replacement.c_str(), _pattern.c_str(), subject.c_str()); @@ -290,11 +294,13 @@ return false; } - matchCount = pcre_exec(_re, nullptr, subject.c_str(), subject.length(), 0, PCRE_NOTEMPTY, ovector, OVECOUNT); + md = pcre2_match_data_create_from_pattern(_re, gen_ctxt); + matchCount = pcre2_match(_re, (PCRE2_SPTR)subject.c_str(), subject.length(), 0, PCRE2_NOTEMPTY, md, match_ctxt); if (matchCount < 0) { - if (matchCount != PCRE_ERROR_NOMATCH) { + if (matchCount != PCRE2_ERROR_NOMATCH) { AccessControlError("matching error %d", matchCount); } + pcre2_match_data_free(md); return false; } @@ -302,15 +308,17 @@ for (int i = 0; i < _tokenCount; i++) { if (_tokens[i] >= matchCount) { AccessControlError("invalid reference in replacement string: $%d", _tokens[i]); + pcre2_match_data_free(md); return false; } } + ovector = pcre2_get_ovector_pointer(md); int previous = 0; for (int i = 0; i < _tokenCount; i++) { int replIndex = _tokens[i]; - int start = ovector[2 * replIndex]; - int length = ovector[2 * replIndex + 1] - ovector[2 * replIndex]; + int start = (int)ovector[2 * replIndex]; + int length = (int)(ovector[2 * replIndex + 1] - ovector[2 * replIndex]); String src(_replacement, _tokenOffset[i], 2); String dst(subject, start, length); @@ -326,6 +334,7 @@ result.append(_replacement, previous, _replacement.length() - previous); AccessControlDebug("replacing '%s' resulted in '%s'", _replacement.c_str(), result.c_str()); + pcre2_match_data_free(md); return true; } @@ -337,31 +346,24 @@ bool Pattern::compile() { - const char *errPtr; /* PCRE error */ - int errOffset; /* PCRE error offset */ + int errCode; /* PCRE error code */ + PCRE2_UCHAR errPtr[120]; /* PCRE error */ + PCRE2_SIZE errOffset; /* PCRE error offset */ AccessControlDebug("compiling pattern:'%s', replace: %s, replacement:'%s'", _pattern.c_str(), _replace ? "true" : "false", _replacement.c_str()); - _re = pcre_compile(_pattern.c_str(), /* the pattern */ - 0, /* options */ - &errPtr, /* for error message */ - &errOffset, /* for error offset */ - nullptr); /* use default character tables */ + _re = pcre2_compile((PCRE2_SPTR)_pattern.c_str(), /* the pattern */ + PCRE2_ZERO_TERMINATED, + 0, /* options */ + &errCode, /* for error code */ + &errOffset, /* for error offset */ + comp_ctxt); /* compile context */ if (nullptr == _re) { - AccessControlError("compile of regex '%s' at char %d: %s", _pattern.c_str(), errOffset, errPtr); - - return false; - } - - _extra = pcre_study(_re, 0, &errPtr); + pcre2_get_error_message(errCode, errPtr, sizeof(errPtr)); + AccessControlError("compile of regex '%s' at char %zu: %s", _pattern.c_str(), errOffset, errPtr); - if ((nullptr == _extra) && (nullptr != errPtr) && (0 != *errPtr)) { - AccessControlError("failed to study regex '%s': %s", _pattern.c_str(), errPtr); - - pcre_free(_re); - _re = nullptr; return false; } --- trafficserver-9.2.3+ds.orig/plugins/experimental/access_control/pattern.h +++ trafficserver-9.2.3+ds/plugins/experimental/access_control/pattern.h @@ -23,11 +23,10 @@ #pragma once -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#include "tscore/ink_memory.h" + +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include "common.h" @@ -56,8 +55,7 @@ bool compile(); void pcreFree(); - pcre *_re = nullptr; /**< @brief PCRE compiled info structure, computed during initialization */ - pcre_extra *_extra = nullptr; /**< @brief PCRE study data block, computed during initialization */ + pcre2_code *_re = nullptr; /**< @brief PCRE compiled info structure, computed during initialization */ String _pattern; /**< @brief PCRE pattern string, containing PCRE patterns and capturing groups. */ String _replacement; /**< @brief PCRE replacement string, containing $0..$9 to be replaced with content of the capturing groups */ --- trafficserver-9.2.3+ds.orig/plugins/experimental/cookie_remap/cookie_remap.cc +++ trafficserver-9.2.3+ds/plugins/experimental/cookie_remap/cookie_remap.cc @@ -24,8 +24,11 @@ #include "cookiejar.h" #include <ts/ts.h> -#include <pcre.h> +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include <ts/remap.h> +#include <tscore/ink_defs.h> +#include <tscore/ink_memory.h> #include <yaml-cpp/yaml.h> #include <string> @@ -261,11 +264,7 @@ { TSDebug(MY_NAME, "subop destructor called"); if (regex) { - pcre_free(regex); - } - - if (regex_extra) { - pcre_free(regex_extra); + pcre2_code_free(regex); } } @@ -378,23 +377,18 @@ bool setRegexMatch(const std::string &s) { - const char *error_comp = nullptr; - const char *error_study = nullptr; - int erroffset; + int error; + PCRE2_SIZE erroffset; op_type = REGEXP; regex_string = s; - regex = pcre_compile(regex_string.c_str(), 0, &error_comp, &erroffset, nullptr); + regex = pcre2_compile((PCRE2_SPTR)regex_string.c_str(), PCRE2_ZERO_TERMINATED, 0, &error, &erroffset, comp_ctxt); if (regex == nullptr) { return false; } - regex_extra = pcre_study(regex, 0, &error_study); - if ((regex_extra == nullptr) && (error_study != nullptr)) { - return false; - } - if (pcre_fullinfo(regex, regex_extra, PCRE_INFO_CAPTURECOUNT, ®ex_ccount) != 0) { + if (pcre2_pattern_info(regex, PCRE2_INFO_CAPTURECOUNT, ®ex_ccount) != 0) { return false; } @@ -414,16 +408,22 @@ } int - regexMatch(const char *str, int len, int ovector[]) const + regexMatch(const char *str, int len, PCRE2_SIZE ovector[] ATS_UNUSED) const { - return pcre_exec(regex, // the compiled pattern - regex_extra, // Extra data from study (maybe) - str, // the subject std::string + pcre2_match_data *md; + int rc; + + md = pcre2_match_data_create(OVECCOUNT, gen_ctxt); + rc = pcre2_match(regex, // the compiled pattern + (PCRE2_SPTR)str, // the subject std::string len, // the length of the subject 0, // start at offset 0 in the subject 0, // default options - ovector, // output vector for substring information - OVECCOUNT); // number of elements in the output vector + md, // match data + match_ctxt); // match context + ovector = pcre2_get_ovector_pointer(md); + pcre2_match_data_free(md); + return rc; }; void @@ -453,10 +453,9 @@ std::string str_match; - pcre *regex = nullptr; - pcre_extra *regex_extra = nullptr; + pcre2_code *regex = nullptr; std::string regex_string; - int regex_ccount = 0; + uint32_t regex_ccount = 0; std::string bucket; unsigned int how_many = 0; @@ -687,7 +686,7 @@ // OPERATION::regex matching if (subop_type == REGEXP) { - int ovector[OVECCOUNT]; + PCRE2_SIZE *ovector = nullptr; int ret = subop->regexMatch(string_to_match.c_str(), string_to_match.length(), ovector); if (ret >= 0) { @@ -720,7 +719,7 @@ if (ix <= subop->getRegexCcount()) { // Just skip an illegal regex group dest += sendto.substr(ppos, pos - ppos); - dest += string_to_match.substr(ovector[ix * 2], ovector[ix * 2 + 1] - ovector[ix * 2]); + dest += string_to_match.substr((int)ovector[ix * 2], (int)(ovector[ix * 2 + 1] - ovector[ix * 2])); ppos = pos + 2; } else { TSDebug(MY_NAME, --- trafficserver-9.2.3+ds.orig/plugins/experimental/geoip_acl/acl.cc +++ trafficserver-9.2.3+ds/plugins/experimental/geoip_acl/acl.cc @@ -170,20 +170,15 @@ bool RegexAcl::compile(const std::string &str, const char *filename, int lineno) { - const char *error; - int erroffset; + PCRE2_UCHAR error[120]; + int errcode; + PCRE2_SIZE erroffset; _regex_s = str; - _rex = pcre_compile(_regex_s.c_str(), 0, &error, &erroffset, nullptr); + _rex = pcre2_compile((PCRE2_SPTR)_regex_s.c_str(), PCRE2_ZERO_TERMINATED, 0, &errcode, &erroffset, comp_ctxt); - if (nullptr != _rex) { - _extra = pcre_study(_rex, 0, &error); - if ((nullptr == _extra) && error && (*error != 0)) { - TSError("[%s] Failed to study regular expression in %s:line %d at offset %d: %s", PLUGIN_NAME, filename, lineno, erroffset, - error); - return false; - } - } else { + if (nullptr == _rex) { + pcre2_get_error_message(errcode, error, sizeof(error)); TSError("[%s] Failed to compile regular expression in %s:line %d: %s", PLUGIN_NAME, filename, lineno, error); return false; } --- trafficserver-9.2.3+ds.orig/plugins/experimental/geoip_acl/acl.h +++ trafficserver-9.2.3+ds/plugins/experimental/geoip_acl/acl.h @@ -26,12 +26,10 @@ #include <ts/remap.h> #include "tscore/ink_defs.h" +#include "tscore/ink_memory.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include <string> #include "lulu.h" @@ -96,7 +94,7 @@ class RegexAcl { public: - RegexAcl(Acl *acl) : _rex(nullptr), _extra(nullptr), _next(nullptr), _acl(acl) {} + RegexAcl(Acl *acl) : _rex(nullptr), _next(nullptr), _acl(acl) {} const std::string & get_regex() const { @@ -116,11 +114,18 @@ bool match(const char *str, int len) const { + pcre2_match_data *md; + int ret; + if (0 == len) { return false; } - return (pcre_exec(_rex, _extra, str, len, 0, PCRE_NOTEMPTY, nullptr, 0) != -1); + md = pcre2_match_data_create_from_pattern(_rex, gen_ctxt); + ret = pcre2_match(_rex, (PCRE2_SPTR)str, len, 0, PCRE2_NOTEMPTY, md, match_ctxt); + pcre2_match_data_free(md); + + return (ret != -1); } void append(RegexAcl *ra); @@ -129,8 +134,7 @@ private: bool compile(const std::string &str, const char *filename, int lineno); std::string _regex_s; - pcre *_rex; - pcre_extra *_extra; + pcre2_code *_rex; RegexAcl *_next; Acl *_acl; }; --- trafficserver-9.2.3+ds.orig/plugins/experimental/maxmind_acl/mmdb.cc +++ trafficserver-9.2.3+ds/plugins/experimental/maxmind_acl/mmdb.cc @@ -294,18 +294,14 @@ plugin_regex temp; auto temprule = i.as<std::vector<std::string>>(); temp._regex_s = temprule.back(); - const char *error; - int erroffset; - temp._rex = pcre_compile(temp._regex_s.c_str(), 0, &error, &erroffset, nullptr); + PCRE2_UCHAR error[120]; + int err; + PCRE2_SIZE erroffset; + temp._rex = pcre2_compile((PCRE2_SPTR)temp._regex_s.c_str(), PCRE2_ZERO_TERMINATED, 0, &err, &erroffset, comp_ctxt); // Compile the regex for this set of countries - if (nullptr != temp._rex) { - temp._extra = pcre_study(temp._rex, 0, &error); - if ((nullptr == temp._extra) && error && (*error != 0)) { - TSError("[%s] Failed to study regular expression in %s:%s", PLUGIN_NAME, temp._regex_s.c_str(), error); - return; - } - } else { + if (nullptr == temp._rex) { + pcre2_get_error_message(err, error, sizeof(error)); TSError("[%s] Failed to compile regular expression in %s: %s", PLUGIN_NAME, temp._regex_s.c_str(), error); return; } @@ -507,6 +503,7 @@ bool ret = false; bool allow = default_allow; char *output = nullptr; + pcre2_match_data *md; // We need to null terminate the iso_code ourselves, they are unterminated in the DBs output = static_cast<char *>(malloc((sizeof(char) * (entry_data->data_size + 1)))); @@ -526,10 +523,11 @@ ret = true; } + md = pcre2_match_data_create(30, gen_ctxt); if (nullptr != path && 0 != path_len) { if (!allow_regex[output].empty()) { for (auto &i : allow_regex[output]) { - if (PCRE_ERROR_NOMATCH != pcre_exec(i._rex, i._extra, path, path_len, 0, PCRE_NOTEMPTY, nullptr, 0)) { + if (PCRE2_ERROR_NOMATCH != pcre2_match(i._rex, (PCRE2_SPTR)path, path_len, 0, PCRE2_NOTEMPTY, md, match_ctxt)) { TSDebug(PLUGIN_NAME, "Got a regex allow hit on regex: %s, country: %s", i._regex_s.c_str(), output); ret = true; } @@ -537,7 +535,7 @@ } if (!deny_regex[output].empty()) { for (auto &i : deny_regex[output]) { - if (PCRE_ERROR_NOMATCH != pcre_exec(i._rex, i._extra, path, path_len, 0, PCRE_NOTEMPTY, nullptr, 0)) { + if (PCRE2_ERROR_NOMATCH != pcre2_match(i._rex, (PCRE2_SPTR)path, path_len, 0, PCRE2_NOTEMPTY, md, match_ctxt)) { TSDebug(PLUGIN_NAME, "Got a regex deny hit on regex: %s, country: %s", i._regex_s.c_str(), output); ret = false; } @@ -546,6 +544,7 @@ } free(output); + pcre2_match_data_free(md); return ret; } --- trafficserver-9.2.3+ds.orig/plugins/experimental/maxmind_acl/mmdb.h +++ trafficserver-9.2.3+ds/plugins/experimental/maxmind_acl/mmdb.h @@ -38,19 +38,15 @@ #include <maxminddb.h> #include "tscore/IpMap.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #define PLUGIN_NAME "maxmind_acl" #define CONFIG_TMOUT 60000 typedef struct { std::string _regex_s; - pcre *_rex; - pcre_extra *_extra; + pcre2_code *_rex; } plugin_regex; typedef enum { ALLOW_IP, DENY_IP, UNKNOWN_IP } ipstate; --- trafficserver-9.2.3+ds.orig/plugins/experimental/slice/Config.cc +++ trafficserver-9.2.3+ds/plugins/experimental/slice/Config.cc @@ -25,6 +25,7 @@ #include <string_view> #include "ts/experimental.h" +#include "tscore/ink_memory.h" namespace { @@ -34,15 +35,8 @@ Config::~Config() { - if (nullptr != m_regex_extra) { -#ifndef PCRE_STUDY_JIT_COMPILE - pcre_free(m_regex_extra); -#else - pcre_free_study(m_regex_extra); -#endif - } if (nullptr != m_regex) { - pcre_free(m_regex); + pcre2_code_free(m_regex); } } @@ -161,15 +155,14 @@ break; } - const char *errptr; - int erroffset; + int err; + PCRE2_SIZE erroffset; m_regexstr = optarg; - m_regex = pcre_compile(m_regexstr.c_str(), 0, &errptr, &erroffset, nullptr); + m_regex = pcre2_compile((PCRE2_SPTR)m_regexstr.c_str(), PCRE2_ZERO_TERMINATED, 0, &err, &erroffset, comp_ctxt); if (nullptr == m_regex) { ERROR_LOG("Invalid regex: '%s'", m_regexstr.c_str()); } else { m_regex_type = Exclude; - m_regex_extra = pcre_study(m_regex, 0, &errptr); DEBUG_LOG("Using regex for url exclude: '%s'", m_regexstr.c_str()); } } break; @@ -179,15 +172,14 @@ break; } - const char *errptr; - int erroffset; + int err; + PCRE2_SIZE erroffset; m_regexstr = optarg; - m_regex = pcre_compile(m_regexstr.c_str(), 0, &errptr, &erroffset, nullptr); + m_regex = pcre2_compile((PCRE2_SPTR)m_regexstr.c_str(), PCRE2_ZERO_TERMINATED, 0, &err, &erroffset, comp_ctxt); if (nullptr == m_regex) { ERROR_LOG("Invalid regex: '%s'", m_regexstr.c_str()); } else { m_regex_type = Include; - m_regex_extra = pcre_study(m_regex, 0, &errptr); DEBUG_LOG("Using regex for url include: '%s'", m_regexstr.c_str()); } } break; @@ -292,22 +284,25 @@ bool Config::matchesRegex(char const *const url, int const urllen) const { + pcre2_match_data *md; bool matches = true; + md = pcre2_match_data_create_from_pattern(m_regex, gen_ctxt); switch (m_regex_type) { case Exclude: { - if (0 <= pcre_exec(m_regex, m_regex_extra, url, urllen, 0, 0, nullptr, 0)) { + if (0 <= pcre2_match(m_regex, (PCRE2_SPTR)url, urllen, 0, 0, md, match_ctxt)) { matches = false; } } break; case Include: { - if (pcre_exec(m_regex, m_regex_extra, url, urllen, 0, 0, nullptr, 0) < 0) { + if (pcre2_match(m_regex, (PCRE2_SPTR)url, urllen, 0, 0, md, match_ctxt) < 0) { matches = false; } } break; default: break; } + pcre2_match_data_free(md); return matches; } --- trafficserver-9.2.3+ds.orig/plugins/experimental/slice/Config.h +++ trafficserver-9.2.3+ds/plugins/experimental/slice/Config.h @@ -20,11 +20,8 @@ #include "slice.h" -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include <string> #include <mutex> @@ -40,8 +37,7 @@ std::string m_regexstr; // regex string for things to slice (default all) enum RegexType { None, Include, Exclude }; RegexType m_regex_type{None}; - pcre *m_regex{nullptr}; - pcre_extra *m_regex_extra{nullptr}; + pcre2_code *m_regex{nullptr}; int m_paceerrsecs{0}; // -1 disable logging, 0 no pacing, max 60s int m_prefetchcount{0}; // 0 disables prefetching enum RefType { First, Relative }; --- trafficserver-9.2.3+ds.orig/plugins/experimental/tls_bridge/regex.cc +++ trafficserver-9.2.3+ds/plugins/experimental/tls_bridge/regex.cc @@ -22,68 +22,57 @@ */ #include "regex.h" +#include "tscore/ink_memory.h" -#ifdef PCRE_CONFIG_JIT #include <pthread.h> struct RegexThreadKey { - RegexThreadKey() { pthread_key_create(&this->key, reinterpret_cast<void (*)(void *)>(&pcre_jit_stack_free)); } + RegexThreadKey() { pthread_key_create(&this->key, reinterpret_cast<void (*)(void *)>(&pcre2_jit_stack_free)); } pthread_key_t key; }; static RegexThreadKey k; -static pcre_jit_stack * +static pcre2_jit_stack * get_jit_stack(void *) { - pcre_jit_stack *jit_stack; + pcre2_jit_stack *jit_stack; - if ((jit_stack = static_cast<pcre_jit_stack *>(pthread_getspecific(k.key))) == nullptr) { - jit_stack = pcre_jit_stack_alloc(8192, 1024 * 1024); // 1 page min and 1MB max + if ((jit_stack = static_cast<pcre2_jit_stack *>(pthread_getspecific(k.key))) == nullptr) { + jit_stack = pcre2_jit_stack_create(8192, 1024 * 1024, gen_ctxt); // 1 page min and 1MB max pthread_setspecific(k.key, (void *)jit_stack); } return jit_stack; } -#endif bool Regex::compile(const char *pattern, const unsigned flags) { - const char *error; - int erroffset; - int options = 0; - int study_opts = 0; + int error; + PCRE2_SIZE erroffset; + uint32_t options = 0; if (regex) { return false; } if (flags & CASE_INSENSITIVE) { - options |= PCRE_CASELESS; + options |= PCRE2_CASELESS; } if (flags & ANCHORED) { - options |= PCRE_ANCHORED; + options |= PCRE2_ANCHORED; } - regex = pcre_compile(pattern, options, &error, &erroffset, nullptr); - if (error) { + regex = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, options, &error, &erroffset, comp_ctxt); + if (0 != erroffset) { regex = nullptr; return false; } -#ifdef PCRE_CONFIG_JIT - study_opts |= PCRE_STUDY_JIT_COMPILE; -#endif - - regex_extra = pcre_study(regex, study_opts, &error); - -#ifdef PCRE_CONFIG_JIT - if (regex_extra) { - pcre_assign_jit_stack(regex_extra, &get_jit_stack, nullptr); - } -#endif + pcre2_jit_compile(regex, PCRE2_JIT_COMPLETE); + pcre2_jit_stack_assign(match_ctxt, &get_jit_stack, nullptr); return true; } @@ -92,7 +81,7 @@ Regex::get_capture_count() { int captures = -1; - if (pcre_fullinfo(regex, regex_extra, PCRE_INFO_CAPTURECOUNT, &captures) != 0) { + if (pcre2_pattern_info(regex, PCRE2_INFO_CAPTURECOUNT, &captures) != 0) { return -1; } @@ -109,22 +98,18 @@ bool Regex::exec(std::string_view src, int *ovector, int ovecsize) const { + pcre2_match_data *md; int rv; - rv = pcre_exec(regex, regex_extra, src.data(), src.size(), 0, 0, ovector, ovecsize); + md = pcre2_match_data_create(ovecsize, gen_ctxt); + rv = pcre2_match(regex, (PCRE2_SPTR)src.data(), src.size(), 0, 0, md, match_ctxt); + pcre2_match_data_free(md); return rv > 0 ? true : false; } Regex::~Regex() { - if (regex_extra) { -#ifdef PCRE_CONFIG_JIT - pcre_free_study(regex_extra); -#else - pcre_free(regex_extra); -#endif - } if (regex) { - pcre_free(regex); + pcre2_code_free(regex); } } --- trafficserver-9.2.3+ds.orig/plugins/experimental/tls_bridge/regex.h +++ trafficserver-9.2.3+ds/plugins/experimental/tls_bridge/regex.h @@ -23,7 +23,8 @@ #pragma once -#include <pcre.h> +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include <algorithm> #include <string_view> @@ -49,12 +50,10 @@ ~Regex(); private: - pcre *regex = nullptr; - pcre_extra *regex_extra = nullptr; + pcre2_code *regex = nullptr; }; inline Regex::Regex(self_type &&that) { std::swap(regex, that.regex); - std::swap(regex_extra, that.regex_extra); } --- trafficserver-9.2.3+ds.orig/plugins/experimental/url_sig/url_sig.c +++ trafficserver-9.2.3+ds/plugins/experimental/url_sig/url_sig.c @@ -24,6 +24,7 @@ }) #include "url_sig.h" +#include "tscore/ink_memory.h" #include <inttypes.h> #include <stdio.h> @@ -41,11 +42,8 @@ #include <stdint.h> #include <stdbool.h> -#ifdef HAVE_PCRE_PCRE_H -#include <pcre/pcre.h> -#else -#include <pcre.h> -#endif +#define PCRE2_CODE_UNIT_WIDTH 8 +#include <pcre2.h> #include <ts/ts.h> #include <ts/remap.h> @@ -56,8 +54,7 @@ TSHttpStatus err_status; char *err_url; char keys[MAX_KEY_NUM][MAX_KEY_LEN]; - pcre *regex; - pcre_extra *regex_extra; + pcre2_code *regex; int pristine_url_flag; char *sig_anchor; bool ignore_expiry; @@ -70,16 +67,8 @@ TSfree(cfg->err_url); TSfree(cfg->sig_anchor); - if (cfg->regex_extra) { -#ifndef PCRE_STUDY_JIT_COMPILE - pcre_free(cfg->regex_extra); -#else - pcre_free_study(cfg->regex_extra); -#endif - } - if (cfg->regex) { - pcre_free(cfg->regex); + pcre2_code_free(cfg->regex); } TSfree(cfg); @@ -213,23 +202,22 @@ cfg->sig_anchor = TSstrndup(value, strlen(value)); } else if (strncmp(line, "excl_regex", 10) == 0) { // compile and study regex - const char *errptr; - int erroffset, options = 0; + PCRE2_UCHAR errptr[120]; + PCRE2_SIZE erroffset; + int error; + uint32_t options = 0; if (cfg->regex) { TSDebug(PLUGIN_NAME, "Skipping duplicate excl_regex"); continue; } - cfg->regex = pcre_compile(value, options, &errptr, &erroffset, NULL); + cfg->regex = pcre2_compile((PCRE2_SPTR)value, PCRE2_ZERO_TERMINATED, options, &error, &erroffset, comp_ctxt); if (cfg->regex == NULL) { - TSDebug(PLUGIN_NAME, "Regex compilation failed with error (%s) at character %d", errptr, erroffset); + pcre2_get_error_message(error, errptr, sizeof(errptr)); + TSDebug(PLUGIN_NAME, "Regex compilation failed with error (%s) at character %zu", errptr, erroffset); } else { -#ifdef PCRE_STUDY_JIT_COMPILE - options = PCRE_STUDY_JIT_COMPILE; -#endif - cfg->regex_extra = pcre_study( - cfg->regex, options, &errptr); // We do not need to check the error here because we can still run without the studying? + pcre2_jit_compile(cfg->regex, PCRE2_JIT_COMPLETE); } } else if (strncmp(line, "ignore_expiry", 13) == 0) { if (strncmp(value, "true", 4) == 0) { @@ -592,7 +580,7 @@ if (cfg->regex) { const int offset = 0, options = 0; - int ovector[30]; + pcre2_match_data *md; /* Only search up to the first ? or # */ const char *base_url_end = url; @@ -601,9 +589,12 @@ } const int len = base_url_end - url; - if (pcre_exec(cfg->regex, cfg->regex_extra, url, len, offset, options, ovector, 30) >= 0) { + md = pcre2_match_data_create(30, gen_ctxt); + if (pcre2_match(cfg->regex, (PCRE2_SPTR)url, len, offset, options, md, match_ctxt) >= 0) { + pcre2_match_data_free(md); goto allow; } + pcre2_match_data_free(md); } const char *query = strchr(url, '?'); --- trafficserver-9.2.3+ds.orig/plugins/experimental/slice/Makefile.inc +++ trafficserver-9.2.3+ds/plugins/experimental/slice/Makefile.inc @@ -51,7 +51,7 @@ experimental/slice/unit-tests/test_content_range.cc \ experimental/slice/ContentRange.cc -experimental_slice_test_content_range_LDADD = @LIBPCRE@ +experimental_slice_test_content_range_LDADD = @LIBPCRE@ $(top_builddir)/src/tscore/libtscore.la check_PROGRAMS += experimental/slice/test_range @@ -60,7 +60,7 @@ experimental/slice/unit-tests/test_range.cc \ experimental/slice/Range.cc -experimental_slice_test_range_LDADD = @LIBPCRE@ +experimental_slice_test_range_LDADD = @LIBPCRE@ $(top_builddir)/src/tscore/libtscore.la check_PROGRAMS += experimental/slice/test_config @@ -69,4 +69,4 @@ experimental/slice/unit-tests/test_config.cc \ experimental/slice/Config.cc -experimental_slice_test_config_LDADD = @LIBPCRE@ +experimental_slice_test_config_LDADD = @LIBPCRE@ $(top_builddir)/src/tscore/libtscore.la
