Remove PCRE references, complete migration to PCRE2 (#12685)

- Remove cmake/FindPCRE.cmake (no longer needed)
- Update all CMakeLists.txt files to use PkgConfig::PCRE2 instead of PCRE::PCRE
- Remove find_package(PCRE) from main CMakeLists.txt
- Update code comments from PCRE to PCRE2 in plugin sources
- Update coverity model from PCRE to PCRE2 API
- Update README example to remove obsolete PCRE_ROOT reference
- Update tools/hrw4u/src/lsp/documentation.py to reference PCRE2
- Note: cripts::Matcher::PCRE class name intentionally unchanged (public API)
- Remove redundant PCRE2 links, use tsutil Regex wrapper
- Use RE_ERROR_NOMATCH from tsutil instead of PCRE2_ERROR_NOMATCH in
  ControlMatcher to properly use the Regex wrapper
- Remove direct pcre2.h include from ControlMatcher.h
- Remove redundant PCRE2 links from components that get it transitively
  through tsutil (tscore, tsapi)
- Remove redundant PCRE2 links from plugins that only use tsutil/Regex.h
  (cachekey, prefetch, slice, tls_bridge)

Only components that directly use PCRE2 APIs (tsutil, cripts, txn_box)
should explicitly link PCRE2. Others get it transitively through tsutil
which publicly exports PCRE2.

Author: bryancall

CMakeLists.txt

@@ -274,7 +274,6 @@ if(LibLZMA_FOUND)
   set(HAVE_LZMA_H TRUE)
 endif()
 
-find_package(PCRE REQUIRED)
 pkg_check_modules(PCRE2 REQUIRED IMPORTED_TARGET libpcre2-8)
 
 include(CheckOpenSSLIsBoringSSL)

README.md

@@ -47,7 +47,7 @@ $ ccmake build
 To specify the location of a dependency (like `--with-*` for autotools builds), you generally set a variable with the `ROOT`. The big exception to this is for openssl. This variable is called `OPENSSL_ROOT_DIR`
 
 ```
-$ cmake -B build -Djemalloc_ROOT=/opt/jemalloc -DPCRE_ROOT=/opt/edge -DOPENSSL_ROOT_DIR=/opt/boringssl
+$ cmake -B build -Djemalloc_ROOT=/opt/jemalloc -DOPENSSL_ROOT_DIR=/opt/boringssl
 ```
 
 #### Using presets to configure the build

ci/coverity-model.cpp

@@ -16,21 +16,15 @@
     limitations under the License.
 */
 
-// This is for PCRE, where the offsets vector is considered uninitialized, but it's an
+// This is for PCRE2, where the offsets vector is considered uninitialized, but it's an
 //  output vector only (input doesn't matter).
 extern "C" {
-#define PCRE_SPTR const char *
-
-struct real_pcre;   /* declaration; the definition is private  */
-typedef struct real_pcre pcre;
-
-typedef struct pcre_extra {
-} pcre_extra;
+#define PCRE2_CODE_UNIT_WIDTH 8
+#include <pcre2.h>
 
 int
-pcre_exec(const pcre *argument_re, const pcre_extra *extra_data,
-          PCRE_SPTR subject, int length, int start_offset, int options, int *offsets,
-          int offsetcount)
+pcre2_match(const pcre2_code *code, PCRE2_SPTR subject, PCRE2_SIZE length, PCRE2_SIZE startoffset, uint32_t options,
+            pcre2_match_data *match_data, pcre2_match_context *mcontext)
 {
   __coverity_panic__();
 }

cmake/FindPCRE.cmake

@@ -97,9 +97,6 @@
 #include "tsutil/Regex.h"
 #include "proxy/hdrs/URL.h"
 
-#define PCRE2_CODE_UNIT_WIDTH 8
-#include "pcre2.h"
-
 #include <swoc/swoc_ip.h>
 
 #ifdef HAVE_CTYPE_H

include/proxy/ControlMatcher.h

@@ -16,8 +16,6 @@
 #######################
 
 add_atsplugin(cachekey cachekey.cc common.cc configs.cc pattern.cc plugin.cc)
-
-target_link_libraries(cachekey PRIVATE PCRE::PCRE)
 verify_global_plugin(cachekey)
 verify_remap_plugin(cachekey)
 
@@ -26,6 +24,6 @@ if(BUILD_TESTING)
   add_executable(pattern_test unit_tests/pattern_test.cc pattern.cc common.cc)
   target_include_directories(pattern_test PRIVATE ${CMAKE_CURRENT_SOURCE_DIR})
   target_compile_definitions(pattern_test PRIVATE CACHEKEY_UNIT_TEST)
-  target_link_libraries(pattern_test PRIVATE Catch2::Catch2WithMain PCRE::PCRE libswoc::libswoc ts::tsutil)
+  target_link_libraries(pattern_test PRIVATE Catch2::Catch2WithMain libswoc::libswoc ts::tsutil)
   add_catch2_test(NAME pattern_test COMMAND pattern_test)
 endif()

plugins/cachekey/CMakeLists.txt

@@ -676,8 +676,8 @@ CacheKey::appendQuery(const ConfigQuery &config)
 /**
  * @brief Append User-Agent header captures specified in the Pattern configuration object.
  *
- * Apply given PCRE pattern/replacement to the first User-Agent value, and append any captured portions to cache key.
- * @param config PCRE pattern which contains capture groups.
+ * Apply given PCRE2 pattern/replacement to the first User-Agent value, and append any captured portions to cache key.
+ * @param config PCRE2 pattern which contains capture groups.
  * @todo: TBD if ignoring the comma in the header as a field separator is generic enough.
  * @note Add the UA captures to hier-part (RFC 3986) in the original order.
  */

plugins/cachekey/cachekey.cc

@@ -134,7 +134,7 @@ Pattern::empty() const
  * @brief Capture or capture-and-replace depending on whether a replacement string is specified.
  * @see replace()
  * @see capture()
- * @param subject PCRE subject string
+ * @param subject PCRE2 subject string
  * @param result vector of strings where the result of captures or the replacements will be returned.
  * @return true if there was a match and capture or replacement succeeded, false if failure.
  */

plugins/cachekey/pattern.cc

@@ -83,7 +83,7 @@ TEST_CASE("Pattern compile and match behavior", "[cachekey][pattern]")
   {
     Pattern p;
     REQUIRE(p.init("^$"));
-    // Pattern::match uses PCRE_NOTEMPTY which prevents empty-string matches.
+    // Pattern::match uses PCRE2_NOTEMPTY which prevents empty-string matches.
     // Therefore '^$' will NOT match an empty subject with the current implementation.
     CHECK(p.match("") == false);
     CHECK(p.match("not-empty") == false);
@@ -92,7 +92,7 @@ TEST_CASE("Pattern compile and match behavior", "[cachekey][pattern]")
   SECTION("Case-insensitive inline flag")
   {
     Pattern p;
-    // PCRE inline flag for case-insensitive
+    // PCRE2 inline flag for case-insensitive
     REQUIRE(p.init("(?i)AbC"));
     CHECK(p.match("aBc") == true);
     CHECK(p.match("ABC") == true);

plugins/cachekey/unit_tests/pattern_test.cc

@@ -45,9 +45,9 @@ Pattern::Pattern() : _pattern(""), _replacement("") {}
 Pattern::~Pattern() = default;
 
 /**
- * @brief Initializes PCRE pattern by providing the subject and replacement strings.
- * @param pattern PCRE pattern, a string containing PCRE patterns, capturing groups.
- * @param replacement PCRE replacement, a string where $0 ... $9 will be replaced with the corresponding capturing groups
+ * @brief Initializes PCRE2 pattern by providing the subject and replacement strings.
+ * @param pattern PCRE2 pattern, a string containing PCRE2 patterns, capturing groups.
+ * @param replacement PCRE2 replacement, a string where $0 ... $9 will be replaced with the corresponding capturing groups
  * @return true if successful, false if failure
  */
 bool
@@ -69,9 +69,9 @@ Pattern::init(const String &pattern, const String &replacement, bool replace)
 }
 
 /**
- * @brief Initializes PCRE pattern by providing the pattern only or pattern+replacement in a single configuration string.
+ * @brief Initializes PCRE2 pattern by providing the pattern only or pattern+replacement in a single configuration string.
  * @see init()
- * @param config PCRE pattern <pattern> or PCRE pattern + replacement in format /<pattern>/<replacement>/
+ * @param config PCRE2 pattern <pattern> or PCRE2 pattern + replacement in format /<pattern>/<replacement>/
  * @return true if successful, false if failure
  */
 bool
@@ -145,14 +145,14 @@ Pattern::empty() const
 }
 
 /**
- * @brief Destructor, frees PCRE related resources.
+ * @brief Destructor, frees PCRE2 related resources.
  */
 
 /**
  * @brief Capture or capture-and-replace depending on whether a replacement string is specified.
  * @see replace()
  * @see capture()
- * @param subject PCRE subject string
+ * @param subject PCRE2 subject string
  * @param result vector of strings where the result of captures or the replacements will be returned.
  * @return true if there was a match and capture or replacement succeeded, false if failure.
  */
@@ -188,8 +188,8 @@ Pattern::process(const String &subject, StringVector &result)
 }
 
 /**
- * @brief PCRE matches a subject string against the regex pattern.
- * @param subject PCRE subject
+ * @brief PCRE2 matches a subject string against the regex pattern.
+ * @param subject PCRE2 subject
  * @return true - matched, false - did not.
  */
 bool
@@ -214,8 +214,8 @@ Pattern::match(const String &subject)
 }
 
 /**
- * @brief Return all PCRE capture groups that matched in the subject string
- * @param subject PCRE subject string
+ * @brief Return all PCRE2 capture groups that matched in the subject string
+ * @param subject PCRE2 subject string
  * @param result reference to vector of strings containing all capture groups
  */
 bool
@@ -249,8 +249,8 @@ Pattern::capture(const String &subject, StringVector &result)
 }
 
 /**
- * @brief Replaces all replacements found in the replacement string with what matched in the PCRE capturing groups.
- * @param subject PCRE subject string
+ * @brief Replaces all replacements found in the replacement string with what matched in the PCRE2 capturing groups.
+ * @param subject PCRE2 subject string
  * @param result reference to A string where the result of the replacement will be stored
  * @return true - success, false - nothing matched or failure.
  */
@@ -305,14 +305,14 @@ Pattern::replace(const String &subject, String &result)
 }
 
 /**
- * @brief PCRE compiles the regex, called only during initialization.
+ * @brief PCRE2 compiles the regex, called only during initialization.
  * @return true if successful, false if not.
  */
 bool
 Pattern::compile()
 {
-  std::string error;     /* PCRE error description */
-  int         errOffset; /* PCRE error offset */
+  std::string error;     /* PCRE2 error description */
+  int         errOffset; /* PCRE2 error offset */
 
   AccessControlDebug("compiling pattern:'%s', replace: %s, replacement:'%s'", _pattern.c_str(), _replace ? "true" : "false",
                      _replacement.c_str());