Compatibility layer for pcre2

This should close #129

15 files changed, 125 insertions, 83 deletions

diff --git a/src/config.m4 b/src/config.m4
index 3d4c1ba..8d5278e 100644
--- a/src/config.m4
+++ b/src/config.m4
@@ -6,6 +6,7 @@ sources="$sources sp_unserialize.c sp_utils.c sp_disable_xxe.c sp_list.c"
 sources="$sources sp_disabled_functions.c sp_execute.c sp_upload_validation.c"
 sources="$sources sp_cookie_encryption.c sp_network_utils.c tweetnacl.c"
 sources="$sources sp_config_keywords.c sp_var_parser.c sp_var_value.c sp_tree.c"
+sources="$sources sp_pcre_compat.c"
 
 PHP_ARG_ENABLE(snuffleupagus, whether to enable snuffleupagus support,
 [  --enable-snuffleupagus           Enable snuffleupagus support])
diff --git a/src/php_snuffleupagus.h b/src/php_snuffleupagus.h
index 95f271d..52bcc93 100644
--- a/src/php_snuffleupagus.h
+++ b/src/php_snuffleupagus.h
@@ -24,6 +24,7 @@
 #include "zend_string.h"
 #include "zend_extensions.h"
 
+#include "sp_pcre_compat.h"
 #include "sp_list.h"
 #include "sp_tree.h"
 #include "sp_var_parser.h"
@@ -71,18 +72,6 @@ ZEND_END_MODULE_GLOBALS(snuffleupagus)
 ZEND_TSRMLS_CACHE_EXTERN()
 #endif
 
-#if HAVE_BUNDLED_PCRE
- #include "ext/pcre/pcrelib/pcre.h"
- #undef pcre_exec
- #undef pcre_compile
- #define sp_pcre_exec pcre_exec
- #define sp_pcre_compile pcre_compile
-#else
- #include "pcre.h"
- #define sp_pcre_exec pcre_exec
- #define sp_pcre_compile pcre_compile
-#endif
-
 PHP_FUNCTION(check_disabled_function);
 PHP_FUNCTION(eval_blacklist_callback);
 
diff --git a/src/sp_config.c b/src/sp_config.c
index 7a38a45..1236ebd 100644
--- a/src/sp_config.c
+++ b/src/sp_config.c
@@ -140,15 +140,9 @@ int parse_regexp(char *restrict line, char *restrict keyword, void *retval) {
   char *value = get_param(&consumed, line, SP_TYPE_STR, keyword);
 
   if (value) {
-    const char *pcre_error;
-    int pcre_error_offset;
-    pcre *compiled_re = sp_pcre_compile(value, PCRE_CASELESS, &pcre_error,
-                                        &pcre_error_offset, NULL);
-    if (NULL == compiled_re) {
-      sp_log_err("config", "Failed to compile '%s': %s on line %zu.", value,
-                 pcre_error, sp_line_no);
-    } else {
-      *(pcre **)retval = compiled_re;
+    sp_pcre *compiled_re = sp_pcre_compile(value);
+                if (NULL != compiled_re) {
+      *(sp_pcre **)retval = compiled_re;
       return consumed;
     }
   }
diff --git a/src/sp_config.h b/src/sp_config.h
index 25963f1..75ee83d 100644
--- a/src/sp_config.h
+++ b/src/sp_config.h
@@ -59,7 +59,7 @@ typedef struct {
   enum samesite_type { strict = 1, lax = 2 } samesite;
   bool encrypt;
   char *name;
-  pcre *name_r;
+  sp_pcre *name_r;
   bool simulation;
 } sp_cookie;
 
@@ -72,29 +72,29 @@ typedef struct {
   char *textual_representation;
 
   char *filename;
-  pcre *r_filename;
+  sp_pcre *r_filename;
 
   char *function;
-  pcre *r_function;
+  sp_pcre *r_function;
   sp_list_node *functions_list;
 
   char *hash;
   int simulation;
 
   sp_tree *param;
-  pcre *r_param;
+  sp_pcre *r_param;
   sp_php_type param_type;
   int pos;
   unsigned int line;
 
   char *ret;
-  pcre *r_ret;
+  sp_pcre *r_ret;
   sp_php_type ret_type;
 
-  pcre *value_r;
+  sp_pcre *value_r;
   char *value;
 
-  pcre *r_key;
+  sp_pcre *r_key;
   char *key;
 
   char *dump;
diff --git a/src/sp_config_keywords.c b/src/sp_config_keywords.c
index 93c3da9..5df3d97 100644
--- a/src/sp_config_keywords.c
+++ b/src/sp_config_keywords.c
@@ -25,7 +25,7 @@ static int get_construct_type(sp_disabled_function const *const df) {
         }
       } else {
         if (true ==
-            is_regexp_matching(df->r_function, CONSTRUCTS_TYPES[i].keys[j])) {
+            sp_is_regexp_matching(df->r_function, CONSTRUCTS_TYPES[i].keys[j])) {
           return CONSTRUCTS_TYPES[i].type;
         }
       }
diff --git a/src/sp_cookie_encryption.c b/src/sp_cookie_encryption.c
index 29e96b1..09cf884 100644
--- a/src/sp_cookie_encryption.c
+++ b/src/sp_cookie_encryption.c
@@ -167,7 +167,7 @@ static zend_string *encrypt_data(char *data, unsigned long long data_len) {
 }
 
 PHP_FUNCTION(sp_setcookie) {
-  zval params[7] = {0};
+  zval params[7] = {{{0}}};
   zend_string *name = NULL, *value = NULL, *path = NULL, *domain = NULL,
               *samesite = NULL;
   zend_long expires = 0;
diff --git a/src/sp_disabled_functions.c b/src/sp_disabled_functions.c
index 8396cae..f8c21d2 100644
--- a/src/sp_disabled_functions.c
+++ b/src/sp_disabled_functions.c
@@ -155,7 +155,7 @@ static bool is_param_matching(zend_execute_data* execute_data,
       }
       const bool pcre_matching =
           config_node->r_param &&
-          (true == is_regexp_matching(config_node->r_param, *arg_name));
+          (true == sp_is_regexp_matching(config_node->r_param, *arg_name));
 
       /* This is the parameter name we're looking for. */
       if (true == pcre_matching || config_node->pos != -1) {
@@ -259,7 +259,7 @@ bool should_disable(zend_execute_data* execute_data, const char* builtin_name,
       }
     } else if (config_node->r_function) {
       if (false ==
-          is_regexp_matching(config_node->r_function, complete_path_function)) {
+          sp_is_regexp_matching(config_node->r_function, complete_path_function)) {
         goto next;
       }
     }
@@ -276,7 +276,7 @@ bool should_disable(zend_execute_data* execute_data, const char* builtin_name,
       }
     } else if (config_node->r_filename) {
       if (false ==
-          is_regexp_matching(config_node->r_filename, current_filename)) {
+          sp_is_regexp_matching(config_node->r_filename, current_filename)) {
         goto next;
       }
     }
@@ -366,7 +366,7 @@ bool should_drop_on_ret(zval* return_value,
       }
     } else if (config_node->r_function) {
       if (false ==
-          is_regexp_matching(config_node->r_function, complete_path_function)) {
+          sp_is_regexp_matching(config_node->r_function, complete_path_function)) {
         goto next;
       }
     }
@@ -377,7 +377,7 @@ bool should_drop_on_ret(zval* return_value,
       }
     } else if (config_node->r_filename) {
       if (false ==
-          is_regexp_matching(config_node->r_filename, current_filename)) {
+          sp_is_regexp_matching(config_node->r_filename, current_filename)) {
         goto next;
       }
     }
@@ -440,7 +440,7 @@ ZEND_FUNCTION(check_disabled_function) {
 static int hook_functions(const sp_list_node* config) {
   while (config && config->data) {
     const char* function_name = ((sp_disabled_function*)config->data)->function;
-    const pcre* function_name_regexp =
+    const sp_pcre* function_name_regexp =
         ((sp_disabled_function*)config->data)->r_function;
 
     if (NULL != function_name) {  // hook function by name
diff --git a/src/sp_pcre_compat.c b/src/sp_pcre_compat.c
new file mode 100644
index 0000000..42a11cb
--- /dev/null
+++ b/src/sp_pcre_compat.c
@@ -0,0 +1,50 @@
+#include "php_snuffleupagus.h"
+
+#include "sp_pcre_compat.h"
+
+sp_pcre* sp_pcre_compile(const char *const pattern) {
+        sp_pcre* ret = NULL;
+        const char *pcre_error = NULL;
+#ifdef SP_HAS_PCRE2
+        int errornumber;
+        PCRE2_SIZE erroroffset;
+        ret = pcre2_compile((PCRE2_SPTR)pattern, PCRE2_ZERO_TERMINATED, PCRE2_CASELESS, &errornumber, &erroroffset, NULL);
+#else
+        int erroroffset;
+        ret = pcre_compile(pattern, PCRE_CASELESS, &pcre_error, &erroroffset, NULL);
+#endif
+        
+        if (NULL == ret) {
+                sp_log_err("config", "Failed to compile '%s': %s on line %zu.", pattern,
+                                pcre_error, sp_line_no);
+        }
+        return ret;
+}
+
+bool sp_is_regexp_matching_len(const sp_pcre* regexp, const char* str, size_t len) {
+  int ret = 0;
+
+  assert(NULL != regexp);
+  assert(NULL != str);
+
+#ifdef SP_HAS_PCRE2
+        pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(regexp, NULL);
+        ret = pcre2_match(regexp, (PCRE2_SPTR)str, len, 0, 0, match_data, NULL);
+#else
+  int vec[30];
+  ret = pcre_exec(regexp, NULL, str, len, 0, 0, vec,
+                     sizeof(vec) / sizeof(int));
+#endif
+
+  if (ret < 0) {
+#ifdef SP_HAS_PCRE2
+    if (ret != PCRE2_ERROR_NOMATCH) {
+#else
+    if (ret != PCRE_ERROR_NOMATCH) {
+#endif
+      sp_log_err("regexp", "Something went wrong with a regexp (%d).", ret);
+    }
+    return false;
+  }
+  return true;
+}
diff --git a/src/sp_pcre_compat.h b/src/sp_pcre_compat.h
new file mode 100644
index 0000000..a9eb253
--- /dev/null
+++ b/src/sp_pcre_compat.h
@@ -0,0 +1,34 @@
+#ifndef SP_PCRE_COMPAT_H
+#define SP_PCRE_COMPAT_H
+
+#include <stdlib.h>
+#include <stdbool.h>
+
+#undef pcre_exec
+#undef pcre_compile
+
+/* We're not supporting pcre2 when it's not bundled with php7,
+ * yet. Pull-requests are welcome. */
+#if HAVE_BUNDLED_PCRE
+  #if PHP_VERSION_ID >= 70300
+    #define SP_HAS_PCRE2
+    #include "ext/pcre/pcre2lib/pcre2.h"
+  #else
+    #include "ext/pcre/pcrelib/pcre.h"
+  #endif
+#else
+ #include "pcre.h"
+#endif
+
+#ifdef SP_HAS_PCRE2
+  #define sp_pcre pcre2_code 
+#else
+  #define sp_pcre pcre
+#endif
+
+sp_pcre* sp_pcre_compile(const char* str);
+#define sp_is_regexp_matching(regexp, str) \
+        sp_is_regexp_matching_len(regexp, str, strlen(str))
+bool sp_is_regexp_matching_len(const sp_pcre* regexp, const char* str, size_t len);
+
+#endif // SP_PCRE_COMPAT_H
diff --git a/src/sp_utils.c b/src/sp_utils.c
index 0625a2f..2979d98 100644
--- a/src/sp_utils.c
+++ b/src/sp_utils.c
@@ -32,25 +32,6 @@ void sp_log_msg(char const* feature, char const* level, const char* fmt, ...) {
               client_ip ? client_ip : "0.0.0.0", feature, level, msg);
 }
 
-zend_always_inline int is_regexp_matching(const pcre* regexp, const char* str) {
-  int vec[30];
-  int ret = 0;
-
-  assert(NULL != regexp);
-  assert(NULL != str);
-
-  ret = sp_pcre_exec(regexp, NULL, str, strlen(str), 0, 0, vec,
-                     sizeof(vec) / sizeof(int));
-
-  if (ret < 0) {
-    if (ret != PCRE_ERROR_NOMATCH) {
-      sp_log_err("regexp", "Something went wrong with a regexp (%d).", ret);
-    }
-    return false;
-  }
-  return true;
-}
-
 int compute_hash(const char* const filename, char* file_hash) {
   unsigned char buf[1024];
   unsigned char digest[SHA256_SIZE];
@@ -192,13 +173,13 @@ char* sp_convert_to_string(zval* zv) {
   return estrdup("");
 }
 
-bool sp_match_value(const char* value, const char* to_match, const pcre* rx) {
+bool sp_match_value(const char* value, const char* to_match, const sp_pcre* rx) {
   if (to_match) {
     if (0 == strcmp(to_match, value)) {
       return true;
     }
   } else if (rx) {
-    return is_regexp_matching(rx, value);
+    return sp_is_regexp_matching(rx, value);
   } else {
     return true;
   }
@@ -274,7 +255,7 @@ void sp_log_disable_ret(const char* restrict path,
   }
 }
 
-bool sp_match_array_key(const zval* zv, const char* to_match, const pcre* rx) {
+bool sp_match_array_key(const zval* zv, const char* to_match, const sp_pcre* rx) {
   zend_string* key;
   zend_ulong idx;
 
@@ -298,7 +279,7 @@ bool sp_match_array_key(const zval* zv, const char* to_match, const pcre* rx) {
 }
 
 bool sp_match_array_value(const zval* arr, const char* to_match,
-                          const pcre* rx) {
+                          const sp_pcre* rx) {
   zval* value;
 
   ZEND_HASH_FOREACH_VAL(Z_ARRVAL_P(arr), value) {
@@ -368,7 +349,7 @@ int hook_function(const char* original_name, HashTable* hook_table,
   return SUCCESS;
 }
 
-int hook_regexp(const pcre* regexp, HashTable* hook_table,
+int hook_regexp(const sp_pcre* regexp, HashTable* hook_table,
                 void (*new_function)(INTERNAL_FUNCTION_PARAMETERS),
                 bool hook_execution_table) {
   zend_string* key;
@@ -377,9 +358,7 @@ int hook_regexp(const pcre* regexp, HashTable* hook_table,
 
   ZEND_HASH_FOREACH_STR_KEY(ht, key) {
     if (key) {
-      int vec[30];
-      int ret = sp_pcre_exec(regexp, NULL, key->val, key->len, 0, 0, vec,
-                             sizeof(vec) / sizeof(int));
+      int ret = sp_is_regexp_matching_len(regexp, key->val, key->len);
       if (ret < 0) { /* Error or no match*/
         if (PCRE_ERROR_NOMATCH != ret) {
           sp_log_err("pcre", "Runtime error with pcre, error code: %d", ret);
diff --git a/src/sp_utils.h b/src/sp_utils.h
index ada7cf6..10a6daa 100644
--- a/src/sp_utils.h
+++ b/src/sp_utils.h
@@ -46,17 +46,16 @@
 void sp_log_msg(char const *feature, char const *level, const char *fmt, ...);
 int compute_hash(const char *const filename, char *file_hash);
 char *sp_convert_to_string(zval *);
-bool sp_match_value(const char *, const char *, const pcre *);
-bool sp_match_array_key(const zval *, const char *, const pcre *);
-bool sp_match_array_value(const zval *, const char *, const pcre *);
+bool sp_match_value(const char *, const char *, const sp_pcre *);
+bool sp_match_array_key(const zval *, const char *, const sp_pcre *);
+bool sp_match_array_value(const zval *, const char *, const sp_pcre *);
 void sp_log_disable(const char *restrict, const char *restrict,
                     const char *restrict, const sp_disabled_function *);
 void sp_log_disable_ret(const char *restrict, const char *restrict,
                         const sp_disabled_function *);
-int is_regexp_matching(const pcre *, const char *);
 int hook_function(const char *, HashTable *,
                   void (*)(INTERNAL_FUNCTION_PARAMETERS), bool);
-int hook_regexp(const pcre *, HashTable *,
+int hook_regexp(const sp_pcre *, HashTable *,
                 void (*)(INTERNAL_FUNCTION_PARAMETERS), bool);
 bool check_is_in_eval_whitelist(const char * const function_name);
 
diff --git a/src/sp_var_parser.c b/src/sp_var_parser.c
index d0ae67c..330fa54 100644
--- a/src/sp_var_parser.c
+++ b/src/sp_var_parser.c
@@ -20,26 +20,22 @@ static sp_list_node *parse_str_tokens(const char *str, const sp_conf_token token
 }
 
 static bool is_var_name_valid(const char *name) {
-  static pcre *regexp_const = NULL;
-  static pcre *regexp_var = NULL;
-  const char *pcre_error;
-  int pcre_error_offset;
+  static sp_pcre *regexp_const = NULL;
+  static sp_pcre *regexp_var = NULL;
 
   if (!name) {
     return false;
   }
   if (NULL == regexp_var || NULL == regexp_const) {
-    regexp_var = sp_pcre_compile(REGEXP_VAR, PCRE_CASELESS, &pcre_error,
-                                 &pcre_error_offset, NULL);
-    regexp_const = sp_pcre_compile(REGEXP_CONST, PCRE_CASELESS, &pcre_error,
-                                   &pcre_error_offset, NULL);
+    regexp_var = sp_pcre_compile(REGEXP_VAR);
+    regexp_const = sp_pcre_compile(REGEXP_CONST);
   }
   if (NULL == regexp_var || NULL == regexp_const) {
     sp_log_err("config", "Could not compile regexp.");
     return false;
   }
-  if (0 > sp_pcre_exec(regexp_var, NULL, name, strlen(name), 0, 0, NULL, 0) &&
-      0 > sp_pcre_exec(regexp_const, NULL, name, strlen(name), 0, 0, NULL, 0)) {
+  if ((false == sp_is_regexp_matching(regexp_var, name)) &&
+      (false == sp_is_regexp_matching(regexp_const, name))) {
     return false;
   }
   return true;
diff --git a/src/tests/broken_conf_config_regexp.phpt b/src/tests/broken_conf_config_regexp.phpt
index d2a379e..e49788c 100644
--- a/src/tests/broken_conf_config_regexp.phpt
+++ b/src/tests/broken_conf_config_regexp.phpt
@@ -5,6 +5,6 @@ Broken configuration
 --INI--
 sp.configuration_file={PWD}/config/broken_config_regexp.ini
 --FILE--
---EXPECT--
-[snuffleupagus][0.0.0.0][config][error] Failed to compile '*.': nothing to repeat on line 1.
+--EXPECTF--
+[snuffleupagus][0.0.0.0][config][error] Failed to compile '*.': %aon line 1.
 [snuffleupagus][0.0.0.0][config][error] '.filename_r()' is expecting a valid regexp, and not '"*."' on line 1.
diff --git a/src/tests/broken_conf_shown_in_phpinfo.phpt b/src/tests/broken_conf_shown_in_phpinfo.phpt
index 61e9512..5053ee0 100644
--- a/src/tests/broken_conf_shown_in_phpinfo.phpt
+++ b/src/tests/broken_conf_shown_in_phpinfo.phpt
@@ -16,7 +16,7 @@ if (strstr($info, 'Valid config => no') !== FALSE) {
         echo "lose";
 }
 ?>
---EXPECT--
-[snuffleupagus][0.0.0.0][config][error] Failed to compile '*.': nothing to repeat on line 1.
+--EXPECTF--
+[snuffleupagus][0.0.0.0][config][error] Failed to compile '*.': %aon line 1.
 [snuffleupagus][0.0.0.0][config][error] '.filename_r()' is expecting a valid regexp, and not '"*."' on line 1.
 win
diff --git a/src/tests/broken_regexp.phpt b/src/tests/broken_regexp.phpt
index 3f027f1..7c7ac9c 100644
--- a/src/tests/broken_regexp.phpt
+++ b/src/tests/broken_regexp.phpt
@@ -6,5 +6,5 @@ Broken regexp
 sp.configuration_file={PWD}/config/broken_regexp.ini
 --FILE--
 --EXPECTF--
-[snuffleupagus][0.0.0.0][config][error] Failed to compile '^$[': missing terminating ] for character class on line 1.
+[snuffleupagus][0.0.0.0][config][error] Failed to compile '^$[': %aon line 1.
 [snuffleupagus][0.0.0.0][config][error] '.value_r()' is expecting a valid regexp, and not '"^$["' on line 1.