update regex module for pcre2

(cherry picked from commit 3cc28d1)

Author: Steve Ayre

modules/regex/Makefile

@@ -21,10 +21,10 @@ ifeq ($(PCRE_BUILDER),)
 	DEFS += -I$(SYSBASE)/include \
 			-I$(LOCALBASE)/include
 	LIBS += -L$(SYSBASE)/lib \
-			-L$(LOCALBASE)/lib -lpcre
+			-L$(LOCALBASE)/lib -lpcre2-8
 else
 	DEFS += $(shell $(PCRE_BUILDER) --cflags)
-	LIBS += $(shell $(PCRE_BUILDER) --libs)
+	LIBS += $(shell $(PCRE_BUILDER) --libs8)
 endif
 
 include ../../Makefile.modules

modules/regex/regex_mod.c

@@ -60,6 +60,8 @@
 #define MAX_GROUPS 20            /*!< Max number of groups */
 #define GROUP_MAX_SIZE 8192      /*!< Max size of a group */
 
+#define ERROR_BUF_SIZE 100
+
 
 /*
  * Locking variables
@@ -82,8 +84,8 @@ static int pcre_extended         = 0;
 /*
  * Module internal parameter variables
  */
-static pcre **pcres;
-static pcre ***pcres_addr;
+static pcre2_code **pcres;
+static pcre2_code ***pcres_addr;
 static int *num_pcres;
 static int pcre_options = 0x00000000;
 
@@ -215,24 +217,24 @@ static int mod_init(void)
 		/* PCRE options */
 		if (pcre_caseless != 0) {
 			LM_DBG("PCRE CASELESS enabled\n");
-			pcre_options = pcre_options | PCRE_CASELESS;
+			pcre_options = pcre_options | PCRE2_CASELESS;
 		}
 		if (pcre_multiline != 0) {
 			LM_DBG("PCRE MULTILINE enabled\n");
-			pcre_options = pcre_options | PCRE_MULTILINE;
+			pcre_options = pcre_options | PCRE2_MULTILINE;
 		}
 		if (pcre_dotall != 0) {
 			LM_DBG("PCRE DOTALL enabled\n");
-			pcre_options = pcre_options | PCRE_DOTALL;
+			pcre_options = pcre_options | PCRE2_DOTALL;
 		}
 		if (pcre_extended != 0) {
 			LM_DBG("PCRE EXTENDED enabled\n");
-			pcre_options = pcre_options | PCRE_EXTENDED;
+			pcre_options = pcre_options | PCRE2_EXTENDED;
 		}
 		LM_DBG("PCRE options: %i\n", pcre_options);
 
 		/* Pointer to pcres */
-		if ((pcres_addr = shm_malloc(sizeof(pcre **))) == 0) {
+		if ((pcres_addr = shm_malloc(sizeof(pcre2_code **))) == 0) {
 			LM_ERR("no memory for pcres_addr\n");
 			goto err;
 		}
@@ -273,13 +275,14 @@ static int load_pcres(int action)
 	FILE *f;
 	char line[FILE_MAX_LINE];
 	char **patterns = NULL;
-	pcre *pcre_tmp = NULL;
-	int pcre_size;
+	pcre2_code *pcre_tmp = NULL;
+	size_t pcre_size;
 	int pcre_rc;
-	const char *pcre_error;
-	int pcre_erroffset;
+	int pcre_error;
+	PCRE2_UCHAR pcre_error_str[ERROR_BUF_SIZE];
+	PCRE2_SIZE pcre_erroffset;
 	int num_pcres_tmp = 0;
-	pcre **pcres_tmp = NULL;
+	pcre2_code **pcres_tmp = NULL;
 
 	/* Get the lock */
 	lock_get(reload_lock);
@@ -414,7 +417,7 @@ static int load_pcres(int action)
 	}
 
 	/* Temporal pointer of pcres */
-	if ((pcres_tmp = pkg_malloc(sizeof(pcre *) * num_pcres_tmp)) == 0) {
+	if ((pcres_tmp = pkg_malloc(sizeof(pcre2_code *) * num_pcres_tmp)) == 0) {
 		LM_ERR("no more memory for pcres_tmp\n");
 		goto err;
 	}
@@ -425,14 +428,15 @@ static int load_pcres(int action)
 	/* Compile the patters */
 	for (i=0; i<num_pcres_tmp; i++) {
 
-		pcre_tmp = pcre_compile(patterns[i], pcre_options, &pcre_error, &pcre_erroffset, NULL);
+		pcre_tmp = pcre2_compile((PCRE2_SPTR)patterns[i], PCRE2_ZERO_TERMINATED, pcre_options, &pcre_error, &pcre_erroffset, NULL);
 		if (pcre_tmp == NULL) {
-			LM_ERR("pcre_tmp compilation of '%s' failed at offset %d: %s\n", patterns[i], pcre_erroffset, pcre_error);
+                	pcre2_get_error_message(pcre_error, pcre_error_str, sizeof(pcre_error_str));
+			LM_ERR("pcre_tmp compilation of '%s' failed at offset %ld: %s\n", patterns[i], pcre_erroffset, pcre_error_str);
 			goto err;
 		}
-		pcre_rc = pcre_fullinfo(pcre_tmp, NULL, PCRE_INFO_SIZE, &pcre_size);
+		pcre_rc = pcre2_pattern_info(pcre_tmp, PCRE2_INFO_SIZE, &pcre_size);
 		if (pcre_rc) {
-			printf("pcre_fullinfo on compiled pattern[%i] yielded error: %d\n", i, pcre_rc);
+			printf("pcre2_pattern_info on compiled pattern[%i] yielded error: %d\n", i, pcre_rc);
 			goto err;
 		}
 
@@ -442,7 +446,7 @@ static int load_pcres(int action)
 		}
 
 		memcpy(pcres_tmp[i], pcre_tmp, pcre_size);
-		pcre_free(pcre_tmp);
+		pcre2_code_free(pcre_tmp);
 		pkg_free(patterns[i]);
 	}
 
@@ -455,15 +459,15 @@ static int load_pcres(int action)
 		}
 		shm_free(pcres);
 	}
-	if ((pcres = shm_malloc(sizeof(pcre *) * num_pcres_tmp)) == 0) {
+	if ((pcres = shm_malloc(sizeof(pcre2_code *) * num_pcres_tmp)) == 0) {
 		LM_ERR("no more memory for pcres\n");
 		goto err;
 	}
 	for (i=0; i<num_pcres_tmp; i++) {
 		pcres[i] = NULL;
 	}
 	for (i=0; i<num_pcres_tmp; i++) {
-		pcre_rc = pcre_fullinfo(pcres_tmp[i], NULL, PCRE_INFO_SIZE, &pcre_size);
+		pcre_rc = pcre2_pattern_info(pcres_tmp[i], PCRE2_INFO_SIZE, &pcre_size);
 		if ((pcres[i] = shm_malloc(pcre_size)) == 0) {
 			LM_ERR("no more memory for pcres[%i]\n", i);
 			goto err;
@@ -549,48 +553,54 @@ static void free_shared_memory(void)
 /*! \brief Return true if the argument matches the regular expression parameter */
 static int w_pcre_match(struct sip_msg* _msg, str* string, str* _regex_s)
 {
-	pcre *pcre_re = NULL;
+	pcre2_code *pcre_re = NULL;
 	int pcre_rc;
-	const char *pcre_error;
-	int pcre_erroffset;
+	int pcre_error;
+	PCRE2_UCHAR pcre_error_str[ERROR_BUF_SIZE];
+	PCRE2_SIZE pcre_erroffset;
+	pcre2_match_data *match_data;
 	str regex;
 
 	if (pkg_nt_str_dup(&regex, _regex_s) < 0)
 		return -1;
 
-	pcre_re = pcre_compile(regex.s, pcre_options, &pcre_error, &pcre_erroffset, NULL);
+	pcre_re = pcre2_compile((PCRE2_SPTR)regex.s, PCRE2_ZERO_TERMINATED, pcre_options, &pcre_error, &pcre_erroffset, NULL);
 	if (pcre_re == NULL) {
-		LM_ERR("pcre_re compilation of '%s' failed at offset %d: %s\n", regex.s, pcre_erroffset, pcre_error);
+                pcre2_get_error_message(pcre_error, pcre_error_str, sizeof(pcre_error_str));
+		LM_ERR("pcre_re compilation of '%s' failed at offset %ld: %s\n", regex.s, pcre_erroffset, pcre_error_str);
 		pkg_free(regex.s);
 		return -4;
 	}
 
-	pcre_rc = pcre_exec(
+	match_data = pcre2_match_data_create(0, NULL); // no captures needed
+
+	pcre_rc = pcre2_match(
 		pcre_re,                    /* the compiled pattern */
-		NULL,                       /* no extra data - we didn't study the pattern */
-		string->s,                   /* the matching string */
-		(int)(string->len),          /* the length of the subject */
+		(PCRE2_SPTR)string->s,                  /* the matching string */
+		(PCRE2_SIZE)(string->len),  /* the length of the subject */
 		0,                          /* start at offset 0 in the string */
 		0,                          /* default options */
-		NULL,                       /* output vector for substring information */
-		0);                         /* number of elements in the output vector */
+		match_data,                 /* match data block */
+		NULL);                      /* match context */
+
+	pcre2_match_data_free(match_data);
 
 	/* Matching failed: handle error cases */
 	if (pcre_rc < 0) {
 		switch(pcre_rc) {
-			case PCRE_ERROR_NOMATCH:
+			case PCRE2_ERROR_NOMATCH:
 				LM_DBG("'%s' doesn't match '%s'\n", string->s, regex.s);
 				break;
 			default:
 				LM_DBG("matching error '%d'\n", pcre_rc);
 				break;
 		}
-		pcre_free(pcre_re);
+		pcre2_code_free(pcre_re);
 		pkg_free(regex.s);
 		return -1;
 	}
 
-	pcre_free(pcre_re);
+	pcre2_code_free(pcre_re);
 	pkg_free(regex.s);
 	LM_DBG("'%s' matches '%s'\n", string->s, regex.s);
 	return 1;
@@ -602,6 +612,7 @@ static int w_pcre_match_group(struct sip_msg* _msg, str* string, int* _num_pcre)
 {
 	int num_pcre;
 	int pcre_rc;
+	pcre2_match_data *match_data;
 
 	/* Check if group matching feature is enabled */
 	if (file == NULL) {
@@ -621,22 +632,25 @@ static int w_pcre_match_group(struct sip_msg* _msg, str* string, int* _num_pcre)
 
 	lock_get(reload_lock);
 
-	pcre_rc = pcre_exec(
+	match_data = pcre2_match_data_create(0, NULL); // no captures needed
+
+	pcre_rc = pcre2_match(
 		(*pcres_addr)[num_pcre],    /* the compiled pattern */
-		NULL,                       /* no extra data - we didn't study the pattern */
-		string->s,                   /* the matching string */
-		(int)(string->len),          /* the length of the subject */
+		(PCRE2_SPTR)string->s,                  /* the matching string */
+		(PCRE2_SIZE)(string->len),  /* the length of the subject */
 		0,                          /* start at offset 0 in the string */
 		0,                          /* default options */
-		NULL,                       /* output vector for substring information */
-		0);                         /* number of elements in the output vector */
+		match_data,                 /* match data block */
+		0);                         /* match context */
+
+	pcre2_match_data_free(match_data);
 
 	lock_release(reload_lock);
 
 	/* Matching failed: handle error cases */
 	if (pcre_rc < 0) {
 		switch(pcre_rc) {
-			case PCRE_ERROR_NOMATCH:
+			case PCRE2_ERROR_NOMATCH:
 				LM_DBG("'%s' doesn't match pcres[%i]\n", string->s, num_pcre);
 				break;
 			default: