Add PCRE2 library #3

Move the oscap_get_substring into the oscap_pcre.c module
and rename it into oscap_pcre_get_substring.

The function imposes implicit dependencies on PCRE/PCRE2 symbols
even for utils.c users that won't use PCRE at all (SCE library).

Author: evgenyz

CMakeLists.txt

@@ -530,6 +530,7 @@ endif()
 
 if (${CMAKE_C_COMPILER_ID} STREQUAL "GNU" OR ${CMAKE_C_COMPILER_ID} STREQUAL "Clang")
 	set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -pipe -W -Wall -Wnonnull -Wshadow -Wformat -Wundef -Wno-unused-parameter -Wmissing-prototypes -Wno-unknown-pragmas -Wno-int-conversion -Werror=implicit-function-declaration -D_GNU_SOURCE -std=c99")
+	add_link_options(-Wl,-z,now)
 endif()
 if(${CMAKE_SYSTEM_NAME} STREQUAL "FreeBSD")
 	add_link_options(-lkvm -lm -lprocstat)

src/OVAL/probes/independent/textfilecontent54_probe.c

@@ -216,7 +216,7 @@ static int process_file(const char *prefix, const char *path, const char *file,
 			want_instance = 0;
 
 		SEXP_free(next_inst);
-		substr_cnt = oscap_get_substrings(buf, &ofs, pfd->compiled_regex, want_instance, &substrs);
+		substr_cnt = oscap_pcre_get_substrings(buf, &ofs, pfd->compiled_regex, want_instance, &substrs);
 
 		if (substr_cnt < 0) {
 			SEXP_t *msg;

src/OVAL/probes/independent/textfilecontent_probe.c

@@ -196,7 +196,7 @@ static int process_file(const char *prefix, const char *path, const char *filena
 	int ofs = 0;
 
 	while (fgets(line, sizeof(line), fp) != NULL) {
-		substr_cnt = oscap_get_substrings(line, &ofs, re, 1, &substrs);
+		substr_cnt = oscap_pcre_get_substrings(line, &ofs, re, 1, &substrs);
 		if (substr_cnt > 0) {
 			int k;
 			SEXP_t *item;

src/common/oscap_pcre.c

@@ -26,6 +26,8 @@
 
 #include <memory.h>
 
+#define OSCAP_PCRE_EXEC_RECURSION_LIMIT_DEFAULT 3500
+
 #ifdef HAVE_PCRE2
 #define PCRE2_CODE_UNIT_WIDTH 8
 #define PCRE2_ERR_BUF_SIZE 127
@@ -243,6 +245,75 @@ void oscap_pcre_free(oscap_pcre_t *opcre)
 	}
 }
 
+int oscap_pcre_get_substrings(char *str, int *ofs, oscap_pcre_t *re, int want_substrs, char ***substrings) {
+	int i, ret, rc;
+	int ovector[60], ovector_len = sizeof (ovector) / sizeof (ovector[0]);
+	char **substrs;
+
+	// todo: max match count check
+
+	for (i = 0; i < ovector_len; ++i) {
+		ovector[i] = -1;
+	}
+
+	unsigned long limit = OSCAP_PCRE_EXEC_RECURSION_LIMIT_DEFAULT;
+	char *limit_str = getenv("OSCAP_PCRE_EXEC_RECURSION_LIMIT");
+	if (limit_str != NULL)
+		if (sscanf(limit_str, "%lu", &limit) <= 0)
+			dW("Unable to parse OSCAP_PCRE_EXEC_RECURSION_LIMIT value");
+	oscap_pcre_set_match_limit_recursion(re, limit);
+	size_t str_len = strlen(str);
+#if defined(OS_SOLARIS)
+	rc = oscap_pcre_exec(re, str, str_len, *ofs, OSCAP_PCRE_OPTS_NO_UTF8_CHECK, ovector, ovector_len);
+#else
+	rc = oscap_pcre_exec(re, str, str_len, *ofs, 0, ovector, ovector_len);
+#endif
+
+	if (rc < OSCAP_PCRE_ERR_NOMATCH) {
+		if (str_len < 100)
+			dE("Function oscap_pcre_exec() failed to match a regular expression with return code %d on string '%s'.", rc, str);
+		else
+			dE("Function oscap_pcre_exec() failed to match a regular expression with return code %d on string '%.100s' (truncated, showing first 100 characters).", rc, str);
+		return rc;
+	} else if (rc == OSCAP_PCRE_ERR_NOMATCH) {
+		/* no match */
+		return 0;
+	}
+
+	*ofs = (*ofs == ovector[1]) ? ovector[1] + 1 : ovector[1];
+
+	if (!want_substrs) {
+		/* just report successful match */
+		return 1;
+	}
+
+	ret = 0;
+	if (rc == 0) {
+		/* vector too small */
+		// todo: report partial results
+		rc = ovector_len / 3;
+	}
+
+	substrs = malloc(rc * sizeof (char *));
+	for (i = 0; i < rc; ++i) {
+		int len;
+		char *buf;
+
+		if (ovector[2 * i] == -1) {
+			continue;
+		}
+		len = ovector[2 * i + 1] - ovector[2 * i];
+		buf = malloc(len + 1);
+		memcpy(buf, str + ovector[2 * i], len);
+		buf[len] = '\0';
+		substrs[ret] = buf;
+		++ret;
+	}
+
+	*substrings = substrs;
+
+	return ret;
+}
 
 void oscap_pcre_err_free(char *err)
 {

src/common/oscap_pcre.h

@@ -96,6 +96,18 @@ void oscap_pcre_set_match_limit_recursion(oscap_pcre_t *opcre, unsigned long lim
  */
 void oscap_pcre_optimize(oscap_pcre_t *opcre);
 
+/**
+ * Match a regular expression and return substrings.
+ * Caller is responsible for freeing the returned array.
+ * @param str subject string
+ * @param ofs starting offset in str
+ * @param re compiled regular expression
+ * @param want_substrs if non-zero, substrings will be returned
+ * @param substrings contains returned substrings
+ * @return count of matched substrings, 0 if no match
+ * negative value on failure
+ */
+int oscap_pcre_get_substrings(char *str, int *ofs, oscap_pcre_t *re, int want_substrs, char ***substrings);
 
 /**
  * Free the error message returned by oscap_pcre_compile. DON'T USE REGULAR free()!

src/common/util.c

@@ -49,7 +49,6 @@
 #endif
 
 #define PATH_SEPARATOR '/'
-#define OSCAP_PCRE_EXEC_RECURSION_LIMIT_DEFAULT 3500
 
 int oscap_string_to_enum(const struct oscap_string_map *map, const char *str)
 {
@@ -361,76 +360,6 @@ char *oscap_path_join(const char *path1, const char *path2)
 	return joined_path;
 }
 
-int oscap_get_substrings(char *str, int *ofs, oscap_pcre_t *re, int want_substrs, char ***substrings) {
-	int i, ret, rc;
-	int ovector[60], ovector_len = sizeof (ovector) / sizeof (ovector[0]);
-	char **substrs;
-
-	// todo: max match count check
-
-	for (i = 0; i < ovector_len; ++i) {
-		ovector[i] = -1;
-	}
-
-	unsigned long limit = OSCAP_PCRE_EXEC_RECURSION_LIMIT_DEFAULT;
-	char *limit_str = getenv("OSCAP_PCRE_EXEC_RECURSION_LIMIT");
-	if (limit_str != NULL)
-		if (sscanf(limit_str, "%lu", &limit) <= 0)
-			dW("Unable to parse OSCAP_PCRE_EXEC_RECURSION_LIMIT value");
-	oscap_pcre_set_match_limit_recursion(re, limit);
-	size_t str_len = strlen(str);
-#if defined(OS_SOLARIS)
-	rc = oscap_pcre_exec(re, str, str_len, *ofs, OSCAP_PCRE_OPTS_NO_UTF8_CHECK, ovector, ovector_len);
-#else
-	rc = oscap_pcre_exec(re, str, str_len, *ofs, 0, ovector, ovector_len);
-#endif
-
-	if (rc < OSCAP_PCRE_ERR_NOMATCH) {
-		if (str_len < 100)
-			dE("Function oscap_pcre_exec() failed to match a regular expression with return code %d on string '%s'.", rc, str);
-		else
-			dE("Function oscap_pcre_exec() failed to match a regular expression with return code %d on string '%.100s' (truncated, showing first 100 characters).", rc, str);
-		return rc;
-	} else if (rc == OSCAP_PCRE_ERR_NOMATCH) {
-		/* no match */
-		return 0;
-	}
-
-	*ofs = (*ofs == ovector[1]) ? ovector[1] + 1 : ovector[1];
-
-	if (!want_substrs) {
-		/* just report successful match */
-		return 1;
-	}
-
-	ret = 0;
-	if (rc == 0) {
-		/* vector too small */
-		// todo: report partial results
-		rc = ovector_len / 3;
-	}
-
-	substrs = malloc(rc * sizeof (char *));
-	for (i = 0; i < rc; ++i) {
-		int len;
-		char *buf;
-
-		if (ovector[2 * i] == -1) {
-			continue;
-		}
-		len = ovector[2 * i + 1] - ovector[2 * i];
-		buf = malloc(len + 1);
-		memcpy(buf, str + ovector[2 * i], len);
-		buf[len] = '\0';
-		substrs[ret] = buf;
-		++ret;
-	}
-
-	*substrings = substrs;
-
-	return ret;
-}
-
 #ifndef OS_WINDOWS
 FILE *oscap_fopen_with_prefix(const char *prefix, const char *path)
 {

src/common/util.h

@@ -487,20 +487,6 @@ int oscap_strncasecmp(const char *s1, const char *s2, size_t n);
  */
 char *oscap_strerror_r(int errnum, char *buf, size_t buflen);
 
-/**
- * Match a regular expression and return substrings.
- * Caller is responsible for freeing the returned array.
- * @param str subject string
- * @param ofs starting offset in str
- * @param re compiled regular expression
- * @param want_substrs if non-zero, substrings will be returned
- * @param substrings contains returned substrings
- * @return count of matched substrings, 0 if no match
- * negative value on failure
- */
-int oscap_get_substrings(char *str, int *ofs, oscap_pcre_t *re, int want_substrs, char ***substrings);
-
-
 #ifndef OS_WINDOWS
 /**
  * Open file for reading with prefix added to its name.