[Rework] Slightly improve old regexp API

author: Vsevolod Stakhov <vsevolod@highsecure.ru> 2021-07-15 14:23:17 +0100
committer: Vsevolod Stakhov <vsevolod@highsecure.ru> 2021-07-15 14:23:17 +0100
commit: 5f80e2e32e9e82c769b3f28be29cff4f62d9ecce (patch)
tree: a18428dc67bae8a113a293b6406cecf76097a336
parent: b2c36feea701c6685d83b8c4e6282fe7e307609d (diff)
download: rspamd-5f80e2e32e9e82c769b3f28be29cff4f62d9ecce.tar.gz
rspamd-5f80e2e32e9e82c769b3f28be29cff4f62d9ecce.zip
2 files changed, 49 insertions, 34 deletions
diff --git a/src/libutil/regexp.c b/src/libutil/regexp.c
index 93b8125e4..6ec5b4d39 100644
--- a/src/libutil/regexp.c
+++ b/src/libutil/regexp.c
@@ -306,10 +306,10 @@ rspamd_regexp_post_process (rspamd_regexp_t *r)
 }
 
 rspamd_regexp_t*
-rspamd_regexp_new (const gchar *pattern, const gchar *flags,
+rspamd_regexp_new_len (const gchar *pattern, gsize len, const gchar *flags,
 		GError **err)
 {
-	const gchar *start = pattern, *end, *flags_str = NULL;
+	const gchar *start = pattern, *end = start + len, *flags_str = NULL;
 	gchar *err_str;
 	rspamd_regexp_t *res;
 	gboolean explicit_utf = FALSE;
@@ -331,7 +331,7 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 		return NULL;
 	}
 
-	if (flags == NULL) {
+	if (flags == NULL && start < end) {
 		/* We need to parse pattern and detect flags set */
 		if (*start == '/') {
 			sep = '/';
@@ -347,14 +347,13 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 
 			rspamd_flags |= RSPAMD_REGEXP_FLAG_FULL_MATCH;
 		}
-		if (sep == '\0' || g_ascii_isalnum (sep)) {
+		if (g_ascii_isalnum (sep)) {
 			/* We have no flags, no separators and just use all line as expr */
 			start = pattern;
-			end = start + strlen (pattern);
 			rspamd_flags &= ~RSPAMD_REGEXP_FLAG_FULL_MATCH;
 		}
 		else {
-			end = strrchr (pattern, sep);
+			end = rspamd_memrchr(pattern, sep, len);
 
 			if (end == NULL || end <= start) {
 				g_set_error (err, rspamd_regexp_quark(), EINVAL,
@@ -370,7 +369,6 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 		/* Strictly check all flags */
 		strict_flags = TRUE;
 		start = pattern;
-		end = pattern + strlen (pattern);
 		flags_str = flags;
 	}
 
@@ -384,7 +382,7 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 #endif
 
 	if (flags_str != NULL) {
-		while (*flags_str) {
+		while (flags_str < end) {
 			switch (*flags_str) {
 			case 'i':
 				regexp_flags |= PCRE_FLAG(CASELESS);
@@ -535,9 +533,16 @@ fin:
 	return res;
 }
 
+rspamd_regexp_t *
+rspamd_regexp_new (const gchar *pattern, const gchar *flags,
+									GError **err)
+{
+	return rspamd_regexp_new_len(pattern, strlen(pattern), flags, err);
+}
+
 #ifndef WITH_PCRE2
 gboolean
-rspamd_regexp_search (rspamd_regexp_t *re, const gchar *text, gsize len,
+rspamd_regexp_search (const rspamd_regexp_t *re, const gchar *text, gsize len,
 		const gchar **start, const gchar **end, gboolean raw,
 		GArray *captures)
 {
@@ -672,7 +677,7 @@ rspamd_regexp_search (rspamd_regexp_t *re, const gchar *text, gsize len,
 #else
 /* PCRE 2 version */
 gboolean
-rspamd_regexp_search (rspamd_regexp_t *re, const gchar *text, gsize len,
+rspamd_regexp_search (const rspamd_regexp_t *re, const gchar *text, gsize len,
 		const gchar **start, const gchar **end, gboolean raw,
 		GArray *captures)
 {
@@ -788,7 +793,7 @@ rspamd_regexp_search (rspamd_regexp_t *re, const gchar *text, gsize len,
 #endif
 
 const char*
-rspamd_regexp_get_pattern (rspamd_regexp_t *re)
+rspamd_regexp_get_pattern (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -808,7 +813,7 @@ rspamd_regexp_set_flags (rspamd_regexp_t *re, guint new_flags)
 }
 
 guint
-rspamd_regexp_get_flags (rspamd_regexp_t *re)
+rspamd_regexp_get_flags (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -816,7 +821,7 @@ rspamd_regexp_get_flags (rspamd_regexp_t *re)
 }
 
 guint
-rspamd_regexp_get_pcre_flags (rspamd_regexp_t *re)
+rspamd_regexp_get_pcre_flags (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -824,7 +829,7 @@ rspamd_regexp_get_pcre_flags (rspamd_regexp_t *re)
 }
 
 gint
-rspamd_regexp_get_nbackrefs (rspamd_regexp_t *re)
+rspamd_regexp_get_nbackrefs (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -832,7 +837,7 @@ rspamd_regexp_get_nbackrefs (rspamd_regexp_t *re)
 }
 
 gint
-rspamd_regexp_get_ncaptures (rspamd_regexp_t *re)
+rspamd_regexp_get_ncaptures (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -840,7 +845,7 @@ rspamd_regexp_get_ncaptures (rspamd_regexp_t *re)
 }
 
 guint
-rspamd_regexp_get_maxhits (rspamd_regexp_t *re)
+rspamd_regexp_get_maxhits (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -860,7 +865,7 @@ rspamd_regexp_set_maxhits (rspamd_regexp_t *re, guint new_maxhits)
 }
 
 guint64
-rspamd_regexp_get_cache_id (rspamd_regexp_t *re)
+rspamd_regexp_get_cache_id (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -880,7 +885,7 @@ rspamd_regexp_set_cache_id (rspamd_regexp_t *re, guint64 id)
 }
 
 gboolean
-rspamd_regexp_match (rspamd_regexp_t *re, const gchar *text, gsize len,
+rspamd_regexp_match (const rspamd_regexp_t *re, const gchar *text, gsize len,
 		gboolean raw)
 {
 	const gchar *start = NULL, *end = NULL;
@@ -888,7 +893,7 @@ rspamd_regexp_match (rspamd_regexp_t *re, const gchar *text, gsize len,
 	g_assert (re != NULL);
 	g_assert (text != NULL);
 
-	if (len == 0 && text != NULL) {
+	if (len == 0) {
 		len = strlen (text);
 	}
 
@@ -926,7 +931,7 @@ rspamd_regexp_set_ud (rspamd_regexp_t *re, gpointer ud)
 }
 
 gpointer
-rspamd_regexp_get_ud (rspamd_regexp_t *re)
+rspamd_regexp_get_ud (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -1169,7 +1174,7 @@ rspamd_regexp_library_init (struct rspamd_config *cfg)
 }
 
 gpointer
-rspamd_regexp_get_id (rspamd_regexp_t *re)
+rspamd_regexp_get_id (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
@@ -1177,7 +1182,7 @@ rspamd_regexp_get_id (rspamd_regexp_t *re)
 }
 
 gpointer
-rspamd_regexp_get_class (rspamd_regexp_t *re)
+rspamd_regexp_get_class (const rspamd_regexp_t *re)
 {
 	g_assert (re != NULL);
 
diff --git a/src/libutil/regexp.h b/src/libutil/regexp.h
index 128edd761..1e98b7b3c 100644
--- a/src/libutil/regexp.h
+++ b/src/libutil/regexp.h
@@ -61,6 +61,16 @@ rspamd_regexp_t *rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 									GError **err);
 
 /**
+ * Create new rspamd regexp
+ * @param pattern regexp pattern
+ * @param flags flags (may be enclosed inside pattern)
+ * @param err error pointer set if compilation failed
+ * @return new regexp object
+ */
+rspamd_regexp_t *rspamd_regexp_new_len (const gchar *pattern, gsize len, const gchar *flags,
+									GError **err);
+
+/**
  * Search the specified regexp in the text
  * @param re
  * @param text
@@ -71,7 +81,7 @@ rspamd_regexp_t *rspamd_regexp_new (const gchar *pattern, const gchar *flags,
  * @param captures array of captured strings of type rspamd_fstring_capture or NULL
  * @return
  */
-gboolean rspamd_regexp_search (rspamd_regexp_t *re,
+gboolean rspamd_regexp_search (const rspamd_regexp_t *re,
 							   const gchar *text, gsize len,
 							   const gchar **start, const gchar **end, gboolean raw,
 							   GArray *captures);
@@ -84,7 +94,7 @@ gboolean rspamd_regexp_search (rspamd_regexp_t *re,
  * @param len
  * @return
  */
-gboolean rspamd_regexp_match (rspamd_regexp_t *re,
+gboolean rspamd_regexp_match (const rspamd_regexp_t *re,
 							  const gchar *text, gsize len, gboolean raw);
 
 /**
@@ -110,31 +120,31 @@ void rspamd_regexp_set_ud (rspamd_regexp_t *re, gpointer ud);
  * @param re regexp object
  * @return opaque pointer
  */
-gpointer rspamd_regexp_get_ud (rspamd_regexp_t *re);
+gpointer rspamd_regexp_get_ud (const rspamd_regexp_t *re);
 
 /**
  * Get regexp ID suitable for hashing
  * @param re
  * @return
  */
-gpointer rspamd_regexp_get_id (rspamd_regexp_t *re);
+gpointer rspamd_regexp_get_id (const rspamd_regexp_t *re);
 
 /**
  * Get pattern for the specified regexp object
  * @param re
  * @return
  */
-const char *rspamd_regexp_get_pattern (rspamd_regexp_t *re);
+const char *rspamd_regexp_get_pattern (const rspamd_regexp_t *re);
 
 /**
  * Get PCRE flags for the regexp
  */
-guint rspamd_regexp_get_pcre_flags (rspamd_regexp_t *re);
+guint rspamd_regexp_get_pcre_flags (const rspamd_regexp_t *re);
 
 /**
  * Get rspamd flags for the regexp
  */
-guint rspamd_regexp_get_flags (rspamd_regexp_t *re);
+guint rspamd_regexp_get_flags (const rspamd_regexp_t *re);
 
 /**
  * Set rspamd flags for the regexp
@@ -144,7 +154,7 @@ guint rspamd_regexp_set_flags (rspamd_regexp_t *re, guint new_flags);
 /**
  * Set regexp maximum hits
  */
-guint rspamd_regexp_get_maxhits (rspamd_regexp_t *re);
+guint rspamd_regexp_get_maxhits (const rspamd_regexp_t *re);
 
 /**
  * Get regexp maximum hits
@@ -154,17 +164,17 @@ guint rspamd_regexp_set_maxhits (rspamd_regexp_t *re, guint new_maxhits);
 /**
  * Returns number of backreferences in a regexp
  */
-gint rspamd_regexp_get_nbackrefs (rspamd_regexp_t *re);
+gint rspamd_regexp_get_nbackrefs (const rspamd_regexp_t *re);
 
 /**
  * Returns number of capture groups in a regexp
  */
-gint rspamd_regexp_get_ncaptures (rspamd_regexp_t *re);
+gint rspamd_regexp_get_ncaptures (const rspamd_regexp_t *re);
 
 /**
  * Returns cache id for a regexp
  */
-guint64 rspamd_regexp_get_cache_id (rspamd_regexp_t *re);
+guint64 rspamd_regexp_get_cache_id (const rspamd_regexp_t *re);
 
 /**
  * Sets cache id for a regexp
@@ -174,7 +184,7 @@ guint64 rspamd_regexp_set_cache_id (rspamd_regexp_t *re, guint64 id);
 /**
  * Get regexp class for the re object
  */
-gpointer rspamd_regexp_get_class (rspamd_regexp_t *re);
+gpointer rspamd_regexp_get_class (const rspamd_regexp_t *re);
 
 /**
  * Set regexp class for the re object
author	Vsevolod Stakhov <vsevolod@highsecure.ru>	2021-07-15 14:23:17 +0100
committer	Vsevolod Stakhov <vsevolod@highsecure.ru>	2021-07-15 14:23:17 +0100
commit	5f80e2e32e9e82c769b3f28be29cff4f62d9ecce (patch)
tree	a18428dc67bae8a113a293b6406cecf76097a336
parent	b2c36feea701c6685d83b8c4e6282fe7e307609d (diff)
download	rspamd-5f80e2e32e9e82c769b3f28be29cff4f62d9ecce.tar.gz rspamd-5f80e2e32e9e82c769b3f28be29cff4f62d9ecce.zip