Improved multi-statement detection in readwritesplit

Now comment blocks and identifiers quoted with backticks are taken into
consideration when multi-statements are processed.

server/core/modutil.c

@@ -801,6 +801,115 @@ void* strnchr_esc(char* ptr, char c, int len)
     return NULL;
 }
 
+/**
+ * Find the first occurrence of a character in a string. This function ignores
+ * escaped characters and all characters that are enclosed in single or double quotes.
+ * Also MySQL style comment blocks are ignored.
+ * @param ptr Pointer to area of memory to inspect
+ * @param c Character to search for
+ * @param len Size of the memory area
+ * @return Pointer to the first non-escaped, non-quoted occurrence of the character.
+ * If the character is not found, NULL is returned.
+ */
+void* strnchr_esc_mysql(char* ptr, char c, int len)
+{
+    char* p = (char*) ptr;
+    char* start = p;
+    bool quoted = false, escaped = false, backtick = false;
+    char qc;
+
+    while (p < start + len)
+    {
+        if (escaped)
+        {
+            escaped = false;
+        }
+        else
+        {
+            switch (*p)
+            {
+                case '\\':
+                    escaped = true;
+                    break;
+
+                case'\'':
+                case '"':
+                    if (!quoted)
+                    {
+                        quoted = true;
+                        qc = *p;
+                    }
+                    else if (*p == qc)
+                    {
+                        quoted = false;
+                    }
+                    break;
+
+                case '`':
+                    backtick = !backtick;
+                    break;
+
+                case '#':
+                    if (!backtick)
+                    {
+                        return NULL;
+                    }
+                    break;
+
+                case '-':
+                    if (!backtick && p + 1 < start + len && *(p + 1) == '-')
+                    {
+                        return NULL;
+                    }
+                    break;
+            }
+
+            if (*p == c && !escaped && !quoted)
+            {
+                return p;
+            }
+            p++;
+        }
+    }
+    return NULL;
+}
+
+/**
+ * Check if the start of the token is the start of a MySQL style comment block
+ * @param ptr String with at least two non-null characters in it
+ * @return True if the token starts a comment block
+ */
+bool is_mysql_comment_start(const char* start, int len)
+{
+    const char *ptr = start;
+
+    while (ptr - start < len && isspace(*ptr))
+    {
+        ptr++;
+    }
+
+    switch (*ptr)
+    {
+        case '-':
+            if (*(ptr + 1) == '-')
+            {
+                return true;
+            }
+            break;
+
+        case '#':
+            return true;
+
+        case '/':
+            if (*(ptr + 1) == '*')
+            {
+                return true;
+            }
+            break;
+    }
+    return false;
+}
+
 /**
  * Create a COM_QUERY packet from a string.
  * @param query Query to create.

server/include/modutil.h

@@ -67,6 +67,10 @@ GWBUF*          modutil_create_mysql_err_msg(int             packet_number,
                                              const char      *msg);
 int modutil_count_signal_packets(GWBUF*,int,int,int*);
 mxs_pcre2_result_t modutil_mysql_wildcard_match(const char* pattern, const char* string);
+
+/** Character and token searching functions */
 void* strnchr_esc(char* ptr, char c, int len);
+void* strnchr_esc_mysql(char* ptr, char c, int len);
+bool is_mysql_comment_start(const char* start, int len);
 
 #endif

server/modules/routing/readwritesplit/readwritesplit.c

@@ -308,6 +308,8 @@ static ROUTER_INSTANCE* instances;
 
 static int hashkeyfun(void* key);
 static int hashcmpfun (void *, void *);
+static void check_for_multi_stmt(ROUTER_CLIENT_SES* rses, GWBUF *buf,
+                                 mysql_server_cmd_t packet_type);
 
 static int hashkeyfun(
 		void* key)
@@ -2156,23 +2158,7 @@ static bool route_single_stmt(
     /** Check for multi-statement queries. We assume here that the client
      * protocol is MySQLClient.
      * TODO: add warnings when incompatible protocols are used */
-    MySQLProtocol *proto = (MySQLProtocol*)rses->client_dcb->protocol;
+    check_for_multi_stmt(rses, querybuf, packet_type);
-    if (proto->client_capabilities & GW_MYSQL_CAPABILITIES_MULTI_STATEMENTS &&
-        packet_type == MYSQL_COM_QUERY && rses->forced_node != rses->rses_master_ref)
-    {
-        for (GWBUF *buf = querybuf; buf != NULL; buf = buf->next)
-        {
-            if (strnchr_esc(GWBUF_DATA(buf), ';', GWBUF_LENGTH(buf)))
-            {
-                /** It is possible that the session state is modified inside
-                 * the multi-statement query which would leave any slave sessions
-                 * in an inconsistent state. Due to this, for the duration of
-                 * this session, all queries will be sent to the master. */
-                rses->forced_node = rses->rses_master_ref;
-                MXS_INFO("Multi-statement query, routing all future queries to master.");
-            }
-        }
-    }
 
     /**
      * Check if the query has anything to do with temporary tables.
@@ -5357,12 +5343,35 @@ static backend_ref_t* get_root_master_bref(
 	return candidate_bref;
 }
 
+/**
+ * @brief Detect multi-statement queries
+ *
+ * It is possible that the session state is modified inside a multi-statement
+ * query which would leave any slave sessions in an inconsistent state. Due to
+ * this, for the duration of this session, all queries will be sent to the master
+ * if the current query contains a multi-statement query.
+ * @param rses Router client session
+ * @param buf Buffer containing the full query
+ */
+static void check_for_multi_stmt(ROUTER_CLIENT_SES* rses, GWBUF *buf,
+                                 mysql_server_cmd_t packet_type)
+{
+    MySQLProtocol *proto = (MySQLProtocol*) rses->client_dcb->protocol;
 
+    if (proto->client_capabilities & GW_MYSQL_CAPABILITIES_MULTI_STATEMENTS &&
+        packet_type == MYSQL_COM_QUERY && rses->forced_node != rses->rses_master_ref)
+    {
+        char *ptr, *data = GWBUF_DATA(buf) + 5;
+        int buflen = GWBUF_LENGTH(buf) - 5;
 
-
+        if ((ptr = strnchr_esc_mysql(data, ';', buflen)))
-
+        {
-
+            ptr++;
-
+            if (ptr - data < buflen && !is_mysql_comment_start(ptr, ptr - data))
-
+            {
-
+                rses->forced_node = rses->rses_master_ref;
-
+                MXS_INFO("Multi-statement query, routing all future queries to master.");
+            }
+        }
+    }
+}