Preparation to extract modifiers from comment (#166)

include/c_tokenizer.h

@@ -5,6 +5,11 @@
 #ifndef C_TOKENIZER_H
 #define C_TOKENIZER_H
 
+
+#ifndef FIRST_COMMENT_MAX_LENGTH
+#define FIRST_COMMENT_MAX_LENGTH  1024
+#endif /* FIRST_COMMENT_MAX_LENGTH */
+
 typedef struct
 {
 	char*       s;
@@ -23,8 +28,7 @@ extern "C" {
 tokenizer_t tokenizer( const char* s, const char* delimiters, int empties );
 const char* free_tokenizer( tokenizer_t* tokenizer );
 const char* tokenize( tokenizer_t* tokenizer );
-char * mysql_query_digest(char *s, int len);
-
+char * mysql_query_digest_and_first_comment(char *s , int len , char *first_comment);
 void c_split_2(const char *in, const char *del, char **out1, char **out2);
 #ifdef __cplusplus
 }

lib/Query_Processor.cpp

@@ -69,6 +69,7 @@ struct __SQP_query_parser_t {
 	sfilter sf;
 	uint64_t digest;
 	char *digest_text;
+	char *first_comment;
 	uint64_t digest_total;
 };
 
@@ -794,9 +795,14 @@ void * Query_Processor::query_parser_init(char *query, int query_length, int fla
 	if (mysql_thread___commands_stats)
 		libinjection_sqli_init(&qp->sf, query, query_length, FLAG_SQL_MYSQL);
 	qp->digest_text=NULL;
+	qp->first_comment=NULL;
+	qp->first_comment=(char *)l_alloc(FIRST_COMMENT_MAX_LENGTH);
 	if (mysql_thread___query_digests) {
-		qp->digest_text=mysql_query_digest(query, query_length);
+		qp->digest_text=mysql_query_digest_and_first_comment(query, query_length, qp->first_comment);
 		qp->digest=SpookyHash::Hash64(qp->digest_text,strlen(qp->digest_text),0);
+		if (strlen(qp->first_comment)) {
+			fprintf(stderr,"Comment= %s \n", qp->first_comment);
+		}
 	}
 	return (void *)qp;
 };
@@ -974,5 +980,8 @@ void Query_Processor::query_parser_free(void *args) {
 		free(qp->digest_text);
 		qp->digest_text=NULL;
 	}
-	free(qp);	
+	if (qp->first_comment) {
+		l_free(FIRST_COMMENT_MAX_LENGTH,qp->first_comment);
+	}
+	free(qp);
 };

lib/c_tokenizer.c

@@ -178,7 +178,7 @@ static char is_digit_string(char *f, char *t)
 }
 
 
-char *mysql_query_digest(char *s, int len){
+char *mysql_query_digest_and_first_comment(char *s, int len, char *first_comment){
 	int i = 0;
 
 	char *r = (char *) malloc(len + SIZECHAR);
@@ -190,6 +190,8 @@ char *mysql_query_digest(char *s, int len){
 	char qutr_char = 0;
 
 	char flag = 0;
+	char fc=0;
+	int fc_len=0;
 
 	while(i < len)
 	{
@@ -250,6 +252,22 @@ char *mysql_query_digest(char *s, int len){
 			// --------
 			// comment
 			// --------
+			if (flag == 1) {
+				if (fc==0) {
+					fc=1;
+				}
+				if (fc==1) {
+					if (fc_len<FIRST_COMMENT_MAX_LENGTH-1) {
+						first_comment[fc_len]= !is_space_char(*s) ? *s : ' ';
+						fc_len++;
+					}
+					if (prev_char == '*' && *s == '/') {
+						if (fc_len>=2) fc_len-=2;
+						first_comment[fc_len]=0;
+						fc=2;
+					}
+				}
+			}
 			if(
 				// comment type 1 - /* .. */
 				(flag == 1 && prev_char == '*' && *s == '/') ||