lexer: Fix multi-line string parsing

Multi-line strings were broken by the generic preprocessing support, as
both double and single quoted strings have their own lexing rules, which
take precedence over the "consume preprocessor tokens" lexer rule.

Reported by Dan Pascu

cfg.lex

@@ -77,11 +77,6 @@
 	#define SCRIPTVAR_S		4
 
 	#define STR_BUF_ALLOC_UNIT	128
-	struct str_buf{
-		char* s;
-		char* crt;
-		int left;
-	};
 
 	static int comment_nest=0;
 	static int state=0;
@@ -795,7 +790,9 @@ SPACE		[ ]
 						memset(&s_buf, 0, sizeof(s_buf));
 						return STRING;
 					}
-<STRING2>.|{EAT_ABLE}|{CR}	{ yymore(); }
+<STRING2>.|{EAT_ABLE}	{ addchar(&s_buf, *yytext); }
+<STRING2>{CR}	{ count(); if (!eatback_pp_tok(&s_buf))
+								addchar(&s_buf, *yytext); }
 
 <STRING1>\\n		{ count(); addchar(&s_buf, '\n'); }
 <STRING1>\\r		{ count(); addchar(&s_buf, '\r'); }
@@ -809,8 +806,9 @@ SPACE		[ ]
     subst_uri if allowed (although everybody should use '' in subt_uri) */
 <STRING1>\\[0-7]{2,3}	{ count(); addchar(&s_buf,
 											(char)strtol(yytext+1, 0, 8));  }
-<STRING1>\\{CR}		{ count(); } /* eat escaped CRs */
-<STRING1>{CR}	{ count();addchar(&s_buf, *yytext); }
+<STRING1>\\{CR}		{ count(); eatback_pp_tok(&s_buf); } /* eat escaped CRs */
+<STRING1>{CR}	{ count(); if (!eatback_pp_tok(&s_buf))
+								addchar(&s_buf, *yytext); }
 <STRING1>.|{EAT_ABLE}|{CR}	{ addchar(&s_buf, *yytext); }
 
 

cfg_pp.c

@@ -366,6 +366,9 @@ int cfg_pop(void)
 		return -1;
 	}
 
+	/* the file path MUST NOT be freed, as the lexer and parser work in tandem,
+	 * so by this point, there are plenty of structures referencing it */
+
 	if (cfg_include_stackp == cfg_include_stack) {
 		cfg_include_stackp = NULL;
 	} else {
@@ -603,3 +606,51 @@ static FILE *exec_preprocessor(FILE *flat_cfg, const char *preproc_cmdline)
 	fclose(flat_cfg);
 	return NULL;
 }
+
+int eatback_pp_tok(struct str_buf *buf)
+{
+	char *p;
+	str last_line;
+
+	if (!buf->s)
+		return 0;
+
+	for (p = buf->crt - 1; p >= buf->s; p--)
+		if (*p == '\n') {
+			p++;
+			goto match_pp_tok;
+		}
+
+	return 0;
+
+match_pp_tok:
+	last_line.s = p;
+	last_line.len = buf->crt - p;
+
+	if (last_line.len < 0) {
+		LM_BUG("negative line len");
+		return 0;
+	}
+
+	if (last_line.len >= cfgtok_line.len &&
+		!memcmp(last_line.s, cfgtok_line.s, cfgtok_line.len))
+		goto clear_last_line;
+
+	if (last_line.len >= cfgtok_filebegin.len &&
+		!memcmp(last_line.s, cfgtok_filebegin.s, cfgtok_filebegin.len))
+		goto clear_last_line;
+
+	if (last_line.len >= cfgtok_fileend.len &&
+		!memcmp(last_line.s, cfgtok_fileend.s, cfgtok_fileend.len))
+		goto clear_last_line;
+
+	/* don't touch anything, this is an actual script line! */
+	return 0;
+
+clear_last_line:
+	LM_DBG("clearing pp token line: '%.*s'\n", (int)(buf->crt - p), p);
+	buf->left += buf->crt - p;
+	*p = '\0';
+	buf->crt = p;
+	return 1;
+}

cfg_pp.h

@@ -29,11 +29,22 @@
 
 #define CFG_MAX_INCLUDE_DEPTH	20
 
+struct str_buf{
+	char* s;
+	char* crt;
+	int left;
+};
+
 int parse_opensips_cfg(const char *cfg_file, const char *preproc_cmdline,
 		FILE **ret_stream);
 int cfg_push(const str *cfg_file);
 int cfg_pop(void);
 void cfg_dump_context(const char *file, int line, int colstart, int colend);
 void cfg_dump_backtrace(void);
 
+/* ultimately helps correctly parse multi-line strings by eating any
+ * additionally inserted preprocessor directive as the last line of the
+ * given string buffer */
+int eatback_pp_tok(struct str_buf *buf);
+
 #endif /* __OSS_CFG_PP_H__ */