Merge pull request #5 from UncoderIO/logsource-and-value-parsing-fixes

UncoderIO · web-flow · commit 06c84ff23a66 · 2023-11-10T15:13:25.000+01:00
splunk fixes
diff --git a/siem-converter/app/converter/backends/splunk/parsers/splunk.py b/siem-converter/app/converter/backends/splunk/parsers/splunk.py
@@ -45,8 +45,8 @@ def _parse_log_sources(self, query: str) -> Tuple[Dict[str, List[str]], str]:
             log_sources.setdefault(source_type, [])
             pattern = self.log_source_pattern.replace('___source_type___', source_type)
             while search := re.search(pattern, query, flags=re.IGNORECASE):
-                results = search.groupdict()
-                value = results.get("value")
+                group_dict = search.groupdict()
+                value = group_dict.get("d_q_value") or group_dict.get("value")
                 log_sources.setdefault(source_type, []).append(value)
                 pos_start = search.start()
                 pos_end = search.end()
diff --git a/siem-converter/app/converter/backends/splunk/tokenizer.py b/siem-converter/app/converter/backends/splunk/tokenizer.py
@@ -29,8 +29,9 @@ class SplunkTokenizer(QueryTokenizer):
     num_value_pattern = r"(?P<num_value>\d+(?:\.\d+)*)\s*"
     double_quotes_value_pattern = r'"(?P<d_q_value>(?:[:a-zA-Z\*0-9=+%#\-_/,\'\.$&^@!\(\)\{\}\s]|\\\"|\\)*)"\s*'
     single_quotes_value_pattern = r"'(?P<s_q_value>(?:[:a-zA-Z\*0-9=+%#\-_/,\"\.$&^@!\(\)\{\}\s]|\\\'|\\)*)'\s*"
-    _value_pattern = fr"{num_value_pattern}|{double_quotes_value_pattern}|{single_quotes_value_pattern}"
-    multi_value_pattern = r"""\((?P<value>[:a-zA-Z\"\*0-9=+%#\-_\/\\'\,.&^@!\(\s]*)\)"""
+    no_quotes_value = r"(?P<no_q_value>(?:[:a-zA-Z\*0-9=+%#\-_/,\.\\$&^@!])+)\s*"
+    _value_pattern = fr"{num_value_pattern}|{no_quotes_value}|{double_quotes_value_pattern}|{single_quotes_value_pattern}"
+    multi_value_pattern = r"""\((?P<value>[:a-zA-Z\"\*0-9=+%#\-_\/\\'\,.&^@!\(\s]+)\)"""
     keyword_pattern = double_quotes_value_pattern
 
     multi_value_operators = ("in",)
@@ -40,6 +41,9 @@ def get_operator_and_value(self, match: re.Match, operator: str = OperatorType.E
         if num_value := get_match_group(match, group_name='num_value'):
             return operator, num_value
 
+        elif no_q_value := get_match_group(match, group_name='no_q_value'):
+            return operator, no_q_value
+
         elif d_q_value := get_match_group(match, group_name='d_q_value'):
             return operator, d_q_value
 
