main
py 219 lines 7.74 KB
Raw
1 from fastapi import HTTPException
2 from loguru import logger
3 from sqlalchemy import select
4 from sqlalchemy.ext.asyncio import AsyncSession
5
6 from app.connectors.wazuh_indexer.utils.universal import AlertsQueryBuilder
7 from app.connectors.wazuh_indexer.utils.universal import (
8 create_wazuh_indexer_client_async,
9 )
10 from app.db.universal_models import EventSources
11 from app.siem.schema.events import EventsQueryParams
12 from app.siem.schema.events import EventsQueryResponse
13 from app.siem.schema.events import FieldMapping
14 from app.siem.schema.events import FieldMappingsResponse
15
16
17 async def get_event_source_by_customer_and_name(
18 customer_code: str,
19 source_name: str,
20 db: AsyncSession,
21 ) -> EventSources:
22 result = await db.execute(
23 select(EventSources).where(
24 EventSources.customer_code == customer_code,
25 EventSources.name == source_name,
26 ),
27 )
28 event_source = result.scalars().first()
29 if not event_source:
30 raise HTTPException(
31 status_code=404,
32 detail=f"Event source '{source_name}' not found for customer {customer_code}",
33 )
34 if not event_source.enabled:
35 raise HTTPException(
36 status_code=400,
37 detail=f"Event source '{source_name}' is disabled",
38 )
39 return event_source
40
41
42 async def query_events(
43 customer_code: str,
44 source_name: str,
45 params: EventsQueryParams,
46 db: AsyncSession,
47 ) -> EventsQueryResponse:
48 logger.info(f"Querying events for customer {customer_code}, source {source_name}")
49
50 # If a scroll_id is provided, continue scrolling
51 if params.scroll_id:
52 return await _scroll_next_page(params.scroll_id)
53
54 # Look up event source to get index_pattern and time_field
55 event_source = await get_event_source_by_customer_and_name(customer_code, source_name, db)
56
57 return await _initial_search(
58 index_pattern=event_source.index_pattern,
59 time_field=event_source.time_field,
60 timerange=params.timerange,
61 page_size=params.page_size,
62 query=params.query,
63 time_from=params.time_from,
64 time_to=params.time_to,
65 )
66
67
68 async def _initial_search(
69 index_pattern: str,
70 time_field: str,
71 timerange: str,
72 page_size: int,
73 query: str = None,
74 time_from: str = None,
75 time_to: str = None,
76 ) -> EventsQueryResponse:
77 es_client = await create_wazuh_indexer_client_async("Wazuh-Indexer")
78 try:
79 query_builder = AlertsQueryBuilder()
80 if time_from and time_to:
81 query_builder.add_absolute_time_range(time_from=time_from, time_to=time_to, timestamp_field=time_field)
82 else:
83 query_builder.add_time_range(timerange=timerange, timestamp_field=time_field)
84 query_builder.add_sort(time_field, order="desc")
85
86 # Add Lucene query_string if provided
87 if query:
88 query_builder.query["query"]["bool"]["must"].append(
89 {"query_string": {"query": query, "default_operator": "AND"}},
90 )
91
92 query = query_builder.build()
93
94 response = await es_client.search(
95 index=index_pattern,
96 body=query,
97 size=page_size,
98 scroll="5m",
99 )
100
101 hits = response["hits"]["hits"]
102 total = response["hits"]["total"]["value"] if isinstance(response["hits"]["total"], dict) else response["hits"]["total"]
103 scroll_id = response.get("_scroll_id")
104
105 # If all results fit in one page, clear the scroll context
106 if len(hits) >= total:
107 if scroll_id:
108 await _clear_scroll(es_client, scroll_id)
109 scroll_id = None
110
111 return EventsQueryResponse(
112 events=[hit["_source"] for hit in hits],
113 total=total,
114 scroll_id=scroll_id,
115 page_size=page_size,
116 success=True,
117 message=f"Retrieved {len(hits)} of {total} events",
118 )
119 except Exception as e:
120 logger.error(f"Error querying events: {e}")
121 raise HTTPException(status_code=500, detail=f"Error querying events: {e}")
122 finally:
123 await es_client.close()
124
125
126 async def _scroll_next_page(scroll_id: str) -> EventsQueryResponse:
127 es_client = await create_wazuh_indexer_client_async("Wazuh-Indexer")
128 try:
129 response = await es_client.scroll(scroll_id=scroll_id, scroll="5m")
130 hits = response["hits"]["hits"]
131 total = response["hits"]["total"]["value"] if isinstance(response["hits"]["total"], dict) else response["hits"]["total"]
132 new_scroll_id = response.get("_scroll_id")
133
134 # If no more results, clear the scroll context
135 if not hits:
136 if new_scroll_id:
137 await _clear_scroll(es_client, new_scroll_id)
138 return EventsQueryResponse(
139 events=[],
140 total=total,
141 scroll_id=None,
142 page_size=0,
143 success=True,
144 message="No more results",
145 )
146
147 return EventsQueryResponse(
148 events=[hit["_source"] for hit in hits],
149 total=total,
150 scroll_id=new_scroll_id,
151 page_size=len(hits),
152 success=True,
153 message=f"Retrieved {len(hits)} of {total} events",
154 )
155 except Exception as e:
156 logger.error(f"Error scrolling events: {e}")
157 raise HTTPException(status_code=500, detail=f"Error scrolling events: {e}")
158 finally:
159 await es_client.close()
160
161
162 async def _clear_scroll(es_client, scroll_id: str) -> None:
163 try:
164 await es_client.clear_scroll(scroll_id=scroll_id)
165 except Exception as e:
166 logger.warning(f"Failed to clear scroll context: {e}")
167
168
169 async def get_field_mappings(
170 customer_code: str,
171 source_name: str,
172 db: AsyncSession,
173 ) -> FieldMappingsResponse:
174 """Retrieve index field name mappings for a customer's event source."""
175 event_source = await get_event_source_by_customer_and_name(customer_code, source_name, db)
176 es_client = await create_wazuh_indexer_client_async("Wazuh-Indexer")
177 try:
178 mapping_response = await es_client.indices.get_mapping(index=event_source.index_pattern)
179
180 # Flatten nested mappings into dot-notation field list
181 fields = []
182 for index_name in mapping_response:
183 properties = mapping_response[index_name].get("mappings", {}).get("properties", {})
184 _flatten_properties(properties, "", fields)
185 break # All indices matching pattern share the same mapping
186
187 # Deduplicate and sort
188 seen = set()
189 unique_fields = []
190 for f in fields:
191 if f.field not in seen:
192 seen.add(f.field)
193 unique_fields.append(f)
194 unique_fields.sort(key=lambda x: x.field)
195
196 return FieldMappingsResponse(
197 fields=unique_fields,
198 total=len(unique_fields),
199 index_pattern=event_source.index_pattern,
200 success=True,
201 message=f"Retrieved {len(unique_fields)} field mappings",
202 )
203 except Exception as e:
204 logger.error(f"Error retrieving field mappings: {e}")
205 raise HTTPException(status_code=500, detail=f"Error retrieving field mappings: {e}")
206 finally:
207 await es_client.close()
208
209
210 def _flatten_properties(properties: dict, prefix: str, fields: list) -> None:
211 """Recursively flatten OpenSearch mapping properties into FieldMapping objects."""
212 for field_name, field_info in properties.items():
213 full_name = f"{prefix}{field_name}" if not prefix else f"{prefix}_{field_name}"
214 field_type = field_info.get("type")
215 if field_type:
216 fields.append(FieldMapping(field=full_name, type=field_type))
217 # Recurse into nested properties
218 if "properties" in field_info:
219 _flatten_properties(field_info["properties"], full_name, fields)