File size: 9,871 Bytes
4851501 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 |
"""
Multi-Step Query Planner
Detects complex queries that require multiple datasets or operations,
decomposes them into atomic steps, and orchestrates execution.
"""
import json
import logging
from dataclasses import dataclass, field
from typing import List, Dict, Any, Optional
from enum import Enum
logger = logging.getLogger(__name__)
class StepType(Enum):
"""Types of query steps."""
DATA_QUERY = "data_query" # Simple data retrieval
AGGREGATION = "aggregation" # COUNT, SUM, GROUP BY
COMPARISON = "comparison" # Comparing results from previous steps
SPATIAL_JOIN = "spatial_join" # Joining datasets spatially
COMBINE = "combine" # Merge/combine step results
@dataclass
class QueryStep:
"""A single atomic step in a query plan."""
step_id: str
step_type: StepType
description: str
tables_needed: List[str]
sql_template: Optional[str] = None
depends_on: List[str] = field(default_factory=list)
result_name: str = "" # Name for intermediate result
def to_dict(self) -> Dict[str, Any]:
return {
"step_id": self.step_id,
"step_type": self.step_type.value,
"description": self.description,
"tables_needed": self.tables_needed,
"sql_template": self.sql_template,
"depends_on": self.depends_on,
"result_name": self.result_name
}
@dataclass
class QueryPlan:
"""Complete execution plan for a complex query."""
original_query: str
is_complex: bool
steps: List[QueryStep] = field(default_factory=list)
parallel_groups: List[List[str]] = field(default_factory=list) # Steps that can run in parallel
final_combination_logic: str = ""
def to_dict(self) -> Dict[str, Any]:
return {
"original_query": self.original_query,
"is_complex": self.is_complex,
"steps": [s.to_dict() for s in self.steps],
"parallel_groups": self.parallel_groups,
"final_combination_logic": self.final_combination_logic
}
class QueryPlanner:
"""
Multi-step query planning service.
Analyzes queries to determine complexity and decomposes
complex queries into executable atomic steps.
"""
_instance = None
# Keywords that often indicate multi-step queries
COMPLEXITY_INDICATORS = [
"compare", "comparison", "versus", "vs",
"more than", "less than", "higher than", "lower than",
"both", "and also", "as well as",
"ratio", "percentage", "proportion",
"correlation", "relationship between",
"combine", "merge", "together with",
"relative to", "compared to",
"difference between", "gap between"
]
# Keywords indicating multiple distinct data types
MULTI_DOMAIN_KEYWORDS = {
"health": ["hospital", "clinic", "healthcare", "health", "medical"],
"education": ["school", "university", "education", "college", "student"],
"infrastructure": ["road", "bridge", "infrastructure", "building"],
"environment": ["forest", "water", "environment", "park", "protected"],
"population": ["population", "demographic", "census", "people", "resident"]
}
def __new__(cls):
if cls._instance is None:
cls._instance = super(QueryPlanner, cls).__new__(cls)
cls._instance.initialized = False
return cls._instance
def __init__(self):
if self.initialized:
return
self.initialized = True
def detect_complexity(self, query: str) -> Dict[str, Any]:
"""
Analyze a query to determine if it requires multi-step planning.
Returns:
{
"is_complex": bool,
"reason": str,
"detected_domains": List[str],
"complexity_indicators": List[str]
}
"""
query_lower = query.lower()
# Check for complexity indicators
found_indicators = [
ind for ind in self.COMPLEXITY_INDICATORS
if ind in query_lower
]
# Check for multiple data domains
found_domains = []
for domain, keywords in self.MULTI_DOMAIN_KEYWORDS.items():
if any(kw in query_lower for kw in keywords):
found_domains.append(domain)
# Determine complexity
is_complex = (
len(found_indicators) > 0 and len(found_domains) >= 2
) or (
len(found_domains) >= 3
) or (
any(x in query_lower for x in ["compare", "ratio", "correlation", "versus", " vs "])
and len(found_domains) >= 2
)
reason = ""
if is_complex:
if len(found_domains) >= 2:
reason = f"Query involves multiple data domains: {', '.join(found_domains)}"
if found_indicators:
reason += f". Contains comparison/aggregation keywords: {', '.join(found_indicators[:3])}"
return {
"is_complex": is_complex,
"reason": reason,
"detected_domains": found_domains,
"complexity_indicators": found_indicators
}
async def plan_query(
self,
query: str,
available_tables: List[str],
llm_gateway
) -> QueryPlan:
"""
Create an execution plan for a complex query.
Uses LLM to decompose the query into atomic steps.
"""
from backend.core.prompts import QUERY_PLANNING_PROMPT
# Build table context
table_list = "\n".join(f"- {t}" for t in available_tables)
prompt = QUERY_PLANNING_PROMPT.format(
user_query=query,
available_tables=table_list
)
try:
response = await llm_gateway.generate_response(prompt, [])
# Parse JSON response
response_clean = response.strip()
if response_clean.startswith("```json"):
response_clean = response_clean[7:]
if response_clean.startswith("```"):
response_clean = response_clean[3:]
if response_clean.endswith("```"):
response_clean = response_clean[:-3]
plan_data = json.loads(response_clean.strip())
# Convert to QueryPlan
steps = []
for i, step_data in enumerate(plan_data.get("steps", [])):
step = QueryStep(
step_id=f"step_{i+1}",
step_type=StepType(step_data.get("type", "data_query")),
description=step_data.get("description", ""),
tables_needed=step_data.get("tables", []),
sql_template=step_data.get("sql_hint", None),
depends_on=step_data.get("depends_on", []),
result_name=step_data.get("result_name", f"result_{i+1}")
)
steps.append(step)
# Determine parallel groups (steps with no dependencies can run together)
parallel_groups = self._compute_parallel_groups(steps)
return QueryPlan(
original_query=query,
is_complex=True,
steps=steps,
parallel_groups=parallel_groups,
final_combination_logic=plan_data.get("combination_logic", "")
)
except Exception as e:
logger.error(f"Query planning failed: {e}")
# Return single-step fallback
return QueryPlan(
original_query=query,
is_complex=False,
steps=[],
parallel_groups=[],
final_combination_logic=""
)
def _compute_parallel_groups(self, steps: List[QueryStep]) -> List[List[str]]:
"""
Compute which steps can be executed in parallel.
Steps with no dependencies (or only completed dependencies)
can run together.
"""
if not steps:
return []
groups = []
executed = set()
remaining = {s.step_id: s for s in steps}
while remaining:
# Find steps whose dependencies are all satisfied
ready = [
step_id for step_id, step in remaining.items()
if all(dep in executed for dep in step.depends_on)
]
if not ready:
# Avoid infinite loop - add remaining as sequential
ready = list(remaining.keys())[:1]
groups.append(ready)
for step_id in ready:
executed.add(step_id)
del remaining[step_id]
return groups
def create_simple_plan(self, query: str) -> QueryPlan:
"""Create a simple single-step plan for non-complex queries."""
return QueryPlan(
original_query=query,
is_complex=False,
steps=[
QueryStep(
step_id="step_1",
step_type=StepType.DATA_QUERY,
description="Execute query directly",
tables_needed=[],
depends_on=[]
)
],
parallel_groups=[["step_1"]]
)
# Singleton accessor
_query_planner: Optional[QueryPlanner] = None
def get_query_planner() -> QueryPlanner:
"""Get the singleton query planner instance."""
global _query_planner
if _query_planner is None:
_query_planner = QueryPlanner()
return _query_planner
|