snowflake-switchboard-ai 0.1.0__tar.gz

snowflake_switchboard_ai-0.1.0/PKG-INFO

@@ -0,0 +1,26 @@
+Metadata-Version: 2.4
+Name: snowflake-switchboard-ai
+Version: 0.1.0
+Summary: A dynamic, metadata-driven query router for Snowflake and Snowpark
+Author-email: Murtaza Mohammed Rafiqi <murtuzarafiqui@gmail.com>
+License: MIT
+Project-URL: Homepage, https://github.com/mmar4/switchboard-ai
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: snowflake-snowpark-python>=1.0.0
+
+# Switchboard AI 🚦
+
+An intelligent query router that automatically optimizes your data workloads by routing lightweight queries to a local embedded engine (`DuckDB`) and heavy workloads to cloud warehouses (`Snowflake`).
+
+## Features
+- **Smart Routing:** Intercepts SQL queries and decides the most cost-effective compute engine.
+- **Low Latency:** Executes minor lookups locally in-memory.
+- **Seamless Snowflake Integration:** Native fallback to Snowflake cloud instances for heavy pipelines.
+
+## Installation
+```bash
+pip install -e .

snowflake_switchboard_ai-0.1.0/README.md

@@ -0,0 +1,12 @@
+# Switchboard AI 🚦
+
+An intelligent query router that automatically optimizes your data workloads by routing lightweight queries to a local embedded engine (`DuckDB`) and heavy workloads to cloud warehouses (`Snowflake`).
+
+## Features
+- **Smart Routing:** Intercepts SQL queries and decides the most cost-effective compute engine.
+- **Low Latency:** Executes minor lookups locally in-memory.
+- **Seamless Snowflake Integration:** Native fallback to Snowflake cloud instances for heavy pipelines.
+
+## Installation
+```bash
+pip install -e .

snowflake_switchboard_ai-0.1.0/pyproject.toml

@@ -0,0 +1,25 @@
+[build-system]
+requires = ["setuptools>=61.0.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "snowflake-switchboard-ai" 
+version = "0.1.0"                 
+description = "A dynamic, metadata-driven query router for Snowflake and Snowpark"
+readme = "README.md"
+authors = [
+    { name = "Murtaza Mohammed Rafiqi", email = "murtuzarafiqui@gmail.com" }
+]
+license = { text = "MIT" }
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: MIT License",
+    "Operating System :: OS Independent",
+]
+requires-python = ">=3.8"
+dependencies = [
+    "snowflake-snowpark-python>=1.0.0"
+]
+
+[project.urls]
+Homepage = "https://github.com/mmar4/switchboard-ai"

snowflake_switchboard_ai-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

snowflake_switchboard_ai-0.1.0/snowflake_switchboard_ai.egg-info/PKG-INFO

@@ -0,0 +1,26 @@
+Metadata-Version: 2.4
+Name: snowflake-switchboard-ai
+Version: 0.1.0
+Summary: A dynamic, metadata-driven query router for Snowflake and Snowpark
+Author-email: Murtaza Mohammed Rafiqi <murtuzarafiqui@gmail.com>
+License: MIT
+Project-URL: Homepage, https://github.com/mmar4/switchboard-ai
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: snowflake-snowpark-python>=1.0.0
+
+# Switchboard AI 🚦
+
+An intelligent query router that automatically optimizes your data workloads by routing lightweight queries to a local embedded engine (`DuckDB`) and heavy workloads to cloud warehouses (`Snowflake`).
+
+## Features
+- **Smart Routing:** Intercepts SQL queries and decides the most cost-effective compute engine.
+- **Low Latency:** Executes minor lookups locally in-memory.
+- **Seamless Snowflake Integration:** Native fallback to Snowflake cloud instances for heavy pipelines.
+
+## Installation
+```bash
+pip install -e .

snowflake_switchboard_ai-0.1.0/snowflake_switchboard_ai.egg-info/SOURCES.txt

@@ -0,0 +1,9 @@
+README.md
+pyproject.toml
+snowflake_switchboard_ai.egg-info/PKG-INFO
+snowflake_switchboard_ai.egg-info/SOURCES.txt
+snowflake_switchboard_ai.egg-info/dependency_links.txt
+snowflake_switchboard_ai.egg-info/requires.txt
+snowflake_switchboard_ai.egg-info/top_level.txt
+switchboard/__init__.py
+switchboard/core.py

snowflake_switchboard_ai-0.1.0/snowflake_switchboard_ai.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

snowflake_switchboard_ai-0.1.0/snowflake_switchboard_ai.egg-info/requires.txt

@@ -0,0 +1 @@
+snowflake-snowpark-python>=1.0.0

snowflake_switchboard_ai-0.1.0/snowflake_switchboard_ai.egg-info/top_level.txt

@@ -0,0 +1 @@
+switchboard

snowflake_switchboard_ai-0.1.0/switchboard/__init__.py

@@ -0,0 +1 @@
+from .core import MultiTierSwitchboard

snowflake_switchboard_ai-0.1.0/switchboard/core.py

@@ -0,0 +1,83 @@
+import json
+from snowflake.snowpark import Session
+from snowflake.snowpark.dataframe import DataFrame
+
+class MultiTierSwitchboard:
+    def __init__(self, session: Session, warehouses: list = None, tiers: list = None):
+        self.session = session
+        
+        if tiers:
+            self.tiers = sorted(tiers, key=lambda x: x["max_bytes"])
+        elif warehouses:
+            standard_limits = [
+                1_000_000_000,      # 1 GB (XS limit)
+                5_000_000_000,      # 5 GB (Small limit)
+                20_000_000_000,     # 20 GB (Medium limit)
+                50_000_000_000,     # 50 GB (Large limit)
+                100_000_000_000,    # 100 GB (X-Large limit)
+                500_000_000_000     # 500 GB (2X-Large limit)
+            ]
+            self.tiers = []
+            for i, wh in enumerate(warehouses):
+                if i == len(warehouses) - 1:
+                    limit = float("inf")
+                else:
+                    limit = standard_limits[i] if i < len(standard_limits) else float("inf")
+                self.tiers.append({"warehouse": wh, "max_bytes": limit})
+        else:
+            self.tiers = [
+                {"warehouse": "COMPUTE_WH", "max_bytes": 1_000_000_000},
+                {"warehouse": "PRACTICE", "max_bytes": float("inf")}
+            ]
+
+    def run(self, query: str) -> DataFrame:
+        print(f"🤖 [Switchboard AI]: Analyzing query structure...")
+        
+        estimated_bytes = 0
+        try:
+            explain_sql = f"SELECT SYSTEM$EXPLAIN_PLAN_JSON($${query}$$)"
+            explain_result = self.session.sql(explain_sql).collect()
+            plan_json_str = explain_result[0][0]
+            
+            if plan_json_str:
+                plan_data = json.loads(plan_json_str)
+                estimated_bytes = self._extract_bytes_assigned(plan_data)
+                
+            print(f"📊 [Switchboard AI]: Estimated telemetry payload -> {estimated_bytes:,} bytes assigned.")
+            
+        except Exception as e:
+            print(f"⚠️ [Switchboard AI]: Pre-check skipped ({e}). Using baseline tier.")
+
+        target_wh = self.tiers[0]["warehouse"]
+        for tier in self.tiers:
+            if estimated_bytes <= tier["max_bytes"]:
+                target_wh = tier["warehouse"]
+                break
+
+        print(f"🚀 [Switchboard AI]: Routing query to optimal tier -> {target_wh}")
+
+        self.session.sql(f"USE WAREHOUSE {target_wh}").collect()
+        
+        print(f"⚙️ [Switchboard AI]: Executing query...")
+        return self.session.sql(query)
+
+    def __call__(self, query: str, auto_show: bool = True) -> DataFrame:
+        df = self.run(query)
+        if auto_show:
+            df.show()
+        return df
+
+    def _extract_bytes_assigned(self, plan_node) -> int:
+        total_bytes = 0
+        if isinstance(plan_node, dict):
+            if "bytesAssigned" in plan_node:
+                try:
+                    total_bytes += int(plan_node["bytesAssigned"])
+                except (ValueError, TypeError):
+                    pass
+            for key, value in plan_node.items():
+                total_bytes += self._extract_bytes_assigned(value)
+        elif isinstance(plan_node, list):
+            for item in plan_node:
+                total_bytes += self._extract_bytes_assigned(item)
+        return total_bytes