dvt-core 0.52.2__cp310-cp310-macosx_10_9_x86_64.whl

dbt/plugins/manager.py

@@ -0,0 +1,163 @@
+import functools
+import importlib
+import pkgutil
+from types import ModuleType
+from typing import Callable, Dict, List, Mapping
+
+import dbt.tracking
+from dbt.contracts.graph.manifest import Manifest
+from dbt.plugins.contracts import PluginArtifacts
+from dbt.plugins.manifest import PluginNodes
+from dbt_common.exceptions import DbtRuntimeError
+from dbt_common.tests import test_caching_enabled
+
+
+def dbt_hook(func):
+    def inner(*args, **kwargs):
+        try:
+            return func(*args, **kwargs)
+        except Exception as e:
+            raise DbtRuntimeError(f"{func.__name__}: {e}")
+
+    setattr(inner, "is_dbt_hook", True)
+    return inner
+
+
+class dbtPlugin:
+    """
+    EXPERIMENTAL: dbtPlugin is the base class for creating plugins.
+    Its interface is **not** stable and will likely change between dbt-core versions.
+    """
+
+    def __init__(self, project_name: str) -> None:
+        self.project_name = project_name
+        try:
+            self.initialize()
+        except DbtRuntimeError as e:
+            # Remove the first line of DbtRuntimeError to avoid redundant "Runtime Error" line
+            raise DbtRuntimeError("\n".join(str(e).split("\n")[1:]))
+        except Exception as e:
+            raise DbtRuntimeError(str(e))
+
+    @property
+    def name(self) -> str:
+        return self.__class__.__name__
+
+    def initialize(self) -> None:
+        """
+        Initialize the plugin. This function may be overridden by subclasses that have
+        additional initialization steps.
+        """
+        pass
+
+    def get_nodes(self) -> PluginNodes:
+        """
+        Provide PluginNodes to dbt for injection into dbt's DAG.
+        Currently the only node types that are accepted are model nodes.
+        """
+        raise NotImplementedError(f"get_nodes hook not implemented for {self.name}")
+
+    def get_manifest_artifacts(self, manifest: Manifest) -> PluginArtifacts:
+        """
+        Given a manifest, provide PluginArtifacts derived for writing by core.
+        PluginArtifacts share the same lifecycle as the manifest.json file -- they
+        will either be written or not depending on whether the manifest is written.
+        """
+        raise NotImplementedError(f"get_manifest_artifacts hook not implemented for {self.name}")
+
+
+@functools.lru_cache(maxsize=None)
+def _get_dbt_modules() -> Mapping[str, ModuleType]:
+    # This is an expensive function, especially in the context of testing, when
+    # it is called repeatedly, so we break it out and cache the result globally.
+    return {
+        name: importlib.import_module(name)
+        for _, name, _ in pkgutil.iter_modules()
+        if name.startswith(PluginManager.PLUGIN_MODULE_PREFIX)
+    }
+
+
+_MODULES_CACHE = None
+
+
+class PluginManager:
+    PLUGIN_MODULE_PREFIX = "dbt_"
+    PLUGIN_ATTR_NAME = "plugins"
+
+    def __init__(self, plugins: List[dbtPlugin]) -> None:
+        self._plugins = plugins
+        self._valid_hook_names = set()
+        # default hook implementations from dbtPlugin
+        for hook_name in dir(dbtPlugin):
+            if not hook_name.startswith("_"):
+                self._valid_hook_names.add(hook_name)
+
+        self.hooks: Dict[str, List[Callable]] = {}
+        for plugin in self._plugins:
+            for hook_name in dir(plugin):
+                hook = getattr(plugin, hook_name)
+                if (
+                    callable(hook)
+                    and hasattr(hook, "is_dbt_hook")
+                    and hook_name in self._valid_hook_names
+                ):
+                    if hook_name in self.hooks:
+                        self.hooks[hook_name].append(hook)
+                    else:
+                        self.hooks[hook_name] = [hook]
+
+    @classmethod
+    def from_modules(cls, project_name: str) -> "PluginManager":
+
+        if test_caching_enabled():
+            global _MODULES_CACHE
+            if _MODULES_CACHE is None:
+                discovered_dbt_modules = cls.get_prefixed_modules()
+                _MODULES_CACHE = discovered_dbt_modules
+            else:
+                discovered_dbt_modules = _MODULES_CACHE
+        else:
+            discovered_dbt_modules = cls.get_prefixed_modules()
+
+        plugins = []
+        for name, module in discovered_dbt_modules.items():
+            if hasattr(module, cls.PLUGIN_ATTR_NAME):
+                available_plugins = getattr(module, cls.PLUGIN_ATTR_NAME, [])
+                for plugin_cls in available_plugins:
+                    assert issubclass(
+                        plugin_cls, dbtPlugin
+                    ), f"'plugin' in {name} must be subclass of dbtPlugin"
+                    plugin = plugin_cls(project_name=project_name)
+                    plugins.append(plugin)
+        return cls(plugins=plugins)
+
+    @classmethod
+    def get_prefixed_modules(cls):
+        return {
+            name: importlib.import_module(name)
+            for _, name, _ in pkgutil.iter_modules()
+            if name.startswith(cls.PLUGIN_MODULE_PREFIX)
+        }
+
+    def get_manifest_artifacts(self, manifest: Manifest) -> PluginArtifacts:
+        all_plugin_artifacts = {}
+        for hook_method in self.hooks.get("get_manifest_artifacts", []):
+            plugin_artifacts = hook_method(manifest)
+            all_plugin_artifacts.update(plugin_artifacts)
+        return all_plugin_artifacts
+
+    def get_nodes(self) -> PluginNodes:
+        all_plugin_nodes = PluginNodes()
+        for hook_method in self.hooks.get("get_nodes", []):
+            plugin_nodes = hook_method()
+            dbt.tracking.track_plugin_get_nodes(
+                {
+                    "plugin_name": hook_method.__self__.name,  # type: ignore
+                    "num_model_nodes": len(plugin_nodes.models),
+                    "num_model_packages": len(
+                        {model.package_name for model in plugin_nodes.models.values()}
+                    ),
+                }
+            )
+            all_plugin_nodes.update(plugin_nodes)
+        return all_plugin_nodes

dbt/plugins/manifest.py

@@ -0,0 +1,21 @@
+from dataclasses import dataclass, field
+from typing import Dict
+
+from dbt.artifacts.resources import NodeVersion  # noqa
+
+# all these are just exports, they need "noqa" so flake8 will not complain.
+from dbt.contracts.graph.manifest import Manifest  # noqa
+from dbt.contracts.graph.node_args import ModelNodeArgs
+from dbt.graph.graph import UniqueId  # noqa
+from dbt.node_types import AccessType, NodeType  # noqa
+
+
+@dataclass
+class PluginNodes:
+    models: Dict[str, ModelNodeArgs] = field(default_factory=dict)
+
+    def add_model(self, model_args: ModelNodeArgs) -> None:
+        self.models[model_args.unique_id] = model_args
+
+    def update(self, other: "PluginNodes") -> None:
+        self.models.update(other.models)

dbt/profiler.py

@@ -0,0 +1,20 @@
+from contextlib import contextmanager
+from cProfile import Profile
+from pstats import Stats
+from typing import Any, Generator
+
+
+@contextmanager
+def profiler(enable: bool, outfile: str) -> Generator[Any, None, None]:
+    try:
+        if enable:
+            profiler = Profile()
+            profiler.enable()
+
+        yield
+    finally:
+        if enable:
+            profiler.disable()
+            stats = Stats(profiler)
+            stats.sort_stats("tottime")
+            stats.dump_stats(str(outfile))

dbt/py.typed

@@ -0,0 +1 @@
+# dummy file, our types are defined inline

dbt/query_analyzer.py

@@ -0,0 +1,410 @@
+"""
+Query Analyzer for Execution Routing
+
+This module analyzes compiled SQL queries to determine the optimal execution strategy:
+- Pushdown: All sources from same connection → execute on source database
+- Federated: Sources from multiple connections → use compute layer
+
+The analyzer respects user configuration overrides while providing intelligent defaults.
+"""
+
+import re
+from dataclasses import dataclass
+from typing import Dict, List, Optional, Set
+from dbt.contracts.graph.manifest import Manifest
+from dbt.contracts.graph.nodes import ManifestNode, SourceDefinition
+from dbt_common.exceptions import DbtRuntimeError
+
+
+@dataclass
+class QueryAnalysisResult:
+    """
+    Result of query analysis.
+
+    Contains information about source connections and recommended execution strategy.
+    """
+
+    # Set of connection names used by this query
+    source_connections: Set[str]
+
+    # Set of source unique_ids referenced
+    source_refs: Set[str]
+
+    # Recommended execution strategy
+    strategy: str  # "pushdown" or "federated"
+
+    # Primary connection (for pushdown) or None (for federated)
+    primary_connection: Optional[str]
+
+    # Reason for the chosen strategy
+    reason: str
+
+    # User override applied (if any)
+    user_override: Optional[str] = None
+
+    @property
+    def is_pushdown(self) -> bool:
+        """Check if pushdown strategy is recommended."""
+        return self.strategy == "pushdown"
+
+    @property
+    def is_federated(self) -> bool:
+        """Check if federated strategy is recommended."""
+        return self.strategy == "federated"
+
+    @property
+    def is_homogeneous(self) -> bool:
+        """Check if all sources are from the same connection."""
+        return len(self.source_connections) <= 1
+
+
+class QueryAnalyzer:
+    """
+    Analyzes compiled SQL queries to determine execution strategy.
+
+    The analyzer:
+    1. Identifies all source() and ref() references in the query
+    2. Determines which connections are involved
+    3. Recommends pushdown (homogeneous) or federated (heterogeneous) execution
+    4. Respects user configuration overrides
+    """
+
+    def __init__(self, manifest: Manifest):
+        """
+        Initialize query analyzer.
+
+        :param manifest: The parsed dbt manifest with all nodes and sources
+        """
+        self.manifest = manifest
+        # Cache for source connection mappings
+        self._source_connection_cache: Dict[str, str] = {}
+
+    def analyze(
+        self,
+        node: ManifestNode,
+        user_compute_override: Optional[str] = None,
+        target_connection: Optional[str] = None
+    ) -> QueryAnalysisResult:
+        """
+        Analyze a compiled node to determine execution strategy.
+
+        :param node: The compiled ManifestNode to analyze
+        :param user_compute_override: User's compute config (if specified)
+        :param target_connection: Target connection for materialization (if different from source)
+        :returns: QueryAnalysisResult with execution strategy
+        """
+        # Skip analysis for seeds - they don't have sources and accessing
+        # node.sources property on SeedNode raises an error
+        if node.resource_type == "seed":
+            return QueryAnalysisResult(
+                source_connections=set(),
+                source_refs=set(),
+                strategy="pushdown",
+                primary_connection=None,
+                reason="Seed node - no source analysis needed",
+                user_override=None
+            )
+
+        # Extract source and ref dependencies
+        source_refs = self._extract_source_references(node)
+
+        # Map sources to their connections
+        source_connections = self._map_sources_to_connections(source_refs)
+
+        # DVT v0.51.6: Determine strategy based on DVT Rules
+        #
+        # Rule 1 (Primary Directive): Pushdown whenever model and ALL inputs are in same Target
+        # Rule 1.1: Federation ONLY when model requires inputs from a Target different than its own
+        # Rule 1.5: Compute engine settings are IGNORED for Pushdown-eligible models
+        # Rule 2.2: CLI --target forces all models to that target; sources NOT in target trigger Federation
+        #
+        # Decision Logic:
+        # 1. If no direct sources → Pushdown (refs resolve to target where upstream is materialized)
+        # 2. If all sources are in target_connection → Pushdown
+        # 3. If any source is NOT in target_connection → Federation
+        # 4. user_compute_override does NOT change this decision (Rule 1.5)
+
+        if len(source_connections) == 0:
+            # No direct source() dependencies
+            # Models with only refs → Pushdown to target (refs are already materialized there)
+            strategy = "pushdown"
+            has_refs = hasattr(node, 'depends_on') and node.depends_on and node.depends_on.nodes
+            if has_refs and target_connection:
+                reason = f"Model refs materialized tables - pushdown to target '{target_connection}'"
+                primary_connection = target_connection
+            else:
+                reason = "No source dependencies - pushdown to default target"
+                primary_connection = target_connection
+            # Rule 1.5: Compute override is noted but doesn't change strategy
+            user_override = user_compute_override if user_compute_override else None
+
+        elif len(source_connections) == 1:
+            single_source_conn = list(source_connections)[0]
+
+            # Rule 2.2: Check if source is in the target connection
+            if target_connection and single_source_conn != target_connection:
+                # Source NOT in target → Federation required (Rule 1.1)
+                strategy = "federated"
+                reason = f"Cross-target: source '{single_source_conn}' != target '{target_connection}' - federation required"
+                primary_connection = None
+                user_override = user_compute_override
+            else:
+                # Source IS in target (or no target override) → Pushdown (Rule 1)
+                strategy = "pushdown"
+                primary_connection = single_source_conn
+                reason = f"All sources in target '{primary_connection}' - pushdown"
+                # Rule 1.5: Compute override ignored for pushdown-eligible models
+                user_override = None
+
+        else:
+            # Multiple source connections → Always requires Federation
+            strategy = "federated"
+            reason = f"Sources span {len(source_connections)} connections: {sorted(source_connections)} - federation required"
+            primary_connection = None
+            user_override = user_compute_override
+
+        return QueryAnalysisResult(
+            source_connections=source_connections,
+            source_refs=source_refs,
+            strategy=strategy,
+            primary_connection=primary_connection,
+            reason=reason,
+            user_override=user_override
+        )
+
+    def _extract_source_references(self, node: ManifestNode) -> Set[str]:
+        """
+        Extract all source unique_ids DIRECTLY referenced by this node.
+
+        DVT v0.51.5: Only direct source() references are considered.
+        ref() dependencies are NOT traced to their underlying sources because:
+        - ref() resolves to the target database where upstream models are materialized
+        - If model_a reads from postgres and is materialized to databricks,
+          then model_b which refs model_a should use databricks (pushdown),
+          NOT trace back to postgres (which would incorrectly force federation)
+
+        :param node: The node to analyze
+        :returns: Set of source unique_ids (direct references only)
+        """
+        source_refs = set()
+
+        # Direct source dependencies ONLY
+        # ref() dependencies resolve to target database, not original sources
+        if hasattr(node, 'sources') and node.sources:
+            # node.sources can be a list or set - convert to list for safety
+            sources = node.sources if isinstance(node.sources, (list, tuple, set)) else []
+            for source in sources:
+                # source can be either:
+                # 1. A string (full unique_id): "source.package.source_name.table_name"
+                # 2. A list/tuple: ["source_name", "table_name"]
+                if isinstance(source, str):
+                    # Full unique_id - use as-is
+                    source_refs.add(source)
+                elif isinstance(source, (list, tuple)) and len(source) == 2:
+                    # Tuple format: ["source_name", "table_name"]
+                    # Need to construct full unique_id: "source.{package}.{source_name}.{table_name}"
+                    source_name, table_name = source
+                    # Build unique_id using node's package_name
+                    package_name = node.package_name if hasattr(node, 'package_name') else self.root_project.project_name
+                    unique_id = f"source.{package_name}.{source_name}.{table_name}"
+                    source_refs.add(unique_id)
+
+        # DVT v0.51.5: DO NOT trace ref() dependencies to their underlying sources
+        # ref() resolves to target database, where upstream models are already materialized
+        # Only direct source() references determine federation vs pushdown
+
+        return source_refs
+
+    def _trace_node_to_sources(self, node_id: str, visited: Optional[Set[str]] = None) -> Set[str]:
+        """
+        Recursively trace a node's dependencies to find all underlying sources.
+
+        :param node_id: Unique ID of the node to trace
+        :param visited: Set of already-visited nodes (for cycle detection)
+        :returns: Set of source unique_ids
+        """
+        if visited is None:
+            visited = set()
+
+        if node_id in visited:
+            return set()
+
+        visited.add(node_id)
+        sources = set()
+
+        # Check if this is a source
+        if node_id.startswith('source.'):
+            sources.add(node_id)
+            return sources
+
+        # Get the node from manifest
+        node = self.manifest.nodes.get(node_id)
+        if not node:
+            # Node not found (could be disabled or external)
+            return sources
+
+        # Skip seeds - they don't have source dependencies and accessing
+        # node.sources property raises an error
+        if hasattr(node, 'resource_type') and node.resource_type == 'seed':
+            # Seeds have no upstream sources to trace
+            return sources
+
+        # Add direct source dependencies
+        if hasattr(node, 'sources') and node.sources:
+            # Handle both list and set types safely
+            node_sources = node.sources if isinstance(node.sources, (list, tuple, set)) else []
+            for source in node_sources:
+                if isinstance(source, str):
+                    sources.add(source)
+
+        # Recursively trace node dependencies
+        if hasattr(node, 'depends_on') and node.depends_on:
+            nodes = node.depends_on.nodes if isinstance(node.depends_on.nodes, (list, tuple, set)) else []
+            for dep_id in nodes:
+                if isinstance(dep_id, str):
+                    dep_sources = self._trace_node_to_sources(dep_id, visited)
+                    sources.update(dep_sources)
+
+        return sources
+
+    def _map_sources_to_connections(self, source_refs: Set[str]) -> Set[str]:
+        """
+        Map source unique_ids to their connection names.
+
+        :param source_refs: Set of source unique_ids
+        :returns: Set of connection names
+        """
+        connections = set()
+
+        for source_id in source_refs:
+            connection = self._get_source_connection(source_id)
+            if connection:
+                connections.add(connection)
+
+        return connections
+
+    def _get_source_connection(self, source_id: str) -> Optional[str]:
+        """
+        Get the connection name for a source.
+
+        Uses caching for performance. Checks multiple locations for backward
+        compatibility with different source definition styles:
+        1. source.connection (direct attribute)
+        2. source.meta.connection (meta dict)
+        3. source.source_meta.connection (source_meta dict)
+
+        :param source_id: Source unique_id
+        :returns: Connection name or None if not specified
+        """
+        # Check cache
+        if source_id in self._source_connection_cache:
+            return self._source_connection_cache[source_id]
+
+        # Look up source in manifest
+        source = self.manifest.sources.get(source_id)
+        if not source:
+            return None
+
+        # Get connection - check multiple locations for backward compatibility
+        connection = None
+
+        # 1. Direct connection attribute (preferred)
+        if hasattr(source, 'connection') and source.connection:
+            connection = source.connection
+
+        # 2. meta.connection (fallback)
+        elif hasattr(source, 'meta') and isinstance(source.meta, dict):
+            connection = source.meta.get('connection')
+
+        # 3. source_meta.connection (legacy fallback)
+        elif hasattr(source, 'source_meta') and isinstance(source.source_meta, dict):
+            connection = source.source_meta.get('connection')
+
+        # Cache and return
+        self._source_connection_cache[source_id] = connection
+        return connection
+
+    def get_execution_summary(self, node: ManifestNode) -> str:
+        """
+        Get a human-readable summary of the execution strategy for a node.
+
+        Useful for logging and debugging.
+
+        :param node: The node to analyze
+        :returns: Summary string
+        """
+        result = self.analyze(node)
+
+        summary_parts = [
+            f"Node: {node.unique_id}",
+            f"Strategy: {result.strategy.upper()}",
+            f"Reason: {result.reason}",
+            f"Source Connections: {sorted(result.source_connections) if result.source_connections else 'None'}",
+            f"Source Count: {len(result.source_refs)}",
+        ]
+
+        if result.primary_connection:
+            summary_parts.append(f"Execution Connection: {result.primary_connection}")
+
+        if result.user_override:
+            summary_parts.append(f"User Override: {result.user_override}")
+
+        return "\n".join(summary_parts)
+
+    def analyze_batch(
+        self,
+        nodes: List[ManifestNode]
+    ) -> Dict[str, QueryAnalysisResult]:
+        """
+        Analyze multiple nodes in batch.
+
+        More efficient than analyzing one at a time due to caching.
+
+        :param nodes: List of nodes to analyze
+        :returns: Dict mapping node unique_id to QueryAnalysisResult
+        """
+        results = {}
+
+        for node in nodes:
+            result = self.analyze(node)
+            results[node.unique_id] = result
+
+        return results
+
+    def get_federated_nodes(
+        self,
+        nodes: List[ManifestNode]
+    ) -> List[ManifestNode]:
+        """
+        Filter nodes that require federated execution.
+
+        :param nodes: List of nodes to filter
+        :returns: List of nodes requiring federated execution
+        """
+        federated = []
+
+        for node in nodes:
+            result = self.analyze(node)
+            if result.is_federated:
+                federated.append(node)
+
+        return federated
+
+    def get_pushdown_nodes(
+        self,
+        nodes: List[ManifestNode]
+    ) -> List[ManifestNode]:
+        """
+        Filter nodes eligible for pushdown execution.
+
+        :param nodes: List of nodes to filter
+        :returns: List of nodes eligible for pushdown
+        """
+        pushdown = []
+
+        for node in nodes:
+            result = self.analyze(node)
+            if result.is_pushdown:
+                pushdown.append(node)
+
+        return pushdown

dbt/runners/__init__.py

@@ -0,0 +1,2 @@
+from .exposure_runner import ExposureRunner
+from .saved_query_runner import SavedQueryRunner

dbt/runners/exposure_runner.py

@@ -0,0 +1,7 @@
+from dbt.runners.no_op_runner import NoOpRunner
+
+
+class ExposureRunner(NoOpRunner):
+    @property
+    def description(self) -> str:
+        return f"exposure {self.node.name}"

dbt/runners/no_op_runner.py

@@ -0,0 +1,45 @@
+import threading
+
+from dbt.artifacts.schemas.results import RunStatus
+from dbt.artifacts.schemas.run import RunResult
+from dbt.contracts.graph.manifest import Manifest
+from dbt.events.types import LogNodeNoOpResult
+from dbt.task.base import BaseRunner
+from dbt_common.events.functions import fire_event
+
+
+class NoOpRunner(BaseRunner):
+    @property
+    def description(self) -> str:
+        raise NotImplementedError("description not implemented")
+
+    def before_execute(self) -> None:
+        pass
+
+    def compile(self, manifest: Manifest):
+        return self.node
+
+    def after_execute(self, result) -> None:
+        fire_event(
+            LogNodeNoOpResult(
+                description=self.description,
+                index=self.node_index,
+                total=self.num_nodes,
+                node_info=self.node.node_info,
+            )
+        )
+
+    def execute(self, compiled_node, manifest):
+        # no-op
+        return RunResult(
+            node=compiled_node,
+            status=RunStatus.NoOp,
+            timing=[],
+            thread_id=threading.current_thread().name,
+            execution_time=0,
+            message="NO-OP",
+            adapter_response={},
+            failures=0,
+            batch_results=None,
+            agate_table=None,
+        )

dbt/runners/saved_query_runner.py

@@ -0,0 +1,7 @@
+from dbt.runners.no_op_runner import NoOpRunner
+
+
+class SavedQueryRunner(NoOpRunner):
+    @property
+    def description(self) -> str:
+        return f"saved query {self.node.name}"