feast/sdk/python/feast/dbt/parser.py at master · feast-dev/feast · GitHub

History

297 lines (243 loc) · 9.78 KB

Raw

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

293

294

295

296

297

"""

dbt manifest parser for Feast integration.

This module provides functionality to parse dbt manifest.json files and extract

model metadata for generating Feast FeatureViews.

Uses dbt-artifacts-parser for typed parsing of manifest versions v1-v12 (dbt 0.19 through 1.11+).

"""

import json

from copy import deepcopy

from dataclasses import dataclass, field

from pathlib import Path

from typing import Any, Dict, List, Optional

@dataclass

class DbtColumn:

"""Represents a column in a dbt model."""

name: str

description: str = ""

data_type: str = "STRING"

tags: List[str] = field(default_factory=list)

meta: Dict[str, Any] = field(default_factory=dict)

@dataclass

class DbtModel:

"""Represents a dbt model."""

name: str

unique_id: str

database: str

schema: str

alias: str

description: str = ""

columns: List[DbtColumn] = field(default_factory=list)

tags: List[str] = field(default_factory=list)

meta: Dict[str, Any] = field(default_factory=dict)

depends_on: List[str] = field(default_factory=list)

@property

def full_table_name(self) -> str:

"""Returns fully qualified table name (database.schema.table)."""

return f"{self.database}.{self.schema}.{self.alias}"

class DbtManifestParser:

"""

Parser for dbt manifest.json files using dbt-artifacts-parser.

Uses dbt-artifacts-parser for typed parsing of manifest versions v1-v12

(dbt versions 0.19 through 1.11+).

Example::

parser = DbtManifestParser("target/manifest.json")

parser.parse()

models = parser.get_models(tag_filter="feast")

for model in models:

print(f"Model: {model.name}, Columns: {len(model.columns)}")

Args:

manifest_path: Path to manifest.json file (typically target/manifest.json)

Raises:

FileNotFoundError: If manifest.json doesn't exist

ValueError: If manifest.json is invalid JSON

"""

def __init__(self, manifest_path: str):

"""

Initialize parser.

Args:

manifest_path: Path to manifest.json file

"""

self.manifest_path = Path(manifest_path)

self._raw_manifest: Optional[Dict[str, Any]] = None

self._parsed_manifest: Optional[Any] = None

@staticmethod

def _sanitize_supported_languages(manifest: Dict[str, Any]) -> Dict[str, Any]:

"""

Remove unsupported macro language values before typed parsing.

dbt may emit values like "javascript" in `macros.*.supported_languages`,

while dbt-artifacts-parser currently accepts only "python" and "sql".

Since Feast only needs model metadata, dropping unknown values is safe.

"""

manifest_copy = deepcopy(manifest)

macros = manifest_copy.get("macros", {})

if not isinstance(macros, dict):

return manifest_copy

for macro in macros.values():

if not isinstance(macro, dict):

continue

supported_languages = macro.get("supported_languages")

if isinstance(supported_languages, list):

macro["supported_languages"] = [

language

for language in supported_languages

if language in {"python", "sql"}

]

return manifest_copy

def parse(self) -> None:

"""

Load and parse the manifest.json file using dbt-artifacts-parser.

Raises:

FileNotFoundError: If manifest.json doesn't exist

ValueError: If manifest.json is invalid JSON

ImportError: If dbt-artifacts-parser is not installed

"""

if not self.manifest_path.exists():

raise FileNotFoundError(

f"dbt manifest not found at {self.manifest_path}.\n"

f"Run 'dbt compile' or 'dbt run' first.\n"

f"Expected path: <dbt_project>/target/manifest.json"

)

try:

with open(self.manifest_path, "r") as f:

self._raw_manifest = json.load(f)

except json.JSONDecodeError as e:

raise ValueError(

f"Invalid JSON in manifest: {e}\nTry: dbt clean && dbt compile"

)

# Parse using dbt-artifacts-parser for typed access

try:

from dbt_artifacts_parser.parser import parse_manifest

assert self._raw_manifest is not None

try:

self._parsed_manifest = parse_manifest(manifest=self._raw_manifest)

except Exception as e:

if "supported_languages" not in str(e):

raise

sanitized_manifest = self._sanitize_supported_languages(

self._raw_manifest

)

self._parsed_manifest = parse_manifest(manifest=sanitized_manifest)

except ImportError:

raise ImportError(

"dbt-artifacts-parser is required for dbt integration.\n"

"Install with: pip install 'feast[dbt]' or pip install dbt-artifacts-parser"

)

def _extract_column_from_node(self, col_name: str, col_data: Any) -> DbtColumn:

"""Extract column info from a parsed node column."""

return DbtColumn(

name=col_name,

description=getattr(col_data, "description", "") or "",

data_type=getattr(col_data, "data_type", "STRING") or "STRING",

tags=list(getattr(col_data, "tags", []) or []),

meta=dict(getattr(col_data, "meta", {}) or {}),

)

def _extract_model_from_node(self, node_id: str, node: Any) -> Optional[DbtModel]:

"""Extract DbtModel from a parsed manifest node."""

# Check resource type

resource_type = getattr(node, "resource_type", None)

if resource_type is None:

if not node_id.startswith("model."):

return None

else:

resource_type_str = (

resource_type.value

if hasattr(resource_type, "value")

else str(resource_type)

)

if resource_type_str != "model":

return None

model_name = getattr(node, "name", "")

node_tags = list(getattr(node, "tags", []) or [])

node_columns = getattr(node, "columns", {}) or {}

depends_on = getattr(node, "depends_on", None)

if depends_on:

depends_on_nodes = list(getattr(depends_on, "nodes", []) or [])

else:

depends_on_nodes = []

# Extract columns

columns = [

self._extract_column_from_node(col_name, col_data)

for col_name, col_data in node_columns.items()

]

# Get schema - dbt-artifacts-parser uses schema_ to avoid Python keyword

schema = getattr(node, "schema_", "") or getattr(node, "schema", "") or ""

return DbtModel(

name=model_name,

unique_id=node_id,

database=getattr(node, "database", "") or "",

schema=schema,

alias=getattr(node, "alias", model_name) or model_name,

description=getattr(node, "description", "") or "",

columns=columns,

tags=node_tags,

meta=dict(getattr(node, "meta", {}) or {}),

depends_on=depends_on_nodes,

)

def get_models(

self,

model_names: Optional[List[str]] = None,

tag_filter: Optional[str] = None,

) -> List[DbtModel]:

"""

Extract dbt models from manifest.

Args:

model_names: Optional list of specific model names to extract

tag_filter: Optional tag to filter models by

Returns:

List of DbtModel objects

Example::

models = parser.get_models(model_names=["driver_stats"])

models = parser.get_models(tag_filter="feast")

"""

if self._parsed_manifest is None:

self.parse()

if self._parsed_manifest is None:

return []

models = []

nodes = getattr(self._parsed_manifest, "nodes", {}) or {}

for node_id, node in nodes.items():

# Only process models (not tests, seeds, snapshots, etc.)

if not node_id.startswith("model."):

continue

model = self._extract_model_from_node(node_id, node)

if model is None:

continue

# Filter by model names if specified

if model_names and model.name not in model_names:

continue

# Filter by tag if specified

if tag_filter and tag_filter not in model.tags:

continue

models.append(model)

return models

def get_model_by_name(self, model_name: str) -> Optional[DbtModel]:

"""

Get a specific model by name.

Args:

model_name: Name of the model to retrieve

Returns:

DbtModel if found, None otherwise

"""

models = self.get_models(model_names=[model_name])

return models[0] if models else None

@property

def dbt_version(self) -> Optional[str]:

"""Get dbt version from manifest metadata."""

if self._parsed_manifest is None:

return None

metadata = getattr(self._parsed_manifest, "metadata", None)

if metadata is None:

return None

return getattr(metadata, "dbt_version", None)

@property

def project_name(self) -> Optional[str]:

"""Get project name from manifest metadata."""

if self._parsed_manifest is None:

return None

metadata = getattr(self._parsed_manifest, "metadata", None)

if metadata is None:

return None

# project_name may not exist in all manifest versions

return getattr(metadata, "project_name", None) or getattr(

metadata, "project_id", None

)

feast/sdk/python/feast/dbt/parser.py at master · feast-dev/feast · GitHub

Uh oh!

Expand file tree

Breadcrumbs

parser.py

Latest commit

History

Breadcrumbs

parser.py

File metadata and controls