|
| 1 | +/** |
| 2 | + * Tests for dbt lineage helper functions: findModel, detectDialect, |
| 3 | + * buildSchemaContext, and the top-level dbtLineage() error paths. |
| 4 | + * |
| 5 | + * These pure functions parse manifest data and build schema contexts |
| 6 | + * for column-level lineage analysis. Zero tests existed previously. |
| 7 | + * A bug in findModel or buildSchemaContext causes lineage to silently |
| 8 | + * return empty results, which users see as "no lineage available". |
| 9 | + */ |
| 10 | + |
| 11 | +import { describe, test, expect, afterEach } from "bun:test" |
| 12 | +import { dbtLineage } from "../../src/altimate/native/dbt/lineage" |
| 13 | +import { writeFileSync, mkdtempSync, rmSync } from "fs" |
| 14 | +import { tmpdir } from "os" |
| 15 | +import { join } from "path" |
| 16 | + |
| 17 | +// --------------------------------------------------------------------------- |
| 18 | +// Helpers |
| 19 | +// --------------------------------------------------------------------------- |
| 20 | + |
| 21 | +let tmpDirs: string[] = [] |
| 22 | + |
| 23 | +function makeTmpDir(): string { |
| 24 | + const dir = mkdtempSync(join(tmpdir(), "dbt-lineage-test-")) |
| 25 | + tmpDirs.push(dir) |
| 26 | + return dir |
| 27 | +} |
| 28 | + |
| 29 | +function writeManifest(dir: string, manifest: Record<string, any>): string { |
| 30 | + const manifestPath = join(dir, "manifest.json") |
| 31 | + writeFileSync(manifestPath, JSON.stringify(manifest)) |
| 32 | + return manifestPath |
| 33 | +} |
| 34 | + |
| 35 | +afterEach(() => { |
| 36 | + for (const dir of tmpDirs) { |
| 37 | + rmSync(dir, { recursive: true, force: true }) |
| 38 | + } |
| 39 | + tmpDirs = [] |
| 40 | +}) |
| 41 | + |
| 42 | +// --------------------------------------------------------------------------- |
| 43 | +// Minimal manifest fixtures |
| 44 | +// --------------------------------------------------------------------------- |
| 45 | + |
| 46 | +const BASE_MANIFEST = { |
| 47 | + metadata: { adapter_type: "snowflake" }, |
| 48 | + nodes: { |
| 49 | + "model.proj.orders": { |
| 50 | + resource_type: "model", |
| 51 | + name: "orders", |
| 52 | + schema: "public", |
| 53 | + database: "analytics", |
| 54 | + config: { materialized: "table" }, |
| 55 | + compiled_code: "SELECT c.id, c.name FROM customers c", |
| 56 | + depends_on: { nodes: ["source.proj.raw.customers"] }, |
| 57 | + columns: { |
| 58 | + id: { name: "id", data_type: "INTEGER" }, |
| 59 | + name: { name: "name", data_type: "VARCHAR" }, |
| 60 | + }, |
| 61 | + }, |
| 62 | + "model.proj.revenue": { |
| 63 | + resource_type: "model", |
| 64 | + name: "revenue", |
| 65 | + compiled_code: "SELECT SUM(amount) AS total FROM orders", |
| 66 | + depends_on: { nodes: ["model.proj.orders"] }, |
| 67 | + columns: {}, |
| 68 | + }, |
| 69 | + "test.proj.not_null": { |
| 70 | + resource_type: "test", |
| 71 | + name: "not_null", |
| 72 | + }, |
| 73 | + }, |
| 74 | + sources: { |
| 75 | + "source.proj.raw.customers": { |
| 76 | + name: "customers", |
| 77 | + source_name: "raw", |
| 78 | + schema: "raw_data", |
| 79 | + database: "analytics", |
| 80 | + columns: { |
| 81 | + id: { name: "id", data_type: "INTEGER" }, |
| 82 | + name: { name: "name", data_type: "VARCHAR" }, |
| 83 | + email: { name: "email", data_type: "VARCHAR" }, |
| 84 | + }, |
| 85 | + }, |
| 86 | + }, |
| 87 | +} |
| 88 | + |
| 89 | +// --------------------------------------------------------------------------- |
| 90 | +// 1. Model lookup (findModel) |
| 91 | +// --------------------------------------------------------------------------- |
| 92 | + |
| 93 | +describe("dbtLineage: model lookup", () => { |
| 94 | + test("finds model by unique_id", () => { |
| 95 | + const dir = makeTmpDir() |
| 96 | + const manifestPath = writeManifest(dir, BASE_MANIFEST) |
| 97 | + |
| 98 | + const result = dbtLineage({ |
| 99 | + manifest_path: manifestPath, |
| 100 | + model: "model.proj.orders", |
| 101 | + }) |
| 102 | + |
| 103 | + expect(result.model_name).toBe("orders") |
| 104 | + expect(result.model_unique_id).toBe("model.proj.orders") |
| 105 | + expect(result.compiled_sql).toContain("SELECT") |
| 106 | + }) |
| 107 | + |
| 108 | + test("finds model by short name", () => { |
| 109 | + const dir = makeTmpDir() |
| 110 | + const manifestPath = writeManifest(dir, BASE_MANIFEST) |
| 111 | + |
| 112 | + const result = dbtLineage({ |
| 113 | + manifest_path: manifestPath, |
| 114 | + model: "orders", |
| 115 | + }) |
| 116 | + |
| 117 | + expect(result.model_name).toBe("orders") |
| 118 | + expect(result.model_unique_id).toBe("model.proj.orders") |
| 119 | + }) |
| 120 | + |
| 121 | + test("returns low confidence when model not found", () => { |
| 122 | + const dir = makeTmpDir() |
| 123 | + const manifestPath = writeManifest(dir, BASE_MANIFEST) |
| 124 | + |
| 125 | + const result = dbtLineage({ |
| 126 | + manifest_path: manifestPath, |
| 127 | + model: "nonexistent_model", |
| 128 | + }) |
| 129 | + |
| 130 | + expect(result.confidence).toBe("low") |
| 131 | + expect(result.confidence_factors).toContain("Model 'nonexistent_model' not found in manifest") |
| 132 | + }) |
| 133 | + |
| 134 | + test("does not match test or seed nodes by name", () => { |
| 135 | + const dir = makeTmpDir() |
| 136 | + const manifestPath = writeManifest(dir, BASE_MANIFEST) |
| 137 | + |
| 138 | + const result = dbtLineage({ |
| 139 | + manifest_path: manifestPath, |
| 140 | + model: "not_null", |
| 141 | + }) |
| 142 | + |
| 143 | + // "not_null" is a test node, not a model — should not be found |
| 144 | + expect(result.confidence).toBe("low") |
| 145 | + expect(result.confidence_factors[0]).toContain("not found in manifest") |
| 146 | + }) |
| 147 | +}) |
| 148 | + |
| 149 | +// --------------------------------------------------------------------------- |
| 150 | +// 2. Dialect detection (detectDialect) |
| 151 | +// --------------------------------------------------------------------------- |
| 152 | + |
| 153 | +describe("dbtLineage: dialect detection", () => { |
| 154 | + test("detects dialect from manifest metadata.adapter_type", () => { |
| 155 | + const dir = makeTmpDir() |
| 156 | + const manifest = { |
| 157 | + ...BASE_MANIFEST, |
| 158 | + metadata: { adapter_type: "bigquery" }, |
| 159 | + } |
| 160 | + const manifestPath = writeManifest(dir, manifest) |
| 161 | + |
| 162 | + const result = dbtLineage({ |
| 163 | + manifest_path: manifestPath, |
| 164 | + model: "orders", |
| 165 | + }) |
| 166 | + |
| 167 | + // We can't directly check dialect, but the result shouldn't error |
| 168 | + // due to dialect mismatch. The model has compiled_code, so confidence |
| 169 | + // should be high if lineage succeeds or reflect the actual error. |
| 170 | + expect(result.model_name).toBe("orders") |
| 171 | + }) |
| 172 | + |
| 173 | + test("explicit dialect param overrides auto-detection", () => { |
| 174 | + const dir = makeTmpDir() |
| 175 | + const manifestPath = writeManifest(dir, BASE_MANIFEST) |
| 176 | + |
| 177 | + const result = dbtLineage({ |
| 178 | + manifest_path: manifestPath, |
| 179 | + model: "orders", |
| 180 | + dialect: "postgres", |
| 181 | + }) |
| 182 | + |
| 183 | + // Should not throw regardless of dialect choice |
| 184 | + expect(result.model_name).toBe("orders") |
| 185 | + }) |
| 186 | + |
| 187 | + test("defaults to snowflake when adapter_type is missing", () => { |
| 188 | + const dir = makeTmpDir() |
| 189 | + const manifest = { |
| 190 | + ...BASE_MANIFEST, |
| 191 | + metadata: {}, |
| 192 | + } |
| 193 | + const manifestPath = writeManifest(dir, manifest) |
| 194 | + |
| 195 | + const result = dbtLineage({ |
| 196 | + manifest_path: manifestPath, |
| 197 | + model: "orders", |
| 198 | + }) |
| 199 | + |
| 200 | + // Should not throw — defaults to snowflake |
| 201 | + expect(result.model_name).toBe("orders") |
| 202 | + }) |
| 203 | +}) |
| 204 | + |
| 205 | +// --------------------------------------------------------------------------- |
| 206 | +// 3. Schema context building (buildSchemaContext) |
| 207 | +// --------------------------------------------------------------------------- |
| 208 | + |
| 209 | +describe("dbtLineage: schema context from upstream deps", () => { |
| 210 | + test("builds context from source with columns", () => { |
| 211 | + const dir = makeTmpDir() |
| 212 | + const manifestPath = writeManifest(dir, BASE_MANIFEST) |
| 213 | + |
| 214 | + const result = dbtLineage({ |
| 215 | + manifest_path: manifestPath, |
| 216 | + model: "orders", |
| 217 | + }) |
| 218 | + |
| 219 | + // The orders model depends on source.proj.raw.customers which has columns. |
| 220 | + // If schema context was built correctly, lineage should have non-empty output. |
| 221 | + expect(result.model_name).toBe("orders") |
| 222 | + // compiled_sql should be present |
| 223 | + expect(result.compiled_sql).toBeDefined() |
| 224 | + expect(result.compiled_sql).toContain("SELECT") |
| 225 | + }) |
| 226 | + |
| 227 | + test("handles model with no upstream columns gracefully", () => { |
| 228 | + const dir = makeTmpDir() |
| 229 | + // Revenue depends on orders, but orders has columns — so context should build. |
| 230 | + // Create a model that depends on a node with no columns. |
| 231 | + const manifest = { |
| 232 | + ...BASE_MANIFEST, |
| 233 | + nodes: { |
| 234 | + ...BASE_MANIFEST.nodes, |
| 235 | + "model.proj.bare": { |
| 236 | + resource_type: "model", |
| 237 | + name: "bare", |
| 238 | + compiled_code: "SELECT 1 AS val", |
| 239 | + depends_on: { nodes: ["model.proj.no_cols"] }, |
| 240 | + columns: {}, |
| 241 | + }, |
| 242 | + "model.proj.no_cols": { |
| 243 | + resource_type: "model", |
| 244 | + name: "no_cols", |
| 245 | + compiled_code: "SELECT 1", |
| 246 | + depends_on: { nodes: [] }, |
| 247 | + columns: {}, |
| 248 | + }, |
| 249 | + }, |
| 250 | + } |
| 251 | + const manifestPath = writeManifest(dir, manifest) |
| 252 | + |
| 253 | + const result = dbtLineage({ |
| 254 | + manifest_path: manifestPath, |
| 255 | + model: "bare", |
| 256 | + }) |
| 257 | + |
| 258 | + // Should not crash — just returns with whatever lineage can determine |
| 259 | + expect(result.model_name).toBe("bare") |
| 260 | + expect(result.compiled_sql).toBe("SELECT 1 AS val") |
| 261 | + }) |
| 262 | +}) |
| 263 | + |
| 264 | +// --------------------------------------------------------------------------- |
| 265 | +// 4. Error paths |
| 266 | +// --------------------------------------------------------------------------- |
| 267 | + |
| 268 | +describe("dbtLineage: error handling", () => { |
| 269 | + test("returns low confidence for non-existent manifest", () => { |
| 270 | + const result = dbtLineage({ |
| 271 | + manifest_path: "/tmp/definitely-not-a-manifest.json", |
| 272 | + model: "orders", |
| 273 | + }) |
| 274 | + |
| 275 | + expect(result.confidence).toBe("low") |
| 276 | + expect(result.confidence_factors).toContain("Manifest file not found") |
| 277 | + expect(result.raw_lineage).toEqual({}) |
| 278 | + }) |
| 279 | + |
| 280 | + test("returns low confidence for invalid JSON manifest", () => { |
| 281 | + const dir = makeTmpDir() |
| 282 | + const manifestPath = join(dir, "manifest.json") |
| 283 | + writeFileSync(manifestPath, "not valid json {{{") |
| 284 | + |
| 285 | + const result = dbtLineage({ |
| 286 | + manifest_path: manifestPath, |
| 287 | + model: "orders", |
| 288 | + }) |
| 289 | + |
| 290 | + expect(result.confidence).toBe("low") |
| 291 | + expect(result.confidence_factors[0]).toContain("Failed to parse manifest") |
| 292 | + }) |
| 293 | + |
| 294 | + test("returns low confidence when model has no compiled SQL", () => { |
| 295 | + const dir = makeTmpDir() |
| 296 | + const manifest = { |
| 297 | + nodes: { |
| 298 | + "model.proj.uncompiled": { |
| 299 | + resource_type: "model", |
| 300 | + name: "uncompiled", |
| 301 | + // No compiled_code or compiled_sql |
| 302 | + depends_on: { nodes: [] }, |
| 303 | + columns: {}, |
| 304 | + }, |
| 305 | + }, |
| 306 | + sources: {}, |
| 307 | + } |
| 308 | + const manifestPath = writeManifest(dir, manifest) |
| 309 | + |
| 310 | + const result = dbtLineage({ |
| 311 | + manifest_path: manifestPath, |
| 312 | + model: "uncompiled", |
| 313 | + }) |
| 314 | + |
| 315 | + expect(result.confidence).toBe("low") |
| 316 | + expect(result.confidence_factors).toContain("No compiled SQL found — run `dbt compile` first") |
| 317 | + }) |
| 318 | + |
| 319 | + test("handles manifest with no nodes key at all", () => { |
| 320 | + const dir = makeTmpDir() |
| 321 | + const manifestPath = writeManifest(dir, { metadata: {} }) |
| 322 | + |
| 323 | + const result = dbtLineage({ |
| 324 | + manifest_path: manifestPath, |
| 325 | + model: "orders", |
| 326 | + }) |
| 327 | + |
| 328 | + expect(result.confidence).toBe("low") |
| 329 | + }) |
| 330 | +}) |
0 commit comments