dataiku-sdk 3.6.1 → 3.8.0

package/dist/src/cli/commands/analysis.js

@@ -1,4 +1,4 @@
-import { requiredStringFlag, } from "../coerce.js";
+import { requiredJsonInput, requiredStringFlag, } from "../coerce.js";
 import { executionMode, } from "../flags.js";
 import { readIfExists, skipResult, } from "../output.js";
 import { commandUsage, withUsage, } from "../syntax.js";
@@ -61,4 +61,22 @@ export const analysisCommands = withUsage("analysis", {
             "dss analysis delete ANALYSIS_ID --dry-run --project-key PROJECT",
         ],
     },
+    update: {
+        handler: async (c, a, f) => {
+            requireArgs(a, 1, commandUsage("analysis", "update"));
+            const definition = requiredJsonInput(f, "The analysis definition is required via --data, --data-file, or --stdin (dss analysis get, edited).");
+            await c.analyses.update(a[0], definition, f["project-key"]);
+            return { updated: a[0], resource: "analysis", };
+        },
+        description: "Replace a visual analysis definition (name, script steps, charts, tags) with the object from dss analysis get, edited.",
+        examples: ["dss analysis update ANALYSIS_ID --data-file analysis.json",],
+    },
+    "list-ml-tasks": {
+        handler: (c, a, f) => {
+            requireArgs(a, 1, commandUsage("analysis", "list-ml-tasks"));
+            return c.analyses.listMlTasks(a[0], f["project-key"]);
+        },
+        description: "List the ML tasks of one visual analysis (whole project: dss ml-task list).",
+        examples: ["dss analysis list-ml-tasks ANALYSIS_ID",],
+    },
 });

package/dist/src/cli/commands/bundle.js

@@ -131,4 +131,26 @@ export const bundleCommands = withUsage("bundle", {
         description: "Delete an imported Automation-node bundle.",
         examples: ["dss bundle delete-imported v1",],
     },
+    "get-exported": {
+        handler: (c, a, f) => c.bundles.getExported(a[0], f["project-key"]),
+        description: "Get an exported bundle's details (content summary, export manifest, changelog).",
+        examples: ["dss bundle get-exported v1",],
+    },
+    "create-project": {
+        handler: async (c, _a, f) => {
+            const file = f["file"];
+            const archive = f["archive-path"];
+            if ((typeof file === "string") === (typeof archive === "string")) {
+                throw new UsageError("Pass exactly one of --file PATH (local bundle zip) or --archive-path PATH (zip on the Automation node).", "missing_required_flag");
+            }
+            const folder = f["project-folder"];
+            const result = await c.bundles.createProjectFromBundle(typeof file === "string" ? { filePath: file, } : { archivePath: archive, }, folder);
+            return result ?? { created: true, resource: "bundle", };
+        },
+        description: "Automation node (global admin): create a new project from a bundle, uploading --file PATH or using --archive-path PATH on the node host; --project-folder places it.",
+        examples: [
+            "dss bundle create-project --file bundle-v1.zip",
+            "dss bundle create-project --archive-path /data/bundles/v1.zip",
+        ],
+    },
 });

package/dist/src/cli/commands/connection.js

@@ -1,3 +1,4 @@
+import { DataikuError, } from "../../errors.js";
 import { sanitizeErrorSecrets, sanitizeSecrets, } from "../../utils/secret-sanitize.js";
 import { requiredJsonInput, } from "../coerce.js";
 import { executionMode, } from "../flags.js";
@@ -26,6 +27,7 @@ const SENSITIVE_CONNECTION_KEYS = {
     refreshtoken: true,
     secret: true,
     secretkey: true,
+    sessiontoken: true,
     sharesecret: true,
     token: true,
 };
@@ -97,12 +99,34 @@ export const connectionCommands = withUsage("connection", {
         handler: async (c, a) => {
             const usage = commandUsage("connection", "get");
             requireArgs(a, 1, usage);
-            const details = await c.connections.adminGet(requireConnectionName(a[0], usage));
-            return sanitizeConnectionSecrets(details);
+            const name = requireConnectionName(a[0], usage);
+            try {
+                const details = await c.connections.adminGet(name);
+                return sanitizeConnectionSecrets(details);
+            }
+            catch (error) {
+                if (error instanceof DataikuError && error.status === 403) {
+                    // The admin endpoint refuses non-admin keys for every connection,
+                    // including ones they can query; status, code, and exit stay as is.
+                    error.retryHint =
+                        "connection get reads the admin definition. For a connection this key can use, run `dss connection info NAME`; `dss doctor` shows which identity the key acts as.";
+                }
+                throw error;
+            }
         },
         description: "Get a connection by name (admin). Credential-bearing params (e.g. params.password) are redacted in output.",
         examples: ["dss connection get postgres",],
     },
+    info: {
+        handler: async (c, a) => {
+            const usage = commandUsage("connection", "info");
+            requireArgs(a, 1, usage);
+            const details = await c.connections.info(requireConnectionName(a[0], usage));
+            return sanitizeConnectionSecrets(details);
+        },
+        description: "Read a connection's usage view (no admin privilege required): type, params, and credentials mode for every connection the key can query, where `connection get` (admin) answers 403. Credential-bearing params (e.g. params.password) are redacted in output.",
+        examples: ["dss connection info postgres",],
+    },
     create: {
         handler: async (c, _a, f) => {
             const usage = commandUsage("connection", "create");

package/dist/src/cli/commands/data-quality.js

@@ -1,5 +1,5 @@
 import { deepMerge, } from "../../utils/deep-merge.js";
-import { num, parseBooleanOption, requiredJsonInput, } from "../coerce.js";
+import { num, parseBooleanOption, requiredJsonInput, splitCsvFlag, } from "../coerce.js";
 import { executionMode, } from "../flags.js";
 import { encodedProjectEndpoint, readIfExists, skipResult, } from "../output.js";
 import { commandUsage, withUsage, } from "../syntax.js";
@@ -41,6 +41,38 @@ export const dataQualityCommands = withUsage("data-quality", {
         description: "Get the aggregate data quality status for a dataset.",
         examples: ["dss data-quality status orders",],
     },
+    "instance-status": {
+        handler: (c) => c.dataQuality.instanceStatus(),
+        description: "Data quality status of every monitored project you can read, keyed by project key (ok/warning/error/empty rule counts, last run).",
+        examples: ["dss data-quality instance-status",],
+    },
+    "partitions-status": {
+        handler: (c, a, f) => c.dataQuality.partitionsStatus(a[0], {
+            projectKey: f["project-key"],
+            partitions: splitCsvFlag(f["partitions"]),
+        }),
+        description: "Last data quality outcome for specific partitions of a dataset (--partitions P1,P2); a non-partitioned dataset needs none.",
+        examples: ["dss data-quality partitions-status orders --partitions 2026-01,2026-02",],
+    },
+    "delete-history": {
+        handler: async (c, a, f) => {
+            const partition = f["partition"] ?? "NP";
+            const pk = f["project-key"];
+            if (executionMode(f).dryRun) {
+                return {
+                    dryRun: true,
+                    action: "delete-history",
+                    resource: "data-quality",
+                    dataset: a[0],
+                    partition,
+                };
+            }
+            await c.dataQuality.deleteHistory(a[0], { projectKey: pk, partition, });
+            return { deleted: "history", resource: "data-quality", dataset: a[0], partition, };
+        },
+        description: "Delete data quality rule history of a dataset partition (--partition, default NP; ALL for every partition).",
+        examples: ["dss data-quality delete-history orders --partition ALL --dry-run",],
+    },
     "create-rule": {
         handler: async (c, a, f) => {
             requireArgs(a, 1, commandUsage("data-quality", "create-rule"));

package/dist/src/cli/commands/dataset.js

@@ -1,9 +1,10 @@
+import { assertDatasetTypeCreatable, buildDatasetCreateBody, } from "../../resources/dataset-create.js";
 import { deepMerge, } from "../../utils/deep-merge.js";
 import { isRecord, } from "../../utils/records.js";
 import { compareStrings, stableHash, } from "../../utils/stable-hash.js";
-import { jsonInput, num, schemaColumnsInput, unknownJsonInput, } from "../coerce.js";
+import { jsonInput, num, schemaColumnsInput, splitCsvFlag, unknownJsonInput, } from "../coerce.js";
 import { executionMode, } from "../flags.js";
-import { datasetSourceSummary, } from "../helpers/dataset.js";
+import { datasetSourceSummary, datasetSqlSourceFromFlags, } from "../helpers/dataset.js";
 import { moveCreatedItemsToZone, resolveFlowZoneIdFromFlags, } from "../helpers/flow-zone.js";
 import { enqueueCliWarning, readIfExists, skipResult, } from "../output.js";
 import { commandUsage, withUsage, } from "../syntax.js";
@@ -96,6 +97,14 @@ export function compareDatasetSchemas(expected, actual) {
         differencesTruncated: acc.items.length < acc.total,
     };
 }
+/** `--columns` / `--partitions` read selection (served by the POST variant of the data endpoint). */
+function dataSelectionFromFlags(f) {
+    const columns = splitCsvFlag(f["columns"]);
+    return {
+        ...(columns.length > 0 ? { columns, } : {}),
+        ...(typeof f["partitions"] === "string" ? { partitions: f["partitions"], } : {}),
+    };
+}
 export const datasetCommands = withUsage("dataset", {
     list: {
         handler: (c, _a, f) => c.datasets.list(f["project-key"]),
@@ -225,6 +234,7 @@ export const datasetCommands = withUsage("dataset", {
                 maxRows,
                 projectKey: f["project-key"],
                 timeoutMs: num(f["timeout"], "--timeout"),
+                ...dataSelectionFromFlags(f),
             });
             if (result.truncated) {
                 enqueueCliWarning({
@@ -239,8 +249,11 @@ export const datasetCommands = withUsage("dataset", {
             }
             return result;
         },
-        description: "Preview dataset rows (--rows is an alias for --max-rows). Returns { columns, rows, rowCount, truncated, limit }; when the dataset has more rows than the cap, truncated is true and a dataset_preview_truncated warning is written to stderr.",
-        examples: ["dss dataset preview orders", "dss dataset preview orders --rows 5",],
+        description: "Preview dataset rows (--rows is an alias for --max-rows); --columns COLS and --partitions SPEC narrow the read. Returns { columns, rows, rowCount, truncated, limit }; a dataset_preview_truncated warning on stderr flags more rows.",
+        examples: [
+            "dss dataset preview orders",
+            "dss dataset preview orders --rows 5 --columns id,amount",
+        ],
     },
     "assert-count": {
         handler: async (c, a, f) => {
@@ -291,6 +304,22 @@ export const datasetCommands = withUsage("dataset", {
         description: "Get dataset-level metadata.",
         examples: ["dss dataset metadata orders",],
     },
+    "sync-hive-metastore": {
+        handler: async (c, a, f) => {
+            await c.datasets.synchronizeHiveMetastore(a[0], f["project-key"]);
+            return { synchronized: a[0], resource: "dataset", };
+        },
+        description: "Create or update the Hive table of an HDFS dataset to match its schema (Hive metastore sync). HDFS datasets only.",
+        examples: ["dss dataset sync-hive-metastore web_logs",],
+    },
+    "update-from-hive": {
+        handler: async (c, a, f) => {
+            await c.datasets.updateFromHive(a[0], f["project-key"]);
+            return { updated: a[0], resource: "dataset", };
+        },
+        description: "Update an HDFS dataset's path, schema, partitioning, and format from its Hive metastore table. HDFS datasets only.",
+        examples: ["dss dataset update-from-hive web_logs --plan",],
+    },
     "metadata-set": {
         handler: async (c, a, f) => {
             const usage = commandUsage("dataset", "metadata-set");
@@ -333,14 +362,18 @@ export const datasetCommands = withUsage("dataset", {
             if (!name || !dsType) {
                 throw new UsageError(`--name and --type are required. Usage: ${commandUsage("dataset", "create")}`);
             }
+            assertDatasetTypeCreatable(dsType);
             if (!connection && dsType.toLowerCase() !== "uploadedfiles") {
                 throw new UsageError("--connection is required unless --type is UploadedFiles.");
             }
+            const sqlSource = datasetSqlSourceFromFlags(f, dsType);
+            const hasSqlSource = sqlSource.table !== undefined || sqlSource.query !== undefined;
             const payload = {
                 datasetName: name,
                 connection,
                 dsType,
                 projectKey: pk,
+                ...sqlSource,
             };
             const zoneId = await resolveFlowZoneIdFromFlags(c, f, pk);
             if (f["if-not-exists"] === true || executionMode(f).dryRun) {
@@ -356,20 +389,53 @@ export const datasetCommands = withUsage("dataset", {
                         resource: "dataset",
                         name,
                         payload,
+                        ...(hasSqlSource
+                            ? {
+                                request: {
+                                    method: "POST",
+                                    endpoint: `/public/api/projects/${encodeURIComponent(c.resolveProjectKey(pk))}/datasets/`,
+                                    body: buildDatasetCreateBody({
+                                        ...payload,
+                                        projectKey: c.resolveProjectKey(pk),
+                                    }),
+                                },
+                                schemaDetection: "After the POST, DSS detects the schema (POST /actions/testAndDetectSettings/externalSQL) and it is stored (PUT /schema).",
+                            }
+                            : {}),
                         ...(existing ? { current: existing, } : {}),
                         ...(zoneId ? { zoneId, zoneMove: [{ objectId: name, objectType: "DATASET", },], } : {}),
                     };
                 }
             }
             await c.datasets.create(payload);
+            const detection = hasSqlSource ? await c.datasets.detectSqlSchema(name, pk) : undefined;
+            if (detection && !detection.detected) {
+                enqueueCliWarning({
+                    code: "dataset_schema_not_detected",
+                    dataset: name,
+                    error: detection.error,
+                    hint: `The dataset was kept without columns. Fix the connection, table, or query, then set the columns with dss dataset refresh-schema ${name}.`,
+                });
+            }
             const moved = await moveCreatedItemsToZone(c, f, [{ objectId: name, objectType: "DATASET", },], pk);
-            return { created: name, resource: "dataset", ...moved, };
+            return {
+                created: name,
+                resource: "dataset",
+                ...(detection
+                    ? {
+                        schemaDetected: detection.detected,
+                        columns: detection.columns,
+                        ...(detection.error !== undefined ? { schemaError: detection.error, } : {}),
+                    }
+                    : {}),
+                ...moved,
+            };
         },
-        description: "Create a new dataset. UploadedFiles uses the server's default upload connection when omitted.",
+        description: "Create a dataset. SQL types: --table or --query, schema auto-detected.",
         examples: [
             "dss dataset create --name uploads --type UploadedFiles",
-            "dss dataset create --name orders --connection filesystem --type Filesystem",
-            "dss dataset create --name orders --connection filesystem --type Filesystem --zone Experiments --dry-run",
+            "dss dataset create --name sales --type Snowflake --connection sf --table SALES --schema PUBLIC",
+            "dss dataset create --name big --type PostgreSQL --connection pg --query-file big.sql --dry-run",
         ],
     },
     "create-managed": {
@@ -472,6 +538,7 @@ export const datasetCommands = withUsage("dataset", {
                 projectKey: f["project-key"],
                 limit: num(f["limit"], "--limit"),
                 rawData: f["raw-data"] === true,
+                ...dataSelectionFromFlags(f),
             });
             if (result.truncated) {
                 enqueueCliWarning({
@@ -495,7 +562,7 @@ export const datasetCommands = withUsage("dataset", {
             }
             return result;
         },
-        description: "Download up to --limit rows (default 100k) as CSV and return { path, rows, truncated, limit }. Formula-like cells are neutralized for spreadsheets; --raw-data preserves exact bytes. Warnings report truncation and the default output path.",
+        description: "Download up to --limit rows (default 100k) as CSV and return { path, rows, truncated, limit }; --columns COLS and --partitions SPEC narrow the read. Formula-like cells are neutralized for spreadsheets; --raw-data preserves exact bytes.",
         examples: [
             "dss dataset download orders",
             "dss dataset download orders --output ./data/",

package/dist/src/cli/commands/discussion.js

@@ -1,3 +1,4 @@
+import { jsonInput, } from "../coerce.js";
 import { commandUsage, withUsage, } from "../syntax.js";
 import { requireArgs, UsageError, } from "../usage.js";
 export const discussionCommands = withUsage("discussion", {
@@ -41,4 +42,18 @@ export const discussionCommands = withUsage("discussion", {
         description: "Add a reply to an existing discussion.",
         examples: ["dss discussion reply DATASET customers d123 --text Done",],
     },
+    update: {
+        handler: async (c, a, f) => {
+            const pk = f["project-key"];
+            const body = jsonInput(f);
+            const topic = f["topic"];
+            if ((body === undefined) === (typeof topic !== "string")) {
+                throw new UsageError("Pass exactly one of --topic TEXT or --data/--data-file/--stdin.", "missing_required_flag");
+            }
+            const next = body ?? { ...await c.discussions.get(a[0], a[1], a[2], pk), topic, };
+            return c.discussions.update(a[0], a[1], a[2], next, pk);
+        },
+        description: "Update a discussion: --topic TEXT edits the topic of the current discussion, or --data sends a full discussion object from get (e.g. to move it to another object of the project).",
+        examples: ['dss discussion update DATASET orders L8gkpoI6 --topic "Schema change (edited)"',],
+    },
 });

package/dist/src/cli/commands/doctor.js

@@ -5,7 +5,7 @@ export const doctorCommands = withUsage("doctor", {
             const { runDoctor, } = await import("../doctor.js");
             return (await runDoctor(f)).result;
         },
-        description: "Run JSON diagnostics for DSS credentials, connectivity, and project access.",
+        description: "Run JSON diagnostics for DSS credentials, connectivity, and project access, and report the identity (whoami: user, groups, auth source) the API key acts as.",
         examples: ["dss doctor", "dss doctor --project-key MYPROJ", "dss doctor --capabilities --fast",],
     },
 });