@erdoai/cli 0.139.0 → 0.140.0

package/dist/index.js

@@ -15,13 +15,34 @@ function advertiserVerificationAction(container) {
 
 // src/refresh.ts
 function describeRefreshTrigger(state) {
-  if (state.disarmed) return "disarmed \u2014 the experiment has settled";
-  if (state.strategy === "live_webhook") return "incoming webhook events \u2014 page views do not trigger this feed";
-  if (state.strategy === "live_native_poll") return "not running \u2014 native polling is unavailable; configure script_js with a schedule or live_webhook";
-  if (!state.job_id) return "not running \u2014 no linked job; configure the refresh again";
-  if (state.job_enabled === false) return "not running \u2014 the linked job is disabled; configure the refresh again";
-  if (state.schedule) return state.schedule_enabled === false ? `cron ${state.schedule} is disabled` : `cron: ${state.schedule}`;
-  return state.refresh_on_view_enabled !== false ? "runs on view \u2014 when a bound page is opened and the data is stale" : "manual only \u2014 page views do not trigger this refresh";
+  const event = describeEventTriggers(state);
+  if (state.disarmed) return event ? `disarmed \u2014 the experiment has settled; ${event}` : "disarmed \u2014 the experiment has settled";
+  if (state.strategy === "live_webhook") return event ? `incoming webhook events \u2014 page views do not trigger this feed; ${event}` : "incoming webhook events \u2014 page views do not trigger this feed";
+  if (state.strategy === "live_native_poll") return event ? `not running \u2014 native polling is unavailable; ${event}` : "not running \u2014 native polling is unavailable; configure script_js with a schedule or live_webhook";
+  if (!state.job_id) return event ? `not running \u2014 no linked job; ${event}` : "not running \u2014 no linked job; configure the refresh again";
+  if (state.job_enabled === false) return event ? `not running \u2014 the linked job is disabled; ${event}` : "not running \u2014 the linked job is disabled; configure the refresh again";
+  const trigger = state.schedule ? state.schedule_enabled === false ? `cron ${state.schedule} is disabled` : `cron: ${state.schedule}` : state.refresh_on_view_enabled !== false ? "runs on view \u2014 when a bound page is opened and the data is stale" : event ? void 0 : "manual only \u2014 page views do not trigger this refresh";
+  return [trigger, event, state.test_run_success === false ? "configuration test failed" : void 0].filter(Boolean).join("; ");
+}
+function describeEventTriggers(state) {
+  if (!state.event_triggers?.length) {
+    if (!state.event_dataset_ids?.length) return void 0;
+    return `event-driven \u2014 configured for ${state.event_dataset_ids.join(", ")} (debounce ${state.event_debounce_seconds ?? 0}s)`;
+  }
+  const triggers = state.event_triggers;
+  const ownerActive = !!state.job_id && state.disarmed !== true && state.job_enabled !== false;
+  const enabled = ownerActive ? triggers.filter((trigger) => trigger.enabled) : [];
+  const disabled = ownerActive ? triggers.filter((trigger) => !trigger.enabled) : triggers;
+  const parts = [];
+  if (enabled.length) {
+    const debounce = new Set(enabled.map((trigger) => trigger.debounce_seconds));
+    parts.push(`event-driven \u2014 enabled for ${enabled.map((trigger) => trigger.dataset_id).join(", ")}${debounce.size === 1 ? ` (debounce ${enabled[0].debounce_seconds}s)` : ""}`);
+  }
+  if (disabled.length || enabled.length !== triggers.length) {
+    const unavailable = disabled.map((trigger) => trigger.dataset_id);
+    parts.push(`event-driven \u2014 disabled for ${unavailable.join(", ")}`);
+  }
+  return parts.join("; ");
 }
 function describeRefreshWrites(strategy, mode, key, liveMode) {
   if (strategy?.startsWith("live_")) return liveMode === "append" ? "append \u2014 each incoming event adds rows" : "replace \u2014 each incoming event replaces the current rows";
@@ -1223,6 +1244,9 @@ var ErdoClient = class {
   getDatasetDeclaredSchema(slug) {
     return this.request("POST", `/v1/datasets/${encodeURIComponent(slug)}/fetch`, { view: "declared_schema" });
   }
+  getDatasetResourceSchema(slug) {
+    return this.request("POST", `/v1/datasets/${encodeURIComponent(slug)}/fetch`, { view: "resource_schema" });
+  }
   // Create an empty dataset with its declared columns. The server refuses a
   // create without a schema unless the purpose's columns are platform-defined.
   createDataset(input) {
@@ -7281,7 +7305,7 @@ Read one with: erdo datasets fetch ${res.dataset_slug} --revision <id> --sql "SE
   }
 });
 datasetsCmd.command("schema <slug>").alias("describe").description(
-  "Show physical columns by default, or the declared write contract with --declared. Physical columns are the names and types you can reference in `datasets fetch --sql`, read from the stored `data` table itself (DuckDB DESCRIBE) so they are what is actually there, not what was declared. For a database or warehouse dataset, DESCRIBE its real table names instead: `datasets fetch <slug> --sql 'DESCRIBE <table>'`."
+  "Show physical columns by default, or the declared write contract with --declared. File datasets read the stored `data` table; integration datasets show their discovered resource tables and columns. For an integration dataset, choose a resource/table explicitly when fetching: `datasets fetch <slug> --sql 'SELECT * FROM <table>'`."
 ).option("--json", "print the raw JSON result instead of a table").option("--declared", "read the declared write contract (column names, type hints, required flags, semantic roles) instead of the physical columns").action(async (slug, opts) => {
   try {
     const client = new ErdoClient();
@@ -7310,7 +7334,35 @@ datasetsCmd.command("schema <slug>").alias("describe").description(
       );
       return;
     }
-    const res = await client.fetchDatasetContents(slug, { sql_query: "DESCRIBE data" });
+    const res = await client.getDatasetResourceSchema(slug);
+    if (!res || typeof res !== "object" || res.view !== "resource_schema")
+      throw new Error("Invalid dataset schema response: the server did not answer the resource_schema view");
+    if (res.schema_provenance === "resource_catalog") {
+      const resources = res.resources ?? [];
+      if (!Array.isArray(resources) || resources.some((resource) => !resource || !Array.isArray(resource.columns)))
+        throw new Error("Invalid dataset schema response: malformed resource catalog");
+      if (opts.json) {
+        print(res);
+        return;
+      }
+      if (!resources.length) {
+        console.log(`${slug} has no discovered integration resources yet.`);
+        return;
+      }
+      const rows = resources.flatMap(
+        (resource) => resource.columns.length ? resource.columns.map((column) => [
+          resource.table_name || resource.key || resource.name,
+          column.name,
+          column.type || ""
+        ]) : [[resource.table_name || resource.key || resource.name, "", "(columns unavailable)"]]
+      );
+      printAlignedTable(["table", "column", "type"], rows);
+      console.log(`
+Fetch a selected table with: erdo datasets fetch ${slug} --sql "SELECT * FROM <table>"`);
+      return;
+    }
+    if (res.schema_provenance !== "dataset_storage" || !Array.isArray(res.columns) || !Array.isArray(res.rows))
+      throw new Error("Invalid dataset schema response: malformed physical schema");
     if (opts.json) {
       print(res);
       return;
@@ -7439,6 +7491,8 @@ refreshCmd.command("show <slug>").description(
     if (res.experiment_run_id) rows.push(["experiment", res.experiment_run_id]);
     rows.push(["writes", describeRefreshWrites(res.strategy, res.refresh_mode, res.key_column, res.live_write_mode)]);
     rows.push(["trigger", describeRefreshTrigger(res)]);
+    const eventTriggers = describeEventTriggers(res);
+    if (eventTriggers) rows.push(["event triggers", eventTriggers]);
     if (res.schedule) {
       const zone = res.schedule_timezone || "UTC";
       rows.push([
@@ -7489,7 +7543,7 @@ refreshCmd.command("show <slug>").description(
     }
     if (res.job_id) rows.push(["job", res.job_id]);
     printAlignedTable(["setting", "value"], rows);
-    if (!res.strategy?.startsWith("live_") && res.job_id && res.job_enabled !== false && !res.disarmed && !res.schedule && !res.refresh_on_view_enabled) {
+    if (!res.strategy?.startsWith("live_") && res.job_id && res.job_enabled !== false && !res.disarmed && !res.schedule && !res.refresh_on_view_enabled && !eventTriggers) {
       console.log(
         `
 Nothing triggers this refresh on its own \u2014 it runs only when asked: erdo datasets refresh run ${res.dataset_slug}`
@@ -7523,7 +7577,7 @@ refreshCmd.command("set <slug>").description(
 ).option("--agent-prompt <textOr@file>", "instructions for the agent_run strategy, inline or @path").option(
   "--context-datasets <csv>",
   "slugs of other datasets the recipe reads besides the one it writes"
-).option(
+).option("--event-datasets <csv>", "dataset slugs whose writes trigger this refresh (up to 16, comma-separated)").option("--event-debounce <seconds>", "wait after the first source write to coalesce a burst (integer 0\u20133600 seconds)").option(
   "--on-view",
   "let a page view trigger this refresh (the platform default); pass --no-on-view to turn it off. Sent only when you pass one of the two"
 ).option("--no-on-view", "stop page views from triggering this refresh").option(
@@ -7586,6 +7640,31 @@ refreshCmd.command("set <slug>").description(
         const slugs = opts.contextDatasets.split(",").map((s) => s.trim()).filter(Boolean);
         if (slugs.length) body.context_dataset_slugs = slugs;
       }
+      let eventDatasetSlugs;
+      if (opts.eventDatasets !== void 0) {
+        eventDatasetSlugs = opts.eventDatasets.split(",").map((s) => s.trim());
+        if (!eventDatasetSlugs.length || eventDatasetSlugs.some((slug2) => !slug2)) {
+          fail(new Error("--event-datasets must include at least one non-blank dataset slug"));
+          return;
+        }
+        if (eventDatasetSlugs.length > 16) {
+          fail(new Error("--event-datasets supports at most 16 dataset slugs"));
+          return;
+        }
+        body.event_dataset_slugs = eventDatasetSlugs;
+      }
+      if (opts.eventDebounce !== void 0) {
+        const value = Number(opts.eventDebounce);
+        if (!opts.eventDebounce.trim() || !Number.isFinite(value) || !Number.isInteger(value) || value < 0 || value > 3600) {
+          fail(new Error("--event-debounce must be a finite integer between 0 and 3600"));
+          return;
+        }
+        if (!eventDatasetSlugs?.length) {
+          fail(new Error("--event-debounce requires non-empty --event-datasets"));
+          return;
+        }
+        body.event_debounce_seconds = value;
+      }
       if (opts.onView !== void 0) body.refresh_on_view_enabled = opts.onView;
       if (opts.staleAfter !== void 0) {
         body.refresh_on_view_stale_after_seconds = opts.staleAfter;
@@ -7606,7 +7685,13 @@ refreshCmd.command("set <slug>").description(
         `Installed a ${res.strategy} refresh on ${res.dataset_slug}${res.previous_strategy ? ` (replacing ${res.previous_strategy})` : ""}`
       );
       console.log(`writes: ${describeRefreshWrites(res.strategy, res.refresh_mode, body.key_column, body.live_write_mode)}`);
-      console.log(describeRefreshTrigger({ ...res, refresh_on_view_enabled: body.refresh_on_view_enabled }));
+      console.log(describeRefreshTrigger({
+        ...res,
+        event_dataset_ids: res.event_dataset_ids,
+        event_debounce_seconds: res.event_debounce_seconds ?? body.event_debounce_seconds,
+        refresh_on_view_enabled: body.refresh_on_view_enabled,
+        test_run_success: res.test_run?.success
+      }));
       if (res.job_id) console.log(`job: ${res.job_id}`);
       if (res.webhook_url) {
         console.log(`webhook url: ${res.webhook_url}`);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@erdoai/cli",
-  "version": "0.139.0",
+  "version": "0.140.0",
   "description": "Erdo CLI \u2014 drive datasets, pages, and evals from the terminal or CI",
   "type": "module",
   "bin": {