fix(azure-data-explorer): handle commas inside quoted properties and empty extent IDs

Two defects in the shared command helpers:

buildWithClause split the property list on every comma before validating,
so a value that legally contains one — a docstring sentence, or a tags
array with more than one entry — was torn in half and rejected. Splitting
is now quote-aware, and an unterminated quote is rejected outright rather
than swallowing the rest of the clause.

transformColumnListResponse dropped empty strings, but `.ingest inline`
reports "no data shards were generated" as a single record carrying an
empty extent ID. A no-op load therefore looked like a missing column
instead of an empty result. Only non-strings are skipped now.

Author: waleedlatif1

apps/docs/content/docs/en/integrations/azure_data_explorer.mdx

@@ -329,7 +329,7 @@ Push rows directly into an Azure Data Explorer table with .ingest inline. Data i
 | `rowCount` | number | Rows carried in this result, after the row cap |
 | `totalRowCount` | number | Rows Kusto returned, before the row cap was applied |
 | `truncated` | boolean | Whether rows were dropped to stay within the row cap — narrow the query if true |
-| `extentIds` | array | Extent IDs created by the ingestion. A single empty ID means no data shard was generated |
+| `extentIds` | array | Extent IDs created by the ingestion — one per data shard. A single empty or zero-valued ID means no data shard was generated |
 
 ### Azure Data Explorer Ingest From Query
 

apps/sim/tools/azure_data_explorer/ingest_inline.ts

@@ -134,7 +134,7 @@ export const azureDataExplorerIngestInlineTool: ToolConfig<
     extentIds: {
       type: 'array',
       description:
-        'Extent IDs created by the ingestion. A single empty ID means no data shard was generated',
+        'Extent IDs created by the ingestion — one per data shard. A single empty or zero-valued ID means no data shard was generated',
       items: { type: 'string' },
     },
   },

apps/sim/tools/azure_data_explorer/utils.test.ts

@@ -8,6 +8,7 @@ import {
   renderEntityName,
   renderIngestMode,
   renderOperationId,
+  transformColumnListResponse,
 } from '@/tools/azure_data_explorer/utils'
 
 describe('renderEntityName', () => {
@@ -55,6 +56,27 @@ describe('buildWithClause', () => {
     )
   })
 
+  it('keeps a comma that sits inside a quoted value', () => {
+    expect(buildWithClause('docstring="Raw logs, archived nightly"', 'format="json"')).toBe(
+      ' with (docstring="Raw logs, archived nightly")'
+    )
+    expect(buildWithClause(`tags="['daily','prod']"`, 'format="json"')).toBe(
+      ` with (tags="['daily','prod']")`
+    )
+  })
+
+  it('still separates properties on the commas between them', () => {
+    expect(
+      buildWithClause('docstring="Logs, raw", folder="Ingest", distributed=true', 'format="json"')
+    ).toBe(' with (docstring="Logs, raw", folder="Ingest", distributed=true)')
+  })
+
+  it('rejects an unterminated quote rather than swallowing the rest of the clause', () => {
+    expect(() => buildWithClause('docstring="never closed', 'format="json"')).toThrow(
+      /Unterminated " quote/
+    )
+  })
+
   it('rejects a value that would close the clause and extend the command', () => {
     expect(() => buildWithClause('format="json") <| evil', 'format="json"')).toThrow(
       /Invalid property/
@@ -130,3 +152,49 @@ describe('renderOperationId', () => {
     expect(() => renderOperationId('abc") | drop table X //')).toThrow(/Invalid operation ID/)
   })
 })
+
+/** Minimal stand-in for the proxy's JSON envelope. */
+function proxyResponse(records: Array<Record<string, unknown>>): Response {
+  return {
+    ok: true,
+    status: 200,
+    json: async () => ({
+      success: true,
+      output: {
+        tableName: 'Table_0',
+        columns: [],
+        rows: [],
+        records,
+        rowCount: records.length,
+        totalRowCount: records.length,
+        truncated: false,
+      },
+    }),
+  } as unknown as Response
+}
+
+describe('transformColumnListResponse', () => {
+  it('collects the values of the named column', async () => {
+    const transform = transformColumnListResponse('TableName', 'tables')
+    const result = await transform(
+      proxyResponse([{ TableName: 'StormEvents' }, { TableName: 'Logs' }])
+    )
+
+    expect(result.output.tables).toEqual(['StormEvents', 'Logs'])
+  })
+
+  it("keeps an empty extent ID, which is how Kusto reports 'no data shard was written'", async () => {
+    const transform = transformColumnListResponse('ExtentId', 'extentIds')
+    const result = await transform(proxyResponse([{ ExtentId: '' }]))
+
+    expect(result.output.extentIds).toEqual([''])
+    expect(result.output.rowCount).toBe(1)
+  })
+
+  it('skips a null value rather than coercing it to a string', async () => {
+    const transform = transformColumnListResponse('TableName', 'tables')
+    const result = await transform(proxyResponse([{ TableName: null }, { TableName: 'Logs' }]))
+
+    expect(result.output.tables).toEqual(['Logs'])
+  })
+})

apps/sim/tools/azure_data_explorer/utils.ts

@@ -36,6 +36,44 @@ export function renderEntityName(name: string): string {
 /** One `name = value` command property, with a quoted, numeric, or bare value. */
 const COMMAND_PROPERTY = /^[A-Za-z_][A-Za-z0-9_]*\s*=\s*(?:"[^"\\]*"|'[^'\\]*'|[A-Za-z0-9_.:+-]+)$/
 
+/**
+ * Splits a property list on the commas that separate properties, ignoring the
+ * ones inside a quoted value.
+ *
+ * A plain `split(',')` would break the documented cases where a value legally
+ * contains a comma — a `docstring` sentence, or a `tags` array with more than
+ * one entry.
+ */
+function splitProperties(input: string): string[] {
+  const parts: string[] = []
+  let current = ''
+  let quote: '"' | "'" | null = null
+
+  for (const char of input) {
+    if (quote) {
+      current += char
+      if (char === quote) quote = null
+      continue
+    }
+    if (char === '"' || char === "'") {
+      quote = char
+      current += char
+      continue
+    }
+    if (char === ',') {
+      parts.push(current)
+      current = ''
+      continue
+    }
+    current += char
+  }
+
+  if (quote) throw new Error(`Unterminated ${quote} quote in property list: ${input}`)
+  parts.push(current)
+
+  return parts.map((part) => part.trim()).filter(Boolean)
+}
+
 /**
  * Builds the `with (...)` clause shared by the create and ingest commands.
  *
@@ -53,7 +91,7 @@ export function buildWithClause(input: string | undefined, example: string): str
     .trim()
   if (!inner) return ''
 
-  const properties = inner.split(',').map((property) => property.trim())
+  const properties = splitProperties(inner)
   for (const property of properties) {
     if (!COMMAND_PROPERTY.test(property)) {
       throw new Error(
@@ -207,13 +245,18 @@ export async function transformAzureDataExplorerResponse(response: Response) {
 /**
  * Adds a flat list of the strings in one documented column, so a `.show` or
  * `.ingest` command surfaces its identifiers alongside the full result table.
+ *
+ * Every string value is kept, including an empty one: `.ingest inline` reports
+ * "no data shards were generated" as a single record carrying an empty
+ * (zero-valued) extent ID, so dropping it would turn a no-op load into what
+ * looks like a missing column.
  */
 export function transformColumnListResponse<K extends string>(columnName: string, outputKey: K) {
   return async (response: Response) => {
     const output = await readProxyEnvelope(response)
     const values = output.records
       .map((record) => record[columnName])
-      .filter((value): value is string => typeof value === 'string' && value.length > 0)
+      .filter((value): value is string => typeof value === 'string')
     return {
       success: true as const,
       output: { ...output, [outputKey]: values } as AzureDataExplorerTable & Record<K, string[]>,