feat(tables): import csv into existing tables

[icecrasher321]

Summary

Import csv into existing tables with configurable mapping of columns. Can append to table [checking uniqueness constraints] or replace rows.

Also added capability to mothership.

Type of Change

• New feature

Testing

Tested manually.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Apr 16, 2026 8:28pm

[cursor]

PR Summary

Medium Risk
Touches table mutation paths by adding transactional row replacement and a new import endpoint, so bugs could lead to data loss or constraint/limit regressions despite added validation and tests.

Overview
Adds a new API endpoint POST /api/table/[tableId]/import-csv to import CSV/TSV data into an existing table with append or transactional replace modes, optional explicit header→column mapping, row-limit checks, archived/workspace validation, and client-friendly 400 errors (including partial insert counts on failed appends).

Introduces shared csv-import utilities (parsing, schema inference, name sanitization, mapping validation, coercion, size/batch constants) and refactors the “create table from CSV” route plus the Copilot user_table tool to reuse them; also adds replaceTableRows in table service with schema/size/unique checks and a locked transaction.

Wires a new ImportCsvDialog into the table header and tables list context menu, adds a useImportCsvIntoTable hook, and updates Copilot tool schemas/catalog to accept mode + mapping for table file imports; includes new unit tests covering route behavior, csv-import helpers, Copilot import, and file-output unwrapping.

^{Reviewed by Cursor Bugbot for commit 1a45b8e. Configure here.}

[greptile-apps]

Greptile Summary

This PR adds CSV import into existing tables — with append/replace modes and configurable column mapping — through a new /api/table/[tableId]/import-csv route, an ImportCsvDialog UI component, and Mothership (copilot) support. It also centralises all CSV parsing/coercion logic that was previously duplicated across the create-from-CSV route and the copilot tool into a shared lib/table/csv-import.ts module.

All remaining findings are P2: two instances of subpath imports instead of the @/lib/table barrel, a duplicated type definition in the query hooks, and a harmless double-unwrap in the copilot file serialiser.

Confidence Score: 5/5

Safe to merge — no blocking issues; only minor style/duplication P2 findings.

All P2. The core implementation is correct: the replace path uses a transactional FOR UPDATE lock, query invalidation covers rows + count + lists, validation catches required columns, unique constraints, and capacity before writing, and test coverage is thorough across both the route and the shared library.

import-csv-dialog.tsx (barrel import style) and hooks/queries/tables.ts (duplicate type declarations)

Important Files Changed

Filename	Overview
apps/sim/app/api/table/[tableId]/import-csv/route.ts	New POST route for importing CSV into an existing table; correctly validates auth, workspace ownership, archived state, mapping, and capacity limits before delegating to shared service helpers.
apps/sim/lib/table/csv-import.ts	New shared CSV parsing, schema inference, validation, and coercion library; well-documented with clear exports and comprehensive test coverage.
apps/sim/lib/table/service.ts	Adds replaceTableRows using a transaction with a FOR UPDATE lock to avoid empty-table intermediate states; validates schema, row size, uniqueness within the batch, and capacity before executing.
apps/sim/app/workspace/[workspaceId]/tables/components/import-csv-dialog/import-csv-dialog.tsx	New dialog for mapping and importing a CSV into an existing table; minor P2 issue — imports come from subpath @/lib/table/csv-import and @/lib/table/types instead of the barrel @/lib/table.
apps/sim/hooks/queries/tables.ts	Adds useImportCsvIntoTable mutation; correctly calls invalidateRowCount (which invalidates rows + detail + lists). Minor issue: re-declares CsvHeaderMapping and CsvImportMode types that already exist in @/lib/table.
apps/sim/lib/copilot/tools/server/table/user-table.ts	Upgrades import_file operation to use shared CSV helpers; adds mode (append/replace) and mapping parameters, workspace ownership check, and archived table guard.
apps/sim/lib/copilot/request/tools/files.ts	Extracts unwrapFunctionExecuteOutput helper; serializeOutputForFile now passes an already-unwrapped value to extractTabularData which calls unwrapFunctionExecuteOutput a second time (harmless no-op but redundant).
apps/sim/app/api/table/[tableId]/import-csv/route.test.ts	Comprehensive test suite for the new import route; covers auth, access control, append, replace, mapping, capacity limits, unique violations, TSV, and invalid file extensions.
apps/sim/lib/table/csv-import.test.ts	Unit tests for all CSV import helpers; good coverage of parseCsvBuffer (including BOM stripping and Uint8Array input), type inference, sanitization, auto-mapping, and coercion.

Sequence Diagram

sequenceDiagram
    actor User
    participant Dialog as ImportCsvDialog
    participant Hook as useImportCsvIntoTable
    participant API as POST /api/table/[tableId]/import-csv
    participant CsvLib as lib/table/csv-import
    participant Service as lib/table/service

    User->>Dialog: Select CSV file
    Dialog->>CsvLib: parseCsvBuffer(Uint8Array)
    CsvLib-->>Dialog: { headers, rows }
    Dialog->>CsvLib: buildAutoMapping(headers, schema)
    CsvLib-->>Dialog: mapping
    Dialog-->>User: Show column mapping UI

    User->>Dialog: Confirm import (append | replace)
    Dialog->>Hook: mutateAsync({ tableId, file, mode, mapping })
    Hook->>API: POST FormData
    API->>API: checkSessionOrInternalAuth
    API->>API: checkAccess(tableId, userId, write)
    API->>CsvLib: parseCsvBuffer(Buffer)
    CsvLib-->>API: { headers, rows }
    API->>CsvLib: buildAutoMapping / validateMapping
    CsvLib-->>API: validation result
    API->>CsvLib: coerceRowsForTable(rows, schema, effectiveMap)
    CsvLib-->>API: coerced RowData[]

    alt mode = append
        API->>Service: batchInsertRows (in batches of 1000)
        Service-->>API: inserted rows
    else mode = replace
        API->>Service: replaceTableRows (transactional DELETE + INSERT)
        Service-->>API: { deletedCount, insertedCount }
    end

    API-->>Hook: { success, data }
    Hook->>Hook: invalidateRowCount (rows + detail + lists)
    Hook-->>Dialog: result
    Dialog-->>User: toast + close

Loading

Comments Outside Diff (3)

1. apps/sim/app/workspace/[workspaceId]/tables/components/import-csv-dialog/import-csv-dialog.tsx, line 893-895 (link)

   Subpath imports bypass the barrel

   @/lib/table/csv-import and @/lib/table/types are re-exported via the @/lib/table barrel (index.ts). Per project conventions, consumers should import from the barrel, not from subpaths directly.

   And remove the import type { TableDefinition } line below (line 894).

   Context Used: Import patterns for the Sim application (source)

   Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

2. apps/sim/hooks/queries/tables.ts, line 1504-1505 (link)

   Duplicate type definition

   CsvHeaderMapping and CsvImportMode are already exported from @/lib/table (via csv-import.ts). Redefining them here creates two identical but independent types — a future change to one won't propagate to the other.

   Or just import them where needed from @/lib/table and remove these local declarations.

   Context Used: Import patterns for the Sim application (source)

3. apps/sim/lib/copilot/request/tools/files.ts, line 136-142 (link)

   Double unwrap in serializeOutputForFile

   serializeOutputForFile already calls unwrapFunctionExecuteOutput at the top and passes unwrapped to extractTabularData. But extractTabularData calls unwrapFunctionExecuteOutput again internally. The second unwrap is a no-op on an already-unwrapped value, but it's redundant and slightly confusing. Consider either having extractTabularData accept an already-unwrapped value (skip internal unwrapping when called from here) or passing output directly and letting extractTabularData own the unwrapping.

_{Reviews (1): Last reviewed commit: "feat(tables): import csv into existing t..." | Re-trigger Greptile}

[icecrasher321]

bugbot run

[icecrasher321]

bugbot run

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 1a45b8e. Configure here.}