|
| 1 | +import { createLogger } from '@sim/logger' |
| 2 | +import { type NextRequest, NextResponse } from 'next/server' |
| 3 | +import { checkSessionOrInternalAuth } from '@/lib/auth/hybrid' |
| 4 | +import { generateRequestId } from '@/lib/core/utils/request' |
| 5 | +import { |
| 6 | + batchInsertRows, |
| 7 | + createTable, |
| 8 | + deleteTable, |
| 9 | + getWorkspaceTableLimits, |
| 10 | + type TableSchema, |
| 11 | +} from '@/lib/table' |
| 12 | +import type { ColumnDefinition, RowData } from '@/lib/table/types' |
| 13 | +import { getUserEntityPermissions } from '@/lib/workspaces/permissions/utils' |
| 14 | +import { normalizeColumn } from '@/app/api/table/utils' |
| 15 | + |
| 16 | +const logger = createLogger('TableImportCSV') |
| 17 | + |
| 18 | +const MAX_CSV_FILE_SIZE = 50 * 1024 * 1024 |
| 19 | +const MAX_BATCH_SIZE = 1000 |
| 20 | +const SCHEMA_SAMPLE_SIZE = 100 |
| 21 | + |
| 22 | +type ColumnType = 'string' | 'number' | 'boolean' | 'date' |
| 23 | + |
| 24 | +async function parseCsvBuffer( |
| 25 | + buffer: Buffer, |
| 26 | + delimiter = ',' |
| 27 | +): Promise<{ headers: string[]; rows: Record<string, unknown>[] }> { |
| 28 | + const { parse } = await import('csv-parse/sync') |
| 29 | + const parsed = parse(buffer.toString('utf-8'), { |
| 30 | + columns: true, |
| 31 | + skip_empty_lines: true, |
| 32 | + trim: true, |
| 33 | + relax_column_count: true, |
| 34 | + relax_quotes: true, |
| 35 | + skip_records_with_error: true, |
| 36 | + cast: false, |
| 37 | + delimiter, |
| 38 | + }) as Record<string, unknown>[] |
| 39 | + |
| 40 | + if (parsed.length === 0) { |
| 41 | + throw new Error('CSV file has no data rows') |
| 42 | + } |
| 43 | + |
| 44 | + const headers = Object.keys(parsed[0]) |
| 45 | + if (headers.length === 0) { |
| 46 | + throw new Error('CSV file has no headers') |
| 47 | + } |
| 48 | + |
| 49 | + return { headers, rows: parsed } |
| 50 | +} |
| 51 | + |
| 52 | +function inferColumnType(values: unknown[]): ColumnType { |
| 53 | + const nonEmpty = values.filter((v) => v !== null && v !== undefined && v !== '') |
| 54 | + if (nonEmpty.length === 0) return 'string' |
| 55 | + |
| 56 | + const allNumber = nonEmpty.every((v) => { |
| 57 | + const n = Number(v) |
| 58 | + return !Number.isNaN(n) && String(v).trim() !== '' |
| 59 | + }) |
| 60 | + if (allNumber) return 'number' |
| 61 | + |
| 62 | + const allBoolean = nonEmpty.every((v) => { |
| 63 | + const s = String(v).toLowerCase() |
| 64 | + return s === 'true' || s === 'false' |
| 65 | + }) |
| 66 | + if (allBoolean) return 'boolean' |
| 67 | + |
| 68 | + const isoDatePattern = /^\d{4}-\d{2}-\d{2}(T\d{2}:\d{2}(:\d{2})?)?/ |
| 69 | + const allDate = nonEmpty.every((v) => { |
| 70 | + const s = String(v) |
| 71 | + return isoDatePattern.test(s) && !Number.isNaN(Date.parse(s)) |
| 72 | + }) |
| 73 | + if (allDate) return 'date' |
| 74 | + |
| 75 | + return 'string' |
| 76 | +} |
| 77 | + |
| 78 | +function inferSchema(headers: string[], rows: Record<string, unknown>[]): ColumnDefinition[] { |
| 79 | + const sample = rows.slice(0, SCHEMA_SAMPLE_SIZE) |
| 80 | + const seen = new Set<string>() |
| 81 | + |
| 82 | + return headers.map((name) => { |
| 83 | + let colName = sanitizeName(name) |
| 84 | + let suffix = 2 |
| 85 | + while (seen.has(colName.toLowerCase())) { |
| 86 | + colName = `${sanitizeName(name)}_${suffix}` |
| 87 | + suffix++ |
| 88 | + } |
| 89 | + seen.add(colName.toLowerCase()) |
| 90 | + |
| 91 | + return { |
| 92 | + name: colName, |
| 93 | + type: inferColumnType(sample.map((r) => r[name])), |
| 94 | + } |
| 95 | + }) |
| 96 | +} |
| 97 | + |
| 98 | +/** |
| 99 | + * Strips non-alphanumeric characters (except underscore), collapses runs of |
| 100 | + * underscores, and ensures the name starts with a letter or underscore. |
| 101 | + * Used for both table names and column names to satisfy NAME_PATTERN. |
| 102 | + */ |
| 103 | +function sanitizeName(raw: string, fallbackPrefix = 'col'): string { |
| 104 | + let name = raw |
| 105 | + .trim() |
| 106 | + .replace(/[^a-zA-Z0-9_]/g, '_') |
| 107 | + .replace(/_+/g, '_') |
| 108 | + .replace(/^_+|_+$/g, '') |
| 109 | + |
| 110 | + if (!name || /^\d/.test(name)) { |
| 111 | + name = `${fallbackPrefix}_${name}` |
| 112 | + } |
| 113 | + |
| 114 | + return name |
| 115 | +} |
| 116 | + |
| 117 | +function coerceValue(value: unknown, colType: ColumnType): string | number | boolean | null { |
| 118 | + if (value === null || value === undefined || value === '') return null |
| 119 | + switch (colType) { |
| 120 | + case 'number': { |
| 121 | + const n = Number(value) |
| 122 | + return Number.isNaN(n) ? null : n |
| 123 | + } |
| 124 | + case 'boolean': { |
| 125 | + const s = String(value).toLowerCase() |
| 126 | + if (s === 'true') return true |
| 127 | + if (s === 'false') return false |
| 128 | + return null |
| 129 | + } |
| 130 | + case 'date': { |
| 131 | + const d = new Date(String(value)) |
| 132 | + return Number.isNaN(d.getTime()) ? String(value) : d.toISOString() |
| 133 | + } |
| 134 | + default: |
| 135 | + return String(value) |
| 136 | + } |
| 137 | +} |
| 138 | + |
| 139 | +function coerceRows( |
| 140 | + rows: Record<string, unknown>[], |
| 141 | + columns: ColumnDefinition[], |
| 142 | + headerToColumn: Map<string, string> |
| 143 | +): RowData[] { |
| 144 | + const colTypeMap = new Map(columns.map((c) => [c.name, c.type as ColumnType])) |
| 145 | + |
| 146 | + return rows.map((row) => { |
| 147 | + const coerced: RowData = {} |
| 148 | + for (const [header, value] of Object.entries(row)) { |
| 149 | + const colName = headerToColumn.get(header) |
| 150 | + if (colName) { |
| 151 | + coerced[colName] = coerceValue(value, colTypeMap.get(colName) ?? 'string') |
| 152 | + } |
| 153 | + } |
| 154 | + return coerced |
| 155 | + }) |
| 156 | +} |
| 157 | + |
| 158 | +export async function POST(request: NextRequest) { |
| 159 | + const requestId = generateRequestId() |
| 160 | + |
| 161 | + try { |
| 162 | + const authResult = await checkSessionOrInternalAuth(request, { requireWorkflowId: false }) |
| 163 | + if (!authResult.success || !authResult.userId) { |
| 164 | + return NextResponse.json({ error: 'Authentication required' }, { status: 401 }) |
| 165 | + } |
| 166 | + |
| 167 | + const formData = await request.formData() |
| 168 | + const file = formData.get('file') |
| 169 | + const workspaceId = formData.get('workspaceId') as string | null |
| 170 | + |
| 171 | + if (!file || !(file instanceof File)) { |
| 172 | + return NextResponse.json({ error: 'CSV file is required' }, { status: 400 }) |
| 173 | + } |
| 174 | + |
| 175 | + if (file.size > MAX_CSV_FILE_SIZE) { |
| 176 | + return NextResponse.json( |
| 177 | + { error: `File exceeds maximum allowed size of ${MAX_CSV_FILE_SIZE / (1024 * 1024)} MB` }, |
| 178 | + { status: 400 } |
| 179 | + ) |
| 180 | + } |
| 181 | + |
| 182 | + if (!workspaceId) { |
| 183 | + return NextResponse.json({ error: 'Workspace ID is required' }, { status: 400 }) |
| 184 | + } |
| 185 | + |
| 186 | + const permission = await getUserEntityPermissions(authResult.userId, 'workspace', workspaceId) |
| 187 | + if (permission !== 'write' && permission !== 'admin') { |
| 188 | + return NextResponse.json({ error: 'Access denied' }, { status: 403 }) |
| 189 | + } |
| 190 | + |
| 191 | + const ext = file.name.split('.').pop()?.toLowerCase() |
| 192 | + if (ext !== 'csv' && ext !== 'tsv') { |
| 193 | + return NextResponse.json({ error: 'Only CSV and TSV files are supported' }, { status: 400 }) |
| 194 | + } |
| 195 | + |
| 196 | + const buffer = Buffer.from(await file.arrayBuffer()) |
| 197 | + const delimiter = ext === 'tsv' ? '\t' : ',' |
| 198 | + const { headers, rows } = await parseCsvBuffer(buffer, delimiter) |
| 199 | + |
| 200 | + const columns = inferSchema(headers, rows) |
| 201 | + const headerToColumn = new Map(headers.map((h, i) => [h, columns[i].name])) |
| 202 | + |
| 203 | + const tableName = sanitizeName(file.name.replace(/\.[^.]+$/, ''), 'imported_table') |
| 204 | + const planLimits = await getWorkspaceTableLimits(workspaceId) |
| 205 | + |
| 206 | + const normalizedSchema: TableSchema = { |
| 207 | + columns: columns.map(normalizeColumn), |
| 208 | + } |
| 209 | + |
| 210 | + const table = await createTable( |
| 211 | + { |
| 212 | + name: tableName, |
| 213 | + description: `Imported from ${file.name}`, |
| 214 | + schema: normalizedSchema, |
| 215 | + workspaceId, |
| 216 | + userId: authResult.userId, |
| 217 | + maxRows: planLimits.maxRowsPerTable, |
| 218 | + maxTables: planLimits.maxTables, |
| 219 | + }, |
| 220 | + requestId |
| 221 | + ) |
| 222 | + |
| 223 | + try { |
| 224 | + const coerced = coerceRows(rows, columns, headerToColumn) |
| 225 | + let inserted = 0 |
| 226 | + for (let i = 0; i < coerced.length; i += MAX_BATCH_SIZE) { |
| 227 | + const batch = coerced.slice(i, i + MAX_BATCH_SIZE) |
| 228 | + const batchRequestId = crypto.randomUUID().slice(0, 8) |
| 229 | + const result = await batchInsertRows( |
| 230 | + { tableId: table.id, rows: batch, workspaceId, userId: authResult.userId }, |
| 231 | + table, |
| 232 | + batchRequestId |
| 233 | + ) |
| 234 | + inserted += result.length |
| 235 | + } |
| 236 | + |
| 237 | + logger.info(`[${requestId}] CSV imported`, { |
| 238 | + tableId: table.id, |
| 239 | + fileName: file.name, |
| 240 | + columns: columns.length, |
| 241 | + rows: inserted, |
| 242 | + }) |
| 243 | + |
| 244 | + return NextResponse.json({ |
| 245 | + success: true, |
| 246 | + data: { |
| 247 | + table: { |
| 248 | + id: table.id, |
| 249 | + name: table.name, |
| 250 | + description: table.description, |
| 251 | + schema: normalizedSchema, |
| 252 | + rowCount: inserted, |
| 253 | + }, |
| 254 | + }, |
| 255 | + }) |
| 256 | + } catch (insertError) { |
| 257 | + await deleteTable(table.id, requestId).catch(() => {}) |
| 258 | + throw insertError |
| 259 | + } |
| 260 | + } catch (error) { |
| 261 | + const message = error instanceof Error ? error.message : String(error) |
| 262 | + logger.error(`[${requestId}] CSV import failed:`, error) |
| 263 | + |
| 264 | + const isClientError = |
| 265 | + message.includes('maximum table limit') || |
| 266 | + message.includes('CSV file has no') || |
| 267 | + message.includes('Invalid table name') || |
| 268 | + message.includes('Invalid schema') || |
| 269 | + message.includes('already exists') |
| 270 | + |
| 271 | + return NextResponse.json( |
| 272 | + { error: isClientError ? message : 'Failed to import CSV' }, |
| 273 | + { status: isClientError ? 400 : 500 } |
| 274 | + ) |
| 275 | + } |
| 276 | +} |
0 commit comments