nocodb/packages/nc-gui/utils/parsers/ExcelTemplateAdapter.ts

import { UITypes } from 'nocodb-sdk'
import TemplateGenerator from './TemplateGenerator'
import {
  extractMultiOrSingleSelectProps,
  getCheckboxValue,
  getDateFormat,
  isCheckboxType,
  isEmailType,
  isMultiLineTextType,
  isUrlType,
} from '#imports'

const excelTypeToUidt: Record<string, UITypes> = {
  d: UITypes.DateTime,
  b: UITypes.Checkbox,
  n: UITypes.Number,
  s: UITypes.SingleLineText,
}

export default class ExcelTemplateAdapter extends TemplateGenerator {
  config: Record<string, any>

  excelData: any

  project: {
    tables: Record<string, any>[]
  }

  data: Record<string, any> = {}

  wb: any

  xlsx: typeof import('xlsx')

  constructor(data = {}, parserConfig = {}) {
    super()
    this.config = parserConfig
    this.excelData = data
    this.project = {
      tables: [],
    }
    this.xlsx = {} as any
  }

  async init() {
    this.xlsx = await import('xlsx')

    const options = {
      cellText: true,
      cellDates: true,
    }

    this.wb = this.xlsx.read(new Uint8Array(this.excelData), {
      type: 'array',
      ...options,
    })
  }

  async parse() {
    const tableNamePrefixRef: Record<string, any> = {}
    this.wb.SheetNames.reduce((acc: any, sheet: any) => {
      return acc.then(
        () =>
          new Promise((resolve) => {
            const columnNamePrefixRef: Record<string, any> = { id: 0 }
            let tn: string = (sheet || 'table').replace(/[` ~!@#$%^&*()_|+\-=?;:'",.<>\{\}\[\]\\\/]/g, '_').trim()

            while (tn in tableNamePrefixRef) {
              tn = `${tn}${++tableNamePrefixRef[tn]}`
            }
            tableNamePrefixRef[tn] = 0

            const table = { table_name: tn, ref_table_name: tn, columns: [] as any[] }
            const ws: any = this.wb.Sheets[sheet]
            const range = this.xlsx.utils.decode_range(ws['!ref'])
            let rows: any = this.xlsx.utils.sheet_to_json(ws, { header: 1, blankrows: false, defval: null })

            // fix precision bug & timezone offset issues introduced by xlsx
            const basedate = new Date(1899, 11, 30, 0, 0, 0)
            // number of milliseconds since base date
            const dnthresh = basedate.getTime() + (new Date().getTimezoneOffset() - basedate.getTimezoneOffset()) * 60000
            // number of milliseconds in a day
            const day_ms = 24 * 60 * 60 * 1000
            // handle date1904 property
            const fixImportedDate = (date: Date) => {
              const parsed = this.xlsx.SSF.parse_date_code((date.getTime() - dnthresh) / day_ms, {
                date1904: this.wb.Workbook.WBProps.date1904,
              })
              return new Date(parsed.y, parsed.m, parsed.d, parsed.H, parsed.M, parsed.S)
            }

            // fix imported date
            rows = rows.map((r: any) =>
              r.map((v: any) => {
                return v instanceof Date ? fixImportedDate(v) : v
              }),
            )

            const columnNameRowExist = +rows[0].every((v: any) => v === null || typeof v === 'string')

            for (let col = 0; col < rows[0].length; col++) {
              let cn: string = (
                (columnNameRowExist && rows[0] && rows[0][col] && rows[0][col].toString().trim()) ||
                `field_${col + 1}`
              )
                .replace(/[` ~!@#$%^&*()_|+\-=?;:'",.<>\{\}\[\]\\\/]/g, '_')
                .trim()

              while (cn in columnNamePrefixRef) {
                cn = `${cn}${++columnNamePrefixRef[cn]}`
              }
              columnNamePrefixRef[cn] = 0

              const column: Record<string, any> = {
                column_name: cn,
                ref_column_name: cn,
                meta: {},
                uidt: UITypes.SingleLineText,
              }

              if (this.config.autoSelectFieldTypes) {
                const cellId = this.xlsx.utils.encode_cell({
                  c: range.s.c + col,
                  r: columnNameRowExist,
                })
                const cellProps = ws[cellId] || {}
                column.uidt = excelTypeToUidt[cellProps.t] || UITypes.SingleLineText

                if (column.uidt === UITypes.SingleLineText) {
                  // check for long text
                  if (isMultiLineTextType(rows)) {
                    column.uidt = UITypes.LongText
                  }

                  if (isEmailType(rows)) {
                    column.uidt = UITypes.Email
                  }

                  if (isUrlType(rows)) {
                    column.uidt = UITypes.URL
                  } else {
                    const vals = rows
                      .slice(columnNameRowExist ? 1 : 0)
                      .map((r: any) => r[col])
                      .filter((v: any) => v !== null && v !== undefined && v.toString().trim() !== '')

                    const checkboxType = isCheckboxType(vals)
                    if (checkboxType.length === 1) {
                      column.uidt = UITypes.Checkbox
                    } else {
                      // Single Select / Multi Select
                      Object.assign(column, extractMultiOrSingleSelectProps(vals))
                    }
                  }
                } else if (column.uidt === UITypes.Number) {
                  if (
                    rows.slice(1, this.config.maxRowsToParse).some((v: any) => {
                      return v && v[col] && parseInt(v[col]) !== +v[col]
                    })
                  ) {
                    column.uidt = UITypes.Decimal
                  }
                  if (
                    rows.slice(1, this.config.maxRowsToParse).every((v: any, i: any) => {
                      const cellId = this.xlsx.utils.encode_cell({
                        c: range.s.c + col,
                        r: i + columnNameRowExist,
                      })

                      const cellObj = ws[cellId]

                      return !cellObj || (cellObj.w && cellObj.w.startsWith('$'))
                    })
                  ) {
                    column.uidt = UITypes.Currency
                  }
                } else if (column.uidt === UITypes.DateTime) {
                  // TODO(import): centralise
                  // hold the possible date format found in the date
                  const dateFormat: Record<string, number> = {}
                  if (
                    rows.slice(1, this.config.maxRowsToParse).every((v: any, i: any) => {
                      const cellId = this.xlsx.utils.encode_cell({
                        c: range.s.c + col,
                        r: i + columnNameRowExist,
                      })

                      const cellObj = ws[cellId]
                      const isDate = !cellObj || (cellObj.w && cellObj.w.split(' ').length === 1)
                      if (isDate && cellObj) {
                        dateFormat[getDateFormat(cellObj.w)] = (dateFormat[getDateFormat(cellObj.w)] || 0) + 1
                      }
                      return isDate
                    })
                  ) {
                    column.uidt = UITypes.Date
                    // take the date format with the max occurrence
                    column.meta.date_format =
                      Object.keys(dateFormat).reduce((x, y) => (dateFormat[x] > dateFormat[y] ? x : y)) || 'YYYY/MM/DD'
                  }
                }
              }
              table.columns.push(column)
            }
            this.project.tables.push(table)

            this.data[tn] = []
            if (this.config.shouldImportData) {
              let rowIndex = 0
              for (const row of rows.slice(1)) {
                const rowData: Record<string, any> = {}
                for (let i = 0; i < table.columns.length; i++) {
                  if (!this.config.autoSelectFieldTypes) {
                    // take raw data instead of data parsed by xlsx
                    const cellId = this.xlsx.utils.encode_cell({
                      c: range.s.c + i,
                      r: rowIndex + columnNameRowExist,
                    })
                    const cellObj = ws[cellId]
                    rowData[table.columns[i].column_name] = (cellObj && cellObj.w) || row[i]
                  } else {
                    if (table.columns[i].uidt === UITypes.Checkbox) {
                      rowData[table.columns[i].column_name] = getCheckboxValue(row[i])
                    } else if (table.columns[i].uidt === UITypes.Currency) {
                      const cellId = this.xlsx.utils.encode_cell({
                        c: range.s.c + i,
                        r: rowIndex + columnNameRowExist,
                      })

                      const cellObj = ws[cellId]
                      rowData[table.columns[i].column_name] =
                        (cellObj && cellObj.w && cellObj.w.replace(/[^\d.]+/g, '')) || row[i]
                    } else if (table.columns[i].uidt === UITypes.SingleSelect || table.columns[i].uidt === UITypes.MultiSelect) {
                      rowData[table.columns[i].column_name] = (row[i] || '').toString().trim() || null
                    } else if (table.columns[i].uidt === UITypes.Date) {
                      const cellId = this.xlsx.utils.encode_cell({
                        c: range.s.c + i,
                        r: rowIndex + columnNameRowExist,
                      })
                      const cellObj = ws[cellId]
                      rowData[table.columns[i].column_name] = (cellObj && cellObj.w) || row[i]
                    } else {
                      // TODO: do parsing if necessary based on type
                      rowData[table.columns[i].column_name] = row[i]
                    }
                  }
                }
                this.data[tn].push(rowData)
                rowIndex++
              }
            }

            resolve(true)
          }),
      )
    }, Promise.resolve())
  }

  getTemplate() {
    return this.project
  }

  getData() {
    return this.data
  }

  getColumns() {
    return this.project.tables.map((t: Record<string, any>) => t.columns)
  }
}