Merge pull request #615 from sasjs/issue-607

Support special missing values

src/SASjs.ts

@@ -736,15 +736,19 @@ export default class SASjs {
     msg: string
   } {
     if (data === null) return { status: true, msg: '' }
+
+    const isSasFormatsTable = (key: string) =>
+      key.match(/^\$.*/) && Object.keys(data).includes(key.replace(/^\$/, ''))
+
     for (const key in data) {
-      if (!key.match(/^[a-zA-Z_]/)) {
+      if (!key.match(/^[a-zA-Z_]/) && !isSasFormatsTable(key)) {
         return {
           status: false,
           msg: 'First letter of table should be alphabet or underscore.'
         }
       }
 
-      if (!key.match(/^[a-zA-Z_][a-zA-Z0-9_]*$/)) {
+      if (!key.match(/^[a-zA-Z_][a-zA-Z0-9_]*$/) && !isSasFormatsTable(key)) {
         return { status: false, msg: 'Table name should be alphanumeric.' }
       }
 
@@ -755,7 +759,7 @@ export default class SASjs {
         }
       }
 
-      if (this.getType(data[key]) !== 'Array') {
+      if (this.getType(data[key]) !== 'Array' && !isSasFormatsTable(key)) {
         return {
           status: false,
           msg: 'Parameter data contains invalid table structure.'
@@ -771,6 +775,7 @@ export default class SASjs {
         }
       }
     }
+
     return { status: true, msg: '' }
   }
 

src/test/utils/formatDataForRequest.spec.ts

@@ -0,0 +1,95 @@
+import { formatDataForRequest } from '../../utils/formatDataForRequest'
+
+describe('formatDataForRequest', () => {
+  const testTable = 'sometable'
+
+  it('should format table with special missing values', () => {
+    const tableWithMissingValues = {
+      [testTable]: [
+        { var1: 'string', var2: 232, nullvar: 'A' },
+        { var1: 'string', var2: 232, nullvar: 'B' },
+        { var1: 'string', var2: 232, nullvar: '_' },
+        { var1: 'string', var2: 232, nullvar: 0 },
+        { var1: 'string', var2: 232, nullvar: 'z' },
+        { var1: 'string', var2: 232, nullvar: null }
+      ],
+      [`$${testTable}`]: { formats: { var1: '$char12.', nullvar: 'best.' } }
+    }
+
+    const expectedOutput = {
+      sasjs1data: `var1:$char12. var2:best. nullvar:best.\r\nstring,232,.a\r\nstring,232,.b\r\nstring,232,._\r\nstring,232,0\r\nstring,232,.z\r\nstring,232,.`,
+      sasjs_tables: testTable
+    }
+
+    expect(formatDataForRequest(tableWithMissingValues)).toEqual(expectedOutput)
+  })
+
+  it('should return error if string is more than 32765 characters', () => {
+    const data = { testTable: [{ var1: 'z'.repeat(32765 + 1) }] }
+
+    expect(() => formatDataForRequest(data)).toThrow(
+      new Error(
+        'The max length of a string value in SASjs is 32765 characters.'
+      )
+    )
+  })
+
+  it('should return error if string is more than 32765 characters', () => {
+    const charsCount = 16 * 1000 + 1
+    const allChars = 'z'.repeat(charsCount)
+    const data = { [testTable]: [{ var1: allChars }] }
+    const firstChunk = `var1:$char${charsCount}.\r\n`
+    const firstChunkChars = 'z'.repeat(16000 - firstChunk.length)
+    const secondChunkChars = 'z'.repeat(
+      charsCount - (16000 - firstChunk.length)
+    )
+
+    const expectedOutput = {
+      sasjs1data0: 2,
+      sasjs1data1: `${firstChunk}${firstChunkChars}`,
+      sasjs1data2: secondChunkChars,
+      sasjs_tables: testTable
+    }
+
+    expect(formatDataForRequest(data)).toEqual(expectedOutput)
+  })
+
+  it('should throw an error if special missing values is not valid', () => {
+    let tableWithMissingValues = {
+      [testTable]: [{ var: 'AA' }, { var: 0 }],
+      [`$${testTable}`]: { formats: { var: 'best.' } }
+    }
+
+    expect(() => formatDataForRequest(tableWithMissingValues)).toThrow(
+      new Error(
+        'Special missing value can only be a single character from A to Z or _'
+      )
+    )
+  })
+
+  it('should auto-detect special missing values type as best.', () => {
+    const tableWithMissingValues = {
+      [testTable]: [{ var: 'a' }, { var: 'A' }, { var: '_' }, { var: 0 }]
+    }
+
+    const expectedOutput = {
+      sasjs1data: `var:best.\r\n.a\r\n.a\r\n._\r\n0`,
+      sasjs_tables: testTable
+    }
+
+    expect(formatDataForRequest(tableWithMissingValues)).toEqual(expectedOutput)
+  })
+
+  it('should auto-detect values type as $char1.', () => {
+    const tableWithMissingValues = {
+      [testTable]: [{ var: 'a' }, { var: 'A' }, { var: '_' }]
+    }
+
+    const expectedOutput = {
+      sasjs1data: `var:$char1.\r\na\r\nA\r\n_`,
+      sasjs_tables: testTable
+    }
+
+    expect(formatDataForRequest(tableWithMissingValues)).toEqual(expectedOutput)
+  })
+})

src/utils/convertToCsv.spec.ts

@@ -167,4 +167,17 @@ describe('convertToCsv', () => {
       convertToCSV([{ slashWithSpecialExtra: '\\\ts\tl\ta\ts\t\th\t' }])
     ).toEqual(`slashWithSpecialExtra:$char13.\r\n\"\\\ts\tl\ta\ts\t\th\t\"`)
   })
+
+  it('should console log error if data has mixed types', () => {
+    const colName = 'var1'
+    const data = [{ [colName]: 'string' }, { [colName]: 232 }]
+
+    jest.spyOn(console, 'error').mockImplementation(() => {})
+
+    convertToCSV(data)
+
+    expect(console.error).toHaveBeenCalledWith(
+      `Row (2), Column (${colName}) has mixed types: ERROR`
+    )
+  })
 })

src/utils/convertToCsv.ts

@@ -2,70 +2,115 @@
  * Converts the given JSON object array to a CSV string.
  * @param data - the array of JSON objects to convert.
  */
-export const convertToCSV = (data: any) => {
-  const replacer = (key: any, value: any) => (value === null ? '' : value)
-  const headerFields = Object.keys(data[0])
+export const convertToCSV = (
+  data: any,
+  sasFormats?: { formats: { [key: string]: string } }
+) => {
+  let formats = sasFormats?.formats
+  let headers: string[] = []
   let csvTest
   let invalidString = false
-  const headers = headerFields.map((field) => {
-    let firstFoundType: string | null = null
-    let hasMixedTypes: boolean = false
-    let rowNumError: number = -1
+  const specialMissingValueRegExp = /^[a-z_]{1}$/i
 
-    const longestValueForField = data
-      .map((row: any, index: number) => {
-        if (row[field] || row[field] === '') {
-          if (firstFoundType) {
-            let currentFieldType =
-              row[field] === '' || typeof row[field] === 'string'
-                ? 'chars'
-                : 'number'
+  if (formats) {
+    headers = Object.keys(formats).map((key) => `${key}:${formats![key]}`)
+  }
 
-            if (!hasMixedTypes) {
-              hasMixedTypes = currentFieldType !== firstFoundType
-              rowNumError = hasMixedTypes ? index + 1 : -1
-            }
-          } else {
-            if (row[field] === '') {
-              firstFoundType = 'chars'
-            } else {
-              firstFoundType =
-                typeof row[field] === 'string' ? 'chars' : 'number'
-            }
-          }
+  const headerFields = Object.keys(data[0])
 
-          let byteSize
+  headerFields.forEach((field) => {
+    if (!formats || !Object.keys(formats).includes(field)) {
+      let hasNullOrNumber = false
+      let hasSpecialMissingString = false
 
-          if (typeof row[field] === 'string') {
-            byteSize = getByteSize(row[field])
-          }
-
-          return byteSize
+      data.forEach((row: { [key: string]: any }) => {
+        if (row[field] === null || typeof row[field] === 'number') {
+          hasNullOrNumber = true
+        } else if (
+          typeof row[field] === 'string' &&
+          specialMissingValueRegExp.test(row[field])
+        ) {
+          hasSpecialMissingString = true
         }
       })
-      .sort((a: number, b: number) => b - a)[0]
-    if (longestValueForField && longestValueForField > 32765) {
-      invalidString = true
-    }
-    if (hasMixedTypes) {
-      console.error(
-        `Row (${rowNumError}), Column (${field}) has mixed types: ERROR`
-      )
-    }
 
-    return `${field}:${firstFoundType === 'chars' ? '$char' : ''}${
-      longestValueForField
-        ? longestValueForField
-        : firstFoundType === 'chars'
-        ? '1'
-        : 'best'
-    }.`
+      if (hasNullOrNumber && hasSpecialMissingString) {
+        headers.push(`${field}:best.`)
+
+        if (!formats) formats = {}
+
+        formats[field] = 'best.'
+      } else {
+        let firstFoundType: string | null = null
+        let hasMixedTypes: boolean = false
+        let rowNumError: number = -1
+
+        const longestValueForField = data
+          .map((row: any, index: number) => {
+            if (row[field] || row[field] === '') {
+              if (firstFoundType) {
+                let currentFieldType =
+                  row[field] === '' || typeof row[field] === 'string'
+                    ? 'chars'
+                    : 'number'
+
+                if (!hasMixedTypes) {
+                  hasMixedTypes = currentFieldType !== firstFoundType
+                  rowNumError = hasMixedTypes ? index + 1 : -1
+                }
+              } else {
+                if (row[field] === '') {
+                  firstFoundType = 'chars'
+                } else {
+                  firstFoundType =
+                    typeof row[field] === 'string' ? 'chars' : 'number'
+                }
+              }
+
+              let byteSize
+
+              if (typeof row[field] === 'string') {
+                byteSize = getByteSize(row[field])
+              }
+
+              return byteSize
+            }
+          })
+          .sort((a: number, b: number) => b - a)[0]
+
+        if (longestValueForField && longestValueForField > 32765) {
+          invalidString = true
+        }
+
+        if (hasMixedTypes) {
+          console.error(
+            `Row (${rowNumError}), Column (${field}) has mixed types: ERROR`
+          )
+        }
+
+        headers.push(
+          `${field}:${firstFoundType === 'chars' ? '$char' : ''}${
+            longestValueForField
+              ? longestValueForField
+              : firstFoundType === 'chars'
+              ? '1'
+              : 'best'
+          }.`
+        )
+      }
+    }
   })
 
-  if (invalidString) {
-    return 'ERROR: LARGE STRING LENGTH'
+  if (sasFormats) {
+    headers = headers.sort(
+      (a, b) =>
+        headerFields.indexOf(a.replace(/:.*/, '')) -
+        headerFields.indexOf(b.replace(/:.*/, ''))
+    )
   }
 
+  if (invalidString) return 'ERROR: LARGE STRING LENGTH'
+
   csvTest = data.map((row: any) => {
     const fields = Object.keys(row).map((fieldName, index) => {
       let value
@@ -76,6 +121,17 @@ export const convertToCSV = (data: any) => {
       // stringify with replacer converts null values to empty strings
       value = currentCell === null ? '' : currentCell
 
+      if (formats && formats[fieldName] === 'best.') {
+        if (value && !specialMissingValueRegExp.test(value)) {
+          console.log(`🤖[value]🤖`, value)
+          throw new Error(
+            'Special missing value can only be a single character from A to Z or _'
+          )
+        }
+
+        return `.${value.toLowerCase()}`
+      }
+
       // if there any present, it should have preceding (") for escaping
       value = value.replace(/"/g, `""`)
 

src/utils/formatDataForRequest.ts

@@ -7,9 +7,17 @@ export const formatDataForRequest = (data: any) => {
   const result: any = {}
 
   for (const tableName in data) {
+    if (
+      tableName.match(/^\$.*/) &&
+      Object.keys(data).includes(tableName.replace(/^\$/, ''))
+    ) {
+      continue
+    }
+
     tableCounter++
     sasjsTables.push(tableName)
-    const csv = convertToCSV(data[tableName])
+    const csv = convertToCSV(data[tableName], data[`$${tableName}`])
+
     if (csv === 'ERROR: LARGE STRING LENGTH') {
       throw new Error(
         'The max length of a string value in SASjs is 32765 characters.'
@@ -27,6 +35,7 @@ export const formatDataForRequest = (data: any) => {
       result[`sasjs${tableCounter}data`] = csv
     }
   }
+
   result['sasjs_tables'] = sasjsTables.join(' ')
 
   return result