dnrops
/
CSV
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Organize data and decode with byte (UInt8) arrays instead of Data. Support quote escapes for newlines and commas

This commit is contained in:
Caleb Kleveter 2018-06-01 17:42:33 -05:00
parent 60eca9a4cc
commit 23bd35d662
No known key found for this signature in database
GPG Key ID: B38DBD5CF2C98D69
6 changed files with 135 additions and 93 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

115
Sources/CSV/CSVParser.swift
View File

@ -1,72 +1,79 @@
import Foundation
extension CSV {
public static func parse(_ data: Data) -> [String: [String?]] {
let rows = data.split(separator: .newLine, omittingEmptySubsequences: false)
var cells = rows.map({ $0.split(separator: .comma, omittingEmptySubsequences: false) })
let rowLength = cells[0].count - 1
public static func parse(_ data: Data, stringEncoding: String.Encoding = .utf8) -> [String: [String?]] {
let end = data.endIndex
for count in 1...cells.count - 1 {
if cells[cells.count - count].count < rowLength {
_ = cells.removeLast()
} else {
break
var columns: [(title: String, cells: [String?])] = []
var columnIndex = 0
var iterator = data.startIndex
var inQuotes = false
var currentCell: Bytes = []
header: while iterator < end {
let byte = data[iterator]
switch byte {
case .quote: inQuotes = !inQuotes
case .comma, .newLine:
if inQuotes { currentCell.append(byte); break }
guard let title = String(data: Data(currentCell), encoding: stringEncoding) else { return [:] }
columns.append((title, []))
currentCell = []
if byte == .newLine { iterator += 1; break header }
default: currentCell.append(byte)
}
iterator += 1
}
var columns: [String: [String?]] = [:]
(0...rowLength).forEach { (cellIndex) in
var column = cells.map({ (row) -> String? in
return row[cellIndex].count > 0 ? String(data: row[cellIndex], encoding: .utf8) : nil
})
let title = column.removeFirst()!
columns[title] = column
while iterator < end {
let byte = data[iterator]
switch byte {
case .quote: inQuotes = !inQuotes
case .comma:
if inQuotes { currentCell.append(.comma); break }
columns[columnIndex].cells.append(currentCell.count > 0 ? nil : String(data: Data(currentCell), encoding: stringEncoding))
columnIndex += 1
currentCell = []
case .newLine:
if inQuotes { currentCell.append(.newLine); break }
columns[columnIndex].cells.append(currentCell.count > 0 ? nil : String(data: Data(currentCell), encoding: stringEncoding))
columnIndex = 0
currentCell = []
default: currentCell.append(byte)
}
iterator += 1
}
return columns
var dictionaryResult: [String: [String?]] = [:]
var resultIterator = columns.startIndex
while resultIterator < columns.endIndex {
let column = columns[resultIterator]
dictionaryResult[column.title] = column.cells
resultIterator += 1
}
return dictionaryResult
}
public static func parse(_ data: Data) -> [String: Column] {
let rows = data.split(separator: .newLine, omittingEmptySubsequences: false)
var cells = rows.map({ $0.split(separator: .comma, omittingEmptySubsequences: false) })
let rowLength = cells[0].count - 1
public static func parse(_ data: Data, stringEncoding: String.Encoding = .utf8) -> [String: Column] {
let elements: [String: [String?]] = self.parse(data, stringEncoding: stringEncoding)
for count in 1...cells.count - 1 {
if cells[cells.count - count].count < rowLength {
_ = cells.removeLast()
} else {
break
}
return elements.reduce(into: [:]) { columns, element in
columns[element.key] = Column(header: element.key, fields: element.value)
}
var columns: [String: Column] = [:]
(0...rowLength).forEach { (cellIndex) in
var column = cells.map({ (row) -> String? in
return row[cellIndex].count > 0 ? String(data: row[cellIndex], encoding: .utf8) : nil
})
let title = column.removeFirst()!
columns[title] = CSV.Column(header: title, fields: column)
}
return columns
}
public static func parse(_ data: Data) -> [Column] {
let rows = data.split(separator: .newLine, omittingEmptySubsequences: false)
var cells = rows.map({ $0.split(separator: .comma, omittingEmptySubsequences: false) })
let rowLength = cells[0].count - 1
public static func parse(_ data: Data, stringEncoding: String.Encoding = .utf8) -> [Column] {
let elements: [String: [String?]] = self.parse(data, stringEncoding: stringEncoding)
for count in 1...cells.count - 1 {
if cells[cells.count - count].count < rowLength {
_ = cells.removeLast()
} else {
break
}
}
return (0...rowLength).map { (cellIndex) -> CSV.Column in
var column = cells.map({ (row) -> String? in
return row[cellIndex].count > 0 ? String(data: row[cellIndex], encoding: .utf8) : nil
})
return CSV.Column(header: column.removeFirst()!, fields: column)
return elements.reduce(into: []) { columns, element in
columns.append(Column(header: element.key, fields: element.value))
}
}
}

2
Sources/CSV/Coder/Decoder/Data+Number.swift
View File

@ -1,6 +1,6 @@
import Foundation
extension Data {
extension Array where Element == UInt8 {
var int: Int? {
let count: Int = self.endIndex
var result: Int = 0

85
Sources/CSV/Coder/Decoder/_CSVDecoder.swift
View File

@ -7,11 +7,11 @@ final class _CSVDecoder: Decoder {
let userInfo: [CodingUserInfoKey : Any]
let stringDecoding: String.Encoding
let csv: [String: [Data?]]?
let row: [String: Data]?
let cell: Data?
let csv: [String: [Bytes?]]?
let row: [String: Bytes]?
let cell: Bytes?
init(csv: [String: [Data?]], path: CodingPath = [], info: [CodingUserInfoKey : Any] = [:], stringDecoding: String.Encoding) {
init(csv: [String: [Bytes?]], path: CodingPath = [], info: [CodingUserInfoKey : Any] = [:], stringDecoding: String.Encoding) {
self.codingPath = path
self.userInfo = info
self.stringDecoding = stringDecoding
@ -20,7 +20,7 @@ final class _CSVDecoder: Decoder {
self.cell = nil
}
init(row: [String: Data], path: CodingPath = [], info: [CodingUserInfoKey : Any] = [:], stringDecoding: String.Encoding) {
init(row: [String: Bytes], path: CodingPath = [], info: [CodingUserInfoKey : Any] = [:], stringDecoding: String.Encoding) {
self.codingPath = path
self.userInfo = info
self.stringDecoding = stringDecoding
@ -29,7 +29,7 @@ final class _CSVDecoder: Decoder {
self.cell = nil
}
init(cell: Data?, path: CodingPath = [], info: [CodingUserInfoKey : Any] = [:], stringDecoding: String.Encoding) {
init(cell: Bytes?, path: CodingPath = [], info: [CodingUserInfoKey : Any] = [:], stringDecoding: String.Encoding) {
self.codingPath = path
self.userInfo = info
self.stringDecoding = stringDecoding
@ -77,34 +77,69 @@ final class _CSVDecoder: Decoder {
}
static func decode<T>(_ type: T.Type, from data: Data, stringDecoding: String.Encoding)throws -> [T] where T: Decodable {
let csv: [String: [Data?]] = try _CSVDecoder.organize(data, stringDecoding: stringDecoding)
let csv: [String: [Bytes?]] = try _CSVDecoder.organize(data, stringDecoding: stringDecoding)
let decoder = _CSVDecoder(csv: csv, stringDecoding: stringDecoding)
return try Array<T>(from: decoder)
}
static func organize(_ data: Data, stringDecoding: String.Encoding)throws -> [String: [Data?]] {
let rows = data.split(separator: .newLine, omittingEmptySubsequences: false)
var cells = rows.map({ $0.split(separator: .comma, omittingEmptySubsequences: false) })
let rowLength = cells[0].count - 1
static func organize(_ data: Data, stringDecoding: String.Encoding)throws -> [String: [Bytes?]] {
let end = data.endIndex
for count in 1...cells.count - 1 {
if cells[cells.count - count].count < rowLength {
_ = cells.removeLast()
} else {
break
var columns: [(title: String, cells: [Bytes?])] = []
var columnIndex = 0
var iterator = data.startIndex
var inQuotes = false
var currentCell: Bytes = []
header: while iterator < end {
let byte = data[iterator]
switch byte {
case .quote: inQuotes = !inQuotes
case .comma, .newLine:
if inQuotes { currentCell.append(byte); break }
guard let title = String(data: Data(currentCell), encoding: stringDecoding) else {
fatalError()
}
columns.append((title, []))
currentCell = []
if byte == .newLine { iterator += 1; break header }
default: currentCell.append(byte)
}
iterator += 1
}
var columns: [String: [Data?]] = [:]
try (0...rowLength).forEach { (cellIndex) in
var column = cells.map({ (row) -> Data? in
return row[cellIndex].count > 0 ? row[cellIndex] : nil
})
guard let title = String(data: column.removeFirst()!, encoding: stringDecoding) else {
throw DecodingError.dataCorrupted(DecodingError.Context(codingPath: [], debugDescription: "Found colunm title with \(stringDecoding) incompatible character"))
while iterator < end {
let byte = data[iterator]
switch byte {
case .quote: inQuotes = !inQuotes
case .comma:
if inQuotes { currentCell.append(.comma); break }
columns[columnIndex].cells.append(currentCell.count > 0 ? nil : currentCell)
columnIndex += 1
currentCell = []
case .newLine:
if inQuotes { currentCell.append(.newLine); break }
columns[columnIndex].cells.append(currentCell.count > 0 ? nil : currentCell)
columnIndex = 0
currentCell = []
default: currentCell.append(byte)
}
columns[title] = column
iterator += 1
}
return columns
var dictionaryResult: [String: [Bytes?]] = [:]
var resultIterator = columns.startIndex
while resultIterator < columns.endIndex {
let column = columns[resultIterator]
dictionaryResult[column.title] = column.cells
resultIterator += 1
}
return dictionaryResult
}
}

10
Sources/CSV/Coder/Decoder/_CSVKeyedDecoder.swift
View File

@ -4,9 +4,9 @@ final class _CSVKeyedDecoder<K>: KeyedDecodingContainerProtocol where K: CodingK
let codingPath: [CodingKey]
let allKeys: [K]
let stringDecoding: String.Encoding
let row: [String: Data]
let row: [String: Bytes]
init(path: CodingPath, row: [String: Data], stringDecoding: String.Encoding) {
init(path: CodingPath, row: [String: Bytes], stringDecoding: String.Encoding) {
self.codingPath = path
self.allKeys = Array(row.keys).compactMap(K.init)
self.stringDecoding = stringDecoding
@ -19,12 +19,12 @@ final class _CSVKeyedDecoder<K>: KeyedDecodingContainerProtocol where K: CodingK
func decodeNil(forKey key: K) throws -> Bool {
let cell = row[key.stringValue]
return cell == nil || cell == Data([.N, .forwardSlash, .A]) || cell == Data([.N, .A])
return cell == nil || cell == [.N, .forwardSlash, .A] || cell == [.N, .A]
}
func decode(_ type: Bool.Type, forKey key: K) throws -> Bool {
guard let cell = row[key.stringValue] else { throw DecodingError.badKey(key, at: self.codingPath + [key]) }
guard let value = String(data: cell, encoding: self.stringDecoding) else {
guard let value = String(data: Data(cell), encoding: self.stringDecoding) else {
throw DecodingError.dataToStringFailed(path: self.codingPath + [key], encoding: self.stringDecoding)
}
switch value.lowercased() {
@ -36,7 +36,7 @@ final class _CSVKeyedDecoder<K>: KeyedDecodingContainerProtocol where K: CodingK
func decode(_ type: String.Type, forKey key: K) throws -> String {
guard let cell = row[key.stringValue] else { throw DecodingError.badKey(key, at: self.codingPath + [key]) }
guard let value = String(data: cell, encoding: self.stringDecoding) else {
guard let value = String(data: Data(cell), encoding: self.stringDecoding) else {
throw DecodingError.dataToStringFailed(path: self.codingPath + [key], encoding: self.stringDecoding)
}
return value

10
Sources/CSV/Coder/Decoder/_CSVSingleValueDecoder.swift
View File

@ -3,21 +3,21 @@ import Foundation
final class _CSVSingleValueDecoder: SingleValueDecodingContainer {
let codingPath: [CodingKey]
let stringDecoding: String.Encoding
let value: Data?
let value: Bytes?
init(value: Data?, path: CodingPath, stringDecoding: String.Encoding) {
init(value: Bytes?, path: CodingPath, stringDecoding: String.Encoding) {
self.codingPath = path
self.stringDecoding = stringDecoding
self.value = value
}
func decodeNil() -> Bool {
return value == nil || value == Data([.N, .forwardSlash, .A]) || value == Data([.N, .A])
return value == nil || value == [.N, .forwardSlash, .A] || value == [.N, .A]
}
func decode(_ type: Bool.Type) throws -> Bool {
guard let cell = self.value else { throw DecodingError.nilValue(type: type, at: self.codingPath) }
guard let value = String(data: cell, encoding: self.stringDecoding) else {
guard let value = String(data: Data(cell), encoding: self.stringDecoding) else {
throw DecodingError.dataToStringFailed(path: self.codingPath, encoding: self.stringDecoding)
}
switch value.lowercased() {
@ -29,7 +29,7 @@ final class _CSVSingleValueDecoder: SingleValueDecodingContainer {
func decode(_ type: String.Type) throws -> String {
guard let cell = self.value else { throw DecodingError.nilValue(type: type, at: self.codingPath) }
guard let value = String(data: cell, encoding: self.stringDecoding) else {
guard let value = String(data: Data(cell), encoding: self.stringDecoding) else {
throw DecodingError.dataToStringFailed(path: self.codingPath, encoding: self.stringDecoding)
}
return value

6
Sources/CSV/Coder/Decoder/_CSVUnKeyedDecoder.swift
View File

@ -7,10 +7,10 @@ final class _CSVUnkeyedDecoder: UnkeyedDecodingContainer {
var currentIndex: Int
let stringDecoding: String.Encoding
let columns: [String: [Data?]]
let next: () -> [String: Data]?
let columns: [String: [Bytes?]]
let next: () -> [String: Bytes]?
init(columns: [String: [Data?]], path: CodingPath = [], stringDecoding: String.Encoding) {
init(columns: [String: [Bytes?]], path: CodingPath = [], stringDecoding: String.Encoding) {
self.codingPath = path
self.count = columns.first?.value.count
self.currentIndex = 0