Replace Trainer.playground with Trainer.swift

Update README.md to reflect improved training time

Author: mattt

Diff for: Classifier Demo.playground/Resources/ProgrammingLanguageClassifier.mlmodelc/coremldata.bin

@@ -47,10 +47,20 @@ $ cd Programming-Language-Classifier
 $ git submodule update --init
 ```
 
-- Open `Trainer.playground` and fill in the placeholder values
-  for `destinationPath` and `corpusPath`.
-- Run the playground and wait for the model to be trained
-  _(on a 2017 MacBook Pro, this took about an hour)_.
+- Open `Trainer.swift` in an editor and fill in the placeholder values
+  for `destinationPath` and `corpusPath`:
+
+```terminal
+$ open ./Trainer.swift
+```
+
+- Run `Trainer.swift` and wait for the model to be trained
+  _(on a 2017 MacBook Pro, this took a few minutes)_:
+
+```terminal
+$ swift ./Trainer.swift
+```
+
 - Compile the generated `.mlmodel` bundle using the following command:
 
 ```terminal

Diff for: ProgrammingLanguageClassifier.mlmodel

@@ -1,3 +1,5 @@
+#!/usr/bin/swift
+
 import Foundation
 
 guard #available(OSX 10.14, *) else {
@@ -7,54 +9,87 @@ guard #available(OSX 10.14, *) else {
 import CreateML
 import NaturalLanguage
 
-//let destinationPath = <#Path to Destination.mlmodel#>
+enum ProgrammingLanguage: String {
+    case c = "C"
+    case cPlusPlus = "C++"
+    case go = "Go"
+    case java = "Java"
+    case javaScript = "JavaScript"
+    case objectiveC = "Objective-C"
+    case php = "PHP"
+    case ruby = "Ruby"
+    case rust = "Rust"
+    case swift = "Swift"
+
+    init?(directory: String, fileExtension: String?) {
+        switch (directory, fileExtension) {
+        case ("c", "h"), (_, "c"): self = .c
+        case ("cc", "h"), (_, "cc"), (_, "cpp"): self = .cPlusPlus
+        case (_, "go"): self = .go
+        case (_, "java"): self = .java
+        case (_, "js"): self = .javaScript
+        case ("objective-c", "h"), (_, "m"): self = .objectiveC
+        case (_, "php"): self = .php
+        case (_, "rb"): self = .ruby
+        case (_, "rs"): self = .rust
+        case (_, "swift"): self = .swift
+        default:
+            return nil
+        }
+    }
+}
 
-//let corpusPath = "<#Path to Corpus Directory#>"
+let destinationPath = "/Users/mattt/Desktop/Classifier.mlmodel"
+
+let corpusPath = "/Users/mattt/Downloads/code-corpora"
 let corpusURL = URL(fileURLWithPath: corpusPath)
 
 let fileManager = FileManager.default
 
-try fileManager.contentsOfDirectory(at: corpusURL, includingPropertiesForKeys: [.isDirectoryKey], options: .skipsHiddenFiles)
-
 do {
-    var corpus = try MLDataTable(dictionary: ["text": [""], "label": [""]])
-    
+    var corpus: [(text: String, label: String)] = []
+
     for directory in try fileManager.contentsOfDirectory(at: corpusURL, includingPropertiesForKeys: [.isDirectoryKey], options: [.skipsHiddenFiles]) {
         guard directory.hasDirectoryPath,
             let enumerator = fileManager.enumerator(at: directory, includingPropertiesForKeys: [.isDirectoryKey])
         else {
             continue
         }
-        
+
         for case let resource as URL in enumerator {
             guard !resource.hasDirectoryPath,
                 let language = ProgrammingLanguage(directory: directory.lastPathComponent, fileExtension: resource.pathExtension),
                 let text = try? String(contentsOf: resource)
             else {
                  continue
             }
-            
-            let dataTable = try MLDataTable(dictionary: ["text": text, "label": language.description])
-            corpus.append(contentsOf: dataTable)
+            corpus.append((text: text, label: language.rawValue))
         }
     }
-    
-    let (trainingData, testingData) = corpus.randomSplit(by: 0.9, seed: 0)
-    
+
+    let (texts, labels) = corpus.reduce(into: ([String](), [String]())) {
+        $0.0.append($1.text)
+        $0.1.append($1.label)
+    }
+
+    let dataTable = try MLDataTable(dictionary: ["text": texts, "label": labels])
+
+    let (trainingData, testingData) = dataTable.randomSplit(by: 0.9, seed: 0)
+
     // As of Xcode 10.0 beta (10L176w),
     // attempted use of CRF algorithm results in EXC_BAD_ACCESS.
     /*
     let parameters = MLTextClassifier.ModelParameters(validationData: validationData, algorithm: .crf(revision: 1), language: .english)
     let classifier = try MLTextClassifier(trainingData: trainingData, textColumn: "text", labelColumn: "label", parameters: parameters)
      */
- 
+
     let classifier = try MLTextClassifier(trainingData: trainingData, textColumn: "text", labelColumn: "label")
-    
+
     classifier.modelParameters.algorithm
-    
+
     let evaluation = classifier.evaluation(on: testingData)
     print(evaluation)
-    
+
     let modelPath = URL(fileURLWithPath: destinationPath)
     try classifier.write(to: modelPath)
 } catch {