This class encapsulates a word in a sentence as a string, with an associated score (floating point value).
public class Word {
var value : String // the word itself
var score : Double // the score for the word
init(value: String, score: Double) {
self.value = value
self.score = score
}
public func get_Value() -> String {
return value
}
public func get_Score() -> Double {
return score
}
}
This class encapsulates a sentence as a list of words, with an aggregate score as well as the order in which it appears in the document (via its index). See the Document class below.
public class Sentence {
var words : [Word] // List of words in one sentence
var score : Double
var original_Index : Int // the original index in the array of sentences from the document;
// will be used later when putting the final summary sentences in order.
init() {
words = []
score = 0.0
original_Index = -1
}
public func add_Word(word: Word) {
self.words.append(word)
score += word.get_Score()
}
public func display() -> String {
var toDisplay : String = ""
for w in self.words {
toDisplay = toDisplay + " " + w.get_Value()
}
return toDisplay
}
public func get_Score() -> Double {
return score
}
public func setOriginal_Index(index: Int) {
original_Index = index
}
}
The Document class allows us to add sentences as we read them from a file, maintaining the original order. It also has a method to sort the sentences by their cumulative scores.
public class Document {
var sentences : [Sentence]
var counter : Int
init() {
sentences = []
counter = 0
}
public func add_Sentence(sentence: Sentence) {
self.sentences.append(sentence)
sentence.setOriginal_Index(index: counter)
counter += 1
}
public func get_Sentence(index: Int) -> Sentence {
return sentences[index]
}
public func get_SentenceScore(index: Int) -> Double {
return sentences[index].get_Score()
}
public func summarize() -> [Sentence] {
var index: Int = 0
sentences.sort { (s1, s2) -> Bool in
return s1.get_Score() > s2.get_Score()
}
// We now have two arrays: one of the document's sentences
// broken up in order, and one in which the sentences are
// sorted based on the score that they have. Now, I am going to finally
let summarySize : Double = 0.1 * Double(sentences.count)
let summarySizeRounded : Int = Int(summarySize.rounded())
var newSentences : [Sentence] = []
var number : Int
for number in 0..<sentences.count {
if (newSentences.count == summarySizeRounded) {
break
} else {
newSentences.append(sentences[number])
}
}
newSentences.sort { (s1, s2) -> Bool in
return s1.original_Index < s2.original_Index
}
return newSentences
}
}
This is the central logic of the program:
let wordlist : WordList = WordList(fullyQualifiedPath: "word_scores.txt")
let document : DocumentReader = DocumentReader(fullyQualifiedPath: "text_document_to_summarize")
let words: [String] = document.get_words()
var fullPageDocument: Document = Document()
var newSentenceObject: Sentence = Sentence()
for var word : String in words {
if (word.range(of: ".") == nil &&
word.range(of: "?") == nil &&
word.range(of: "!") == nil) {
var newWordObject: Word = Word(value: word, score: wordlist.lookup_word(word: word))
newSentenceObject.add_Word(word: newWordObject)
} else if (word.range(of: "\"") != nil ||
word.range(of: ",") != nil ||
word.range(of: ":") != nil ||
word.range(of: "'") != nil) {
let index = word.index(word.startIndex, offsetBy: word.characters.count - 1)
var newWordObject: Word = Word(value: word,
score: wordlist.lookup_word(word: word.substring(to: index)))
newSentenceObject.add_Word(word: newWordObject)
} else {
let index = word.index(word.startIndex, offsetBy: word.characters.count - 1)
var newWordObject: Word = Word(value: word,
score: wordlist.lookup_word(word: word.substring(to: index)))
newSentenceObject.add_Word(word: newWordObject)
fullPageDocument.add_Sentence(sentence: newSentenceObject)
newSentenceObject = Sentence() /*This creates a new object under the same variable name;
if a bucket is full, you need to get a new one of the same
kind that holds the same stuff. */
}
}
var summary : [Sentence] = fullPageDocument.summarize()
for s:Sentence in summary {
print(s.display())
}