Example #1
0
        private static IEnumerable <Word> Parse(HtmlDocument html, string wordName)
        {
            var centerDiv = html.GetElementbyId("center");
            var nodes     = centerDiv.ChildNodes
                            .Where(x => x.Name == "span")
                            .Where(x => x.Attributes.Any(a => a.Name == "style" && a.Value == "color:black;background-color:#FFCCCC"))
                            .ToList();

            var words = new List <Word>();

            foreach (var item in nodes)
            {
                var      type = item.InnerText;
                Word     word;
                string[] forms;
                #region switch
                switch (type)
                {
                case "(Article)":
                    forms = ParseSimple(item, html);
                    word  = new Article(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Verb)":
                case "(Defective verb)":
                    string[] v1, v2, v3, v4;
                    var      name = ParseVerb(item, html, out v1, out v2, out v3, out v4);
                    var      verb = new Verb(name);
                    verb.AddForms(v1, VerbType.Infinitive);
                    verb.AddForms(v2, VerbType.Past);
                    verb.AddForms(v3, VerbType.PastParticiple);
                    verb.AddForms(v4, VerbType.PresentParticiple);
                    word = verb;
                    break;

                case "(Modal verb)":
                    forms = ParseSimple(item, html);
                    word  = new ModalVerb(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Noun)":
                case "(Feminine noun)":
                case "(Masculine noun)":
                    forms = ParseNoun(item, html);
                    word  = new Noun(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Plural noun)":
                    forms = ParseSimple(item, html);
                    word  = new Noun(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Adjective)":
                case "(Uninflected adjective)":
                    forms = ParseSimple(item, html);
                    word  = new Adjective(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Preposition)":
                    forms = ParseSimple(item, html);
                    word  = new Preposition(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Adverb)":
                case "(Uninflected adverb)":
                    forms = ParseSimple(item, html);
                    word  = new Adverb(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Conjunction)":
                    forms = ParseSimple(item, html);
                    word  = new Conjunction(forms[0]);
                    word.AddForms(forms);
                    break;

                case "(Interjection)":
                    forms = ParseSimple(item, html);
                    word  = new Interjection(forms[0]);
                    word.AddForm(forms[0]);
                    break;

                case "(Pronoun)":
                case "(Quantitative plural pronoun)":
                case "(Uninflected interrogative pronoun)":
                case "(Demonstrative pronoun)":
                case "(Plural pronoun)":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.Pronoun);
                    word.AddForms(forms);
                    break;

                case "(Quantitative pronoun)":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.QuantitativePronoun);
                    word.AddForms(forms);
                    break;

                case "(Indefinite pronoun)":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.PronounIndefinite);
                    word.AddForms(forms);
                    break;

                case "(Uninflected pronoun)":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.PronounUninflected);
                    word.AddForms(forms);
                    break;

                case "(Relative pronoun)":
                case "(Relative pronoun 2)":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.PronounRelative);
                    word.AddForms(forms);
                    break;

                case "(1st person singular pronoun)":
                case "()":      // (1st person plural pronoun )
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.PronounPerson1);
                    word.AddForms(forms);
                    break;

                case "(2nd person singular pronoun)":
                case "(2nd person plural pronoun )":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.PronounPerson2);
                    word.AddForms(forms);
                    break;

                case "(3rd person masculine singular pronoun)":
                case "(3rd person feminine singular pronoun )":
                case "(3rd person plural pronoun)":
                case "(3rd person neuter singular pronoun)":
                    forms = ParseSimple(item, html);
                    word  = new Pronoun(forms[0], WordType.PronounPerson2);
                    word.AddForms(forms);
                    break;

                default:
                    throw new NotSupportedException(string.Format("The type {0} is not defined", type));
                }
                #endregion
                words.Add(word);
            }
            return(words);
        }