private Task<bool> ArticleToArticleModel(NzzArticle na, ArticleModel am)
{
return ExecuteSafe(async () =>
{
am.Content.Clear();
for (int i = 0; i < na.body.Length; i++)
{
if (na.body[i].style == "h4")
na.body[i].style = "h2";
if (na.body[i].style == "h3")
na.body[i].style = "h1";
string starttag = "<" + na.body[i].style + ">";
string endtag = "</" + na.body[i].style + ">";
if (string.IsNullOrWhiteSpace(na.body[i].text))
{
foreach (var nzzBox in na.body[i].boxes)
{
if (nzzBox.type == "image")
{
var uri = ParseImageUri(nzzBox.path);
if (uri != null)
am.Content.Add(new ImageContentModel()
{
Url = uri,
Text = TextHelper.TextToTextModel(nzzBox.caption)
});
}
else if (nzzBox.type == "video" || nzzBox.type == "html")
{
//dont do shit
}
else if (nzzBox.type == "infobox")
{
var newContent = HtmlConverter.CreateOnce(am.Feed.Source.PublicBaseUrl).HtmlToParagraph("<p>" + nzzBox.body + "</p>");
foreach (var paragraphModel in newContent)
{
var ntm = new TextModel()
{
Children = paragraphModel.Children,
TextType = TextType.Cursive
};
paragraphModel.Children = new List<TextModel> { ntm };
}
if (!string.IsNullOrWhiteSpace(nzzBox.title))
newContent.Insert(0, new ParagraphModel()
{
ParagraphType = ParagraphType.Title,
Children = new List<TextModel>()
{
new TextModel()
{
Text = nzzBox.title,
TextType = TextType.Cursive
}
}
});
if (newContent.Any())
am.Content.Add(new TextContentModel()
{
Content = newContent
});
}
else
LogHelper.Instance.LogInfo("nzz content type not found: " + nzzBox.mimeType, this);
}
}
else
{
if (!na.body[i].text.StartsWith("Mehr zum Thema"))
{
var content = HtmlConverter.CreateOnce(am.Feed.Source.PublicBaseUrl).HtmlToParagraph(starttag + na.body[i].text + endtag);
if (content != null && content.Count > 0)
am.Content.Add(new TextContentModel()
{
Content = content
});
}
}
}
if (!am.Content.Any())
am.Content.Add(TextHelper.TextToTextModel("Der Inhalt dieses Artikels wird nicht unterstützt. Öffne den Artikel im Browser um mehr zu sehen."));
if (na.authors != null)
foreach (var nzzAuthor in na.authors)
{
if (!string.IsNullOrEmpty(nzzAuthor.name))
{
am.Author = nzzAuthor.name;
if (!string.IsNullOrEmpty(nzzAuthor.abbreviation))
am.Author += ", " + nzzAuthor.abbreviation;
}
else
am.Author = nzzAuthor.abbreviation;
}
if (!string.IsNullOrEmpty(na.agency))
am.Author += " " + na.agency;
if (string.IsNullOrWhiteSpace(am.Author))
am.Author = "NZZ";
if (!string.IsNullOrEmpty(na.leadText))
am.Teaser = na.leadText;
am.Themes.Clear();
await AddThemesAsync(am, na.departments);
return true;
});
}
private static void AddInlineChildren(Span span, TextModel model)
{
if (model.Children != null)
{
foreach (var textModel in model.Children)
{
var span2 = RenderTextContent(textModel);
if (span2 != null)
span.Inlines.Add(span2);
}
}
}
private static Span RenderTextContent(TextModel text)
{
if (text.TextType == TextType.Bold)
{
var span = new Bold();
if (!string.IsNullOrWhiteSpace(text.Text))
span.Inlines.Add(new Run()
{
Text = text.Text
});
AddInlineChildren(span, text);
foreach (var inline in span.Inlines)
{
inline.FontWeight = FontWeights.Bold;
}
return span;
}
if (text.TextType == TextType.Cursive)
{
var span = new Span();
if (!string.IsNullOrWhiteSpace(text.Text))
span.Inlines.Add(new Run()
{
Text = text.Text
});
AddInlineChildren(span, text);
foreach (var inline in span.Inlines)
{
inline.FontStyle = FontStyle.Italic;
}
return span;
}
if (text.TextType == TextType.Hyperlink)
{
var span = new Hyperlink()
{
NavigateUri = new Uri(text.Text)
};
AddInlineChildren(span, text);
return span;
}
if (text.TextType == TextType.Underline)
{
var span = new Underline();
if (!string.IsNullOrWhiteSpace(text.Text))
span.Inlines.Add(new Run()
{
Text = text.Text
});
AddInlineChildren(span, text);
return span;
}
else//(text.TextType == TextType.Normal)
{
var span = new Span();
span.Inlines.Add(
new Run()
{
Text = text.Text
});
AddInlineChildren(span, text);
return span;
}
}
private static List<SpritzWord> ToSpritzWords(TextModel model)
{
var words = new List<SpritzWord>();
if (model.TextType == TextType.Hyperlink)
{
var list = ToSpritzWords(model.Children);
words.AddRange(list);
}
else
{
string[] splitresult = model.Text?.Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
if (splitresult != null)
{
var wordlist = new List<string>(splitresult);
for (int i = 0; i < wordlist.Count; i++)
{
if (wordlist[i].Contains("-") && wordlist[i].IndexOf("-", StringComparison.Ordinal) != wordlist[i].Length - 1 && !Regex.IsMatch(wordlist[i], "{0-9}-{0-9}"))
{
int index = wordlist[i].IndexOf("-", StringComparison.Ordinal) + 1;
wordlist.Insert(i + 1, wordlist[i].Substring(index));
wordlist[i] = wordlist[i].Substring(0, index);
}
//ab 14 muss getrennt werden, versuche dann, das word zu splitten
if (wordlist[i].Length > 13)
{
//wenn wort länger als 13 + 13 werden einfach die ersten 13 buchstaben genommen
if (wordlist[i].Length > 26)
{
wordlist.Insert(i + 1, wordlist[i].Substring(13));
wordlist[i] = wordlist[i].Substring(0, 13);
}
else
{
//wordlist wird zweigeteilt
int count = wordlist[i].Length / 2;
wordlist.Insert(i + 1, wordlist[i].Substring(count));
wordlist[i] = wordlist[i].Substring(0, count);
}
}
var sw = new SpritzWord();
if (wordlist[i].Length == 1)
sw.Middle = wordlist[i][0];
else if (wordlist[i].Length >= 2 && wordlist[i].Length <= 5)
{
sw.Before = wordlist[i][0].ToString();
sw.Middle = wordlist[i][1];
sw.After = wordlist[i].Substring(2);
}
else if (wordlist[i].Length >= 6 && wordlist[i].Length <= 9)
{
sw.Before = wordlist[i].Substring(0, 2);
sw.Middle = wordlist[i][2];
sw.After = wordlist[i].Substring(3);
}
else //(wordlist[i].Length >= 10)
{
sw.Before = wordlist[i].Substring(0, 3);
sw.Middle = wordlist[i][3];
sw.After = wordlist[i].Substring(4);
}
if (sw.After != null)
{
if (sw.After.Contains("."))
{
sw.Lenght = 4;
words.Add(sw);
words.Add(new SpritzWord() { Lenght = 5 });
}
else if (sw.After.Contains(";") || sw.After.Contains(",") || sw.After.Contains(":") || sw.Middle == '-')
{
sw.Lenght = 4;
words.Add(sw);
}
else
{
sw.Lenght = 1;
words.Add(sw);
}
}
else
{
sw.Lenght = 1;
words.Add(sw);
}
}
}
}
return words;
}
private TextModel ParseText(HtmlNode parentNode)
{
var model = new TextModel();
var texts = new[] { "h1", "h2", "h3", "h4", "p" };
var bolds = new[] { "b", "strong", "em" };
var cursives = new[] { "i" };
var underlines = new[] { "u" };
var hyperlink = new[] { "a" };
if (!parentNode.ChildNodes.Any() && parentNode.NodeType == HtmlNodeType.Text)
{
model.Text = TextHelper.NormalizeString(TextHelper.StripHtml(parentNode.InnerText));
if (string.IsNullOrWhiteSpace(model.Text))
return null;
return model;
}
if (texts.Any(predicate => predicate == parentNode.Name))
model.TextType = TextType.Normal;
else if (bolds.Any(predicate => predicate == parentNode.Name))
model.TextType = TextType.Bold;
else if (cursives.Any(predicate => predicate == parentNode.Name))
model.TextType = TextType.Cursive;
else if (underlines.Any(predicate => predicate == parentNode.Name))
model.TextType = TextType.Underline;
else if (hyperlink.Any(predicate => predicate == parentNode.Name))
{
model.TextType = TextType.Hyperlink;
model.Text = TextHelper.NormalizeString(parentNode.Attributes["href"]?.Value);
if (string.IsNullOrWhiteSpace(model.Text))
model.TextType = TextType.Normal;
else
{
if (model.Text.StartsWith("www"))
model.Text = "http://" + model.Text;
if (!model.Text.StartsWith("http://"))
{
if (model.Text.StartsWith("/"))
model.Text = _baseUrl + model.Text.Substring(1);
else
model.Text = _baseUrl + model.Text;
}
if (!Uri.IsWellFormedUriString(model.Text, UriKind.Absolute))
{
//todo: do additional repair stuff
//parse utf8 caracters like: http://www.ragnar%C3%B6k-spektakel.ch
model.Text = _baseUrl;
}
}
}
else
return null;
//shortcut for once node stuff
if (parentNode.ChildNodes.Count() == 1 && parentNode.ChildNodes.FirstOrDefault().NodeType == HtmlNodeType.Text && model.TextType != TextType.Hyperlink)
{
model.Text = TextHelper.NormalizeString(parentNode.ChildNodes.FirstOrDefault().InnerText.Trim());
if (string.IsNullOrWhiteSpace(model.Text))
return null;
return model;
}
foreach (var node in parentNode.ChildNodes)
{
var tm = ParseText(node);
if (tm != null)
model.Children.Add(tm);
}
if (model.TextType == TextType.Hyperlink && model.Children.Count == 0)
return null;
return !string.IsNullOrEmpty(model.Text) || model.Children.Any() ? model : null;
}
private void CollapseModelsIfNecessary(TextModel model, List<TextType> knownTextTypes)
{
while (model.Children.Count == 1 && knownTextTypes.Contains(model.TextType))
{
model.Text = model.Children[0].Text;
model.TextType = model.Children[0].TextType;
model.Children = model.Children[0].Children;
}
knownTextTypes.Add(model.TextType);
foreach (var textModel in model.Children)
{
CollapseModelsIfNecessary(textModel, knownTextTypes);
}
}
