public void Add(Uri baseUri, QuoteCollectorRules rules)
{
// checking preconditions
if (baseUri == null || rules == null)
{
throw new ArgumentNullException();
}
_dataCollection.Add(new CultureData(rules.Culture, baseUri, rules));
}
public void Add(QuoteCollectorRules rules)
{
// checking preconditions
if (rules == null)
{
throw new ArgumentNullException();
}
Add(new Uri(String.Format(WikiquoteBaseUrl, rules.Culture.TwoLetterISOLanguageName)), rules);
}
private static IEnumerable <TopicChoice> GetTopicChoices(string contents, QuoteCollectorRules rules)
{
var returnValue = from Match match in rules.DisambiguationTemplateRegex.Matches(contents)
where match.Groups[TopicIdRegexGroup].Success
let topicId = HttpUtility.HtmlDecode(match.Groups[TopicIdRegexGroup].Value)
let topicName = match.Groups[TopicNameRegexGroup].Success ?
HttpUtility.HtmlDecode(match.Groups[TopicNameRegexGroup].Value) :
topicId
let topicDescription = match.Groups[TopicDescriptionRegexGroup].Success ?
HttpUtility.HtmlDecode(ReplaceMiscSequences(match.Groups[TopicDescriptionRegexGroup].Value)) :
null
where !topicId.StartsWith(WikipediaLinkPrefix)
select new TopicChoice(topicId, topicName, topicDescription);
return(returnValue);
}
private SelectableQuoteCollection ParsePageContentsAndExtractQuotes(string topic, CultureInfo culture, string contents, SelectableQuoteCollection oldQuotes)
{
QuoteCollectorRules rules = _cultureMapper[culture].Rules as QuoteCollectorRules;
if (rules.DisambiguationTemplateIdentifiers != null && rules.DisambiguationTemplateRegex != null)
{
for (int i = 0; i < rules.DisambiguationTemplateIdentifiers.Length; i++)
{
if (contents.Contains(rules.DisambiguationTemplateIdentifiers[i]))
{
throw new TopicAmbiguousException(topic, GetTopicChoices(contents, rules));
}
}
}
// removing sections that do not contain quotes (only if required, i.e. quotes have to be determined heuristically)
if (rules.WikiSectionsToSkipRegex != null)
{
contents = rules.WikiSectionsToSkipRegex.Replace(contents, String.Empty);
}
// replacing xhtml tags by their contents
contents = ReplaceXhtmlTags(contents);
// various string replacements
contents = HttpUtility.HtmlDecode(ReplaceMiscSequences(contents));
// extracting quotes
MatchCollection quoteMatches = rules.QuoteRegex.Matches(contents);
var quotes = from Match match in quoteMatches
where match.Groups[QuoteRegexGroup].Success
let quoteHelper = match.Groups[QuoteRegexGroup].Value
let quote = (rules.NewlineRegex != null) ?
rules.NewlineRegex.Replace(quoteHelper, "\n") :
quoteHelper
let additionalInformation = match.Groups[AdditionalInformationRegexGroup].Success ?
rules.NewlineRegex.Replace(match.Groups[AdditionalInformationRegexGroup].Value, " ") :
String.Empty
select GetQuote(oldQuotes, quote.TrimEnd(), additionalInformation.TrimEnd());
return(quotes.ToSelectableQuoteCollection(culture));
}
public CultureData(CultureInfo cultureInfo, Uri uri, QuoteCollectorRules rules)
{
_cultureInfo = cultureInfo;
_baseUri = uri;
_rules = rules;
}
