// Parses a single book.
Book ParseBook(XmlReader reader)
{
var book = new Book();
while (reader.Read())
if (reader.NodeType == XmlNodeType.Element) {
string parseType;
switch (reader.LocalName) {
case "etext":
book.Number = int.Parse(reader.GetAttribute("ID", RDF).Substring(5),
CultureInfo.InvariantCulture);
break;
case "title":
parseType = reader.GetAttribute("parseType", RDF);
if (parseType != "Literal")
throw new ApplicationException("Unrecognized parse type.");
book.Title = reader.ReadElementContentAsString();
break;
case "description":
book.Notes = ParseText(reader);
break;
case "creator":
book.Authors = ParsePersonsAndEra(book, reader);
break;
case "contributor":
// Eras of contributors do not affect the era of the book intentionally.
// Contributors might be not only illustrators, who would live at the
// same time as authors, but also people who retyped the book and
// uploaded it to the Project Gutenberg web site. They would mess the
// era of the book which is supposed to embrace the time when the actual
// paper book was written.
book.Contributors = ParsePersons(reader);
break;
case "language":
if (!reader.ReadToDescendant("value", RDF))
throw new ApplicationException("Missing language value.");
book.Language = reader.ReadElementContentAsString();
break;
case "subject":
using (var subreader = reader.ReadSubtree())
book.Tags = ParseTags(subreader).ToArray();
break;
case "created":
if (!reader.ReadToDescendant("value", RDF))
throw new ApplicationException("Missing creation date value.");
book.Included = reader.ReadElementContentAsDate();
break;
case "downloads":
if (!reader.ReadToDescendant("value", RDF))
throw new ApplicationException("Missing download count value.");
book.Downloads = reader.ReadElementContentAsInt();
break;
}
}
return book;
}
// Parses information about single or multiple persons (authors or contributors) which
// contain era (year span) of their life. Names of the persons are returned and the
// era in the book is updated to span across all life spans of all persons.
string[] ParsePersonsAndEra(Book book, XmlReader reader)
{
// Checking the attribute parseType is a quick and dirty way to detect a multivalue
// property without inspecting the entire element structure and luckily works here.
var parseType = reader.GetAttribute("parseType", RDF);
if (parseType == "Literal") {
var person = reader.ReadElementContentAsString();
ParsePersonEra(book, person);
return new[] { person };
}
using (var subreader = reader.ReadSubtree())
return ParseMultiplePersonsAndEra(book, subreader);
}
// Extracts the era of a person's life from the line with the person's name and updates
// the book so that its era includes the person's one.
void ParsePersonEra(Book book, string value)
{
YearSpan era;
var comma = value.LastIndexOf(',');
if (comma > 0 && YearSpan.TryParse(value.Substring(comma + 1).Trim(),
CultureInfo.InvariantCulture, out era))
book.Era = book.Era.Union(era);
}
// Parses the subtree of multiple persons. It is to be called from the ParsePersonsAndEra
// method only.
string[] ParseMultiplePersonsAndEra(Book book, XmlReader reader)
{
var persons = new List<string>();
while (reader.ReadToFollowing("li", RDF)) {
// So far, I haven't noticed other multivalues than those consisting of literals.
// When they occur I'll implement better vaue parsing here. Now I save time.
var parseType = reader.GetAttribute("parseType", RDF);
if (parseType != "Literal")
throw new ApplicationException("Unrecognized parse type.");
string person = reader.ReadElementContentAsString();
ParsePersonEra(book, person);
persons.Add(person);
}
return persons.Any() ? persons.ToArray() : null;
}
// Gets a book by its name or number if it exists, not throwing and exception if it does
// not. (If the string contains only digits it will be interpreted as a number.)
public bool TryGetBook(string name, out Book book)
{
int number;
if (int.TryParse(name, NumberStyles.Integer,
CultureInfo.InvariantCulture, out number)) {
if (number >= 0 && number < BooksByNumber.Length) {
book = BooksByNumber[number];
return book != null;
}
book = null;
return false;
}
return BooksByName.TryGetValue(name, out book);
}