public static void CSWordSenses()
{
var fn = WikiRawConsts.loadStat().First(f => f.lang == "cs" && f.type == WikiRawConsts.wiktionary).fileNameDump();
var names = WikiRawConsts.csWordSenses.ToHashSet();
var lines = new List <string>();
JsonNew.DeserializeEnum <Sections>(fn + ".sec.json", sect => {
if (sect.subsections == null)
{
return;
}
var cs = sect.subsections.FirstOrDefault(s => s.title == "čeština");
if (cs == null || cs.subsections == null)
{
return;
}
var senses = cs.subsections.Select(scs => scs.title).Where(s => names.Contains(s)).Distinct().ToArray();
if (senses.Length == 0)
{
return;
}
lines.Add($"{sect.title}={string.Join(",", senses)}");
});
//using (var rdr = new JsonStreamReader(fn + ".sec.json")) {
// foreach (var sect in rdr.Deserialize<Sections>()) {
// if (sect.subsections == null) continue;
// var cs = sect.subsections.FirstOrDefault(s => s.title == "čeština");
// if (cs == null || cs.subsections == null) continue;
// var senses = cs.subsections.Select(scs => scs.title).Where(s => names.Contains(s)).Distinct().ToArray();
// if (senses.Length == 0) continue;
// lines.Add($"{sect.title}={string.Join(",", senses)}");
// }
//}
File.WriteAllLines(fn + ".cs-senses.txt", lines.OrderBy(s => s));
}
public static void ExtractSections()
{
var stat = WikiRawConsts.loadStat();
Parallel.ForEach(WikiRawConsts.getRawFiles(WikiRawConsts.wiktionary).Where(rf => rf.pages >= wiktPageNumLimit), new ParallelOptions {
MaxDegreeOfParallelism = 6
}, rf => {
IEnumerable <WikimediaPage> pages = new Wikimedia(rf.fileName()).Articles.Where(article => !article.IsDisambiguation && !article.IsRedirect && !article.IsSpecialPage);
var cnt = 0;
JsonNew.SerializeEnum <Sections>(rf.fileNameDump() + ".sec.json", pages.Select(p => new Sections(p)).identityEnum(sect => {
if (cnt % 100000 == 0)
{
Console.WriteLine($"{rf.lang} {cnt}");
}
cnt++;
}));
//using (var wr = new JsonStreamWriter(rf.fileNameDump() + ".sec.json"))
// foreach (var sect in pages.Select(p => new Sections(p))) {
// if (cnt % 100000 == 0) Console.WriteLine($"{rf.lang} {cnt}");
// cnt++;
// wr.Serialize(sect);
// }
lock (stat) {
stat.First(s => s.type == WikiRawConsts.wiktionary && s.lang == rf.lang).pages = cnt;
}
});
WikiRawConsts.saveStat();
}