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 SectionStats()
{
Parallel.ForEach(WikiRawConsts.getRawFiles(WikiRawConsts.wiktionary), new ParallelOptions {
MaxDegreeOfParallelism = 4
}, rf => {
var sectStat = new Dictionary <string, int>();
void add(string l) => sectStat[l] = sectStat.TryGetValue(l, out int c) ? c + 1 : 1;
JsonNew.DeserializeEnum <Sections>(rf.fileNameDump() + ".sec.json", sect => {
if (sectStat.Count > 5000)
{
return;
}
foreach (var s in sect.lines(0, ""))
{
add(s);
}
});
//using (var rdr = new JsonStreamReader(rf.fileNameDump() + ".sec.json")) {
// foreach (var sect in rdr.Deserialize<Sections>()) {
// if (sectStat.Count > 5000) break;
// foreach (var s in sect.lines(0, "")) add(s);
// }
//}
File.WriteAllLines(rf.fileNameDump() + ".sec-stat.txt", sectStat.Where(kv => kv.Value >= 10).OrderBy(s => s.Key).Select(s => $"{s.Key} #{s.Value}"));
});
}
public void Test1()
{
JsonNew.SerializeEnum(@"c:\temp\pom.json", objs());
Parallel.ForEach(Enumerable.Range(0, 100), idx => {
var count = 0;
JsonNew.DeserializeEnum(typeof(X), @"c:\temp\pom.json", x => {
count++;
});
Assert.Equal(cnt * 2, count);
count = 0;
});
}