public static void MakeWordList() {
using (var imp = new Impersonator("pavel", "LANGMaster", "zvahov88_"))
DictLib.RunStemming<List<string>>(
Dict.allLangs,
lng => File.ReadAllLines(string.Format(Machines.rootPath + @"RwDicts\Sources\LingeaOld\design\wordList_{0}.txt", lng)).Concat(
File.ReadAllLines(string.Format(Machines.rootPath + @"RwDicts\Sources\Wiktionary.back\{0}.txt", lng)).Select(l => l.Split(' ')[0]).Take(50000)).
Where(w => !string.IsNullOrEmpty(w)).
Select(w => w.ToLower()).
Distinct().ToArray(),
lng => new List<string>(),
(lng, word, row, res) => res.AddRange(row.Select(r => doNormalize(r)).Where(r => !string.IsNullOrEmpty(r))),
(lng, res) => File.WriteAllLines(string.Format(Machines.rootPath + @"RwDicts\Sources\Ultralingua.back\design\WordsStems_{0}.txt", lng), res.Distinct().OrderBy(w => w)),
imp
);
}
public static void MakeWordList_Other() {
using (var imp = new Impersonator("pavel", "LANGMaster", "zvahov88_")) {
DictLib.RunStemming<List<string>>(
otherLangsStr.Values,
lng => {
Thread.CurrentThread.CurrentCulture = Thread.CurrentThread.CurrentUICulture = CultureInfo.CreateSpecificCulture("ar-sa");
var list = File.ReadAllLines(string.Format(@"d:\LMCom\rew\Web4\RwDicts\Sources\Wiktionary.back\src\{0}.txt", otherLangsStr.First(kv => kv.Value == lng).Key)).ToArray();
return list.Select(l => l.Split(' ')[0]).Take(200000).
Where(w => !string.IsNullOrEmpty(w)).
Select(w => w.ToLower()).
Distinct().ToArray();
},
lng => new List<string>(),
(lng, word, row, res) => res.AddRange(row.Select(r => doNormalize(r)).Where(r => !string.IsNullOrEmpty(r))),
(lng, res) => File.WriteAllLines(string.Format(Machines.rootPath + @"RwDicts\Sources\Ultralingua.back\design\WordsStems_{0}.txt", lng), res.Distinct().OrderBy(w => w)),
imp
);
}
}
//LingeaOld\temp1 + LingeaOld\design\entriesInfo.xml => LingeaOld
public static void OldToNew3() {
using (var imp = new Impersonator("pavel", "LANGMaster", "zvahov88_")) {
var files = repairLingea(Machines.rootPath + @"RwDicts\Sources\LingeaOld\temp1\", imp);
foreach (var fn in Directory.EnumerateFiles(Machines.rootPath + @"RwDicts\Sources\LingeaOld", "*.xml")) File.Delete(fn);
foreach (var f in files) XmlUtils.ObjectToFile(string.Format(Machines.rootPath + @"RwDicts\Sources\LingeaOld\{1}_{0}.xml", f.crsLang, f.natLang), f);
}
}
public DictEntryObj toNew(Impersonator imp) {
using (WindowsIdentity.Impersonate(imp.token)) {
return new DictEntryObj {
entryId = entryId,
entry = entry.Elements().First(),
soundMaster = soundMaster,
type = DictEntryType.lingeaOld,
headWords = headWords,
courseWords = entry.AttributeValue("courseUsed", "").Split('|').Select(w => w.ToLower().Replace("'", null)).Except(headWords.Select(h => h.ToLower())).Where(w => !string.IsNullOrEmpty(w)).ToArray(),
//headWords.Select(hw => new CourseDictionary2.DictStem { type = CourseDictionary2.DictStemType.wordId, word = hw }).
// Concat(headWords.ToArray().SelectMany(w => CourseDictionary.RunStemming(okCrs, w)).Distinct().Select(w => new CourseDictionary2.DictStem { type = CourseDictionary2.DictStemType.stem, word = w })).
// Concat(entry.AttributeValue("courseUsed", "").Split('|').Where(w => !string.IsNullOrEmpty(w)).Select(w => new CourseDictionary2.DictStem { type = CourseDictionary2.DictStemType.courseUses, word = w })).
// ToArray()
};
};
}
public static IEnumerable<DictObj> repairLingea(string srcDir, Impersonator imp) {
var allnew = XmlUtils.FileToObject<LingeaDictionary.DictEntry[]>(@"d:\LMCom\rew\Web4\RwDicts\Sources\LingeaOld\design\entriesInfo.xml");
var allOld = LingeaDictionary.getOldLingeaEntries(srcDir).SelectMany(kv => kv.Value).SelectMany(kv => kv.Value.entryIdToEntry).ToDictionary(kv => kv.Key, kv => kv.Value);
//k jednomu zvuku jeden headword
Dictionary<Langs, Dictionary<string, string>> soundMasterToHeadword = allnew.Where(de => de.soundMaster != null).GroupBy(de => de.okCrs).ToDictionary(
g => g.Key,
g => g.GroupBy(de => de.soundMaster).ToDictionary(
sg => sg.Key,
sg => sg.Select(d => d.headWords[0].ToLower()).Distinct().Single()));
//pripojeni .info souboru s headword k .mp3 souboru
foreach (var lKv in soundMasterToHeadword) foreach (var hKv in lKv.Value) {
string right = @"d:\LMCom\rew\Web4\RwDicts\LingeaSound\" + hKv.Key.Replace('/', '\\') + ".mp3";
if (!File.Exists(right)) throw new Exception();
File.WriteAllText(right.Replace(".mp3", ".info"), lKv.Key.ToString() + "=" + hKv.Value);
}
//vytvor nova hesla - merge entriesInfo.xml s LingeaOld.back (LingeaOld\temp1). Pridej a napln sound tag v entry
List<DictObj> res = new List<DictObj>();
Parallel.ForEach(allnew.Where(de => de.okCrs != Langs.no).GroupBy(dn => dn.okCrs),
//new ParallelOptions() { MaxDegreeOfParallelism = 1 },
crs => {
Parallel.ForEach(crs.GroupBy(dc => dc.crsLang == dc.okCrs ? dc.natLang : dc.crsLang),
//new ParallelOptions() { MaxDegreeOfParallelism = 1 },
nat => {
foreach (var newDe in nat) {
//merge
newDe.entry = allOld[newDe.entryId];
//sound tag
var snd = newDe.entry.Descendants("sound").FirstOrDefault();
if (snd == null) {
snd = new XElement("sound");
var head = newDe.entry.DescendantsAttr("class", "head").First();
var pron = head.Elements().FirstOrDefault(e => e.AttributeValue("class") == "pron");
if (pron != null) pron.AddAfterSelf(snd);
else {
var morf = head.Elements().FirstOrDefault(e => e.AttributeValue("class") == "morf");
if (morf != null) morf.AddAfterSelf(snd);
else head.Add(snd);
}
}
snd.Value = (string.IsNullOrEmpty(newDe.soundMaster) ? "@" + newDe.headWords[0].ToLower() : @"RwDicts/LingeaSound/" + newDe.soundMaster + ".mp3").ToLower();
}
lock (res) res.Add(new DictObj { crsLang = crs.Key, natLang = nat.Key, entries = nat.Select(n => n.toNew(imp)).DistinctBy(n => n.entry.ToString()).ToArray() });
});
});
return res;
}
