public IActionResult OnPostTag(string sequenceOfParallelTokens)
{
var parallels = sequenceOfParallelTokens.Trim().Split("[[Delete]]").Where(parallel => parallel != "").ToList();
ParallelDocument parallelSubcorpus = null;
string parallelSubcorpusFilePath = null;
foreach (var parallel in parallels)
{
var sequencesSplit = parallel.Split("};").Where(token => token != "").ToList();
var sequencesIDs = sequencesSplit.Select(x => x.Trim().Trim('{').Split(" - ")[0]).ToList();
if (parallelSubcorpus is null)
{
string documentID = sequencesIDs[0].Split('|')[0];
var dirParallelCorpus = Path.Combine(_environment.ContentRootPath, "database", "parallelizedDocuments");
DirectoryInfo directoryParallelCorpusInfo = new DirectoryInfo(dirParallelCorpus);
parallelSubcorpusFilePath = directoryParallelCorpusInfo.GetFiles().Where(f => f.Name.Split('_')[0] == documentID).FirstOrDefault().FullName;
using (StreamReader r = new StreamReader(new FileStream(parallelSubcorpusFilePath, FileMode.Open)))
{
parallelSubcorpus = JsonConvert.DeserializeObject <ParallelDocument>(r.ReadToEnd());
}
parallelSubcorpus.parallelTokens = new List <ParallelToken>();
}
var token2Add = new ParallelToken();
foreach (var token in sequencesIDs)
{
Console.WriteLine(token);
var splitID = token.Split('|').Where(x => x != "").ToList();
RealizationGroup currentGroup;
if (token2Add.Where(g => g.documentID == splitID[0] && g.textID == splitID[2] && g.clauseID == splitID[1]).ToList().Count == 0)
{
currentGroup = new RealizationGroup();
currentGroup.documentID = splitID[0];
currentGroup.textID = splitID[2];
currentGroup.clauseID = splitID[1];
token2Add.Add(currentGroup);
}
currentGroup = token2Add.Where(g => g.documentID == splitID[0] && g.textID == splitID[2] && g.clauseID == splitID[1]).FirstOrDefault();
var singleToken = parallelSubcorpus.parallelClauses[Convert.ToInt32(splitID[2]), Convert.ToInt32(splitID[1])].clause.realizations.Where(r => r.realizationID == splitID[3]).FirstOrDefault();
singleToken.documentID = splitID[0];
singleToken.textID = splitID[2];
singleToken.clauseID = splitID[1];
if (!currentGroup.Contains(singleToken))
{
currentGroup.Add(singleToken);
}
}
if (!parallelSubcorpus.parallelTokens.Contains(token2Add))
{
parallelSubcorpus.parallelTokens.Add(token2Add);
}
}
if (!(parallelSubcorpus is null) && !(parallelSubcorpusFilePath is null))
{
var parallelSubcorpusInJSON = parallelSubcorpus.Jsonize();
using (StreamWriter w = new StreamWriter(new FileStream(parallelSubcorpusFilePath, FileMode.Create)))
{
w.Write(parallelSubcorpusInJSON);
}
}
return(RedirectToPage());
}
public void OnPostParallel()
{
var files = new DirectoryInfo(Path.Combine(_environment.ContentRootPath, "database", "parallelizedDocuments")).GetFiles();
var documents = new List <ParallelDocument>();
if (files.Length > 0)
{
for (int f = 0; f < files.Length; f++)
{
using (StreamReader r = new StreamReader(new FileStream(files[f].FullName, FileMode.Open, FileAccess.Read)))
{
documents.Add(JsonConvert.DeserializeObject <ParallelDocument>(r.ReadToEnd()));
}
}
List <string> lemmata = new List <string>();
chosenTexts.ForEach(x =>
{
var docName = x.Split(": ")[0];
var textID = x.Split(": ")[1].Split(" (")[1].Split(')')[0];
var currentLemmata = documents
.Where(d => d.name == docName)
.SelectMany(d => d.parallelTokens)
.SelectMany(t => t.ToList())
.SelectMany(rg => rg.ToList())
.Where(r => r.clauseID == textID)
.Select(r => r.realizationFields is null ? r.lexemeTwo : r.realizationFields.Any(f => f.ContainsKey("Lemma")) ? r.realizationFields.Where(t => t.ContainsKey("Lemma")).SelectMany(t => t["Lemma"]).Select(v => v.name).FirstOrDefault() : r.lexemeTwo)
.Distinct()
.ToList();
currentLemmata.ForEach(l => lemmata.Add(l));
});
lemmata = lemmata.Distinct().ToList();
var finalDictionary = new List <ParallelDictionaryUnit>();
lemmata.ForEach(lemma => finalDictionary.Add(new ParallelDictionaryUnit(lemma, documents
.SelectMany(d => d.parallelTokens)
.Where(t => t.Any(rg => rg.Any(r => r.lexemeTwo == lemma || (r.realizationFields != null ? (r.realizationFields.Any(f => f.ContainsKey("Lemma")) ? r.realizationFields.Where(f => f.ContainsKey("Lemma")).SelectMany(kvp => kvp["Lemma"]).Any(v => v.name == lemma) : false) : false))))
.ToList())));
finalDictionary = finalDictionary.OrderBy(unit => unit.lemma).ToList();
if (parallelDictsToFiles == true)
{
string path = Path.Combine(_environment.ContentRootPath, "database");
if (!Directory.Exists(path))
{
Directory.CreateDirectory(path);
}
path = Path.Combine(path, "parallelDictionary");
if (!Directory.Exists(path))
{
Directory.CreateDirectory(path);
}
using (StreamWriter w = new StreamWriter(new FileStream(Path.Combine(path, Regex.Replace(string.Join("_", chosenTexts), @"[*""><:/\\|?\.,]", "") + "_parallelDict.json"), FileMode.CreateNew, FileAccess.Write)))
{
for (int u = 0; u < finalDictionary.Count; u++)
{
w.Write(JsonConvert.SerializeObject(finalDictionary[u], Formatting.Indented));
}
}
}
foreach (var unit in finalDictionary)
{
string output = "<b>" + unit.lemma + "</b>";
output += " (" + unit.realizations.Count.ToString() + ")<br />";
output += "<ul>";
foreach (var token in unit.realizations)
{
RealizationGroup coreGroup = null;
Realization coreRealization = null;
foreach (var rg in token)
{
foreach (var realization in rg)
{
if (realization.realizationFields != null)
{
if (realization.realizationFields.Any(f => f.ContainsKey("Lemma")))
{
if (realization.realizationFields.Where(f => f.ContainsKey("Lemma")).SelectMany(kvp => kvp["Lemma"]).Any(v => v.name == unit.lemma))
{
coreGroup = rg;
coreRealization = realization;
break;
}
}
}
if (realization.lexemeTwo == unit.lemma)
{
coreGroup = rg;
coreRealization = realization;
break;
}
}
}
output += "<li>";
var currentDocument = documents.Where(d => d.id == coreRealization.documentID).Single();
output += coreRealization.lexemeTwo + " (" + String.Join(' ', coreGroup.Select(r => r.lexemeTwo)) + ", " + currentDocument.name + " - " + currentDocument.parallelClauses[Convert.ToInt32(coreRealization.textID), Convert.ToInt32(coreRealization.clauseID)].textName + "); <span class=\"clause\" id=\"clauseExtractionButton\" clause=\"clause: " + currentDocument.parallelClauses[Convert.ToInt32(coreRealization.textID), Convert.ToInt32(coreRealization.clauseID)].clause.clauseText + "\">see text segment</span><br />";
output += ((coreRealization.realizationFields != null && coreRealization.realizationFields.Count > 0) ? String.Join(" - ", coreRealization.realizationFields.SelectMany(t => t).Where(a => a.Key != "Lemma").Select(a => String.Join("", a.Value.SelectMany(v => v.name).ToList())).Distinct().ToList()) + "<br />" : "");
output += coreRealization.letters.Any(l => !(l.graphemeFields is null) && l.graphemeFields.Count > 0) ? ("<ul class=\"graphemeFeatures\">" + String.Join("", coreRealization.letters.Select(l => ((l.graphemeFields != null && l.graphemeFields.Count > 0) ? "<li>" + l.grapheme + ": " + String.Join(" - ", l.graphemeFields.SelectMany(t => t).Select(a => a.Key + "(" + String.Join("", a.Value.SelectMany(v => v.name).ToList()) + ")").Distinct().ToList()) + "<br />" : ""))) + "</ul>") : "";
output += "<ul class=\"parallels\"><li>" + String.Join(";<br/><li>", token.GetParallels(coreGroup).Select(rg => String.Join(' ', rg.Select(r => r.lexemeTwo).ToList()) + "(" + currentDocument.name + " - " + currentDocument.parallelClauses[Convert.ToInt32(rg[0].textID), Convert.ToInt32(rg[0].clauseID)].textName + ")" + "; <span class=\"clause\" id=\"clauseExtractionButton\" clause=\"text segment:\r" + String.Join('\r', rg.Select(r => r.clauseID).Distinct().ToList().OrderBy(id => Convert.ToInt32(id)).Select(id => currentDocument.parallelClauses[Convert.ToInt32(rg.textID), Convert.ToInt32(id)].clause.clauseText)) + "\">see text segment</span>")) + "</ul></li>";
}
output += "</ul><br /><br />";
convertedTexts.Add(output);
}
}
}
