/// <summary>
/// Gets sense display string from entry, leaving traditional/simplified away for monolingual searches.
/// </summary>
private string getSense(int senseIx)
{
var cs = entry.GetSenseAt(senseIx);
string res = "";
if (!cs.Domain.EqualsPlainText("CL:"))
{
string domain = CedictFormatter.HybridToHtml(cs.Domain, script);
domain = CedictFormatter.StripPlainFromHtml(domain);
res += domain;
}
if (cs.Equiv != HybridText.Empty)
{
if (res != string.Empty)
{
res += " ";
}
string equiv = CedictFormatter.HybridToHtml(cs.Equiv, script);
equiv = CedictFormatter.StripPlainFromHtml(equiv);
res += equiv;
}
if (cs.Note != HybridText.Empty)
{
if (res != string.Empty)
{
res += " ";
}
string note = CedictFormatter.HybridToHtml(cs.Note, script);
note = CedictFormatter.StripPlainFromHtml(note);
res += note;
}
return(res);
}
public static void Write(CedictEntry entry, out string head, out string trg)
{
StringBuilder sbHead = new StringBuilder();
sbHead.Append(entry.ChTrad);
sbHead.Append(' ');
sbHead.Append(entry.ChSimpl);
sbHead.Append(" [");
for (int i = 0; i != entry.PinyinCount; ++i)
{
if (i != 0)
{
sbHead.Append(' ');
}
sbHead.Append(entry.GetPinyinAt(i).GetDisplayString(false));
}
sbHead.Append("]");
head = sbHead.ToString();
StringBuilder sbTrg = new StringBuilder();
sbTrg.Append('/');
for (int i = 0; i != entry.SenseCount; ++i)
{
string sense = entry.GetSenseAt(i).GetPlainText();
if (sense.Contains('/'))
{
sense = sense.Replace('/', '\\');
}
sbTrg.Append(sense);
sbTrg.Append('/');
}
trg = sbTrg.ToString();
}
public static string Write(CedictEntry entry)
{
StringBuilder sb = new StringBuilder();
sb.Append(entry.ChTrad);
sb.Append(' ');
sb.Append(entry.ChSimpl);
sb.Append(" [");
for (int i = 0; i != entry.PinyinCount; ++i)
{
if (i != 0)
{
sb.Append(' ');
}
sb.Append(entry.GetPinyinAt(i).GetDisplayString(false));
}
sb.Append("] /");
for (int i = 0; i != entry.SenseCount; ++i)
{
string sense = entry.GetSenseAt(i).GetPlainText();
if (sense.Contains('/'))
{
sense = sense.Replace('/', '\\');
}
sb.Append(sense);
sb.Append('/');
}
return(sb.ToString());
}
/// <summary>
/// Returns true of display font covers all Hanzi in entry; false otherwise.
/// </summary>
private bool areHanziCovered(CedictEntry entry)
{
// Simplified and traditional headword
if (!areHanziCovered(entry.ChSimpl, entry.ChTrad))
{
return(false);
}
// Hanzi in hybrid text of senses
for (int i = 0; i != entry.SenseCount; ++i)
{
CedictSense cs = entry.GetSenseAt(i);
if (!areHanziCovered(cs.Domain))
{
return(false);
}
if (!areHanziCovered(cs.Equiv))
{
return(false);
}
if (!areHanziCovered(cs.Note))
{
return(false);
}
}
// We're good to go.
return(true);
}
private void renderAnnotation(HtmlTextWriter writer)
{
CedictEntry entry = prov.GetEntry(ann.EntryId);
string entryClass = "entry";
if (tones == UiTones.Pleco)
{
entryClass += " toneColorsPleco";
}
else if (tones == UiTones.Dummitt)
{
entryClass += " toneColorsDummitt";
}
writer.AddAttribute(HtmlTextWriterAttribute.Class, entryClass);
writer.RenderBeginTag(HtmlTextWriterTag.Div); // <div class="entry">
writer.AddAttribute(HtmlTextWriterAttribute.Class, "hw-ann");
writer.RenderBeginTag(HtmlTextWriterTag.Span); // <span class="hw-simp">
renderHanzi(query, entry, ann.StartInQuery, ann.LengthInQuery, writer);
writer.RenderEndTag(); // <span class="hw-ann">
writer.AddAttribute(HtmlTextWriterAttribute.Class, "hw-pinyin");
writer.RenderBeginTag(HtmlTextWriterTag.Span); // <span class="hw-pinyin">
bool firstSyll = true;
foreach (var pinyin in entry.Pinyin)
{
if (!firstSyll)
{
writer.WriteEncodedText(" ");
}
firstSyll = false;
writer.WriteEncodedText(pinyin.GetDisplayString(true));
}
writer.RenderEndTag(); // <span class="hw-pinyin">
writer.AddAttribute(HtmlTextWriterAttribute.Class, "senses");
writer.RenderBeginTag(HtmlTextWriterTag.Div); // <div class="senses">
for (int i = 0; i != entry.SenseCount; ++i)
{
renderSense(writer, entry.GetSenseAt(i), i, null);
}
writer.RenderEndTag(); // <div class="senses">
writer.RenderEndTag(); // <div class="entry">
}
private void renderResult(StringBuilder sb, string lang, bool renderEntryDiv, string extraSensesClass = "")
{
CedictEntry entry = entryToRender;
if (entry == null)
{
entry = res.Entry;
}
Dictionary <int, CedictTargetHighlight> senseHLs = new Dictionary <int, CedictTargetHighlight>();
if (res != null)
{
foreach (CedictTargetHighlight hl in res.TargetHilites)
{
senseHLs[hl.SenseIx] = hl;
}
}
string entryClass = "entry";
if (tones == UiTones.Pleco)
{
entryClass += " toneColorsPleco";
}
else if (tones == UiTones.Dummitt)
{
entryClass += " toneColorsDummitt";
}
if (extraEntryClass != "")
{
entryClass += " " + extraEntryClass;
}
if (renderEntryDiv)
{
sb.Append("<div class='" + entryClass + "'>"); // <div class="entry">
}
if (entryId != null && lang != null)
{
sb.Append("<a class='ajax' href='/" + lang + "/edit/existing/" + entryId + "'>");
sb.Append("<i class='fa fa-pencil entryAction edit'></i></a>");
}
XRenderStatus(sb);
XRenderHanzi(sb);
XRenderPinyin(sb);
string sensesClass = "senses";
if (!string.IsNullOrEmpty(extraSensesClass))
{
sensesClass += " " + extraSensesClass;
}
sb.Append("<div class='" + sensesClass + "'>"); // <div class="senses">
int senseIx = 0;
for (int i = 0; i != entry.SenseCount; ++i)
{
CedictTargetHighlight thl = null;
if (senseHLs.ContainsKey(i))
{
thl = senseHLs[i];
}
var sense = entry.GetSenseAt(i);
if (isSpecialSense(sense))
{
if (i != 0)
{
sb.Append("<br/>");
}
}
else if (i != 0)
{
if (isSpecialSense(entry.GetSenseAt(i - 1)))
{
sb.Append("<br/>");
}
else
{
sb.Append(' ');
}
}
renderSense(sb, sense, ref senseIx, thl);
}
sb.Append("</div>"); // <div class="senses">
if (renderEntryDiv)
{
sb.Append("</div>"); // <div class="entry">
}
}
private void renderAnnotation(StringBuilder sb)
{
CedictEntry entry = ann.Entry;
string entryClass = "entry";
if (tones == UiTones.Pleco)
{
entryClass += " toneColorsPleco";
}
else if (tones == UiTones.Dummitt)
{
entryClass += " toneColorsDummitt";
}
if (extraEntryClass != "")
{
entryClass += " " + extraEntryClass;
}
sb.Append("<div class='" + entryClass + "'>"); // <div class="entry">
sb.Append("<span class='hw-ann'>"); // <span class="hw-simp">
renderHanzi(query, entry, ann.StartInQuery, ann.LengthInQuery, sb);
sb.Append("</span>"); // <span class="hw-ann">
sb.Append("<span class='hw-pinyin'>"); // <span class="hw-pinyin">
bool firstSyll = true;
foreach (var pinyin in entry.Pinyin)
{
if (!firstSyll)
{
sb.Append(" ");
}
firstSyll = false;
sb.Append(HtmlEncoder.Default.Encode(pinyin.GetDisplayString(true)));
}
sb.Append("</span>"); // <span class="hw-pinyin">
sb.Append("<div class='senses'>"); // <div class="senses">
int senseIx = 0;
for (int i = 0; i != entry.SenseCount; ++i)
{
var sense = entry.GetSenseAt(i);
if (isSpecialSense(sense))
{
if (i != 0)
{
sb.Append("<br/>");
}
}
else if (i != 0)
{
if (isSpecialSense(entry.GetSenseAt(i - 1)))
{
sb.Append("<br/>");
}
else
{
sb.Append(' ');
}
}
renderSense(sb, sense, ref senseIx, null);
}
sb.Append("</div>"); // <div class="senses">
sb.Append("</div>"); // <div class="entry">
}
private bool verifyTrg(Tokenizer tokenizer, CedictEntry entry, int entryId, int senseIx,
List <Token> qtoks, List <CedictResult> res)
{
if (entry == null)
{
return(false);
}
// Tokenize indicated sense's equiv; see if it matches query
string equiv = entry.GetSenseAt(senseIx).Equiv;
List <Token> rtoks = tokenizer.Tokenize(equiv);
for (int i = 0; i != rtoks.Count; ++i)
{
int j = 0;
bool startSplit = false;
bool endSplit = false;
for (; j != qtoks.Count; ++j)
{
if (i + j >= rtoks.Count)
{
break;
}
bool ok = false;
Token rtok = rtoks[i + j];
Token qtok = qtoks[j];
if (rtok.Norm == qtok.Norm)
{
// Stopwords: only OK if token is an entire sub-sense in retrieved sense
if (!trgStopWords.Contains(rtok.Norm))
{
ok = true;
}
else
{
ok = rtok.SubSeq == 0 && (i == rtoks.Count - 1 || rtoks[i + 1].SubSeq == 0);
}
}
// First query word: can be second half of word in retrieved sense
if (j == 0 && rtok.SplitPosNorm != 0)
{
string rTwo = rtok.Norm.Substring(rtok.SplitPosNorm);
if (rTwo == qtok.Norm)
{
// Stopwords: only if query is min 2 tokens, or result is one token only
if (!trgStopWords.Contains(rTwo) || qtoks.Count > 1)
{
ok = true;
}
startSplit = true;
}
}
// Last query word: can be first half of word in retrieved sense
if (j == qtoks.Count - 1 && rtok.SplitPosNorm != 0)
{
string rOne = rtok.Norm.Substring(0, rtok.SplitPosNorm);
if (rOne == qtok.Norm)
{
// Stopwords: only if query is min 2 tokens, or result is one token only
if (!trgStopWords.Contains(rOne) || qtoks.Count > 1)
{
ok = true;
}
endSplit = true;
}
}
if (!ok)
{
break;
}
}
if (j != qtoks.Count)
{
continue;
}
// We got a match starting at i!
CedictTargetHighlight[] hlarr = new CedictTargetHighlight[1];
int start = rtoks[i].Start;
if (startSplit)
{
start += rtoks[i].SplitPosSurf;
}
int end = rtoks[i + j - 1].Start + rtoks[i + j - 1].Surf.Length;
if (endSplit)
{
end -= (rtoks[i + j - 1].Surf.Length - rtoks[i + j - 1].SplitPosSurf);
}
hlarr[0] = new CedictTargetHighlight(senseIx, start, end - start);
ReadOnlyCollection <CedictTargetHighlight> hlcoll = new ReadOnlyCollection <CedictTargetHighlight>(hlarr);
CedictResult cr = new CedictResult(entry, hlcoll);
// Stop right here
res.Add(cr);
return(true);
}
// Not a match
return(false);
}
/// <summary>
/// Gets the entry formatted in HTML.
/// </summary>
public static string GetHtml(ITextProvider tprov, CedictEntry entry, SearchScript script)
{
StringBuilder bodyHtml = new StringBuilder();
// Are we showing one or two Hanzi headwords?
string hanzi1 = script == SearchScript.Traditional ? entry.ChTrad : entry.ChSimpl;
string hanzi2 = null;
if (script == SearchScript.Both && entry.ChSimpl != entry.ChTrad)
{
hanzi2 = entry.ChTrad;
}
// Find simplest possible template, work with that
// Only one hanzi, no longer than 2 chars, only one sense
bool mustDoSenses = true;
if (hanzi2 == null && hanzi1.Length <= 2 && entry.SenseCount == 1)
{
mustDoSenses = false;
bodyHtml.Append(template1);
bodyHtml.Replace("{hanzi}", escape(hanzi1));
bodyHtml.Replace("{pinyin}", escape(GetPinyinString(entry.GetPinyinForDisplay(true))));
bodyHtml.Replace("{sense}", getSenseHtmlPure(tprov, entry.GetSenseAt(0), script));
}
// Only one script, no more than 6 chars
else if (hanzi2 == null && hanzi1.Length <= 6)
{
bodyHtml.Append(template2);
bodyHtml.Replace("{hanzi}", escape(hanzi1));
bodyHtml.Replace("{pinyin}", escape(GetPinyinString(entry.GetPinyinForDisplay(true))));
}
// Only one script
else if (hanzi2 == null)
{
bodyHtml.Append(template3);
bodyHtml.Replace("{hanzi}", escape(hanzi1));
bodyHtml.Replace("{pinyin}", escape(GetPinyinString(entry.GetPinyinForDisplay(true))));
}
// Everything else: very full-fledged entry
else
{
bodyHtml.Append(template4);
bodyHtml.Replace("{hanzi1}", escape(hanzi1));
bodyHtml.Replace("{hanzi2}", escape(hanzi2));
bodyHtml.Replace("{pinyin}", escape(GetPinyinString(entry.GetPinyinForDisplay(true))));
}
// In all but the first, simplest case, dealing with senses is the same
if (mustDoSenses)
{
StringBuilder sbSenses = new StringBuilder();
foreach (CedictSense sense in entry.Senses)
{
string senseHtml = "";
if (!sense.Domain.EqualsPlainText("CL:"))
{
senseHtml += templateDiamond;
senseHtml += " ";
}
senseHtml += getSenseHtmlPure(tprov, sense, script);
senseHtml = templateSense.Replace("{sense}", senseHtml);
sbSenses.Append(senseHtml);
}
bodyHtml.Replace("{senses}", sbSenses.ToString());
}
// Assemble the whole HTML
StringBuilder sb = new StringBuilder();
sb.Append(templateOuter);
sb.Replace("{body}", bodyHtml.ToString());
// Purge new lines and tabs: this avoids extra spaces e.g. when pasting into Word
sb.Replace("\r\n", "");
sb.Replace("\t", "");
// Done
return(sb.ToString());
}
private void renderResult(HtmlTextWriter writer)
{
int hanziLimit = isMobile ? 4 : 6;
CedictEntry entry = prov.GetEntry(res.EntryId);
Dictionary <int, CedictTargetHighlight> senseHLs = new Dictionary <int, CedictTargetHighlight>();
foreach (CedictTargetHighlight hl in res.TargetHilites)
{
senseHLs[hl.SenseIx] = hl;
}
string entryClass = "entry";
if (tones == UiTones.Pleco)
{
entryClass += " toneColorsPleco";
}
else if (tones == UiTones.Dummitt)
{
entryClass += " toneColorsDummitt";
}
writer.AddAttribute(HtmlTextWriterAttribute.Class, entryClass);
writer.RenderBeginTag(HtmlTextWriterTag.Div); // <div class="entry">
if (script != UiScript.Trad)
{
writer.AddAttribute(HtmlTextWriterAttribute.Class, "hw-simp");
writer.RenderBeginTag(HtmlTextWriterTag.Span); // <span class="hw-simp">
renderHanzi(entry, true, false, writer);
writer.RenderEndTag(); // <span class="hw-simp">
}
if (script == UiScript.Both)
{
// Up to 6 hanzi: on a single line
if (entry.ChSimpl.Length <= hanziLimit)
{
string clsSep = "hw-sep";
if (tones != UiTones.None)
{
clsSep = "hw-sep faint";
}
writer.AddAttribute(HtmlTextWriterAttribute.Class, clsSep);
writer.RenderBeginTag(HtmlTextWriterTag.Span); // <span class="hw-sep">
writer.WriteEncodedText("•");
writer.RenderEndTag(); // <span class="hw-sep">
}
// Otherwise, line break
else
{
writer.RenderBeginTag(HtmlTextWriterTag.Br);
writer.RenderEndTag();
}
}
if (script != UiScript.Simp)
{
string clsTrad = "hw-trad";
// Need special class so traditional floats left after line break
if (script == UiScript.Both && entry.ChSimpl.Length > hanziLimit)
{
clsTrad = "hw-trad break";
}
writer.AddAttribute(HtmlTextWriterAttribute.Class, clsTrad);
writer.RenderBeginTag(HtmlTextWriterTag.Span); // <span class="hw-trad">
renderHanzi(entry, false, script == UiScript.Both, writer);
writer.RenderEndTag(); // <span class="hw-trad">
}
writer.AddAttribute(HtmlTextWriterAttribute.Class, "hw-pinyin");
writer.RenderBeginTag(HtmlTextWriterTag.Span); // <span class="hw-pinyin">
bool firstSyll = true;
foreach (var pinyin in entry.Pinyin)
{
if (!firstSyll)
{
writer.WriteEncodedText(" ");
}
firstSyll = false;
writer.WriteEncodedText(pinyin.GetDisplayString(true));
}
writer.RenderEndTag(); // <span class="hw-pinyin">
writer.AddAttribute(HtmlTextWriterAttribute.Class, "senses");
writer.RenderBeginTag(HtmlTextWriterTag.Div); // <div class="senses">
for (int i = 0; i != entry.SenseCount; ++i)
{
CedictTargetHighlight thl = null;
if (senseHLs.ContainsKey(i))
{
thl = senseHLs[i];
}
renderSense(writer, entry.GetSenseAt(i), i, thl);
}
writer.RenderEndTag(); // <div class="senses">
writer.RenderEndTag(); // <div class="entry">
}
