private static ArchiveDatePoint ParseDecade(string text)
{
ArchiveDatePoint point = new ArchiveDatePoint
{
ValueType = DateValueType.Decade
};
switch (text.ToLowerInvariant())
{
case "dieci":
point.Value = 191;
break;
case "venti":
point.Value = 192;
break;
case "trenta":
point.Value = 193;
break;
case "quaranta":
point.Value = 194;
break;
case "cinquanta":
point.Value = 195;
break;
case "sessanta":
point.Value = 196;
break;
case "settanta":
point.Value = 197;
break;
case "ottanta":
point.Value = 198;
break;
case "novanta":
point.Value = 199;
break;
default:
// 'NN or NNNN
point.Value = text[0] == '\''
? (short)(short.Parse(text.Substring(1),
CultureInfo.InvariantCulture) / 10 + 190)
: (short)
(short.Parse(text, CultureInfo.InvariantCulture) / 10);
break;
}
return(point);
}
/// <summary>
/// Parses the specified text, representing one or more dates or dates
/// ranges.
/// </summary>
/// <param name="text">The text.</param>
/// <returns>date(s) parsed</returns>
/// <exception cref="ArgumentNullException">null text</exception>
public IList <ArchiveDate> Parse(string text)
{
if (text == null)
{
throw new ArgumentNullException(nameof(text));
}
// split at `;` (TODO: prefilter to replace `,` as ranges
// separator with `;`)
DateMonthStyle monthStyle = DateMonthStyle.Undefined;
char ymdSep = '\0';
List <ArchiveDate> dates = new List <ArchiveDate>();
// several dates (or dates ranges) are separated by ;
foreach (string part in text.Split(new[] { ';' },
StringSplitOptions.RemoveEmptyEntries))
{
// remove [ ] to simplify styles detection
string purgedPart = _squaresRegex.Replace(part, "");
// detect month style (unless already detected in another part)
if (monthStyle == DateMonthStyle.Undefined)
{
monthStyle = DetectMonthStyle(purgedPart);
}
// detect numeric M style (unless already detected in another part)
if ((monthStyle == DateMonthStyle.Numeric ||
monthStyle == DateMonthStyle.Undefined) &&
ymdSep == '\0')
{
ymdSep = DetectYmdSeparator(purgedPart);
}
// detect YMD order
bool dmy = _dmyRegex.IsMatch(purgedPart) ||
_dmyNamedRegex.IsMatch(purgedPart);
// split range pair if any
var t = SplitRange(part, ymdSep);
if (t != null)
{
ArchiveDate date = new ArchiveDate
{
A = ParsePoint(t.Item1, dmy, monthStyle),
B = ParsePoint(t.Item2, dmy, monthStyle)
};
// corner case: shortened ranges
AdjustForShortenedRange(t, date, dmy, monthStyle);
dates.Add(date);
} // rng
else
{
// ante/post can appear only before single-point dates
string pointText = part;
Match m = _antePostRegex.Match(part);
bool min = true, max = true;
if (m.Success)
{
pointText = part.Substring(m.Length);
if (string.Equals(m.Groups[1].Value, "ante",
StringComparison.InvariantCultureIgnoreCase))
{
min = false;
}
else
{
max = false;
}
}
ArchiveDatePoint point = ParsePoint(pointText, dmy, monthStyle);
if (point != null)
{
ArchiveDate date = new ArchiveDate();
if (min && max)
{
date.A = point;
date.B = point.Clone();
}
else
{
if (min)
{
date.A = point;
}
else
{
date.B = point;
}
}
dates.Add(date);
}
} // !rng
}
return(dates);
}
private ArchiveDatePoint ParsePoint(string text, bool dmy,
DateMonthStyle monthStyle)
{
// senza data or s.d.
if (_sineDataRegex.IsMatch(text))
{
return(null);
}
// century
if (text.IndexOf("secolo", StringComparison.OrdinalIgnoreCase) > -1 ||
text.IndexOf("sec.", StringComparison.CurrentCultureIgnoreCase) > -1)
{
return(ParseCenturyPoint(text));
}
// decade
Match match = _decadeRegex.Match(text);
if (match.Success)
{
return(ParseDecade(match.Groups[1].Value));
}
// DMY/YMD
ArchiveDatePoint point = new ArchiveDatePoint();
// about
if (_aboutRegex.IsMatch(text))
{
point.Approximation = ApproximationType.About;
}
int i = (int)(monthStyle == 0 ? 0 : monthStyle - 1);
Regex r = dmy ? _dmyRegexes[i] : _ymdRegexes[i];
match = r.Match(text);
if (!match.Success)
{
return(null);
}
YmdToken y = YmdToken.Parse(match.Groups["y"].Value, 'y');
YmdToken m = YmdToken.Parse(match.Groups["m"].Value, 'm');
YmdToken d = YmdToken.Parse(match.Groups["d"].Value, 'd');
// take square brackets into account: this depends on the DMY/YMD order
MarkInferred(dmy ? new[] { d, m, y } : new[] { y, m, d });
point.Value = y.Value;
point.IsYearInferred = y.IsInferred;
if (m != null)
{
point.Month = m.Value;
point.IsMonthInferred = m.IsInferred;
}
if (d != null)
{
point.Day = d.Value;
point.IsDayInferred = d.IsInferred;
}
return(point);
}
private ArchiveDatePoint ParseCenturyPoint(string text)
{
ArchiveDatePoint point = new ArchiveDatePoint
{
ValueType = DateValueType.Century
};
// prefix modifiers
Match m = _centPrefixModRegex.Match(text);
if (m.Success)
{
switch (_wsRegex.Replace(m.Groups[1].Value, " ").ToLowerInvariant())
{
case "ca.":
case "circa":
point.Approximation = ApproximationType.About;
break;
case "inizio":
point.Approximation = ApproximationType.Beginning;
break;
case "i metà":
point.Approximation = ApproximationType.FirstHalf;
break;
case "metà":
point.Approximation = ApproximationType.Mid;
break;
case "ii metà":
point.Approximation = ApproximationType.SecondHalf;
break;
case "fine":
point.Approximation = ApproximationType.End;
break;
}
// remove matched prefix
text = text.Substring(m.Length);
}
// postfix modifiers
if (point.Approximation == ApproximationType.None)
{
m = _centSuffixModRegex.Match(text);
if (m.Success)
{
switch (m.Groups[1].Value.ToLowerInvariant())
{
case "in.":
point.Approximation = ApproximationType.Beginning;
break;
case "ex.":
point.Approximation = ApproximationType.End;
break;
}
// remove matched postfix
text = text.Remove(m.Groups[1].Index, m.Groups[1].Length);
}
}
// parse value
m = _centuryRegex.Match(text);
if (!m.Success)
{
throw new ArgumentException(
LocalizedStrings.Format(
Properties.Resources.NoCenturyValue, text));
}
point.Value = (short)RomanNumber.FromRoman(m.Value);
if (_acRegex.IsMatch(text))
{
point.Value = (short)-point.Value;
}
return(point);
}
