private void GetMatchedStartAndLength(Match match, string type, string source, out int start, out int length)
{
start = NumberRangeConstants.INVALID_NUM;
length = NumberRangeConstants.INVALID_NUM;
var numberStr1 = match.Groups["number1"].Value;
var numberStr2 = match.Groups["number2"].Value;
if (type.Contains(NumberRangeConstants.TWONUM))
{
var extractNumList1 = ExtractNumberAndOrdinalFromStr(numberStr1);
var extractNumList2 = ExtractNumberAndOrdinalFromStr(numberStr2);
if (extractNumList1 != null && extractNumList2 != null)
{
if (type.Contains(NumberRangeConstants.TWONUMTILL))
{
// num1 must less than num2
var num1 = (double)(numberParser.Parse(extractNumList1[0]).Value ?? 0);
var num2 = (double)(numberParser.Parse(extractNumList2[0]).Value ?? 0);
if (num1 > num2)
{
return;
}
}
bool validNum1 = false, validNum2 = false;
start = match.Index;
length = match.Length;
validNum1 = ValidateMatchAndGetStartAndLength(extractNumList1, numberStr1, match, source, ref start, ref length);
validNum2 = ValidateMatchAndGetStartAndLength(extractNumList2, numberStr2, match, source, ref start, ref length);
if (!validNum1 || !validNum2)
{
start = NumberRangeConstants.INVALID_NUM;
length = NumberRangeConstants.INVALID_NUM;
}
}
}
else
{
var numberStr = string.IsNullOrEmpty(numberStr1) ? numberStr2 : numberStr1;
var extractNumList = ExtractNumberAndOrdinalFromStr(numberStr);
if (extractNumList != null)
{
start = match.Index;
length = match.Length;
if (!ValidateMatchAndGetStartAndLength(extractNumList, numberStr, match, source, ref start, ref length))
{
start = NumberRangeConstants.INVALID_NUM;
length = NumberRangeConstants.INVALID_NUM;
}
}
}
}
public virtual List <ExtractResult> Extract(string source)
{
if (string.IsNullOrEmpty(source))
{
return(new List <ExtractResult>());
}
var results = new List <ExtractResult>();
var matchSource = new Dictionary <Tuple <int, int>, string>();
var matched = new bool[source.Length];
var collections = Regexes.ToDictionary(o => o.Key.Matches(source), p => p.Value);
foreach (var collection in collections)
{
foreach (Match m in collection.Key)
{
GetMatchedStartAndLength(m, collection.Value, source, out int start, out int length);
if (start >= 0 && length > 0)
{
// Keep Source Data for extra information
matchSource.Add(new Tuple <int, int>(start, length), collection.Value);
}
}
}
foreach (var match in matchSource)
{
var start = match.Key.Item1;
var length = match.Key.Item2;
// Filter wrong two number ranges such as "more than 20 and less than 10" and "大于20小于10".
if (match.Value.Equals(NumberRangeConstants.TWONUM, StringComparison.Ordinal))
{
int moreIndex = 0, lessIndex = 0;
var text = source.Substring(match.Key.Item1, match.Key.Item2);
var er = numberExtractor.Extract(text);
if (er.Count != 2)
{
er = ordinalExtractor.Extract(text);
if (er.Count != 2)
{
continue;
}
}
var nums = er.Select(r => (double)(numberParser.Parse(r).Value ?? 0)).ToList();
// Order matchSource by decreasing match length so that "no less than x" is before "less than x"
var matchList = matchSource.ToList();
matchList.Sort((pair1, pair2) => pair2.Key.Item2.CompareTo(pair1.Key.Item2));
moreIndex = matchList.First(r =>
r.Value.Equals(NumberRangeConstants.MORE, StringComparison.Ordinal) &&
r.Key.Item1 >= start && r.Key.Item1 + r.Key.Item2 <= start + length).Key.Item1;
lessIndex = matchList.First(r =>
r.Value.Equals(NumberRangeConstants.LESS, StringComparison.Ordinal) &&
r.Key.Item1 >= start && r.Key.Item1 + r.Key.Item2 <= start + length).Key.Item1;
if (!((nums[0] < nums[1] && moreIndex <= lessIndex) || (nums[0] > nums[1] && moreIndex >= lessIndex)))
{
continue;
}
}
// The entity is longer than 1, so don't mark the last char to represent the end.
// To avoid no connector cases like "大于20小于10" being marked as a whole entity.
for (var j = 0; j < length - 1; j++)
{
matched[start + j] = true;
}
}
var last = -1;
for (var i = 0; i < source.Length; i++)
{
if (matched[i])
{
if (i + 1 == source.Length || !matched[i + 1])
{
var start = last + 1;
var length = i - last + 1;
var substr = source.Substring(start, length);
if (matchSource.Keys.Any(o => o.Item1 == start && o.Item2 == length))
{
var srcMatch = matchSource.Keys.First(o => o.Item1 == start && o.Item2 == length);
var er = new ExtractResult
{
Start = start,
Length = length,
Text = substr,
Type = ExtractType,
Data = matchSource.ContainsKey(srcMatch) ? matchSource[srcMatch] : null,
};
results.Add(er);
}
}
}
else
{
last = i;
}
}
// In ExperimentalMode, cases like "from 3 to 5" and "between 10 and 15" are set to closed at both start and end
if ((Config.Options & NumberOptions.ExperimentalMode) != 0)
{
foreach (var result in results)
{
var data = result.Data.ToString();
if (data == NumberRangeConstants.TWONUMBETWEEN ||
data == NumberRangeConstants.TWONUMTILL)
{
result.Data = NumberRangeConstants.TWONUMCLOSED;
}
}
}
return(results);
}