private List <ExtractResult> MergeMultipleDuration(string text, List <ExtractResult> extractorResults)
{
if (extractorResults.Count <= 1)
{
return(extractorResults);
}
var unitMap = this.config.UnitMap;
var unitValueMap = this.config.UnitValueMap;
var unitRegex = this.config.DurationUnitRegex;
List <ExtractResult> ret = new List <ExtractResult>();
var firstExtractionIndex = 0;
var timeUnit = 0;
var totalUnit = 0;
while (firstExtractionIndex < extractorResults.Count)
{
string curUnit = null;
var unitMatch = unitRegex.Match(extractorResults[firstExtractionIndex].Text);
if (unitMatch.Success && unitMap.ContainsKey(unitMatch.Groups["unit"].ToString()))
{
curUnit = unitMatch.Groups["unit"].ToString();
totalUnit++;
if (DurationParsingUtil.IsTimeDurationUnit(unitMap[curUnit]))
{
timeUnit++;
}
}
if (string.IsNullOrEmpty(curUnit))
{
firstExtractionIndex++;
continue;
}
var secondExtractionIndex = firstExtractionIndex + 1;
while (secondExtractionIndex < extractorResults.Count)
{
var valid = false;
var midStrBegin = extractorResults[secondExtractionIndex - 1].Start + extractorResults[secondExtractionIndex - 1].Length ?? 0;
var midStrEnd = extractorResults[secondExtractionIndex].Start ?? 0;
if (midStrBegin > midStrEnd)
{
return(extractorResults);
}
var midStr = text.Substring(midStrBegin, midStrEnd - midStrBegin);
var match = this.config.DurationConnectorRegex.Match(midStr);
if (match.Success)
{
unitMatch = unitRegex.Match(extractorResults[secondExtractionIndex].Text);
if (unitMatch.Success && unitMap.ContainsKey(unitMatch.Groups["unit"].ToString()))
{
var nextUnitStr = unitMatch.Groups["unit"].ToString();
if (unitValueMap[unitMap[nextUnitStr]] != unitValueMap[unitMap[curUnit]])
{
valid = true;
if (unitValueMap[unitMap[nextUnitStr]] < unitValueMap[unitMap[curUnit]])
{
curUnit = nextUnitStr;
}
}
totalUnit++;
if (DurationParsingUtil.IsTimeDurationUnit(unitMap[nextUnitStr]))
{
timeUnit++;
}
}
}
if (!valid)
{
break;
}
secondExtractionIndex++;
}
if (secondExtractionIndex - 1 > firstExtractionIndex)
{
var node = new ExtractResult();
node.Start = extractorResults[firstExtractionIndex].Start;
node.Length = extractorResults[secondExtractionIndex - 1].Start + extractorResults[secondExtractionIndex - 1].Length - node.Start;
node.Text = text.Substring(node.Start ?? 0, node.Length ?? 0);
node.Type = extractorResults[firstExtractionIndex].Type;
// Add multiple duration type to extract result
string type = Constants.MultipleDuration_DateTime; // Default type
if (timeUnit == totalUnit)
{
type = Constants.MultipleDuration_Time;
}
else if (timeUnit == 0)
{
type = Constants.MultipleDuration_Date;
}
node.Data = type;
ret.Add(node);
timeUnit = 0;
totalUnit = 0;
}
else
{
ret.Add(extractorResults[firstExtractionIndex]);
}
firstExtractionIndex = secondExtractionIndex;
}
return(ret);
}
private List <ExtractResult> MergeMultipleDuration(string text, List <ExtractResult> extractorResults)
{
if (extractorResults.Count <= 1)
{
return(extractorResults);
}
var unitMap = this.config.UnitMap;
var unitValueMap = this.config.UnitValueMap;
var unitRegex = this.config.DurationUnitRegex;
List <ExtractResult> results = new List <ExtractResult>();
List <List <ExtractResult> > separateResults = new List <List <ExtractResult> >();
var firstExtractionIndex = 0;
var timeUnit = 0;
var totalUnit = 0;
while (firstExtractionIndex < extractorResults.Count)
{
string curUnit = null;
var unitMatch = unitRegex.Match(extractorResults[firstExtractionIndex].Text);
if (unitMatch.Success && unitMap.ContainsKey(unitMatch.Groups["unit"].ToString()))
{
curUnit = unitMatch.Groups["unit"].ToString();
totalUnit++;
if (DurationParsingUtil.IsTimeDurationUnit(unitMap[curUnit]))
{
timeUnit++;
}
}
if (string.IsNullOrEmpty(curUnit))
{
firstExtractionIndex++;
continue;
}
// Add extraction to list of separate results (needed in case the extractions should not be merged)
List <ExtractResult> separateList = new List <ExtractResult>()
{
extractorResults[firstExtractionIndex]
};
var secondExtractionIndex = firstExtractionIndex + 1;
while (secondExtractionIndex < extractorResults.Count)
{
var valid = false;
var midStrBegin = extractorResults[secondExtractionIndex - 1].Start + extractorResults[secondExtractionIndex - 1].Length ?? 0;
var midStrEnd = extractorResults[secondExtractionIndex].Start ?? 0;
var midStr = text.Substring(midStrBegin, midStrEnd - midStrBegin);
var match = this.config.DurationConnectorRegex.Match(midStr);
if (match.Success)
{
unitMatch = unitRegex.Match(extractorResults[secondExtractionIndex].Text);
if (unitMatch.Success && unitMap.ContainsKey(unitMatch.Groups["unit"].ToString()))
{
var nextUnitStr = unitMatch.Groups["unit"].ToString();
if (unitValueMap[nextUnitStr] != unitValueMap[curUnit])
{
valid = true;
if (unitValueMap[nextUnitStr] < unitValueMap[curUnit])
{
curUnit = nextUnitStr;
}
}
totalUnit++;
if (DurationParsingUtil.IsTimeDurationUnit(unitMap[nextUnitStr]))
{
timeUnit++;
}
}
}
if (!valid)
{
break;
}
// Add extraction to list of separate results (needed in case the extractions should not be merged)
separateList.Add(extractorResults[secondExtractionIndex]);
secondExtractionIndex++;
}
if (secondExtractionIndex - 1 > firstExtractionIndex)
{
var node = new ExtractResult();
node.Start = extractorResults[firstExtractionIndex].Start;
node.Length = extractorResults[secondExtractionIndex - 1].Start + extractorResults[secondExtractionIndex - 1].Length - node.Start;
node.Text = text.Substring(node.Start ?? 0, node.Length ?? 0);
node.Type = extractorResults[firstExtractionIndex].Type;
// Add multiple duration type to extract result
string type = Constants.MultipleDuration_DateTime; // Default type
if (timeUnit == totalUnit)
{
type = Constants.MultipleDuration_Time;
}
else if (timeUnit == 0)
{
type = Constants.MultipleDuration_Date;
}
node.Data = type;
results.Add(node);
timeUnit = 0;
totalUnit = 0;
}
else
{
results.Add(extractorResults[firstExtractionIndex]);
}
// Add list of separate extractions to separateResults, so that there is a 1 to 1 correspondence
// between results (list of merged extractions) and separateResults (list of unmerged extractions)
separateResults.Add(separateList);
firstExtractionIndex = secondExtractionIndex;
}
// If the first and last elements of a group of contiguous extractions are both preceded/followed by modifiers,
// they should not be merged, e.g. "last 2 weeks and 3 days ago"
for (int i = results.Count - 1; i >= 0; i--)
{
var start = (int)results[i].Start;
var end = start + (int)results[i].Length;
var beforeStr = text.Substring(0, start);
var afterStr = text.Substring(end);
var beforeMod = this.config.ModPrefixRegex.MatchEnd(beforeStr, trim: true);
var afterMod = this.config.ModSuffixRegex.MatchBegin(afterStr, trim: true);
if (beforeMod.Success && afterMod.Success)
{
results.RemoveAt(i);
results.InsertRange(i, separateResults[i]);
}
}
return(results);
}
