Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[JA DateTimeV2] Merged refinements #2950

Merged
merged 3 commits into from
May 16, 2022
Merged
Show file tree
Hide file tree
Changes from 2 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view

Large diffs are not rendered by default.

Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,8 @@ public ChineseDurationExtractorConfiguration(IDateTimeOptionsConfiguration confi

public Dictionary<string, long> UnitValueMap { get; }

public Dictionary<Regex, Regex> AmbiguityDurationFiltersDict => null;

Regex ICJKDurationExtractorConfiguration.DurationUnitRegex => DurationUnitRegex;

Regex ICJKDurationExtractorConfiguration.DurationConnectorRegex => DurationConnectorRegex;
Expand Down
2 changes: 2 additions & 0 deletions .NET/Microsoft.Recognizers.Text.DateTime/Constants.cs
Original file line number Diff line number Diff line change
Expand Up @@ -53,6 +53,7 @@ public static class Constants

// AmPm time representation for time parser
public const string Comment_AmPm = "ampm";
public const string Comment_Am = "am";

// Prefix early/late for time parser
public const string Comment_Early = "early";
Expand Down Expand Up @@ -267,6 +268,7 @@ public static class Constants
public const string TimexFuzzyDay = "XX";
public const string DateTimexConnector = "-";
public const string TimeTimexConnector = ":";
public const string TimexSeparator = ",";
public const string GeneralPeriodPrefix = "P";
public const string TimeTimexPrefix = "T";

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,8 @@
// Licensed under the MIT License.

using System.Collections.Generic;

using System.Linq;
using System.Text.RegularExpressions;
using Microsoft.Recognizers.Text.Utilities;
using DateObject = System.DateTime;

Expand Down Expand Up @@ -54,9 +55,32 @@ public List<ExtractResult> Extract(string source, DateObject referenceTime)
res = MergeMultipleDuration(source, res);
}

res = FilterAmbiguity(res, source);

return res;
}

private List<ExtractResult> FilterAmbiguity(List<ExtractResult> extractResults, string text)
{
if (this.config.AmbiguityDurationFiltersDict != null)
{
foreach (var regex in this.config.AmbiguityDurationFiltersDict)
{
foreach (var extractResult in extractResults)
{
if (regex.Key.IsMatch(text))
{
var matches = regex.Value.Matches(text).Cast<Match>();
extractResults = extractResults.Where(er => !matches.Any(m => m.Index < er.Start + er.Length && m.Index + m.Length > er.Start))
.ToList();
}
}
}
}

return extractResults;
}

private List<ExtractResult> MergeMultipleDuration(string text, List<ExtractResult> extractorResults)
{
if (extractorResults.Count <= 1)
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -129,6 +129,17 @@ private void AddMod(List<ExtractResult> ers, string text)
er.Metadata = AssignModMetadata(er.Metadata);
}

match = this.config.UntilRegex.MatchBegin(afterStr, trim: true);

if (match.Success)
{
var modLength = match.Index + match.Length;
er.Length += modLength;
er.Text = text.Substring(er.Start ?? 0, er.Length ?? 0);

er.Metadata = AssignModMetadata(er.Metadata);
}

match = this.config.SincePrefixRegex.MatchEnd(beforeStr, trim: true);

if (match.Success && AmbiguousRangeChecker(beforeStr, text, er))
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -32,5 +32,7 @@ public interface ICJKDurationExtractorConfiguration : IDateTimeOptionsConfigurat

Dictionary<string, long> UnitValueMap { get; }

Dictionary<Regex, Regex> AmbiguityDurationFiltersDict { get; }

}
}
Original file line number Diff line number Diff line change
Expand Up @@ -6,8 +6,8 @@
using System.Globalization;
using System.Linq;
using System.Text.RegularExpressions;

using Microsoft.Recognizers.Definitions.Japanese;
using Microsoft.Recognizers.Definitions.Utilities;
using Microsoft.Recognizers.Text.NumberWithUnit;
using Microsoft.Recognizers.Text.NumberWithUnit.Japanese;

Expand Down Expand Up @@ -48,6 +48,8 @@ public JapaneseDurationExtractorConfiguration(IDateTimeOptionsConfiguration conf

UnitMap = DateTimeDefinitions.ParserConfigurationUnitMap.ToDictionary(k => k.Key, k => k.Value);
UnitValueMap = DateTimeDefinitions.DurationUnitValueMap;
AmbiguityDurationFiltersDict = DefinitionLoader.LoadAmbiguityFilters(DateTimeDefinitions.AmbiguityDurationFiltersDict);

}

public IExtractor InternalExtractor { get; }
Expand All @@ -56,6 +58,8 @@ public JapaneseDurationExtractorConfiguration(IDateTimeOptionsConfiguration conf

public Dictionary<string, long> UnitValueMap { get; }

public Dictionary<Regex, Regex> AmbiguityDurationFiltersDict { get; }

Regex ICJKDurationExtractorConfiguration.DurationUnitRegex => DurationUnitRegex;

Regex ICJKDurationExtractorConfiguration.DurationConnectorRegex => DurationConnectorRegex;
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,8 @@ public KoreanDurationExtractorConfiguration(IDateTimeOptionsConfiguration config

public Dictionary<string, long> UnitValueMap { get; }

public Dictionary<Regex, Regex> AmbiguityDurationFiltersDict => null;

Regex ICJKDurationExtractorConfiguration.DurationUnitRegex => DurationUnitRegex;

Regex ICJKDurationExtractorConfiguration.DurationConnectorRegex => DurationConnectorRegex;
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -318,8 +318,9 @@ protected DateTimeResolutionResult ParseImplicitDate(string text, DateObject ref
// handle "明日から3週間" (3 weeks from tomorrow)
var durationResult = this.config.DurationExtractor.Extract(text, referenceDate);
var unitMatch = this.config.DurationRelativeDurationUnitRegex.Match(text);
var isWithin = this.config.DurationRelativeDurationUnitRegex.MatchEnd(text, trim: true).Groups[Constants.WithinGroupName].Success;

if (exactMatch.Success && unitMatch.Success && (durationResult.Count > 0) &&
if ((exactMatch.Success || isWithin) && unitMatch.Success && (durationResult.Count > 0) &&
string.IsNullOrEmpty(unitMatch.Groups["few"].Value))
{
var pr = this.config.DurationParser.Parse(durationResult[0], referenceDate);
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -158,6 +158,14 @@ private DateTimeResolutionResult MergeDateAndTime(string text, DateObject refere
var pastDate = (DateObject)((DateTimeResolutionResult)pr1.Value).PastValue;
var time = (DateObject)((DateTimeResolutionResult)pr2.Value).FutureValue;

// handle cases with time like 25時 which resolve to the next day
var timexHour = TimexUtility.ParseHourFromTimeTimex(pr2.TimexStr);
if (timexHour > Constants.DayHourCount)
{
futureDate = futureDate.AddDays(1);
pastDate = pastDate.AddDays(1);
}

var hour = time.Hour;
var min = time.Minute;
var sec = time.Second;
Expand All @@ -178,7 +186,6 @@ private DateTimeResolutionResult MergeDateAndTime(string text, DateObject refere
timeStr = timeStr.Substring(0, timeStr.Length - 4);
}

timeStr = "T" + hour.ToString("D2", CultureInfo.InvariantCulture) + timeStr.Substring(3);
ret.Timex = pr1.TimexStr + timeStr;

var val = (DateTimeResolutionResult)pr2.Value;
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -126,15 +126,31 @@ private DateTimeResolutionResult MergeDateAndTimePeriod(string text, DateObject
var futureDate = (DateObject)((DateTimeResolutionResult)pr1.Value).FutureValue;
var pastDate = (DateObject)((DateTimeResolutionResult)pr1.Value).PastValue;

// handle cases with time like 25時 which resolve to the next day
var swiftDay = 0;
var timexHours = TimexUtility.ParseHoursFromTimePeriodTimex(pr2.TimexStr);
if (timexHours.Item1 > Constants.DayHourCount)
{
pastDate = pastDate.AddDays(1);
futureDate = futureDate.AddDays(1);
}
else if (timexHours.Item2 > Constants.DayHourCount)
{
swiftDay++;
}

var pastDate1 = pastDate.AddDays(swiftDay);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe rename to pastDateAlt as it's an alternative resolution?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

done

var futureDate1 = futureDate.AddDays(swiftDay);

ret.FutureValue =
new Tuple<DateObject, DateObject>(
DateObject.MinValue.SafeCreateFromValue(futureDate.Year, futureDate.Month, futureDate.Day, beginTime.Hour, beginTime.Minute, beginTime.Second),
DateObject.MinValue.SafeCreateFromValue(futureDate.Year, futureDate.Month, futureDate.Day, endTime.Hour, endTime.Minute, endTime.Second));
DateObject.MinValue.SafeCreateFromValue(futureDate1.Year, futureDate1.Month, futureDate1.Day, endTime.Hour, endTime.Minute, endTime.Second));

ret.PastValue =
new Tuple<DateObject, DateObject>(
DateObject.MinValue.SafeCreateFromValue(pastDate.Year, pastDate.Month, pastDate.Day, beginTime.Hour, beginTime.Minute, beginTime.Second),
DateObject.MinValue.SafeCreateFromValue(pastDate.Year, pastDate.Month, pastDate.Day, endTime.Hour, endTime.Minute, endTime.Second));
DateObject.MinValue.SafeCreateFromValue(pastDate1.Year, pastDate1.Month, pastDate1.Day, endTime.Hour, endTime.Minute, endTime.Second));

ret.Timex = TimexUtility.GenerateSplitDateTimePeriodTimex(pr1.TimexStr, pr2.TimexStr);
ret.Success = !string.IsNullOrEmpty(ret.Timex);
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -79,12 +79,12 @@ public DateTimeResolutionResult PackTimeResult(DateTimeExtra<TimeType> extra, Ti
noDesc = false;
}

int hour = timeResult.Hour > 0 ? timeResult.Hour % Constants.DayHourCount : 0,
// Hours > 24 (e.g. 25時 which resolves to the next day) are kept unnormalized in the timex
// to avoid ambiguity in other entities. For example, "on the 30th at 25" is resolved to
// "XXXX-XX-30T25" because with "XXXX-XX-30+1T01" it is not known if the day should be "31" or "01".
int hour = timeResult.Hour > 0 && timeResult.Hour != Constants.DayHourCount ? timeResult.Hour : 0,
min = timeResult.Minute > 0 ? timeResult.Minute : 0,
second = timeResult.Second > 0 ? timeResult.Second : 0,
day = referenceTime.Day,
month = referenceTime.Month,
year = referenceTime.Year;
second = timeResult.Second > 0 ? timeResult.Second : 0;

var dateTimeResult = new DateTimeResolutionResult();

Expand All @@ -109,6 +109,18 @@ public DateTimeResolutionResult PackTimeResult(DateTimeExtra<TimeType> extra, Ti
build.Append(":" + second.ToString("D2", CultureInfo.InvariantCulture));
}

// handle cases with time like 25時 (the hour is normalized in the past/future values)
if (timeResult.Hour > Constants.DayHourCount)
{
hour = timeResult.Hour - Constants.DayHourCount;
referenceTime = referenceTime.AddDays(1);
if (noDesc)
{
dateTimeResult.Comment = Constants.Comment_Am;
noDesc = false;
}
}

if (noDesc && hour <= Constants.HalfDayHourCount)
{
// build.Append("ampm");
Expand All @@ -117,6 +129,7 @@ public DateTimeResolutionResult PackTimeResult(DateTimeExtra<TimeType> extra, Ti

dateTimeResult.Timex = build.ToString();

int day = referenceTime.Day, month = referenceTime.Month, year = referenceTime.Year;
dateTimeResult.FutureValue = dateTimeResult.PastValue =
DateObject.MinValue.SafeCreateFromValue(year, month, day, hour, min, second);
dateTimeResult.Success = true;
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -91,31 +91,50 @@ public static DateTimeResolutionResult Handle(IDateTimeParser timeParser, DateTi

int day = refTime.Day,
month = refTime.Month,
year = refTime.Year;
year = refTime.Year,
rightSwiftDay = 0,
leftSwiftDay = 0;

// determine if the right side time is smaller than the left side, if yes, add one day
int hour = leftResult.Hour > 0 ? leftResult.Hour : 0,
min = leftResult.Minute > 0 ? leftResult.Minute : 0,
second = leftResult.Second > 0 ? leftResult.Second : 0;

// handle cases with time like 25時 which resolve to the next day
if (hour > Constants.DayHourCount)
{
hour -= Constants.DayHourCount;
leftSwiftDay++;
}

var leftTime = DateObject.MinValue.SafeCreateFromValue(year, month, day, hour, min, second);

hour = rightResult.Hour > 0 ? rightResult.Hour : 0;
min = rightResult.Minute > 0 ? rightResult.Minute : 0;
second = rightResult.Second > 0 ? rightResult.Second : 0;

// handle cases with time like 25時 which resolve to the next day
if (hour > Constants.DayHourCount)
{
hour -= Constants.DayHourCount;
rightSwiftDay++;
}

var rightTime = DateObject.MinValue.SafeCreateFromValue(year, month, day, hour, min, second);

if (rightTime.Hour < leftTime.Hour)
if (rightResult.Hour < leftResult.Hour)
{
rightTime = rightTime.AddDays(1);
}

ret.FutureValue = ret.PastValue = new Tuple<DateObject, DateObject>(leftTime, rightTime);

var leftTimex = BuildTimex(leftResult);
var rightTimex = BuildTimex(rightResult);
ret.Timex = $"({leftTimex},{rightTimex},{BuildSpan(leftResult, rightResult)})";

leftTime = leftTime.AddDays(leftSwiftDay);
rightTime = rightTime.AddDays(rightSwiftDay);

ret.FutureValue = ret.PastValue = new Tuple<DateObject, DateObject>(leftTime, rightTime);
return ret;
}

Expand Down
24 changes: 24 additions & 0 deletions .NET/Microsoft.Recognizers.Text.DateTime/Utilities/TimexUtility.cs
Original file line number Diff line number Diff line change
Expand Up @@ -569,6 +569,30 @@ public static float ParseNumberFromDurationTimex(string timex)
return float.Parse(numberStr);
}

public static int ParseHourFromTimeTimex(string timex)
{
var start = timex.IndexOf(Constants.TimeTimexPrefix) + 1;
var end = timex.IndexOf(Constants.TimeTimexConnector);
end = end > 0 ? end : timex.Length;
var hourStr = timex.Substring(start, end - start);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

From now on, let's try to write such functions using Spans, to avoid creating a lot of small string objects garbage collected all the time. Please open an issue to refactor existing Timex parsing code to do the same.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@aitelint Please ACK this comment. Thanks.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@tellarin, issue created

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@tellarin, the problem is that net462 does not support parsing of spans, so we would need to convert the span back to string when passing it to int.TryParse

int.TryParse(hourStr, out int hour);

return hour;
}

public static Tuple<int, int> ParseHoursFromTimePeriodTimex(string timex)
{
int hour1 = 0, hour2 = 0;
var timeList = timex.Split(Constants.TimexSeparator[0]);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same.

if (timeList.Length > 2)
{
hour1 = ParseHourFromTimeTimex(timeList[0]);
hour2 = ParseHourFromTimeTimex(timeList[1]);
}

return new Tuple<int, int>(hour1, hour2);
}

private static bool IsTimeDurationTimex(string timex)
{
return timex.StartsWith($"{Constants.GeneralPeriodPrefix}{Constants.TimeTimexPrefix}", StringComparison.Ordinal);
Expand Down
Loading